File: regexec.c

package info (click to toggle)
perl 5.24.1-3+deb9u5
  • links: PTS, VCS
  • area: main
  • in suites: stretch
  • size: 103,716 kB
  • sloc: perl: 559,611; ansic: 293,886; sh: 67,316; pascal: 7,632; cpp: 3,895; makefile: 2,436; xml: 2,410; yacc: 989; sed: 6; lisp: 1
file content (9522 lines) | stat: -rw-r--r-- 357,504 bytes parent folder | download
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
2965
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
2989
2990
2991
2992
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
3004
3005
3006
3007
3008
3009
3010
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
3024
3025
3026
3027
3028
3029
3030
3031
3032
3033
3034
3035
3036
3037
3038
3039
3040
3041
3042
3043
3044
3045
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
3109
3110
3111
3112
3113
3114
3115
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
3140
3141
3142
3143
3144
3145
3146
3147
3148
3149
3150
3151
3152
3153
3154
3155
3156
3157
3158
3159
3160
3161
3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
3185
3186
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
3208
3209
3210
3211
3212
3213
3214
3215
3216
3217
3218
3219
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
3233
3234
3235
3236
3237
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
3284
3285
3286
3287
3288
3289
3290
3291
3292
3293
3294
3295
3296
3297
3298
3299
3300
3301
3302
3303
3304
3305
3306
3307
3308
3309
3310
3311
3312
3313
3314
3315
3316
3317
3318
3319
3320
3321
3322
3323
3324
3325
3326
3327
3328
3329
3330
3331
3332
3333
3334
3335
3336
3337
3338
3339
3340
3341
3342
3343
3344
3345
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356
3357
3358
3359
3360
3361
3362
3363
3364
3365
3366
3367
3368
3369
3370
3371
3372
3373
3374
3375
3376
3377
3378
3379
3380
3381
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396
3397
3398
3399
3400
3401
3402
3403
3404
3405
3406
3407
3408
3409
3410
3411
3412
3413
3414
3415
3416
3417
3418
3419
3420
3421
3422
3423
3424
3425
3426
3427
3428
3429
3430
3431
3432
3433
3434
3435
3436
3437
3438
3439
3440
3441
3442
3443
3444
3445
3446
3447
3448
3449
3450
3451
3452
3453
3454
3455
3456
3457
3458
3459
3460
3461
3462
3463
3464
3465
3466
3467
3468
3469
3470
3471
3472
3473
3474
3475
3476
3477
3478
3479
3480
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
3496
3497
3498
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
3514
3515
3516
3517
3518
3519
3520
3521
3522
3523
3524
3525
3526
3527
3528
3529
3530
3531
3532
3533
3534
3535
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
3558
3559
3560
3561
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
3580
3581
3582
3583
3584
3585
3586
3587
3588
3589
3590
3591
3592
3593
3594
3595
3596
3597
3598
3599
3600
3601
3602
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
3616
3617
3618
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
3642
3643
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
3671
3672
3673
3674
3675
3676
3677
3678
3679
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
3693
3694
3695
3696
3697
3698
3699
3700
3701
3702
3703
3704
3705
3706
3707
3708
3709
3710
3711
3712
3713
3714
3715
3716
3717
3718
3719
3720
3721
3722
3723
3724
3725
3726
3727
3728
3729
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
3741
3742
3743
3744
3745
3746
3747
3748
3749
3750
3751
3752
3753
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
3769
3770
3771
3772
3773
3774
3775
3776
3777
3778
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
3789
3790
3791
3792
3793
3794
3795
3796
3797
3798
3799
3800
3801
3802
3803
3804
3805
3806
3807
3808
3809
3810
3811
3812
3813
3814
3815
3816
3817
3818
3819
3820
3821
3822
3823
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
3866
3867
3868
3869
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
3885
3886
3887
3888
3889
3890
3891
3892
3893
3894
3895
3896
3897
3898
3899
3900
3901
3902
3903
3904
3905
3906
3907
3908
3909
3910
3911
3912
3913
3914
3915
3916
3917
3918
3919
3920
3921
3922
3923
3924
3925
3926
3927
3928
3929
3930
3931
3932
3933
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
3951
3952
3953
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
3972
3973
3974
3975
3976
3977
3978
3979
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
3991
3992
3993
3994
3995
3996
3997
3998
3999
4000
4001
4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
4038
4039
4040
4041
4042
4043
4044
4045
4046
4047
4048
4049
4050
4051
4052
4053
4054
4055
4056
4057
4058
4059
4060
4061
4062
4063
4064
4065
4066
4067
4068
4069
4070
4071
4072
4073
4074
4075
4076
4077
4078
4079
4080
4081
4082
4083
4084
4085
4086
4087
4088
4089
4090
4091
4092
4093
4094
4095
4096
4097
4098
4099
4100
4101
4102
4103
4104
4105
4106
4107
4108
4109
4110
4111
4112
4113
4114
4115
4116
4117
4118
4119
4120
4121
4122
4123
4124
4125
4126
4127
4128
4129
4130
4131
4132
4133
4134
4135
4136
4137
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
4153
4154
4155
4156
4157
4158
4159
4160
4161
4162
4163
4164
4165
4166
4167
4168
4169
4170
4171
4172
4173
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187
4188
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
4204
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
4225
4226
4227
4228
4229
4230
4231
4232
4233
4234
4235
4236
4237
4238
4239
4240
4241
4242
4243
4244
4245
4246
4247
4248
4249
4250
4251
4252
4253
4254
4255
4256
4257
4258
4259
4260
4261
4262
4263
4264
4265
4266
4267
4268
4269
4270
4271
4272
4273
4274
4275
4276
4277
4278
4279
4280
4281
4282
4283
4284
4285
4286
4287
4288
4289
4290
4291
4292
4293
4294
4295
4296
4297
4298
4299
4300
4301
4302
4303
4304
4305
4306
4307
4308
4309
4310
4311
4312
4313
4314
4315
4316
4317
4318
4319
4320
4321
4322
4323
4324
4325
4326
4327
4328
4329
4330
4331
4332
4333
4334
4335
4336
4337
4338
4339
4340
4341
4342
4343
4344
4345
4346
4347
4348
4349
4350
4351
4352
4353
4354
4355
4356
4357
4358
4359
4360
4361
4362
4363
4364
4365
4366
4367
4368
4369
4370
4371
4372
4373
4374
4375
4376
4377
4378
4379
4380
4381
4382
4383
4384
4385
4386
4387
4388
4389
4390
4391
4392
4393
4394
4395
4396
4397
4398
4399
4400
4401
4402
4403
4404
4405
4406
4407
4408
4409
4410
4411
4412
4413
4414
4415
4416
4417
4418
4419
4420
4421
4422
4423
4424
4425
4426
4427
4428
4429
4430
4431
4432
4433
4434
4435
4436
4437
4438
4439
4440
4441
4442
4443
4444
4445
4446
4447
4448
4449
4450
4451
4452
4453
4454
4455
4456
4457
4458
4459
4460
4461
4462
4463
4464
4465
4466
4467
4468
4469
4470
4471
4472
4473
4474
4475
4476
4477
4478
4479
4480
4481
4482
4483
4484
4485
4486
4487
4488
4489
4490
4491
4492
4493
4494
4495
4496
4497
4498
4499
4500
4501
4502
4503
4504
4505
4506
4507
4508
4509
4510
4511
4512
4513
4514
4515
4516
4517
4518
4519
4520
4521
4522
4523
4524
4525
4526
4527
4528
4529
4530
4531
4532
4533
4534
4535
4536
4537
4538
4539
4540
4541
4542
4543
4544
4545
4546
4547
4548
4549
4550
4551
4552
4553
4554
4555
4556
4557
4558
4559
4560
4561
4562
4563
4564
4565
4566
4567
4568
4569
4570
4571
4572
4573
4574
4575
4576
4577
4578
4579
4580
4581
4582
4583
4584
4585
4586
4587
4588
4589
4590
4591
4592
4593
4594
4595
4596
4597
4598
4599
4600
4601
4602
4603
4604
4605
4606
4607
4608
4609
4610
4611
4612
4613
4614
4615
4616
4617
4618
4619
4620
4621
4622
4623
4624
4625
4626
4627
4628
4629
4630
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
4643
4644
4645
4646
4647
4648
4649
4650
4651
4652
4653
4654
4655
4656
4657
4658
4659
4660
4661
4662
4663
4664
4665
4666
4667
4668
4669
4670
4671
4672
4673
4674
4675
4676
4677
4678
4679
4680
4681
4682
4683
4684
4685
4686
4687
4688
4689
4690
4691
4692
4693
4694
4695
4696
4697
4698
4699
4700
4701
4702
4703
4704
4705
4706
4707
4708
4709
4710
4711
4712
4713
4714
4715
4716
4717
4718
4719
4720
4721
4722
4723
4724
4725
4726
4727
4728
4729
4730
4731
4732
4733
4734
4735
4736
4737
4738
4739
4740
4741
4742
4743
4744
4745
4746
4747
4748
4749
4750
4751
4752
4753
4754
4755
4756
4757
4758
4759
4760
4761
4762
4763
4764
4765
4766
4767
4768
4769
4770
4771
4772
4773
4774
4775
4776
4777
4778
4779
4780
4781
4782
4783
4784
4785
4786
4787
4788
4789
4790
4791
4792
4793
4794
4795
4796
4797
4798
4799
4800
4801
4802
4803
4804
4805
4806
4807
4808
4809
4810
4811
4812
4813
4814
4815
4816
4817
4818
4819
4820
4821
4822
4823
4824
4825
4826
4827
4828
4829
4830
4831
4832
4833
4834
4835
4836
4837
4838
4839
4840
4841
4842
4843
4844
4845
4846
4847
4848
4849
4850
4851
4852
4853
4854
4855
4856
4857
4858
4859
4860
4861
4862
4863
4864
4865
4866
4867
4868
4869
4870
4871
4872
4873
4874
4875
4876
4877
4878
4879
4880
4881
4882
4883
4884
4885
4886
4887
4888
4889
4890
4891
4892
4893
4894
4895
4896
4897
4898
4899
4900
4901
4902
4903
4904
4905
4906
4907
4908
4909
4910
4911
4912
4913
4914
4915
4916
4917
4918
4919
4920
4921
4922
4923
4924
4925
4926
4927
4928
4929
4930
4931
4932
4933
4934
4935
4936
4937
4938
4939
4940
4941
4942
4943
4944
4945
4946
4947
4948
4949
4950
4951
4952
4953
4954
4955
4956
4957
4958
4959
4960
4961
4962
4963
4964
4965
4966
4967
4968
4969
4970
4971
4972
4973
4974
4975
4976
4977
4978
4979
4980
4981
4982
4983
4984
4985
4986
4987
4988
4989
4990
4991
4992
4993
4994
4995
4996
4997
4998
4999
5000
5001
5002
5003
5004
5005
5006
5007
5008
5009
5010
5011
5012
5013
5014
5015
5016
5017
5018
5019
5020
5021
5022
5023
5024
5025
5026
5027
5028
5029
5030
5031
5032
5033
5034
5035
5036
5037
5038
5039
5040
5041
5042
5043
5044
5045
5046
5047
5048
5049
5050
5051
5052
5053
5054
5055
5056
5057
5058
5059
5060
5061
5062
5063
5064
5065
5066
5067
5068
5069
5070
5071
5072
5073
5074
5075
5076
5077
5078
5079
5080
5081
5082
5083
5084
5085
5086
5087
5088
5089
5090
5091
5092
5093
5094
5095
5096
5097
5098
5099
5100
5101
5102
5103
5104
5105
5106
5107
5108
5109
5110
5111
5112
5113
5114
5115
5116
5117
5118
5119
5120
5121
5122
5123
5124
5125
5126
5127
5128
5129
5130
5131
5132
5133
5134
5135
5136
5137
5138
5139
5140
5141
5142
5143
5144
5145
5146
5147
5148
5149
5150
5151
5152
5153
5154
5155
5156
5157
5158
5159
5160
5161
5162
5163
5164
5165
5166
5167
5168
5169
5170
5171
5172
5173
5174
5175
5176
5177
5178
5179
5180
5181
5182
5183
5184
5185
5186
5187
5188
5189
5190
5191
5192
5193
5194
5195
5196
5197
5198
5199
5200
5201
5202
5203
5204
5205
5206
5207
5208
5209
5210
5211
5212
5213
5214
5215
5216
5217
5218
5219
5220
5221
5222
5223
5224
5225
5226
5227
5228
5229
5230
5231
5232
5233
5234
5235
5236
5237
5238
5239
5240
5241
5242
5243
5244
5245
5246
5247
5248
5249
5250
5251
5252
5253
5254
5255
5256
5257
5258
5259
5260
5261
5262
5263
5264
5265
5266
5267
5268
5269
5270
5271
5272
5273
5274
5275
5276
5277
5278
5279
5280
5281
5282
5283
5284
5285
5286
5287
5288
5289
5290
5291
5292
5293
5294
5295
5296
5297
5298
5299
5300
5301
5302
5303
5304
5305
5306
5307
5308
5309
5310
5311
5312
5313
5314
5315
5316
5317
5318
5319
5320
5321
5322
5323
5324
5325
5326
5327
5328
5329
5330
5331
5332
5333
5334
5335
5336
5337
5338
5339
5340
5341
5342
5343
5344
5345
5346
5347
5348
5349
5350
5351
5352
5353
5354
5355
5356
5357
5358
5359
5360
5361
5362
5363
5364
5365
5366
5367
5368
5369
5370
5371
5372
5373
5374
5375
5376
5377
5378
5379
5380
5381
5382
5383
5384
5385
5386
5387
5388
5389
5390
5391
5392
5393
5394
5395
5396
5397
5398
5399
5400
5401
5402
5403
5404
5405
5406
5407
5408
5409
5410
5411
5412
5413
5414
5415
5416
5417
5418
5419
5420
5421
5422
5423
5424
5425
5426
5427
5428
5429
5430
5431
5432
5433
5434
5435
5436
5437
5438
5439
5440
5441
5442
5443
5444
5445
5446
5447
5448
5449
5450
5451
5452
5453
5454
5455
5456
5457
5458
5459
5460
5461
5462
5463
5464
5465
5466
5467
5468
5469
5470
5471
5472
5473
5474
5475
5476
5477
5478
5479
5480
5481
5482
5483
5484
5485
5486
5487
5488
5489
5490
5491
5492
5493
5494
5495
5496
5497
5498
5499
5500
5501
5502
5503
5504
5505
5506
5507
5508
5509
5510
5511
5512
5513
5514
5515
5516
5517
5518
5519
5520
5521
5522
5523
5524
5525
5526
5527
5528
5529
5530
5531
5532
5533
5534
5535
5536
5537
5538
5539
5540
5541
5542
5543
5544
5545
5546
5547
5548
5549
5550
5551
5552
5553
5554
5555
5556
5557
5558
5559
5560
5561
5562
5563
5564
5565
5566
5567
5568
5569
5570
5571
5572
5573
5574
5575
5576
5577
5578
5579
5580
5581
5582
5583
5584
5585
5586
5587
5588
5589
5590
5591
5592
5593
5594
5595
5596
5597
5598
5599
5600
5601
5602
5603
5604
5605
5606
5607
5608
5609
5610
5611
5612
5613
5614
5615
5616
5617
5618
5619
5620
5621
5622
5623
5624
5625
5626
5627
5628
5629
5630
5631
5632
5633
5634
5635
5636
5637
5638
5639
5640
5641
5642
5643
5644
5645
5646
5647
5648
5649
5650
5651
5652
5653
5654
5655
5656
5657
5658
5659
5660
5661
5662
5663
5664
5665
5666
5667
5668
5669
5670
5671
5672
5673
5674
5675
5676
5677
5678
5679
5680
5681
5682
5683
5684
5685
5686
5687
5688
5689
5690
5691
5692
5693
5694
5695
5696
5697
5698
5699
5700
5701
5702
5703
5704
5705
5706
5707
5708
5709
5710
5711
5712
5713
5714
5715
5716
5717
5718
5719
5720
5721
5722
5723
5724
5725
5726
5727
5728
5729
5730
5731
5732
5733
5734
5735
5736
5737
5738
5739
5740
5741
5742
5743
5744
5745
5746
5747
5748
5749
5750
5751
5752
5753
5754
5755
5756
5757
5758
5759
5760
5761
5762
5763
5764
5765
5766
5767
5768
5769
5770
5771
5772
5773
5774
5775
5776
5777
5778
5779
5780
5781
5782
5783
5784
5785
5786
5787
5788
5789
5790
5791
5792
5793
5794
5795
5796
5797
5798
5799
5800
5801
5802
5803
5804
5805
5806
5807
5808
5809
5810
5811
5812
5813
5814
5815
5816
5817
5818
5819
5820
5821
5822
5823
5824
5825
5826
5827
5828
5829
5830
5831
5832
5833
5834
5835
5836
5837
5838
5839
5840
5841
5842
5843
5844
5845
5846
5847
5848
5849
5850
5851
5852
5853
5854
5855
5856
5857
5858
5859
5860
5861
5862
5863
5864
5865
5866
5867
5868
5869
5870
5871
5872
5873
5874
5875
5876
5877
5878
5879
5880
5881
5882
5883
5884
5885
5886
5887
5888
5889
5890
5891
5892
5893
5894
5895
5896
5897
5898
5899
5900
5901
5902
5903
5904
5905
5906
5907
5908
5909
5910
5911
5912
5913
5914
5915
5916
5917
5918
5919
5920
5921
5922
5923
5924
5925
5926
5927
5928
5929
5930
5931
5932
5933
5934
5935
5936
5937
5938
5939
5940
5941
5942
5943
5944
5945
5946
5947
5948
5949
5950
5951
5952
5953
5954
5955
5956
5957
5958
5959
5960
5961
5962
5963
5964
5965
5966
5967
5968
5969
5970
5971
5972
5973
5974
5975
5976
5977
5978
5979
5980
5981
5982
5983
5984
5985
5986
5987
5988
5989
5990
5991
5992
5993
5994
5995
5996
5997
5998
5999
6000
6001
6002
6003
6004
6005
6006
6007
6008
6009
6010
6011
6012
6013
6014
6015
6016
6017
6018
6019
6020
6021
6022
6023
6024
6025
6026
6027
6028
6029
6030
6031
6032
6033
6034
6035
6036
6037
6038
6039
6040
6041
6042
6043
6044
6045
6046
6047
6048
6049
6050
6051
6052
6053
6054
6055
6056
6057
6058
6059
6060
6061
6062
6063
6064
6065
6066
6067
6068
6069
6070
6071
6072
6073
6074
6075
6076
6077
6078
6079
6080
6081
6082
6083
6084
6085
6086
6087
6088
6089
6090
6091
6092
6093
6094
6095
6096
6097
6098
6099
6100
6101
6102
6103
6104
6105
6106
6107
6108
6109
6110
6111
6112
6113
6114
6115
6116
6117
6118
6119
6120
6121
6122
6123
6124
6125
6126
6127
6128
6129
6130
6131
6132
6133
6134
6135
6136
6137
6138
6139
6140
6141
6142
6143
6144
6145
6146
6147
6148
6149
6150
6151
6152
6153
6154
6155
6156
6157
6158
6159
6160
6161
6162
6163
6164
6165
6166
6167
6168
6169
6170
6171
6172
6173
6174
6175
6176
6177
6178
6179
6180
6181
6182
6183
6184
6185
6186
6187
6188
6189
6190
6191
6192
6193
6194
6195
6196
6197
6198
6199
6200
6201
6202
6203
6204
6205
6206
6207
6208
6209
6210
6211
6212
6213
6214
6215
6216
6217
6218
6219
6220
6221
6222
6223
6224
6225
6226
6227
6228
6229
6230
6231
6232
6233
6234
6235
6236
6237
6238
6239
6240
6241
6242
6243
6244
6245
6246
6247
6248
6249
6250
6251
6252
6253
6254
6255
6256
6257
6258
6259
6260
6261
6262
6263
6264
6265
6266
6267
6268
6269
6270
6271
6272
6273
6274
6275
6276
6277
6278
6279
6280
6281
6282
6283
6284
6285
6286
6287
6288
6289
6290
6291
6292
6293
6294
6295
6296
6297
6298
6299
6300
6301
6302
6303
6304
6305
6306
6307
6308
6309
6310
6311
6312
6313
6314
6315
6316
6317
6318
6319
6320
6321
6322
6323
6324
6325
6326
6327
6328
6329
6330
6331
6332
6333
6334
6335
6336
6337
6338
6339
6340
6341
6342
6343
6344
6345
6346
6347
6348
6349
6350
6351
6352
6353
6354
6355
6356
6357
6358
6359
6360
6361
6362
6363
6364
6365
6366
6367
6368
6369
6370
6371
6372
6373
6374
6375
6376
6377
6378
6379
6380
6381
6382
6383
6384
6385
6386
6387
6388
6389
6390
6391
6392
6393
6394
6395
6396
6397
6398
6399
6400
6401
6402
6403
6404
6405
6406
6407
6408
6409
6410
6411
6412
6413
6414
6415
6416
6417
6418
6419
6420
6421
6422
6423
6424
6425
6426
6427
6428
6429
6430
6431
6432
6433
6434
6435
6436
6437
6438
6439
6440
6441
6442
6443
6444
6445
6446
6447
6448
6449
6450
6451
6452
6453
6454
6455
6456
6457
6458
6459
6460
6461
6462
6463
6464
6465
6466
6467
6468
6469
6470
6471
6472
6473
6474
6475
6476
6477
6478
6479
6480
6481
6482
6483
6484
6485
6486
6487
6488
6489
6490
6491
6492
6493
6494
6495
6496
6497
6498
6499
6500
6501
6502
6503
6504
6505
6506
6507
6508
6509
6510
6511
6512
6513
6514
6515
6516
6517
6518
6519
6520
6521
6522
6523
6524
6525
6526
6527
6528
6529
6530
6531
6532
6533
6534
6535
6536
6537
6538
6539
6540
6541
6542
6543
6544
6545
6546
6547
6548
6549
6550
6551
6552
6553
6554
6555
6556
6557
6558
6559
6560
6561
6562
6563
6564
6565
6566
6567
6568
6569
6570
6571
6572
6573
6574
6575
6576
6577
6578
6579
6580
6581
6582
6583
6584
6585
6586
6587
6588
6589
6590
6591
6592
6593
6594
6595
6596
6597
6598
6599
6600
6601
6602
6603
6604
6605
6606
6607
6608
6609
6610
6611
6612
6613
6614
6615
6616
6617
6618
6619
6620
6621
6622
6623
6624
6625
6626
6627
6628
6629
6630
6631
6632
6633
6634
6635
6636
6637
6638
6639
6640
6641
6642
6643
6644
6645
6646
6647
6648
6649
6650
6651
6652
6653
6654
6655
6656
6657
6658
6659
6660
6661
6662
6663
6664
6665
6666
6667
6668
6669
6670
6671
6672
6673
6674
6675
6676
6677
6678
6679
6680
6681
6682
6683
6684
6685
6686
6687
6688
6689
6690
6691
6692
6693
6694
6695
6696
6697
6698
6699
6700
6701
6702
6703
6704
6705
6706
6707
6708
6709
6710
6711
6712
6713
6714
6715
6716
6717
6718
6719
6720
6721
6722
6723
6724
6725
6726
6727
6728
6729
6730
6731
6732
6733
6734
6735
6736
6737
6738
6739
6740
6741
6742
6743
6744
6745
6746
6747
6748
6749
6750
6751
6752
6753
6754
6755
6756
6757
6758
6759
6760
6761
6762
6763
6764
6765
6766
6767
6768
6769
6770
6771
6772
6773
6774
6775
6776
6777
6778
6779
6780
6781
6782
6783
6784
6785
6786
6787
6788
6789
6790
6791
6792
6793
6794
6795
6796
6797
6798
6799
6800
6801
6802
6803
6804
6805
6806
6807
6808
6809
6810
6811
6812
6813
6814
6815
6816
6817
6818
6819
6820
6821
6822
6823
6824
6825
6826
6827
6828
6829
6830
6831
6832
6833
6834
6835
6836
6837
6838
6839
6840
6841
6842
6843
6844
6845
6846
6847
6848
6849
6850
6851
6852
6853
6854
6855
6856
6857
6858
6859
6860
6861
6862
6863
6864
6865
6866
6867
6868
6869
6870
6871
6872
6873
6874
6875
6876
6877
6878
6879
6880
6881
6882
6883
6884
6885
6886
6887
6888
6889
6890
6891
6892
6893
6894
6895
6896
6897
6898
6899
6900
6901
6902
6903
6904
6905
6906
6907
6908
6909
6910
6911
6912
6913
6914
6915
6916
6917
6918
6919
6920
6921
6922
6923
6924
6925
6926
6927
6928
6929
6930
6931
6932
6933
6934
6935
6936
6937
6938
6939
6940
6941
6942
6943
6944
6945
6946
6947
6948
6949
6950
6951
6952
6953
6954
6955
6956
6957
6958
6959
6960
6961
6962
6963
6964
6965
6966
6967
6968
6969
6970
6971
6972
6973
6974
6975
6976
6977
6978
6979
6980
6981
6982
6983
6984
6985
6986
6987
6988
6989
6990
6991
6992
6993
6994
6995
6996
6997
6998
6999
7000
7001
7002
7003
7004
7005
7006
7007
7008
7009
7010
7011
7012
7013
7014
7015
7016
7017
7018
7019
7020
7021
7022
7023
7024
7025
7026
7027
7028
7029
7030
7031
7032
7033
7034
7035
7036
7037
7038
7039
7040
7041
7042
7043
7044
7045
7046
7047
7048
7049
7050
7051
7052
7053
7054
7055
7056
7057
7058
7059
7060
7061
7062
7063
7064
7065
7066
7067
7068
7069
7070
7071
7072
7073
7074
7075
7076
7077
7078
7079
7080
7081
7082
7083
7084
7085
7086
7087
7088
7089
7090
7091
7092
7093
7094
7095
7096
7097
7098
7099
7100
7101
7102
7103
7104
7105
7106
7107
7108
7109
7110
7111
7112
7113
7114
7115
7116
7117
7118
7119
7120
7121
7122
7123
7124
7125
7126
7127
7128
7129
7130
7131
7132
7133
7134
7135
7136
7137
7138
7139
7140
7141
7142
7143
7144
7145
7146
7147
7148
7149
7150
7151
7152
7153
7154
7155
7156
7157
7158
7159
7160
7161
7162
7163
7164
7165
7166
7167
7168
7169
7170
7171
7172
7173
7174
7175
7176
7177
7178
7179
7180
7181
7182
7183
7184
7185
7186
7187
7188
7189
7190
7191
7192
7193
7194
7195
7196
7197
7198
7199
7200
7201
7202
7203
7204
7205
7206
7207
7208
7209
7210
7211
7212
7213
7214
7215
7216
7217
7218
7219
7220
7221
7222
7223
7224
7225
7226
7227
7228
7229
7230
7231
7232
7233
7234
7235
7236
7237
7238
7239
7240
7241
7242
7243
7244
7245
7246
7247
7248
7249
7250
7251
7252
7253
7254
7255
7256
7257
7258
7259
7260
7261
7262
7263
7264
7265
7266
7267
7268
7269
7270
7271
7272
7273
7274
7275
7276
7277
7278
7279
7280
7281
7282
7283
7284
7285
7286
7287
7288
7289
7290
7291
7292
7293
7294
7295
7296
7297
7298
7299
7300
7301
7302
7303
7304
7305
7306
7307
7308
7309
7310
7311
7312
7313
7314
7315
7316
7317
7318
7319
7320
7321
7322
7323
7324
7325
7326
7327
7328
7329
7330
7331
7332
7333
7334
7335
7336
7337
7338
7339
7340
7341
7342
7343
7344
7345
7346
7347
7348
7349
7350
7351
7352
7353
7354
7355
7356
7357
7358
7359
7360
7361
7362
7363
7364
7365
7366
7367
7368
7369
7370
7371
7372
7373
7374
7375
7376
7377
7378
7379
7380
7381
7382
7383
7384
7385
7386
7387
7388
7389
7390
7391
7392
7393
7394
7395
7396
7397
7398
7399
7400
7401
7402
7403
7404
7405
7406
7407
7408
7409
7410
7411
7412
7413
7414
7415
7416
7417
7418
7419
7420
7421
7422
7423
7424
7425
7426
7427
7428
7429
7430
7431
7432
7433
7434
7435
7436
7437
7438
7439
7440
7441
7442
7443
7444
7445
7446
7447
7448
7449
7450
7451
7452
7453
7454
7455
7456
7457
7458
7459
7460
7461
7462
7463
7464
7465
7466
7467
7468
7469
7470
7471
7472
7473
7474
7475
7476
7477
7478
7479
7480
7481
7482
7483
7484
7485
7486
7487
7488
7489
7490
7491
7492
7493
7494
7495
7496
7497
7498
7499
7500
7501
7502
7503
7504
7505
7506
7507
7508
7509
7510
7511
7512
7513
7514
7515
7516
7517
7518
7519
7520
7521
7522
7523
7524
7525
7526
7527
7528
7529
7530
7531
7532
7533
7534
7535
7536
7537
7538
7539
7540
7541
7542
7543
7544
7545
7546
7547
7548
7549
7550
7551
7552
7553
7554
7555
7556
7557
7558
7559
7560
7561
7562
7563
7564
7565
7566
7567
7568
7569
7570
7571
7572
7573
7574
7575
7576
7577
7578
7579
7580
7581
7582
7583
7584
7585
7586
7587
7588
7589
7590
7591
7592
7593
7594
7595
7596
7597
7598
7599
7600
7601
7602
7603
7604
7605
7606
7607
7608
7609
7610
7611
7612
7613
7614
7615
7616
7617
7618
7619
7620
7621
7622
7623
7624
7625
7626
7627
7628
7629
7630
7631
7632
7633
7634
7635
7636
7637
7638
7639
7640
7641
7642
7643
7644
7645
7646
7647
7648
7649
7650
7651
7652
7653
7654
7655
7656
7657
7658
7659
7660
7661
7662
7663
7664
7665
7666
7667
7668
7669
7670
7671
7672
7673
7674
7675
7676
7677
7678
7679
7680
7681
7682
7683
7684
7685
7686
7687
7688
7689
7690
7691
7692
7693
7694
7695
7696
7697
7698
7699
7700
7701
7702
7703
7704
7705
7706
7707
7708
7709
7710
7711
7712
7713
7714
7715
7716
7717
7718
7719
7720
7721
7722
7723
7724
7725
7726
7727
7728
7729
7730
7731
7732
7733
7734
7735
7736
7737
7738
7739
7740
7741
7742
7743
7744
7745
7746
7747
7748
7749
7750
7751
7752
7753
7754
7755
7756
7757
7758
7759
7760
7761
7762
7763
7764
7765
7766
7767
7768
7769
7770
7771
7772
7773
7774
7775
7776
7777
7778
7779
7780
7781
7782
7783
7784
7785
7786
7787
7788
7789
7790
7791
7792
7793
7794
7795
7796
7797
7798
7799
7800
7801
7802
7803
7804
7805
7806
7807
7808
7809
7810
7811
7812
7813
7814
7815
7816
7817
7818
7819
7820
7821
7822
7823
7824
7825
7826
7827
7828
7829
7830
7831
7832
7833
7834
7835
7836
7837
7838
7839
7840
7841
7842
7843
7844
7845
7846
7847
7848
7849
7850
7851
7852
7853
7854
7855
7856
7857
7858
7859
7860
7861
7862
7863
7864
7865
7866
7867
7868
7869
7870
7871
7872
7873
7874
7875
7876
7877
7878
7879
7880
7881
7882
7883
7884
7885
7886
7887
7888
7889
7890
7891
7892
7893
7894
7895
7896
7897
7898
7899
7900
7901
7902
7903
7904
7905
7906
7907
7908
7909
7910
7911
7912
7913
7914
7915
7916
7917
7918
7919
7920
7921
7922
7923
7924
7925
7926
7927
7928
7929
7930
7931
7932
7933
7934
7935
7936
7937
7938
7939
7940
7941
7942
7943
7944
7945
7946
7947
7948
7949
7950
7951
7952
7953
7954
7955
7956
7957
7958
7959
7960
7961
7962
7963
7964
7965
7966
7967
7968
7969
7970
7971
7972
7973
7974
7975
7976
7977
7978
7979
7980
7981
7982
7983
7984
7985
7986
7987
7988
7989
7990
7991
7992
7993
7994
7995
7996
7997
7998
7999
8000
8001
8002
8003
8004
8005
8006
8007
8008
8009
8010
8011
8012
8013
8014
8015
8016
8017
8018
8019
8020
8021
8022
8023
8024
8025
8026
8027
8028
8029
8030
8031
8032
8033
8034
8035
8036
8037
8038
8039
8040
8041
8042
8043
8044
8045
8046
8047
8048
8049
8050
8051
8052
8053
8054
8055
8056
8057
8058
8059
8060
8061
8062
8063
8064
8065
8066
8067
8068
8069
8070
8071
8072
8073
8074
8075
8076
8077
8078
8079
8080
8081
8082
8083
8084
8085
8086
8087
8088
8089
8090
8091
8092
8093
8094
8095
8096
8097
8098
8099
8100
8101
8102
8103
8104
8105
8106
8107
8108
8109
8110
8111
8112
8113
8114
8115
8116
8117
8118
8119
8120
8121
8122
8123
8124
8125
8126
8127
8128
8129
8130
8131
8132
8133
8134
8135
8136
8137
8138
8139
8140
8141
8142
8143
8144
8145
8146
8147
8148
8149
8150
8151
8152
8153
8154
8155
8156
8157
8158
8159
8160
8161
8162
8163
8164
8165
8166
8167
8168
8169
8170
8171
8172
8173
8174
8175
8176
8177
8178
8179
8180
8181
8182
8183
8184
8185
8186
8187
8188
8189
8190
8191
8192
8193
8194
8195
8196
8197
8198
8199
8200
8201
8202
8203
8204
8205
8206
8207
8208
8209
8210
8211
8212
8213
8214
8215
8216
8217
8218
8219
8220
8221
8222
8223
8224
8225
8226
8227
8228
8229
8230
8231
8232
8233
8234
8235
8236
8237
8238
8239
8240
8241
8242
8243
8244
8245
8246
8247
8248
8249
8250
8251
8252
8253
8254
8255
8256
8257
8258
8259
8260
8261
8262
8263
8264
8265
8266
8267
8268
8269
8270
8271
8272
8273
8274
8275
8276
8277
8278
8279
8280
8281
8282
8283
8284
8285
8286
8287
8288
8289
8290
8291
8292
8293
8294
8295
8296
8297
8298
8299
8300
8301
8302
8303
8304
8305
8306
8307
8308
8309
8310
8311
8312
8313
8314
8315
8316
8317
8318
8319
8320
8321
8322
8323
8324
8325
8326
8327
8328
8329
8330
8331
8332
8333
8334
8335
8336
8337
8338
8339
8340
8341
8342
8343
8344
8345
8346
8347
8348
8349
8350
8351
8352
8353
8354
8355
8356
8357
8358
8359
8360
8361
8362
8363
8364
8365
8366
8367
8368
8369
8370
8371
8372
8373
8374
8375
8376
8377
8378
8379
8380
8381
8382
8383
8384
8385
8386
8387
8388
8389
8390
8391
8392
8393
8394
8395
8396
8397
8398
8399
8400
8401
8402
8403
8404
8405
8406
8407
8408
8409
8410
8411
8412
8413
8414
8415
8416
8417
8418
8419
8420
8421
8422
8423
8424
8425
8426
8427
8428
8429
8430
8431
8432
8433
8434
8435
8436
8437
8438
8439
8440
8441
8442
8443
8444
8445
8446
8447
8448
8449
8450
8451
8452
8453
8454
8455
8456
8457
8458
8459
8460
8461
8462
8463
8464
8465
8466
8467
8468
8469
8470
8471
8472
8473
8474
8475
8476
8477
8478
8479
8480
8481
8482
8483
8484
8485
8486
8487
8488
8489
8490
8491
8492
8493
8494
8495
8496
8497
8498
8499
8500
8501
8502
8503
8504
8505
8506
8507
8508
8509
8510
8511
8512
8513
8514
8515
8516
8517
8518
8519
8520
8521
8522
8523
8524
8525
8526
8527
8528
8529
8530
8531
8532
8533
8534
8535
8536
8537
8538
8539
8540
8541
8542
8543
8544
8545
8546
8547
8548
8549
8550
8551
8552
8553
8554
8555
8556
8557
8558
8559
8560
8561
8562
8563
8564
8565
8566
8567
8568
8569
8570
8571
8572
8573
8574
8575
8576
8577
8578
8579
8580
8581
8582
8583
8584
8585
8586
8587
8588
8589
8590
8591
8592
8593
8594
8595
8596
8597
8598
8599
8600
8601
8602
8603
8604
8605
8606
8607
8608
8609
8610
8611
8612
8613
8614
8615
8616
8617
8618
8619
8620
8621
8622
8623
8624
8625
8626
8627
8628
8629
8630
8631
8632
8633
8634
8635
8636
8637
8638
8639
8640
8641
8642
8643
8644
8645
8646
8647
8648
8649
8650
8651
8652
8653
8654
8655
8656
8657
8658
8659
8660
8661
8662
8663
8664
8665
8666
8667
8668
8669
8670
8671
8672
8673
8674
8675
8676
8677
8678
8679
8680
8681
8682
8683
8684
8685
8686
8687
8688
8689
8690
8691
8692
8693
8694
8695
8696
8697
8698
8699
8700
8701
8702
8703
8704
8705
8706
8707
8708
8709
8710
8711
8712
8713
8714
8715
8716
8717
8718
8719
8720
8721
8722
8723
8724
8725
8726
8727
8728
8729
8730
8731
8732
8733
8734
8735
8736
8737
8738
8739
8740
8741
8742
8743
8744
8745
8746
8747
8748
8749
8750
8751
8752
8753
8754
8755
8756
8757
8758
8759
8760
8761
8762
8763
8764
8765
8766
8767
8768
8769
8770
8771
8772
8773
8774
8775
8776
8777
8778
8779
8780
8781
8782
8783
8784
8785
8786
8787
8788
8789
8790
8791
8792
8793
8794
8795
8796
8797
8798
8799
8800
8801
8802
8803
8804
8805
8806
8807
8808
8809
8810
8811
8812
8813
8814
8815
8816
8817
8818
8819
8820
8821
8822
8823
8824
8825
8826
8827
8828
8829
8830
8831
8832
8833
8834
8835
8836
8837
8838
8839
8840
8841
8842
8843
8844
8845
8846
8847
8848
8849
8850
8851
8852
8853
8854
8855
8856
8857
8858
8859
8860
8861
8862
8863
8864
8865
8866
8867
8868
8869
8870
8871
8872
8873
8874
8875
8876
8877
8878
8879
8880
8881
8882
8883
8884
8885
8886
8887
8888
8889
8890
8891
8892
8893
8894
8895
8896
8897
8898
8899
8900
8901
8902
8903
8904
8905
8906
8907
8908
8909
8910
8911
8912
8913
8914
8915
8916
8917
8918
8919
8920
8921
8922
8923
8924
8925
8926
8927
8928
8929
8930
8931
8932
8933
8934
8935
8936
8937
8938
8939
8940
8941
8942
8943
8944
8945
8946
8947
8948
8949
8950
8951
8952
8953
8954
8955
8956
8957
8958
8959
8960
8961
8962
8963
8964
8965
8966
8967
8968
8969
8970
8971
8972
8973
8974
8975
8976
8977
8978
8979
8980
8981
8982
8983
8984
8985
8986
8987
8988
8989
8990
8991
8992
8993
8994
8995
8996
8997
8998
8999
9000
9001
9002
9003
9004
9005
9006
9007
9008
9009
9010
9011
9012
9013
9014
9015
9016
9017
9018
9019
9020
9021
9022
9023
9024
9025
9026
9027
9028
9029
9030
9031
9032
9033
9034
9035
9036
9037
9038
9039
9040
9041
9042
9043
9044
9045
9046
9047
9048
9049
9050
9051
9052
9053
9054
9055
9056
9057
9058
9059
9060
9061
9062
9063
9064
9065
9066
9067
9068
9069
9070
9071
9072
9073
9074
9075
9076
9077
9078
9079
9080
9081
9082
9083
9084
9085
9086
9087
9088
9089
9090
9091
9092
9093
9094
9095
9096
9097
9098
9099
9100
9101
9102
9103
9104
9105
9106
9107
9108
9109
9110
9111
9112
9113
9114
9115
9116
9117
9118
9119
9120
9121
9122
9123
9124
9125
9126
9127
9128
9129
9130
9131
9132
9133
9134
9135
9136
9137
9138
9139
9140
9141
9142
9143
9144
9145
9146
9147
9148
9149
9150
9151
9152
9153
9154
9155
9156
9157
9158
9159
9160
9161
9162
9163
9164
9165
9166
9167
9168
9169
9170
9171
9172
9173
9174
9175
9176
9177
9178
9179
9180
9181
9182
9183
9184
9185
9186
9187
9188
9189
9190
9191
9192
9193
9194
9195
9196
9197
9198
9199
9200
9201
9202
9203
9204
9205
9206
9207
9208
9209
9210
9211
9212
9213
9214
9215
9216
9217
9218
9219
9220
9221
9222
9223
9224
9225
9226
9227
9228
9229
9230
9231
9232
9233
9234
9235
9236
9237
9238
9239
9240
9241
9242
9243
9244
9245
9246
9247
9248
9249
9250
9251
9252
9253
9254
9255
9256
9257
9258
9259
9260
9261
9262
9263
9264
9265
9266
9267
9268
9269
9270
9271
9272
9273
9274
9275
9276
9277
9278
9279
9280
9281
9282
9283
9284
9285
9286
9287
9288
9289
9290
9291
9292
9293
9294
9295
9296
9297
9298
9299
9300
9301
9302
9303
9304
9305
9306
9307
9308
9309
9310
9311
9312
9313
9314
9315
9316
9317
9318
9319
9320
9321
9322
9323
9324
9325
9326
9327
9328
9329
9330
9331
9332
9333
9334
9335
9336
9337
9338
9339
9340
9341
9342
9343
9344
9345
9346
9347
9348
9349
9350
9351
9352
9353
9354
9355
9356
9357
9358
9359
9360
9361
9362
9363
9364
9365
9366
9367
9368
9369
9370
9371
9372
9373
9374
9375
9376
9377
9378
9379
9380
9381
9382
9383
9384
9385
9386
9387
9388
9389
9390
9391
9392
9393
9394
9395
9396
9397
9398
9399
9400
9401
9402
9403
9404
9405
9406
9407
9408
9409
9410
9411
9412
9413
9414
9415
9416
9417
9418
9419
9420
9421
9422
9423
9424
9425
9426
9427
9428
9429
9430
9431
9432
9433
9434
9435
9436
9437
9438
9439
9440
9441
9442
9443
9444
9445
9446
9447
9448
9449
9450
9451
9452
9453
9454
9455
9456
9457
9458
9459
9460
9461
9462
9463
9464
9465
9466
9467
9468
9469
9470
9471
9472
9473
9474
9475
9476
9477
9478
9479
9480
9481
9482
9483
9484
9485
9486
9487
9488
9489
9490
9491
9492
9493
9494
9495
9496
9497
9498
9499
9500
9501
9502
9503
9504
9505
9506
9507
9508
9509
9510
9511
9512
9513
9514
9515
9516
9517
9518
9519
9520
9521
9522
/*    regexec.c
 */

/*
 *	One Ring to rule them all, One Ring to find them
 *
 *     [p.v of _The Lord of the Rings_, opening poem]
 *     [p.50 of _The Lord of the Rings_, I/iii: "The Shadow of the Past"]
 *     [p.254 of _The Lord of the Rings_, II/ii: "The Council of Elrond"]
 */

/* This file contains functions for executing a regular expression.  See
 * also regcomp.c which funnily enough, contains functions for compiling
 * a regular expression.
 *
 * This file is also copied at build time to ext/re/re_exec.c, where
 * it's built with -DPERL_EXT_RE_BUILD -DPERL_EXT_RE_DEBUG -DPERL_EXT.
 * This causes the main functions to be compiled under new names and with
 * debugging support added, which makes "use re 'debug'" work.
 */

/* NOTE: this is derived from Henry Spencer's regexp code, and should not
 * confused with the original package (see point 3 below).  Thanks, Henry!
 */

/* Additional note: this code is very heavily munged from Henry's version
 * in places.  In some spots I've traded clarity for efficiency, so don't
 * blame Henry for some of the lack of readability.
 */

/* The names of the functions have been changed from regcomp and
 * regexec to  pregcomp and pregexec in order to avoid conflicts
 * with the POSIX routines of the same names.
*/

#ifdef PERL_EXT_RE_BUILD
#include "re_top.h"
#endif

/*
 * pregcomp and pregexec -- regsub and regerror are not used in perl
 *
 *	Copyright (c) 1986 by University of Toronto.
 *	Written by Henry Spencer.  Not derived from licensed software.
 *
 *	Permission is granted to anyone to use this software for any
 *	purpose on any computer system, and to redistribute it freely,
 *	subject to the following restrictions:
 *
 *	1. The author is not responsible for the consequences of use of
 *		this software, no matter how awful, even if they arise
 *		from defects in it.
 *
 *	2. The origin of this software must not be misrepresented, either
 *		by explicit claim or by omission.
 *
 *	3. Altered versions must be plainly marked as such, and must not
 *		be misrepresented as being the original software.
 *
 ****    Alterations to Henry's code are...
 ****
 ****    Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
 ****    2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
 ****    by Larry Wall and others
 ****
 ****    You may distribute under the terms of either the GNU General Public
 ****    License or the Artistic License, as specified in the README file.
 *
 * Beware that some of this code is subtly aware of the way operator
 * precedence is structured in regular expressions.  Serious changes in
 * regular-expression syntax might require a total rethink.
 */
#include "EXTERN.h"
#define PERL_IN_REGEXEC_C
#include "perl.h"

#ifdef PERL_IN_XSUB_RE
#  include "re_comp.h"
#else
#  include "regcomp.h"
#endif

#include "invlist_inline.h"
#include "unicode_constants.h"

#define B_ON_NON_UTF8_LOCALE_IS_WRONG            \
 "Use of \\b{} or \\B{} for non-UTF-8 locale is wrong.  Assuming a UTF-8 locale"

static const char utf8_locale_required[] =
      "Use of (?[ ]) for non-UTF-8 locale is wrong.  Assuming a UTF-8 locale";

#ifdef DEBUGGING
/* At least one required character in the target string is expressible only in
 * UTF-8. */
static const char* const non_utf8_target_but_utf8_required
                = "Can't match, because target string needs to be in UTF-8\n";
#endif

#define NON_UTF8_TARGET_BUT_UTF8_REQUIRED(target) STMT_START {           \
    DEBUG_EXECUTE_r(Perl_re_printf( aTHX_  "%s", non_utf8_target_but_utf8_required));\
    goto target;                                                         \
} STMT_END

#define HAS_NONLATIN1_FOLD_CLOSURE(i) _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(i)

#ifndef STATIC
#define	STATIC	static
#endif

/* Valid only if 'c', the character being looke-up, is an invariant under
 * UTF-8: it avoids the reginclass call if there are no complications: i.e., if
 * everything matchable is straight forward in the bitmap */
#define REGINCLASS(prog,p,c,u)  (ANYOF_FLAGS(p)                             \
                                ? reginclass(prog,p,c,c+1,u)                \
                                : ANYOF_BITMAP_TEST(p,*(c)))

/*
 * Forwards.
 */

#define CHR_SVLEN(sv) (utf8_target ? sv_len_utf8(sv) : SvCUR(sv))
#define CHR_DIST(a,b) (reginfo->is_utf8_target ? utf8_distance(a,b) : a - b)

#define HOPc(pos,off) \
	(char *)(reginfo->is_utf8_target \
	    ? reghop3((U8*)pos, off, \
                    (U8*)(off >= 0 ? reginfo->strend : reginfo->strbeg)) \
	    : (U8*)(pos + off))

/* like HOPMAYBE3 but backwards. lim must be +ve. Returns NULL on overshoot */
#define HOPBACK3(pos, off, lim) \
	(reginfo->is_utf8_target                          \
	    ? reghopmaybe3((U8*)pos, (SSize_t)0-off, (U8*)(lim)) \
	    : (pos - off >= lim)	                         \
		? (U8*)pos - off		                 \
		: NULL)

#define HOPBACKc(pos, off) ((char*)HOPBACK3(pos, off, reginfo->strbeg))

#define HOP3(pos,off,lim) (reginfo->is_utf8_target  ? reghop3((U8*)(pos), off, (U8*)(lim)) : (U8*)(pos + off))
#define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim))

/* lim must be +ve. Returns NULL on overshoot */
#define HOPMAYBE3(pos,off,lim) \
	(reginfo->is_utf8_target                        \
	    ? reghopmaybe3((U8*)pos, off, (U8*)(lim))   \
	    : ((U8*)pos + off <= lim)                   \
		? (U8*)pos + off                        \
		: NULL)

/* like HOP3, but limits the result to <= lim even for the non-utf8 case.
 * off must be >=0; args should be vars rather than expressions */
#define HOP3lim(pos,off,lim) (reginfo->is_utf8_target \
    ? reghop3((U8*)(pos), off, (U8*)(lim)) \
    : (U8*)((pos + off) > lim ? lim : (pos + off)))

#define HOP4(pos,off,llim, rlim) (reginfo->is_utf8_target \
    ? reghop4((U8*)(pos), off, (U8*)(llim), (U8*)(rlim)) \
    : (U8*)(pos + off))
#define HOP4c(pos,off,llim, rlim) ((char*)HOP4(pos,off,llim, rlim))

#define NEXTCHR_EOS -10 /* nextchr has fallen off the end */
#define NEXTCHR_IS_EOS (nextchr < 0)

#define SET_nextchr \
    nextchr = ((locinput < reginfo->strend) ? UCHARAT(locinput) : NEXTCHR_EOS)

#define SET_locinput(p) \
    locinput = (p);  \
    SET_nextchr


#define LOAD_UTF8_CHARCLASS(swash_ptr, property_name, invlist) STMT_START {   \
        if (!swash_ptr) {                                                     \
            U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;                       \
            swash_ptr = _core_swash_init("utf8", property_name, &PL_sv_undef, \
                                         1, 0, invlist, &flags);              \
            assert(swash_ptr);                                                \
        }                                                                     \
    } STMT_END

/* If in debug mode, we test that a known character properly matches */
#ifdef DEBUGGING
#   define LOAD_UTF8_CHARCLASS_DEBUG_TEST(swash_ptr,                          \
                                          property_name,                      \
                                          invlist,                            \
                                          utf8_char_in_property)              \
        LOAD_UTF8_CHARCLASS(swash_ptr, property_name, invlist);               \
        assert(swash_fetch(swash_ptr, (U8 *) utf8_char_in_property, TRUE));
#else
#   define LOAD_UTF8_CHARCLASS_DEBUG_TEST(swash_ptr,                          \
                                          property_name,                      \
                                          invlist,                            \
                                          utf8_char_in_property)              \
        LOAD_UTF8_CHARCLASS(swash_ptr, property_name, invlist)
#endif

#define LOAD_UTF8_CHARCLASS_ALNUM() LOAD_UTF8_CHARCLASS_DEBUG_TEST(           \
                                        PL_utf8_swash_ptrs[_CC_WORDCHAR],     \
                                        "",                                   \
                                        PL_XPosix_ptrs[_CC_WORDCHAR],         \
                                        LATIN_SMALL_LIGATURE_LONG_S_T_UTF8);

#define PLACEHOLDER	/* Something for the preprocessor to grab onto */
/* TODO: Combine JUMPABLE and HAS_TEXT to cache OP(rn) */

/* for use after a quantifier and before an EXACT-like node -- japhy */
/* it would be nice to rework regcomp.sym to generate this stuff. sigh
 *
 * NOTE that *nothing* that affects backtracking should be in here, specifically
 * VERBS must NOT be included. JUMPABLE is used to determine  if we can ignore a
 * node that is in between two EXACT like nodes when ascertaining what the required
 * "follow" character is. This should probably be moved to regex compile time
 * although it may be done at run time beause of the REF possibility - more
 * investigation required. -- demerphq
*/
#define JUMPABLE(rn) (                                                             \
    OP(rn) == OPEN ||                                                              \
    (OP(rn) == CLOSE &&                                                            \
     !EVAL_CLOSE_PAREN_IS(cur_eval,ARG(rn)) ) ||                                   \
    OP(rn) == EVAL ||                                                              \
    OP(rn) == SUSPEND || OP(rn) == IFMATCH ||                                      \
    OP(rn) == PLUS || OP(rn) == MINMOD ||                                          \
    OP(rn) == KEEPS ||                                                             \
    (PL_regkind[OP(rn)] == CURLY && ARG1(rn) > 0)                                  \
)
#define IS_EXACT(rn) (PL_regkind[OP(rn)] == EXACT)

#define HAS_TEXT(rn) ( IS_EXACT(rn) || PL_regkind[OP(rn)] == REF )

#if 0 
/* Currently these are only used when PL_regkind[OP(rn)] == EXACT so
   we don't need this definition.  XXX These are now out-of-sync*/
#define IS_TEXT(rn)   ( OP(rn)==EXACT   || OP(rn)==REF   || OP(rn)==NREF   )
#define IS_TEXTF(rn)  ( OP(rn)==EXACTFU || OP(rn)==EXACTFU_SS || OP(rn)==EXACTFA || OP(rn)==EXACTFA_NO_TRIE || OP(rn)==EXACTF || OP(rn)==REFF  || OP(rn)==NREFF )
#define IS_TEXTFL(rn) ( OP(rn)==EXACTFL || OP(rn)==REFFL || OP(rn)==NREFFL )

#else
/* ... so we use this as its faster. */
#define IS_TEXT(rn)   ( OP(rn)==EXACT || OP(rn)==EXACTL )
#define IS_TEXTFU(rn)  ( OP(rn)==EXACTFU || OP(rn)==EXACTFLU8 || OP(rn)==EXACTFU_SS || OP(rn) == EXACTFA || OP(rn) == EXACTFA_NO_TRIE)
#define IS_TEXTF(rn)  ( OP(rn)==EXACTF  )
#define IS_TEXTFL(rn) ( OP(rn)==EXACTFL )

#endif

/*
  Search for mandatory following text node; for lookahead, the text must
  follow but for lookbehind (rn->flags != 0) we skip to the next step.
*/
#define FIND_NEXT_IMPT(rn) STMT_START {                                   \
    while (JUMPABLE(rn)) { \
	const OPCODE type = OP(rn); \
	if (type == SUSPEND || PL_regkind[type] == CURLY) \
	    rn = NEXTOPER(NEXTOPER(rn)); \
	else if (type == PLUS) \
	    rn = NEXTOPER(rn); \
	else if (type == IFMATCH) \
	    rn = (rn->flags == 0) ? NEXTOPER(NEXTOPER(rn)) : rn + ARG(rn); \
	else rn += NEXT_OFF(rn); \
    } \
} STMT_END 

#define SLAB_FIRST(s) (&(s)->states[0])
#define SLAB_LAST(s)  (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1])

static void S_setup_eval_state(pTHX_ regmatch_info *const reginfo);
static void S_cleanup_regmatch_info_aux(pTHX_ void *arg);
static regmatch_state * S_push_slab(pTHX);

#define REGCP_PAREN_ELEMS 3
#define REGCP_OTHER_ELEMS 3
#define REGCP_FRAME_ELEMS 1
/* REGCP_FRAME_ELEMS are not part of the REGCP_OTHER_ELEMS and
 * are needed for the regexp context stack bookkeeping. */

STATIC CHECKPOINT
S_regcppush(pTHX_ const regexp *rex, I32 parenfloor, U32 maxopenparen)
{
    const int retval = PL_savestack_ix;
    const int paren_elems_to_push =
                (maxopenparen - parenfloor) * REGCP_PAREN_ELEMS;
    const UV total_elems = paren_elems_to_push + REGCP_OTHER_ELEMS;
    const UV elems_shifted = total_elems << SAVE_TIGHT_SHIFT;
    I32 p;
    GET_RE_DEBUG_FLAGS_DECL;

    PERL_ARGS_ASSERT_REGCPPUSH;

    if (paren_elems_to_push < 0)
        Perl_croak(aTHX_ "panic: paren_elems_to_push, %i < 0, maxopenparen: %i parenfloor: %i REGCP_PAREN_ELEMS: %u",
                   (int)paren_elems_to_push, (int)maxopenparen,
                   (int)parenfloor, (unsigned)REGCP_PAREN_ELEMS);

    if ((elems_shifted >> SAVE_TIGHT_SHIFT) != total_elems)
	Perl_croak(aTHX_ "panic: paren_elems_to_push offset %"UVuf
		   " out of range (%lu-%ld)",
		   total_elems,
                   (unsigned long)maxopenparen,
                   (long)parenfloor);

    SSGROW(total_elems + REGCP_FRAME_ELEMS);
    
    DEBUG_BUFFERS_r(
	if ((int)maxopenparen > (int)parenfloor)
            Perl_re_printf( aTHX_
		"rex=0x%"UVxf" offs=0x%"UVxf": saving capture indices:\n",
		PTR2UV(rex),
		PTR2UV(rex->offs)
	    );
    );
    for (p = parenfloor+1; p <= (I32)maxopenparen;  p++) {
/* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */
	SSPUSHIV(rex->offs[p].end);
	SSPUSHIV(rex->offs[p].start);
	SSPUSHINT(rex->offs[p].start_tmp);
        DEBUG_BUFFERS_r(Perl_re_printf( aTHX_
	    "    \\%"UVuf": %"IVdf"(%"IVdf")..%"IVdf"\n",
	    (UV)p,
	    (IV)rex->offs[p].start,
	    (IV)rex->offs[p].start_tmp,
	    (IV)rex->offs[p].end
	));
    }
/* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */
    SSPUSHINT(maxopenparen);
    SSPUSHINT(rex->lastparen);
    SSPUSHINT(rex->lastcloseparen);
    SSPUSHUV(SAVEt_REGCONTEXT | elems_shifted); /* Magic cookie. */

    return retval;
}

/* These are needed since we do not localize EVAL nodes: */
#define REGCP_SET(cp)                                           \
    DEBUG_STATE_r(                                              \
        Perl_re_exec_indentf( aTHX_                                         \
            "Setting an EVAL scope, savestack=%"IVdf",\n",      \
            depth, (IV)PL_savestack_ix                          \
        )                                                       \
    );                                                          \
    cp = PL_savestack_ix

#define REGCP_UNWIND(cp)                                        \
    DEBUG_STATE_r(                                              \
        if (cp != PL_savestack_ix)                              \
            Perl_re_exec_indentf( aTHX_                                     \
                "Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n",\
                depth, (IV)(cp), (IV)PL_savestack_ix            \
            )                                                   \
    );                                                          \
    regcpblow(cp)

#define UNWIND_PAREN(lp, lcp)               \
    for (n = rex->lastparen; n > lp; n--)   \
        rex->offs[n].end = -1;              \
    rex->lastparen = n;                     \
    rex->lastcloseparen = lcp;


STATIC void
S_regcppop(pTHX_ regexp *rex, U32 *maxopenparen_p)
{
    UV i;
    U32 paren;
    GET_RE_DEBUG_FLAGS_DECL;

    PERL_ARGS_ASSERT_REGCPPOP;

    /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */
    i = SSPOPUV;
    assert((i & SAVE_MASK) == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */
    i >>= SAVE_TIGHT_SHIFT; /* Parentheses elements to pop. */
    rex->lastcloseparen = SSPOPINT;
    rex->lastparen = SSPOPINT;
    *maxopenparen_p = SSPOPINT;

    i -= REGCP_OTHER_ELEMS;
    /* Now restore the parentheses context. */
    DEBUG_BUFFERS_r(
	if (i || rex->lastparen + 1 <= rex->nparens)
            Perl_re_printf( aTHX_
		"rex=0x%"UVxf" offs=0x%"UVxf": restoring capture indices to:\n",
		PTR2UV(rex),
		PTR2UV(rex->offs)
	    );
    );
    paren = *maxopenparen_p;
    for ( ; i > 0; i -= REGCP_PAREN_ELEMS) {
	SSize_t tmps;
	rex->offs[paren].start_tmp = SSPOPINT;
	rex->offs[paren].start = SSPOPIV;
	tmps = SSPOPIV;
	if (paren <= rex->lastparen)
	    rex->offs[paren].end = tmps;
        DEBUG_BUFFERS_r( Perl_re_printf( aTHX_
	    "    \\%"UVuf": %"IVdf"(%"IVdf")..%"IVdf"%s\n",
	    (UV)paren,
	    (IV)rex->offs[paren].start,
	    (IV)rex->offs[paren].start_tmp,
	    (IV)rex->offs[paren].end,
	    (paren > rex->lastparen ? "(skipped)" : ""));
	);
	paren--;
    }
#if 1
    /* It would seem that the similar code in regtry()
     * already takes care of this, and in fact it is in
     * a better location to since this code can #if 0-ed out
     * but the code in regtry() is needed or otherwise tests
     * requiring null fields (pat.t#187 and split.t#{13,14}
     * (as of patchlevel 7877)  will fail.  Then again,
     * this code seems to be necessary or otherwise
     * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/
     * --jhi updated by dapm */
    for (i = rex->lastparen + 1; i <= rex->nparens; i++) {
	if (i > *maxopenparen_p)
	    rex->offs[i].start = -1;
	rex->offs[i].end = -1;
        DEBUG_BUFFERS_r( Perl_re_printf( aTHX_
	    "    \\%"UVuf": %s   ..-1 undeffing\n",
	    (UV)i,
	    (i > *maxopenparen_p) ? "-1" : "  "
	));
    }
#endif
}

/* restore the parens and associated vars at savestack position ix,
 * but without popping the stack */

STATIC void
S_regcp_restore(pTHX_ regexp *rex, I32 ix, U32 *maxopenparen_p)
{
    I32 tmpix = PL_savestack_ix;
    PL_savestack_ix = ix;
    regcppop(rex, maxopenparen_p);
    PL_savestack_ix = tmpix;
}

#define regcpblow(cp) LEAVE_SCOPE(cp)	/* Ignores regcppush()ed data. */

STATIC bool
S_isFOO_lc(pTHX_ const U8 classnum, const U8 character)
{
    /* Returns a boolean as to whether or not 'character' is a member of the
     * Posix character class given by 'classnum' that should be equivalent to a
     * value in the typedef '_char_class_number'.
     *
     * Ideally this could be replaced by a just an array of function pointers
     * to the C library functions that implement the macros this calls.
     * However, to compile, the precise function signatures are required, and
     * these may vary from platform to to platform.  To avoid having to figure
     * out what those all are on each platform, I (khw) am using this method,
     * which adds an extra layer of function call overhead (unless the C
     * optimizer strips it away).  But we don't particularly care about
     * performance with locales anyway. */

    switch ((_char_class_number) classnum) {
        case _CC_ENUM_ALPHANUMERIC: return isALPHANUMERIC_LC(character);
        case _CC_ENUM_ALPHA:     return isALPHA_LC(character);
        case _CC_ENUM_ASCII:     return isASCII_LC(character);
        case _CC_ENUM_BLANK:     return isBLANK_LC(character);
        case _CC_ENUM_CASED:     return isLOWER_LC(character)
                                        || isUPPER_LC(character);
        case _CC_ENUM_CNTRL:     return isCNTRL_LC(character);
        case _CC_ENUM_DIGIT:     return isDIGIT_LC(character);
        case _CC_ENUM_GRAPH:     return isGRAPH_LC(character);
        case _CC_ENUM_LOWER:     return isLOWER_LC(character);
        case _CC_ENUM_PRINT:     return isPRINT_LC(character);
        case _CC_ENUM_PUNCT:     return isPUNCT_LC(character);
        case _CC_ENUM_SPACE:     return isSPACE_LC(character);
        case _CC_ENUM_UPPER:     return isUPPER_LC(character);
        case _CC_ENUM_WORDCHAR:  return isWORDCHAR_LC(character);
        case _CC_ENUM_XDIGIT:    return isXDIGIT_LC(character);
        default:    /* VERTSPACE should never occur in locales */
            Perl_croak(aTHX_ "panic: isFOO_lc() has an unexpected character class '%d'", classnum);
    }

    NOT_REACHED; /* NOTREACHED */
    return FALSE;
}

STATIC bool
S_isFOO_utf8_lc(pTHX_ const U8 classnum, const U8* character)
{
    /* Returns a boolean as to whether or not the (well-formed) UTF-8-encoded
     * 'character' is a member of the Posix character class given by 'classnum'
     * that should be equivalent to a value in the typedef
     * '_char_class_number'.
     *
     * This just calls isFOO_lc on the code point for the character if it is in
     * the range 0-255.  Outside that range, all characters use Unicode
     * rules, ignoring any locale.  So use the Unicode function if this class
     * requires a swash, and use the Unicode macro otherwise. */

    PERL_ARGS_ASSERT_ISFOO_UTF8_LC;

    if (UTF8_IS_INVARIANT(*character)) {
        return isFOO_lc(classnum, *character);
    }
    else if (UTF8_IS_DOWNGRADEABLE_START(*character)) {
        return isFOO_lc(classnum,
                        EIGHT_BIT_UTF8_TO_NATIVE(*character, *(character + 1)));
    }

    _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(character, character + UTF8SKIP(character));

    if (classnum < _FIRST_NON_SWASH_CC) {

        /* Initialize the swash unless done already */
        if (! PL_utf8_swash_ptrs[classnum]) {
            U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;
            PL_utf8_swash_ptrs[classnum] =
                    _core_swash_init("utf8",
                                     "",
                                     &PL_sv_undef, 1, 0,
                                     PL_XPosix_ptrs[classnum], &flags);
        }

        return cBOOL(swash_fetch(PL_utf8_swash_ptrs[classnum], (U8 *)
                                 character,
                                 TRUE /* is UTF */ ));
    }

    switch ((_char_class_number) classnum) {
        case _CC_ENUM_SPACE:     return is_XPERLSPACE_high(character);
        case _CC_ENUM_BLANK:     return is_HORIZWS_high(character);
        case _CC_ENUM_XDIGIT:    return is_XDIGIT_high(character);
        case _CC_ENUM_VERTSPACE: return is_VERTWS_high(character);
        default:                 break;
    }

    return FALSE; /* Things like CNTRL are always below 256 */
}

/*
 * pregexec and friends
 */

#ifndef PERL_IN_XSUB_RE
/*
 - pregexec - match a regexp against a string
 */
I32
Perl_pregexec(pTHX_ REGEXP * const prog, char* stringarg, char *strend,
	 char *strbeg, SSize_t minend, SV *screamer, U32 nosave)
/* stringarg: the point in the string at which to begin matching */
/* strend:    pointer to null at end of string */
/* strbeg:    real beginning of string */
/* minend:    end of match must be >= minend bytes after stringarg. */
/* screamer:  SV being matched: only used for utf8 flag, pos() etc; string
 *            itself is accessed via the pointers above */
/* nosave:    For optimizations. */
{
    PERL_ARGS_ASSERT_PREGEXEC;

    return
	regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL,
		      nosave ? 0 : REXEC_COPY_STR);
}
#endif



/* re_intuit_start():
 *
 * Based on some optimiser hints, try to find the earliest position in the
 * string where the regex could match.
 *
 *   rx:     the regex to match against
 *   sv:     the SV being matched: only used for utf8 flag; the string
 *           itself is accessed via the pointers below. Note that on
 *           something like an overloaded SV, SvPOK(sv) may be false
 *           and the string pointers may point to something unrelated to
 *           the SV itself.
 *   strbeg: real beginning of string
 *   strpos: the point in the string at which to begin matching
 *   strend: pointer to the byte following the last char of the string
 *   flags   currently unused; set to 0
 *   data:   currently unused; set to NULL
 *
 * The basic idea of re_intuit_start() is to use some known information
 * about the pattern, namely:
 *
 *   a) the longest known anchored substring (i.e. one that's at a
 *      constant offset from the beginning of the pattern; but not
 *      necessarily at a fixed offset from the beginning of the
 *      string);
 *   b) the longest floating substring (i.e. one that's not at a constant
 *      offset from the beginning of the pattern);
 *   c) Whether the pattern is anchored to the string; either
 *      an absolute anchor: /^../, or anchored to \n: /^.../m,
 *      or anchored to pos(): /\G/;
 *   d) A start class: a real or synthetic character class which
 *      represents which characters are legal at the start of the pattern;
 *
 * to either quickly reject the match, or to find the earliest position
 * within the string at which the pattern might match, thus avoiding
 * running the full NFA engine at those earlier locations, only to
 * eventually fail and retry further along.
 *
 * Returns NULL if the pattern can't match, or returns the address within
 * the string which is the earliest place the match could occur.
 *
 * The longest of the anchored and floating substrings is called 'check'
 * and is checked first. The other is called 'other' and is checked
 * second. The 'other' substring may not be present.  For example,
 *
 *    /(abc|xyz)ABC\d{0,3}DEFG/
 *
 * will have
 *
 *   check substr (float)    = "DEFG", offset 6..9 chars
 *   other substr (anchored) = "ABC",  offset 3..3 chars
 *   stclass = [ax]
 *
 * Be aware that during the course of this function, sometimes 'anchored'
 * refers to a substring being anchored relative to the start of the
 * pattern, and sometimes to the pattern itself being anchored relative to
 * the string. For example:
 *
 *   /\dabc/:   "abc" is anchored to the pattern;
 *   /^\dabc/:  "abc" is anchored to the pattern and the string;
 *   /\d+abc/:  "abc" is anchored to neither the pattern nor the string;
 *   /^\d+abc/: "abc" is anchored to neither the pattern nor the string,
 *                    but the pattern is anchored to the string.
 */

char *
Perl_re_intuit_start(pTHX_
                    REGEXP * const rx,
                    SV *sv,
                    const char * const strbeg,
                    char *strpos,
                    char *strend,
                    const U32 flags,
                    re_scream_pos_data *data)
{
    struct regexp *const prog = ReANY(rx);
    SSize_t start_shift = prog->check_offset_min;
    /* Should be nonnegative! */
    SSize_t end_shift   = 0;
    /* current lowest pos in string where the regex can start matching */
    char *rx_origin = strpos;
    SV *check;
    const bool utf8_target = (sv && SvUTF8(sv)) ? 1 : 0; /* if no sv we have to assume bytes */
    U8   other_ix = 1 - prog->substrs->check_ix;
    bool ml_anch = 0;
    char *other_last = strpos;/* latest pos 'other' substr already checked to */
    char *check_at = NULL;		/* check substr found at this pos */
    const I32 multiline = prog->extflags & RXf_PMf_MULTILINE;
    RXi_GET_DECL(prog,progi);
    regmatch_info reginfo_buf;  /* create some info to pass to find_byclass */
    regmatch_info *const reginfo = &reginfo_buf;
    GET_RE_DEBUG_FLAGS_DECL;

    PERL_ARGS_ASSERT_RE_INTUIT_START;
    PERL_UNUSED_ARG(flags);
    PERL_UNUSED_ARG(data);

    DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
                "Intuit: trying to determine minimum start position...\n"));

    /* for now, assume that all substr offsets are positive. If at some point
     * in the future someone wants to do clever things with lookbehind and
     * -ve offsets, they'll need to fix up any code in this function
     * which uses these offsets. See the thread beginning
     * <20140113145929.GF27210@iabyn.com>
     */
    assert(prog->substrs->data[0].min_offset >= 0);
    assert(prog->substrs->data[0].max_offset >= 0);
    assert(prog->substrs->data[1].min_offset >= 0);
    assert(prog->substrs->data[1].max_offset >= 0);
    assert(prog->substrs->data[2].min_offset >= 0);
    assert(prog->substrs->data[2].max_offset >= 0);

    /* for now, assume that if both present, that the floating substring
     * doesn't start before the anchored substring.
     * If you break this assumption (e.g. doing better optimisations
     * with lookahead/behind), then you'll need to audit the code in this
     * function carefully first
     */
    assert(
            ! (  (prog->anchored_utf8 || prog->anchored_substr)
              && (prog->float_utf8    || prog->float_substr))
           || (prog->float_min_offset >= prog->anchored_offset));

    /* byte rather than char calculation for efficiency. It fails
     * to quickly reject some cases that can't match, but will reject
     * them later after doing full char arithmetic */
    if (prog->minlen > strend - strpos) {
        DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
			      "  String too short...\n"));
	goto fail;
    }

    RX_MATCH_UTF8_set(rx,utf8_target);
    reginfo->is_utf8_target = cBOOL(utf8_target);
    reginfo->info_aux = NULL;
    reginfo->strbeg = strbeg;
    reginfo->strend = strend;
    reginfo->is_utf8_pat = cBOOL(RX_UTF8(rx));
    reginfo->intuit = 1;
    /* not actually used within intuit, but zero for safety anyway */
    reginfo->poscache_maxiter = 0;

    if (utf8_target) {
	if (!prog->check_utf8 && prog->check_substr)
	    to_utf8_substr(prog);
	check = prog->check_utf8;
    } else {
	if (!prog->check_substr && prog->check_utf8) {
	    if (! to_byte_substr(prog)) {
                NON_UTF8_TARGET_BUT_UTF8_REQUIRED(fail);
            }
        }
	check = prog->check_substr;
    }

    /* dump the various substring data */
    DEBUG_OPTIMISE_MORE_r({
        int i;
        for (i=0; i<=2; i++) {
            SV *sv = (utf8_target ? prog->substrs->data[i].utf8_substr
                                  : prog->substrs->data[i].substr);
            if (!sv)
                continue;

            Perl_re_printf( aTHX_
                "  substrs[%d]: min=%"IVdf" max=%"IVdf" end shift=%"IVdf
                " useful=%"IVdf" utf8=%d [%s]\n",
                i,
                (IV)prog->substrs->data[i].min_offset,
                (IV)prog->substrs->data[i].max_offset,
                (IV)prog->substrs->data[i].end_shift,
                BmUSEFUL(sv),
                utf8_target ? 1 : 0,
                SvPEEK(sv));
        }
    });

    if (prog->intflags & PREGf_ANCH) { /* Match at \G, beg-of-str or after \n */

        /* ml_anch: check after \n?
         *
         * A note about PREGf_IMPLICIT: on an un-anchored pattern beginning
         * with /.*.../, these flags will have been added by the
         * compiler:
         *   /.*abc/, /.*abc/m:  PREGf_IMPLICIT | PREGf_ANCH_MBOL
         *   /.*abc/s:           PREGf_IMPLICIT | PREGf_ANCH_SBOL
         */
	ml_anch =      (prog->intflags & PREGf_ANCH_MBOL)
                   && !(prog->intflags & PREGf_IMPLICIT);

	if (!ml_anch && !(prog->intflags & PREGf_IMPLICIT)) {
            /* we are only allowed to match at BOS or \G */

            /* trivially reject if there's a BOS anchor and we're not at BOS.
             *
             * Note that we don't try to do a similar quick reject for
             * \G, since generally the caller will have calculated strpos
             * based on pos() and gofs, so the string is already correctly
             * anchored by definition; and handling the exceptions would
             * be too fiddly (e.g. REXEC_IGNOREPOS).
             */
            if (   strpos != strbeg
                && (prog->intflags & PREGf_ANCH_SBOL))
            {
                DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
                                "  Not at start...\n"));
	        goto fail;
	    }

            /* in the presence of an anchor, the anchored (relative to the
             * start of the regex) substr must also be anchored relative
             * to strpos. So quickly reject if substr isn't found there.
             * This works for \G too, because the caller will already have
             * subtracted gofs from pos, and gofs is the offset from the
             * \G to the start of the regex. For example, in /.abc\Gdef/,
             * where substr="abcdef", pos()=3, gofs=4, offset_min=1:
             * caller will have set strpos=pos()-4; we look for the substr
             * at position pos()-4+1, which lines up with the "a" */

	    if (prog->check_offset_min == prog->check_offset_max) {
	        /* Substring at constant offset from beg-of-str... */
	        SSize_t slen = SvCUR(check);
                char *s = HOP3c(strpos, prog->check_offset_min, strend);
	    
                DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
                    "  Looking for check substr at fixed offset %"IVdf"...\n",
                    (IV)prog->check_offset_min));

	        if (SvTAIL(check)) {
                    /* In this case, the regex is anchored at the end too.
                     * Unless it's a multiline match, the lengths must match
                     * exactly, give or take a \n.  NB: slen >= 1 since
                     * the last char of check is \n */
		    if (!multiline
                        && (   strend - s > slen
                            || strend - s < slen - 1
                            || (strend - s == slen && strend[-1] != '\n')))
                    {
                        DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
                                            "  String too long...\n"));
                        goto fail_finish;
                    }
                    /* Now should match s[0..slen-2] */
                    slen--;
                }
                if (slen && (*SvPVX_const(check) != *s
                    || (slen > 1 && memNE(SvPVX_const(check), s, slen))))
                {
                    DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
                                    "  String not equal...\n"));
                    goto fail_finish;
                }

                check_at = s;
                goto success_at_start;
	    }
	}
    }

    end_shift = prog->check_end_shift;

#ifdef DEBUGGING	/* 7/99: reports of failure (with the older version) */
    if (end_shift < 0)
	Perl_croak(aTHX_ "panic: end_shift: %"IVdf" pattern:\n%s\n ",
		   (IV)end_shift, RX_PRECOMP(prog));
#endif

  restart:
    
    /* This is the (re)entry point of the main loop in this function.
     * The goal of this loop is to:
     * 1) find the "check" substring in the region rx_origin..strend
     *    (adjusted by start_shift / end_shift). If not found, reject
     *    immediately.
     * 2) If it exists, look for the "other" substr too if defined; for
     *    example, if the check substr maps to the anchored substr, then
     *    check the floating substr, and vice-versa. If not found, go
     *    back to (1) with rx_origin suitably incremented.
     * 3) If we find an rx_origin position that doesn't contradict
     *    either of the substrings, then check the possible additional
     *    constraints on rx_origin of /^.../m or a known start class.
     *    If these fail, then depending on which constraints fail, jump
     *    back to here, or to various other re-entry points further along
     *    that skip some of the first steps.
     * 4) If we pass all those tests, update the BmUSEFUL() count on the
     *    substring. If the start position was determined to be at the
     *    beginning of the string  - so, not rejected, but not optimised,
     *    since we have to run regmatch from position 0 - decrement the
     *    BmUSEFUL() count. Otherwise increment it.
     */


    /* first, look for the 'check' substring */

    {
        U8* start_point;
        U8* end_point;

        DEBUG_OPTIMISE_MORE_r({
            Perl_re_printf( aTHX_
                "  At restart: rx_origin=%"IVdf" Check offset min: %"IVdf
                " Start shift: %"IVdf" End shift %"IVdf
                " Real end Shift: %"IVdf"\n",
                (IV)(rx_origin - strbeg),
                (IV)prog->check_offset_min,
                (IV)start_shift,
                (IV)end_shift,
                (IV)prog->check_end_shift);
        });
        
        end_point = HOPBACK3(strend, end_shift, rx_origin);
        if (!end_point)
            goto fail_finish;
        start_point = HOPMAYBE3(rx_origin, start_shift, end_point);
        if (!start_point)
            goto fail_finish;


        /* If the regex is absolutely anchored to either the start of the
         * string (SBOL) or to pos() (ANCH_GPOS), then
         * check_offset_max represents an upper bound on the string where
         * the substr could start. For the ANCH_GPOS case, we assume that
         * the caller of intuit will have already set strpos to
         * pos()-gofs, so in this case strpos + offset_max will still be
         * an upper bound on the substr.
         */
        if (!ml_anch
            && prog->intflags & PREGf_ANCH
            && prog->check_offset_max != SSize_t_MAX)
        {
            SSize_t len = SvCUR(check) - !!SvTAIL(check);
            const char * const anchor =
                        (prog->intflags & PREGf_ANCH_GPOS ? strpos : strbeg);

            /* do a bytes rather than chars comparison. It's conservative;
             * so it skips doing the HOP if the result can't possibly end
             * up earlier than the old value of end_point.
             */
            if ((char*)end_point - anchor > prog->check_offset_max) {
                end_point = HOP3lim((U8*)anchor,
                                prog->check_offset_max,
                                end_point -len)
                            + len;
            }
        }

	check_at = fbm_instr( start_point, end_point,
		      check, multiline ? FBMrf_MULTILINE : 0);

        DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
            "  doing 'check' fbm scan, [%"IVdf"..%"IVdf"] gave %"IVdf"\n",
            (IV)((char*)start_point - strbeg),
            (IV)((char*)end_point   - strbeg),
            (IV)(check_at ? check_at - strbeg : -1)
        ));

        /* Update the count-of-usability, remove useless subpatterns,
            unshift s.  */

        DEBUG_EXECUTE_r({
            RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
                SvPVX_const(check), RE_SV_DUMPLEN(check), 30);
            Perl_re_printf( aTHX_  "  %s %s substr %s%s%s",
                              (check_at ? "Found" : "Did not find"),
                (check == (utf8_target ? prog->anchored_utf8 : prog->anchored_substr)
                    ? "anchored" : "floating"),
                quoted,
                RE_SV_TAIL(check),
                (check_at ? " at offset " : "...\n") );
        });

        if (!check_at)
            goto fail_finish;
        /* set rx_origin to the minimum position where the regex could start
         * matching, given the constraint of the just-matched check substring.
         * But don't set it lower than previously.
         */

        if (check_at - rx_origin > prog->check_offset_max)
            rx_origin = HOP3c(check_at, -prog->check_offset_max, rx_origin);
        /* Finish the diagnostic message */
        DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
            "%ld (rx_origin now %"IVdf")...\n",
            (long)(check_at - strbeg),
            (IV)(rx_origin - strbeg)
        ));
    }


    /* now look for the 'other' substring if defined */

    if (utf8_target ? prog->substrs->data[other_ix].utf8_substr
                    : prog->substrs->data[other_ix].substr)
    {
	/* Take into account the "other" substring. */
        char *last, *last1;
        char *s;
        SV* must;
        struct reg_substr_datum *other;

      do_other_substr:
        other = &prog->substrs->data[other_ix];

        /* if "other" is anchored:
         * we've previously found a floating substr starting at check_at.
         * This means that the regex origin must lie somewhere
         * between min (rx_origin): HOP3(check_at, -check_offset_max)
         * and max:                 HOP3(check_at, -check_offset_min)
         * (except that min will be >= strpos)
         * So the fixed  substr must lie somewhere between
         *  HOP3(min, anchored_offset)
         *  HOP3(max, anchored_offset) + SvCUR(substr)
         */

        /* if "other" is floating
         * Calculate last1, the absolute latest point where the
         * floating substr could start in the string, ignoring any
         * constraints from the earlier fixed match. It is calculated
         * as follows:
         *
         * strend - prog->minlen (in chars) is the absolute latest
         * position within the string where the origin of the regex
         * could appear. The latest start point for the floating
         * substr is float_min_offset(*) on from the start of the
         * regex.  last1 simply combines thee two offsets.
         *
         * (*) You might think the latest start point should be
         * float_max_offset from the regex origin, and technically
         * you'd be correct. However, consider
         *    /a\d{2,4}bcd\w/
         * Here, float min, max are 3,5 and minlen is 7.
         * This can match either
         *    /a\d\dbcd\w/
         *    /a\d\d\dbcd\w/
         *    /a\d\d\d\dbcd\w/
         * In the first case, the regex matches minlen chars; in the
         * second, minlen+1, in the third, minlen+2.
         * In the first case, the floating offset is 3 (which equals
         * float_min), in the second, 4, and in the third, 5 (which
         * equals float_max). In all cases, the floating string bcd
         * can never start more than 4 chars from the end of the
         * string, which equals minlen - float_min. As the substring
         * starts to match more than float_min from the start of the
         * regex, it makes the regex match more than minlen chars,
         * and the two cancel each other out. So we can always use
         * float_min - minlen, rather than float_max - minlen for the
         * latest position in the string.
         *
         * Note that -minlen + float_min_offset is equivalent (AFAIKT)
         * to CHR_SVLEN(must) - !!SvTAIL(must) + prog->float_end_shift
         */

        assert(prog->minlen >= other->min_offset);
        last1 = HOP3c(strend,
                        other->min_offset - prog->minlen, strbeg);

        if (other_ix) {/* i.e. if (other-is-float) */
            /* last is the latest point where the floating substr could
             * start, *given* any constraints from the earlier fixed
             * match. This constraint is that the floating string starts
             * <= float_max_offset chars from the regex origin (rx_origin).
             * If this value is less than last1, use it instead.
             */
            assert(rx_origin <= last1);
            last =
                /* this condition handles the offset==infinity case, and
                 * is a short-cut otherwise. Although it's comparing a
                 * byte offset to a char length, it does so in a safe way,
                 * since 1 char always occupies 1 or more bytes,
                 * so if a string range is  (last1 - rx_origin) bytes,
                 * it will be less than or equal to  (last1 - rx_origin)
                 * chars; meaning it errs towards doing the accurate HOP3
                 * rather than just using last1 as a short-cut */
                (last1 - rx_origin) < other->max_offset
                    ? last1
                    : (char*)HOP3lim(rx_origin, other->max_offset, last1);
        }
        else {
            assert(strpos + start_shift <= check_at);
            last = HOP4c(check_at, other->min_offset - start_shift,
                        strbeg, strend);
        }

        s = HOP3c(rx_origin, other->min_offset, strend);
        if (s < other_last)	/* These positions already checked */
            s = other_last;

        must = utf8_target ? other->utf8_substr : other->substr;
        assert(SvPOK(must));
        {
            char *from = s;
            char *to   = last + SvCUR(must) - (SvTAIL(must)!=0);

            if (from > to) {
                s = NULL;
                DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
                    "  skipping 'other' fbm scan: %"IVdf" > %"IVdf"\n",
                    (IV)(from - strbeg),
                    (IV)(to   - strbeg)
                ));
            }
            else {
                s = fbm_instr(
                    (unsigned char*)from,
                    (unsigned char*)to,
                    must,
                    multiline ? FBMrf_MULTILINE : 0
                );
                DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
                    "  doing 'other' fbm scan, [%"IVdf"..%"IVdf"] gave %"IVdf"\n",
                    (IV)(from - strbeg),
                    (IV)(to   - strbeg),
                    (IV)(s ? s - strbeg : -1)
                ));
            }
        }

        DEBUG_EXECUTE_r({
            RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
                SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
            Perl_re_printf( aTHX_  "  %s %s substr %s%s",
                s ? "Found" : "Contradicts",
                other_ix ? "floating" : "anchored",
                quoted, RE_SV_TAIL(must));
        });


        if (!s) {
            /* last1 is latest possible substr location. If we didn't
             * find it before there, we never will */
            if (last >= last1) {
                DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
                                        "; giving up...\n"));
                goto fail_finish;
            }

            /* try to find the check substr again at a later
             * position. Maybe next time we'll find the "other" substr
             * in range too */
            other_last = HOP3c(last, 1, strend) /* highest failure */;
            rx_origin =
                other_ix /* i.e. if other-is-float */
                    ? HOP3c(rx_origin, 1, strend)
                    : HOP4c(last, 1 - other->min_offset, strbeg, strend);
            DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
                "; about to retry %s at offset %ld (rx_origin now %"IVdf")...\n",
                (other_ix ? "floating" : "anchored"),
                (long)(HOP3c(check_at, 1, strend) - strbeg),
                (IV)(rx_origin - strbeg)
            ));
            goto restart;
        }
        else {
            if (other_ix) { /* if (other-is-float) */
                /* other_last is set to s, not s+1, since its possible for
                 * a floating substr to fail first time, then succeed
                 * second time at the same floating position; e.g.:
                 *     "-AB--AABZ" =~ /\wAB\d*Z/
                 * The first time round, anchored and float match at
                 * "-(AB)--AAB(Z)" then fail on the initial \w character
                 * class. Second time round, they match at "-AB--A(AB)(Z)".
                 */
                other_last = s;
            }
            else {
                rx_origin = HOP3c(s, -other->min_offset, strbeg);
                other_last = HOP3c(s, 1, strend);
            }
            DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
                " at offset %ld (rx_origin now %"IVdf")...\n",
                  (long)(s - strbeg),
                (IV)(rx_origin - strbeg)
              ));

        }
    }
    else {
        DEBUG_OPTIMISE_MORE_r(
            Perl_re_printf( aTHX_
                "  Check-only match: offset min:%"IVdf" max:%"IVdf
                " check_at:%"IVdf" rx_origin:%"IVdf" rx_origin-check_at:%"IVdf
                " strend:%"IVdf"\n",
                (IV)prog->check_offset_min,
                (IV)prog->check_offset_max,
                (IV)(check_at-strbeg),
                (IV)(rx_origin-strbeg),
                (IV)(rx_origin-check_at),
                (IV)(strend-strbeg)
            )
        );
    }

  postprocess_substr_matches:

    /* handle the extra constraint of /^.../m if present */

    if (ml_anch && rx_origin != strbeg && rx_origin[-1] != '\n') {
        char *s;

        DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
                        "  looking for /^/m anchor"));

        /* we have failed the constraint of a \n before rx_origin.
         * Find the next \n, if any, even if it's beyond the current
         * anchored and/or floating substrings. Whether we should be
         * scanning ahead for the next \n or the next substr is debatable.
         * On the one hand you'd expect rare substrings to appear less
         * often than \n's. On the other hand, searching for \n means
         * we're effectively flipping between check_substr and "\n" on each
         * iteration as the current "rarest" string candidate, which
         * means for example that we'll quickly reject the whole string if
         * hasn't got a \n, rather than trying every substr position
         * first
         */

        s = HOP3c(strend, - prog->minlen, strpos);
        if (s <= rx_origin ||
            ! ( rx_origin = (char *)memchr(rx_origin, '\n', s - rx_origin)))
        {
            DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
                            "  Did not find /%s^%s/m...\n",
                            PL_colors[0], PL_colors[1]));
            goto fail_finish;
        }

        /* earliest possible origin is 1 char after the \n.
         * (since *rx_origin == '\n', it's safe to ++ here rather than
         * HOP(rx_origin, 1)) */
        rx_origin++;

        if (prog->substrs->check_ix == 0  /* check is anchored */
            || rx_origin >= HOP3c(check_at,  - prog->check_offset_min, strpos))
        {
            /* Position contradicts check-string; either because
             * check was anchored (and thus has no wiggle room),
             * or check was float and rx_origin is above the float range */
            DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
                "  Found /%s^%s/m, about to restart lookup for check-string with rx_origin %ld...\n",
                PL_colors[0], PL_colors[1], (long)(rx_origin - strbeg)));
            goto restart;
        }

        /* if we get here, the check substr must have been float,
         * is in range, and we may or may not have had an anchored
         * "other" substr which still contradicts */
        assert(prog->substrs->check_ix); /* check is float */

        if (utf8_target ? prog->anchored_utf8 : prog->anchored_substr) {
            /* whoops, the anchored "other" substr exists, so we still
             * contradict. On the other hand, the float "check" substr
             * didn't contradict, so just retry the anchored "other"
             * substr */
            DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
                "  Found /%s^%s/m, rescanning for anchored from offset %"IVdf" (rx_origin now %"IVdf")...\n",
                PL_colors[0], PL_colors[1],
                (IV)(rx_origin - strbeg + prog->anchored_offset),
                (IV)(rx_origin - strbeg)
            ));
            goto do_other_substr;
        }

        /* success: we don't contradict the found floating substring
         * (and there's no anchored substr). */
        DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
            "  Found /%s^%s/m with rx_origin %ld...\n",
            PL_colors[0], PL_colors[1], (long)(rx_origin - strbeg)));
    }
    else {
        DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
            "  (multiline anchor test skipped)\n"));
    }

  success_at_start:


    /* if we have a starting character class, then test that extra constraint.
     * (trie stclasses are too expensive to use here, we are better off to
     * leave it to regmatch itself) */

    if (progi->regstclass && PL_regkind[OP(progi->regstclass)]!=TRIE) {
        const U8* const str = (U8*)STRING(progi->regstclass);

        /* XXX this value could be pre-computed */
        const int cl_l = (PL_regkind[OP(progi->regstclass)] == EXACT
		    ?  (reginfo->is_utf8_pat
                        ? utf8_distance(str + STR_LEN(progi->regstclass), str)
                        : STR_LEN(progi->regstclass))
		    : 1);
	char * endpos;
        char *s;
        /* latest pos that a matching float substr constrains rx start to */
        char *rx_max_float = NULL;

        /* if the current rx_origin is anchored, either by satisfying an
         * anchored substring constraint, or a /^.../m constraint, then we
         * can reject the current origin if the start class isn't found
         * at the current position. If we have a float-only match, then
         * rx_origin is constrained to a range; so look for the start class
         * in that range. if neither, then look for the start class in the
         * whole rest of the string */

        /* XXX DAPM it's not clear what the minlen test is for, and why
         * it's not used in the floating case. Nothing in the test suite
         * causes minlen == 0 here. See <20140313134639.GS12844@iabyn.com>.
         * Here are some old comments, which may or may not be correct:
         *
	 *   minlen == 0 is possible if regstclass is \b or \B,
	 *   and the fixed substr is ''$.
         *   Since minlen is already taken into account, rx_origin+1 is
         *   before strend; accidentally, minlen >= 1 guaranties no false
         *   positives at rx_origin + 1 even for \b or \B.  But (minlen? 1 :
         *   0) below assumes that regstclass does not come from lookahead...
	 *   If regstclass takes bytelength more than 1: If charlength==1, OK.
         *   This leaves EXACTF-ish only, which are dealt with in
         *   find_byclass().
         */

	if (prog->anchored_substr || prog->anchored_utf8 || ml_anch)
            endpos= HOP3c(rx_origin, (prog->minlen ? cl_l : 0), strend);
        else if (prog->float_substr || prog->float_utf8) {
	    rx_max_float = HOP3c(check_at, -start_shift, strbeg);
	    endpos= HOP3c(rx_max_float, cl_l, strend);
        }
        else 
            endpos= strend;
		    
        DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
            "  looking for class: start_shift: %"IVdf" check_at: %"IVdf
            " rx_origin: %"IVdf" endpos: %"IVdf"\n",
              (IV)start_shift, (IV)(check_at - strbeg),
              (IV)(rx_origin - strbeg), (IV)(endpos - strbeg)));

        s = find_byclass(prog, progi->regstclass, rx_origin, endpos,
                            reginfo);
	if (!s) {
	    if (endpos == strend) {
                DEBUG_EXECUTE_r( Perl_re_printf( aTHX_
				"  Could not match STCLASS...\n") );
		goto fail;
	    }
            DEBUG_EXECUTE_r( Perl_re_printf( aTHX_
                               "  This position contradicts STCLASS...\n") );
            if ((prog->intflags & PREGf_ANCH) && !ml_anch
                        && !(prog->intflags & PREGf_IMPLICIT))
		goto fail;

	    /* Contradict one of substrings */
	    if (prog->anchored_substr || prog->anchored_utf8) {
                if (prog->substrs->check_ix == 1) { /* check is float */
                    /* Have both, check_string is floating */
                    assert(rx_origin + start_shift <= check_at);
                    if (rx_origin + start_shift != check_at) {
                        /* not at latest position float substr could match:
                         * Recheck anchored substring, but not floating.
                         * The condition above is in bytes rather than
                         * chars for efficiency. It's conservative, in
                         * that it errs on the side of doing 'goto
                         * do_other_substr'. In this case, at worst,
                         * an extra anchored search may get done, but in
                         * practice the extra fbm_instr() is likely to
                         * get skipped anyway. */
                        DEBUG_EXECUTE_r( Perl_re_printf( aTHX_
                            "  about to retry anchored at offset %ld (rx_origin now %"IVdf")...\n",
                            (long)(other_last - strbeg),
                            (IV)(rx_origin - strbeg)
                        ));
                        goto do_other_substr;
                    }
                }
            }
	    else {
                /* float-only */

                if (ml_anch) {
                    /* In the presence of ml_anch, we might be able to
                     * find another \n without breaking the current float
                     * constraint. */

                    /* strictly speaking this should be HOP3c(..., 1, ...),
                     * but since we goto a block of code that's going to
                     * search for the next \n if any, its safe here */
                    rx_origin++;
                    DEBUG_EXECUTE_r( Perl_re_printf( aTHX_
                              "  about to look for /%s^%s/m starting at rx_origin %ld...\n",
                              PL_colors[0], PL_colors[1],
                              (long)(rx_origin - strbeg)) );
                    goto postprocess_substr_matches;
                }

                /* strictly speaking this can never be true; but might
                 * be if we ever allow intuit without substrings */
                if (!(utf8_target ? prog->float_utf8 : prog->float_substr))
                    goto fail;

                rx_origin = rx_max_float;
            }

            /* at this point, any matching substrings have been
             * contradicted. Start again... */

            rx_origin = HOP3c(rx_origin, 1, strend);

            /* uses bytes rather than char calculations for efficiency.
             * It's conservative: it errs on the side of doing 'goto restart',
             * where there is code that does a proper char-based test */
            if (rx_origin + start_shift + end_shift > strend) {
                DEBUG_EXECUTE_r( Perl_re_printf( aTHX_
                                       "  Could not match STCLASS...\n") );
                goto fail;
            }
            DEBUG_EXECUTE_r( Perl_re_printf( aTHX_
                "  about to look for %s substr starting at offset %ld (rx_origin now %"IVdf")...\n",
                (prog->substrs->check_ix ? "floating" : "anchored"),
                (long)(rx_origin + start_shift - strbeg),
                (IV)(rx_origin - strbeg)
            ));
            goto restart;
	}

        /* Success !!! */

	if (rx_origin != s) {
            DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
			"  By STCLASS: moving %ld --> %ld\n",
                                  (long)(rx_origin - strbeg), (long)(s - strbeg))
                   );
        }
        else {
            DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
                                  "  Does not contradict STCLASS...\n");
                   );
        }
    }

    /* Decide whether using the substrings helped */

    if (rx_origin != strpos) {
	/* Fixed substring is found far enough so that the match
	   cannot start at strpos. */

        DEBUG_EXECUTE_r(Perl_re_printf( aTHX_  "  try at offset...\n"));
	++BmUSEFUL(utf8_target ? prog->check_utf8 : prog->check_substr);	/* hooray/5 */
    }
    else {
        /* The found rx_origin position does not prohibit matching at
         * strpos, so calling intuit didn't gain us anything. Decrement
         * the BmUSEFUL() count on the check substring, and if we reach
         * zero, free it.  */
	if (!(prog->intflags & PREGf_NAUGHTY)
	    && (utf8_target ? (
		prog->check_utf8		/* Could be deleted already */
		&& --BmUSEFUL(prog->check_utf8) < 0
		&& (prog->check_utf8 == prog->float_utf8)
	    ) : (
		prog->check_substr		/* Could be deleted already */
		&& --BmUSEFUL(prog->check_substr) < 0
		&& (prog->check_substr == prog->float_substr)
	    )))
	{
	    /* If flags & SOMETHING - do not do it many times on the same match */
            DEBUG_EXECUTE_r(Perl_re_printf( aTHX_  "  ... Disabling check substring...\n"));
	    /* XXX Does the destruction order has to change with utf8_target? */
	    SvREFCNT_dec(utf8_target ? prog->check_utf8 : prog->check_substr);
	    SvREFCNT_dec(utf8_target ? prog->check_substr : prog->check_utf8);
	    prog->check_substr = prog->check_utf8 = NULL;	/* disable */
	    prog->float_substr = prog->float_utf8 = NULL;	/* clear */
	    check = NULL;			/* abort */
	    /* XXXX This is a remnant of the old implementation.  It
	            looks wasteful, since now INTUIT can use many
	            other heuristics. */
	    prog->extflags &= ~RXf_USE_INTUIT;
	}
    }

    DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
            "Intuit: %sSuccessfully guessed:%s match at offset %ld\n",
             PL_colors[4], PL_colors[5], (long)(rx_origin - strbeg)) );

    return rx_origin;

  fail_finish:				/* Substring not found */
    if (prog->check_substr || prog->check_utf8)		/* could be removed already */
	BmUSEFUL(utf8_target ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */
  fail:
    DEBUG_EXECUTE_r(Perl_re_printf( aTHX_  "%sMatch rejected by optimizer%s\n",
			  PL_colors[4], PL_colors[5]));
    return NULL;
}


#define DECL_TRIE_TYPE(scan) \
    const enum { trie_plain, trie_utf8, trie_utf8_fold, trie_latin_utf8_fold,       \
                 trie_utf8_exactfa_fold, trie_latin_utf8_exactfa_fold,              \
                 trie_utf8l, trie_flu8, trie_flu8_latin }                           \
                    trie_type = ((scan->flags == EXACT)                             \
                                 ? (utf8_target ? trie_utf8 : trie_plain)           \
                                 : (scan->flags == EXACTL)                          \
                                    ? (utf8_target ? trie_utf8l : trie_plain)       \
                                    : (scan->flags == EXACTFA)                      \
                                      ? (utf8_target                                \
                                         ? trie_utf8_exactfa_fold                   \
                                         : trie_latin_utf8_exactfa_fold)            \
                                      : (scan->flags == EXACTFLU8                   \
                                         ? (utf8_target                             \
                                           ? trie_flu8                              \
                                           : trie_flu8_latin)                       \
                                         : (utf8_target                             \
                                           ? trie_utf8_fold                         \
                                           : trie_latin_utf8_fold)))

/* 'uscan' is set to foldbuf, and incremented, so below the end of uscan is
 * 'foldbuf+sizeof(foldbuf)' */
#define REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, uc_end, uscan, len, uvc, charid, foldlen, foldbuf, uniflags) \
STMT_START {                                                                        \
    STRLEN skiplen;                                                                 \
    U8 flags = FOLD_FLAGS_FULL;                                                     \
    switch (trie_type) {                                                            \
    case trie_flu8:                                                                 \
        _CHECK_AND_WARN_PROBLEMATIC_LOCALE;                                         \
        if (UTF8_IS_ABOVE_LATIN1(*uc)) {                                            \
            _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(uc, uc_end - uc);                \
        }                                                                           \
        goto do_trie_utf8_fold;                                                     \
    case trie_utf8_exactfa_fold:                                                    \
        flags |= FOLD_FLAGS_NOMIX_ASCII;                                            \
        /* FALLTHROUGH */                                                           \
    case trie_utf8_fold:                                                            \
      do_trie_utf8_fold:                                                            \
        if ( foldlen>0 ) {                                                          \
            uvc = utf8n_to_uvchr( (const U8*) uscan, foldlen, &len, uniflags );     \
            foldlen -= len;                                                         \
            uscan += len;                                                           \
            len=0;                                                                  \
        } else {                                                                    \
            uvc = _to_utf8_fold_flags( (const U8*) uc, foldbuf, &foldlen, flags);   \
            len = UTF8SKIP(uc);                                                     \
            skiplen = UVCHR_SKIP( uvc );                                            \
            foldlen -= skiplen;                                                     \
            uscan = foldbuf + skiplen;                                              \
        }                                                                           \
        break;                                                                      \
    case trie_flu8_latin:                                                           \
        _CHECK_AND_WARN_PROBLEMATIC_LOCALE;                                         \
        goto do_trie_latin_utf8_fold;                                               \
    case trie_latin_utf8_exactfa_fold:                                              \
        flags |= FOLD_FLAGS_NOMIX_ASCII;                                            \
        /* FALLTHROUGH */                                                           \
    case trie_latin_utf8_fold:                                                      \
      do_trie_latin_utf8_fold:                                                      \
        if ( foldlen>0 ) {                                                          \
            uvc = utf8n_to_uvchr( (const U8*) uscan, foldlen, &len, uniflags );     \
            foldlen -= len;                                                         \
            uscan += len;                                                           \
            len=0;                                                                  \
        } else {                                                                    \
            len = 1;                                                                \
            uvc = _to_fold_latin1( (U8) *uc, foldbuf, &foldlen, flags);             \
            skiplen = UVCHR_SKIP( uvc );                                            \
            foldlen -= skiplen;                                                     \
            uscan = foldbuf + skiplen;                                              \
        }                                                                           \
        break;                                                                      \
    case trie_utf8l:                                                                \
        _CHECK_AND_WARN_PROBLEMATIC_LOCALE;                                         \
        if (utf8_target && UTF8_IS_ABOVE_LATIN1(*uc)) {                             \
            _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(uc, uc + UTF8SKIP(uc));          \
        }                                                                           \
        /* FALLTHROUGH */                                                           \
    case trie_utf8:                                                                 \
        uvc = utf8n_to_uvchr( (const U8*) uc, uc_end - uc, &len, uniflags );        \
        break;                                                                      \
    case trie_plain:                                                                \
        uvc = (UV)*uc;                                                              \
        len = 1;                                                                    \
    }                                                                               \
    if (uvc < 256) {                                                                \
        charid = trie->charmap[ uvc ];                                              \
    }                                                                               \
    else {                                                                          \
        charid = 0;                                                                 \
        if (widecharmap) {                                                          \
            SV** const svpp = hv_fetch(widecharmap,                                 \
                        (char*)&uvc, sizeof(UV), 0);                                \
            if (svpp)                                                               \
                charid = (U16)SvIV(*svpp);                                          \
        }                                                                           \
    }                                                                               \
} STMT_END

#define DUMP_EXEC_POS(li,s,doutf8,depth)                    \
    dump_exec_pos(li,s,(reginfo->strend),(reginfo->strbeg), \
                startpos, doutf8, depth)

#define REXEC_FBC_EXACTISH_SCAN(COND)                     \
STMT_START {                                              \
    while (s <= e) {                                      \
	if ( (COND)                                       \
	     && (ln == 1 || folder(s, pat_string, ln))    \
	     && (reginfo->intuit || regtry(reginfo, &s)) )\
	    goto got_it;                                  \
	s++;                                              \
    }                                                     \
} STMT_END

#define REXEC_FBC_UTF8_SCAN(CODE)                     \
STMT_START {                                          \
    while (s < strend) {                              \
	CODE                                          \
	s += UTF8SKIP(s);                             \
    }                                                 \
} STMT_END

#define REXEC_FBC_SCAN(CODE)                          \
STMT_START {                                          \
    while (s < strend) {                              \
	CODE                                          \
	s++;                                          \
    }                                                 \
} STMT_END

#define REXEC_FBC_UTF8_CLASS_SCAN(COND)                        \
REXEC_FBC_UTF8_SCAN( /* Loops while (s < strend) */            \
    if (COND) {                                                \
	if (tmp && (reginfo->intuit || regtry(reginfo, &s)))   \
	    goto got_it;                                       \
	else                                                   \
	    tmp = doevery;                                     \
    }                                                          \
    else                                                       \
	tmp = 1;                                               \
)

#define REXEC_FBC_CLASS_SCAN(COND)                             \
REXEC_FBC_SCAN( /* Loops while (s < strend) */                 \
    if (COND) {                                                \
	if (tmp && (reginfo->intuit || regtry(reginfo, &s)))   \
	    goto got_it;                                       \
	else                                                   \
	    tmp = doevery;                                     \
    }                                                          \
    else                                                       \
	tmp = 1;                                               \
)

#define REXEC_FBC_CSCAN(CONDUTF8,COND)                         \
    if (utf8_target) {                                         \
	REXEC_FBC_UTF8_CLASS_SCAN(CONDUTF8);                   \
    }                                                          \
    else {                                                     \
	REXEC_FBC_CLASS_SCAN(COND);                            \
    }

/* The three macros below are slightly different versions of the same logic.
 *
 * The first is for /a and /aa when the target string is UTF-8.  This can only
 * match ascii, but it must advance based on UTF-8.   The other two handle the
 * non-UTF-8 and the more generic UTF-8 cases.   In all three, we are looking
 * for the boundary (or non-boundary) between a word and non-word character.
 * The utf8 and non-utf8 cases have the same logic, but the details must be
 * different.  Find the "wordness" of the character just prior to this one, and
 * compare it with the wordness of this one.  If they differ, we have a
 * boundary.  At the beginning of the string, pretend that the previous
 * character was a new-line.
 *
 * All these macros uncleanly have side-effects with each other and outside
 * variables.  So far it's been too much trouble to clean-up
 *
 * TEST_NON_UTF8 is the macro or function to call to test if its byte input is
 *               a word character or not.
 * IF_SUCCESS    is code to do if it finds that we are at a boundary between
 *               word/non-word
 * IF_FAIL       is code to do if we aren't at a boundary between word/non-word
 *
 * Exactly one of the two IF_FOO parameters is a no-op, depending on whether we
 * are looking for a boundary or for a non-boundary.  If we are looking for a
 * boundary, we want IF_FAIL to be the no-op, and for IF_SUCCESS to go out and
 * see if this tentative match actually works, and if so, to quit the loop
 * here.  And vice-versa if we are looking for a non-boundary.
 *
 * 'tmp' below in the next three macros in the REXEC_FBC_SCAN and
 * REXEC_FBC_UTF8_SCAN loops is a loop invariant, a bool giving the return of
 * TEST_NON_UTF8(s-1).  To see this, note that that's what it is defined to be
 * at entry to the loop, and to get to the IF_FAIL branch, tmp must equal
 * TEST_NON_UTF8(s), and in the opposite branch, IF_SUCCESS, tmp is that
 * complement.  But in that branch we complement tmp, meaning that at the
 * bottom of the loop tmp is always going to be equal to TEST_NON_UTF8(s),
 * which means at the top of the loop in the next iteration, it is
 * TEST_NON_UTF8(s-1) */
#define FBC_UTF8_A(TEST_NON_UTF8, IF_SUCCESS, IF_FAIL)                         \
    tmp = (s != reginfo->strbeg) ? UCHARAT(s - 1) : '\n';                      \
    tmp = TEST_NON_UTF8(tmp);                                                  \
    REXEC_FBC_UTF8_SCAN( /* advances s while s < strend */                     \
        if (tmp == ! TEST_NON_UTF8((U8) *s)) {                                 \
            tmp = !tmp;                                                        \
            IF_SUCCESS; /* Is a boundary if values for s-1 and s differ */     \
        }                                                                      \
        else {                                                                 \
            IF_FAIL;                                                           \
        }                                                                      \
    );                                                                         \

/* Like FBC_UTF8_A, but TEST_UV is a macro which takes a UV as its input, and
 * TEST_UTF8 is a macro that for the same input code points returns identically
 * to TEST_UV, but takes a pointer to a UTF-8 encoded string instead */
#define FBC_UTF8(TEST_UV, TEST_UTF8, IF_SUCCESS, IF_FAIL)                      \
    if (s == reginfo->strbeg) {                                                \
        tmp = '\n';                                                            \
    }                                                                          \
    else { /* Back-up to the start of the previous character */                \
        U8 * const r = reghop3((U8*)s, -1, (U8*)reginfo->strbeg);              \
        tmp = utf8n_to_uvchr(r, (U8*) reginfo->strend - r,                     \
                                                       0, UTF8_ALLOW_DEFAULT); \
    }                                                                          \
    tmp = TEST_UV(tmp);                                                        \
    LOAD_UTF8_CHARCLASS_ALNUM();                                               \
    REXEC_FBC_UTF8_SCAN( /* advances s while s < strend */                     \
        if (tmp == ! (TEST_UTF8((U8 *) s))) {                                  \
            tmp = !tmp;                                                        \
            IF_SUCCESS;                                                        \
        }                                                                      \
        else {                                                                 \
            IF_FAIL;                                                           \
        }                                                                      \
    );

/* Like the above two macros.  UTF8_CODE is the complete code for handling
 * UTF-8.  Common to the BOUND and NBOUND cases, set-up by the FBC_BOUND, etc
 * macros below */
#define FBC_BOUND_COMMON(UTF8_CODE, TEST_NON_UTF8, IF_SUCCESS, IF_FAIL)        \
    if (utf8_target) {                                                         \
        UTF8_CODE                                                              \
    }                                                                          \
    else {  /* Not utf8 */                                                     \
	tmp = (s != reginfo->strbeg) ? UCHARAT(s - 1) : '\n';                  \
	tmp = TEST_NON_UTF8(tmp);                                              \
	REXEC_FBC_SCAN( /* advances s while s < strend */                      \
	    if (tmp == ! TEST_NON_UTF8((U8) *s)) {                             \
		IF_SUCCESS;                                                    \
		tmp = !tmp;                                                    \
	    }                                                                  \
	    else {                                                             \
		IF_FAIL;                                                       \
	    }                                                                  \
	);                                                                     \
    }                                                                          \
    /* Here, things have been set up by the previous code so that tmp is the   \
     * return of TEST_NON_UTF(s-1) or TEST_UTF8(s-1) (depending on the         \
     * utf8ness of the target).  We also have to check if this matches against \
     * the EOS, which we treat as a \n (which is the same value in both UTF-8  \
     * or non-UTF8, so can use the non-utf8 test condition even for a UTF-8    \
     * string */                                                               \
    if (tmp == ! TEST_NON_UTF8('\n')) {                                        \
        IF_SUCCESS;                                                            \
    }                                                                          \
    else {                                                                     \
        IF_FAIL;                                                               \
    }

/* This is the macro to use when we want to see if something that looks like it
 * could match, actually does, and if so exits the loop */
#define REXEC_FBC_TRYIT                            \
    if ((reginfo->intuit || regtry(reginfo, &s)))  \
        goto got_it

/* The only difference between the BOUND and NBOUND cases is that
 * REXEC_FBC_TRYIT is called when matched in BOUND, and when non-matched in
 * NBOUND.  This is accomplished by passing it as either the if or else clause,
 * with the other one being empty (PLACEHOLDER is defined as empty).
 *
 * The TEST_FOO parameters are for operating on different forms of input, but
 * all should be ones that return identically for the same underlying code
 * points */
#define FBC_BOUND(TEST_NON_UTF8, TEST_UV, TEST_UTF8)                           \
    FBC_BOUND_COMMON(                                                          \
          FBC_UTF8(TEST_UV, TEST_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER),          \
          TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER)

#define FBC_BOUND_A(TEST_NON_UTF8)                                             \
    FBC_BOUND_COMMON(                                                          \
            FBC_UTF8_A(TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER),           \
            TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER)

#define FBC_NBOUND(TEST_NON_UTF8, TEST_UV, TEST_UTF8)                          \
    FBC_BOUND_COMMON(                                                          \
          FBC_UTF8(TEST_UV, TEST_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT),          \
          TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT)

#define FBC_NBOUND_A(TEST_NON_UTF8)                                            \
    FBC_BOUND_COMMON(                                                          \
            FBC_UTF8_A(TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT),           \
            TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT)

#ifdef DEBUGGING
static IV
S_get_break_val_cp_checked(SV* const invlist, const UV cp_in) {
  IV cp_out = Perl__invlist_search(invlist, cp_in);
  assert(cp_out >= 0);
  return cp_out;
}
#  define _generic_GET_BREAK_VAL_CP_CHECKED(invlist, invmap, cp) \
	invmap[S_get_break_val_cp_checked(invlist, cp)]
#else
#  define _generic_GET_BREAK_VAL_CP_CHECKED(invlist, invmap, cp) \
	invmap[_invlist_search(invlist, cp)]
#endif

/* Takes a pointer to an inversion list, a pointer to its corresponding
 * inversion map, and a code point, and returns the code point's value
 * according to the two arrays.  It assumes that all code points have a value.
 * This is used as the base macro for macros for particular properties */
#define _generic_GET_BREAK_VAL_CP(invlist, invmap, cp)              \
	_generic_GET_BREAK_VAL_CP_CHECKED(invlist, invmap, cp)

/* Same as above, but takes begin, end ptrs to a UTF-8 encoded string instead
 * of a code point, returning the value for the first code point in the string.
 * And it takes the particular macro name that finds the desired value given a
 * code point.  Merely convert the UTF-8 to code point and call the cp macro */
#define _generic_GET_BREAK_VAL_UTF8(cp_macro, pos, strend)                     \
             (__ASSERT_(pos < strend)                                          \
                 /* Note assumes is valid UTF-8 */                             \
             (cp_macro(utf8_to_uvchr_buf((pos), (strend), NULL))))

/* Returns the GCB value for the input code point */
#define getGCB_VAL_CP(cp)                                                      \
          _generic_GET_BREAK_VAL_CP(                                           \
                                    PL_GCB_invlist,                            \
                                    _Perl_GCB_invmap,                          \
                                    (cp))

/* Returns the GCB value for the first code point in the UTF-8 encoded string
 * bounded by pos and strend */
#define getGCB_VAL_UTF8(pos, strend)                                           \
    _generic_GET_BREAK_VAL_UTF8(getGCB_VAL_CP, pos, strend)

/* Returns the LB value for the input code point */
#define getLB_VAL_CP(cp)                                                       \
          _generic_GET_BREAK_VAL_CP(                                           \
                                    PL_LB_invlist,                             \
                                    _Perl_LB_invmap,                           \
                                    (cp))

/* Returns the LB value for the first code point in the UTF-8 encoded string
 * bounded by pos and strend */
#define getLB_VAL_UTF8(pos, strend)                                            \
    _generic_GET_BREAK_VAL_UTF8(getLB_VAL_CP, pos, strend)


/* Returns the SB value for the input code point */
#define getSB_VAL_CP(cp)                                                       \
          _generic_GET_BREAK_VAL_CP(                                           \
                                    PL_SB_invlist,                             \
                                    _Perl_SB_invmap,                     \
                                    (cp))

/* Returns the SB value for the first code point in the UTF-8 encoded string
 * bounded by pos and strend */
#define getSB_VAL_UTF8(pos, strend)                                            \
    _generic_GET_BREAK_VAL_UTF8(getSB_VAL_CP, pos, strend)

/* Returns the WB value for the input code point */
#define getWB_VAL_CP(cp)                                                       \
          _generic_GET_BREAK_VAL_CP(                                           \
                                    PL_WB_invlist,                             \
                                    _Perl_WB_invmap,                         \
                                    (cp))

/* Returns the WB value for the first code point in the UTF-8 encoded string
 * bounded by pos and strend */
#define getWB_VAL_UTF8(pos, strend)                                            \
    _generic_GET_BREAK_VAL_UTF8(getWB_VAL_CP, pos, strend)

/* We know what class REx starts with.  Try to find this position... */
/* if reginfo->intuit, its a dryrun */
/* annoyingly all the vars in this routine have different names from their counterparts
   in regmatch. /grrr */
STATIC char *
S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s, 
    const char *strend, regmatch_info *reginfo)
{
    dVAR;
    const I32 doevery = (prog->intflags & PREGf_SKIP) == 0;
    char *pat_string;   /* The pattern's exactish string */
    char *pat_end;	    /* ptr to end char of pat_string */
    re_fold_t folder;	/* Function for computing non-utf8 folds */
    const U8 *fold_array;   /* array for folding ords < 256 */
    STRLEN ln;
    STRLEN lnc;
    U8 c1;
    U8 c2;
    char *e;
    I32 tmp = 1;	/* Scratch variable? */
    const bool utf8_target = reginfo->is_utf8_target;
    UV utf8_fold_flags = 0;
    const bool is_utf8_pat = reginfo->is_utf8_pat;
    bool to_complement = FALSE; /* Invert the result?  Taking the xor of this
                                   with a result inverts that result, as 0^1 =
                                   1 and 1^1 = 0 */
    _char_class_number classnum;

    RXi_GET_DECL(prog,progi);

    PERL_ARGS_ASSERT_FIND_BYCLASS;

    /* We know what class it must start with. */
    switch (OP(c)) {
    case ANYOFL:
        _CHECK_AND_WARN_PROBLEMATIC_LOCALE;

        if (ANYOFL_UTF8_LOCALE_REQD(FLAGS(c)) && ! IN_UTF8_CTYPE_LOCALE) {
            Perl_ck_warner(aTHX_ packWARN(WARN_LOCALE), utf8_locale_required);
        }

        /* FALLTHROUGH */
    case ANYOFD:
    case ANYOF:
        if (utf8_target) {
            REXEC_FBC_UTF8_CLASS_SCAN(
                      reginclass(prog, c, (U8*)s, (U8*) strend, utf8_target));
        }
        else {
            REXEC_FBC_CLASS_SCAN(REGINCLASS(prog, c, (U8*)s, 0));
        }
        break;

    case EXACTFA_NO_TRIE:   /* This node only generated for non-utf8 patterns */
        assert(! is_utf8_pat);
	/* FALLTHROUGH */
    case EXACTFA:
        if (is_utf8_pat || utf8_target) {
            utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII;
            goto do_exactf_utf8;
        }
        fold_array = PL_fold_latin1;    /* Latin1 folds are not affected by */
        folder = foldEQ_latin1;	        /* /a, except the sharp s one which */
        goto do_exactf_non_utf8;	/* isn't dealt with by these */

    case EXACTF:   /* This node only generated for non-utf8 patterns */
        assert(! is_utf8_pat);
        if (utf8_target) {
            utf8_fold_flags = 0;
            goto do_exactf_utf8;
        }
        fold_array = PL_fold;
        folder = foldEQ;
        goto do_exactf_non_utf8;

    case EXACTFL:
        _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
        if (is_utf8_pat || utf8_target || IN_UTF8_CTYPE_LOCALE) {
            utf8_fold_flags = FOLDEQ_LOCALE;
            goto do_exactf_utf8;
        }
        fold_array = PL_fold_locale;
        folder = foldEQ_locale;
        goto do_exactf_non_utf8;

    case EXACTFU_SS:
        if (is_utf8_pat) {
            utf8_fold_flags = FOLDEQ_S2_ALREADY_FOLDED;
        }
        goto do_exactf_utf8;

    case EXACTFLU8:
            if (! utf8_target) {    /* All code points in this node require
                                       UTF-8 to express.  */
                break;
            }
            utf8_fold_flags =  FOLDEQ_LOCALE | FOLDEQ_S2_ALREADY_FOLDED
                                             | FOLDEQ_S2_FOLDS_SANE;
            goto do_exactf_utf8;

    case EXACTFU:
        if (is_utf8_pat || utf8_target) {
            utf8_fold_flags = is_utf8_pat ? FOLDEQ_S2_ALREADY_FOLDED : 0;
            goto do_exactf_utf8;
        }

        /* Any 'ss' in the pattern should have been replaced by regcomp,
         * so we don't have to worry here about this single special case
         * in the Latin1 range */
        fold_array = PL_fold_latin1;
        folder = foldEQ_latin1;

        /* FALLTHROUGH */

      do_exactf_non_utf8: /* Neither pattern nor string are UTF8, and there
                           are no glitches with fold-length differences
                           between the target string and pattern */

        /* The idea in the non-utf8 EXACTF* cases is to first find the
         * first character of the EXACTF* node and then, if necessary,
         * case-insensitively compare the full text of the node.  c1 is the
         * first character.  c2 is its fold.  This logic will not work for
         * Unicode semantics and the german sharp ss, which hence should
         * not be compiled into a node that gets here. */
        pat_string = STRING(c);
        ln  = STR_LEN(c);	/* length to match in octets/bytes */

        /* We know that we have to match at least 'ln' bytes (which is the
         * same as characters, since not utf8).  If we have to match 3
         * characters, and there are only 2 availabe, we know without
         * trying that it will fail; so don't start a match past the
         * required minimum number from the far end */
        e = HOP3c(strend, -((SSize_t)ln), s);

        if (reginfo->intuit && e < s) {
            e = s;			/* Due to minlen logic of intuit() */
        }

        c1 = *pat_string;
        c2 = fold_array[c1];
        if (c1 == c2) { /* If char and fold are the same */
            REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1);
        }
        else {
            REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1 || *(U8*)s == c2);
        }
        break;

      do_exactf_utf8:
      {
        unsigned expansion;

        /* If one of the operands is in utf8, we can't use the simpler folding
         * above, due to the fact that many different characters can have the
         * same fold, or portion of a fold, or different- length fold */
        pat_string = STRING(c);
        ln  = STR_LEN(c);	/* length to match in octets/bytes */
        pat_end = pat_string + ln;
        lnc = is_utf8_pat       /* length to match in characters */
                ? utf8_length((U8 *) pat_string, (U8 *) pat_end)
                : ln;

        /* We have 'lnc' characters to match in the pattern, but because of
         * multi-character folding, each character in the target can match
         * up to 3 characters (Unicode guarantees it will never exceed
         * this) if it is utf8-encoded; and up to 2 if not (based on the
         * fact that the Latin 1 folds are already determined, and the
         * only multi-char fold in that range is the sharp-s folding to
         * 'ss'.  Thus, a pattern character can match as little as 1/3 of a
         * string character.  Adjust lnc accordingly, rounding up, so that
         * if we need to match at least 4+1/3 chars, that really is 5. */
        expansion = (utf8_target) ? UTF8_MAX_FOLD_CHAR_EXPAND : 2;
        lnc = (lnc + expansion - 1) / expansion;

        /* As in the non-UTF8 case, if we have to match 3 characters, and
         * only 2 are left, it's guaranteed to fail, so don't start a
         * match that would require us to go beyond the end of the string
         */
        e = HOP3c(strend, -((SSize_t)lnc), s);

        if (reginfo->intuit && e < s) {
            e = s;			/* Due to minlen logic of intuit() */
        }

        /* XXX Note that we could recalculate e to stop the loop earlier,
         * as the worst case expansion above will rarely be met, and as we
         * go along we would usually find that e moves further to the left.
         * This would happen only after we reached the point in the loop
         * where if there were no expansion we should fail.  Unclear if
         * worth the expense */

        while (s <= e) {
            char *my_strend= (char *)strend;
            if (foldEQ_utf8_flags(s, &my_strend, 0,  utf8_target,
                  pat_string, NULL, ln, is_utf8_pat, utf8_fold_flags)
                && (reginfo->intuit || regtry(reginfo, &s)) )
            {
                goto got_it;
            }
            s += (utf8_target) ? UTF8SKIP(s) : 1;
        }
        break;
    }

    case BOUNDL:
        _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
        if (FLAGS(c) != TRADITIONAL_BOUND) {
            if (! IN_UTF8_CTYPE_LOCALE) {
                Perl_ck_warner(aTHX_ packWARN(WARN_LOCALE),
                                                B_ON_NON_UTF8_LOCALE_IS_WRONG);
            }
            goto do_boundu;
        }

        FBC_BOUND(isWORDCHAR_LC, isWORDCHAR_LC_uvchr, isWORDCHAR_LC_utf8);
        break;

    case NBOUNDL:
        _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
        if (FLAGS(c) != TRADITIONAL_BOUND) {
            if (! IN_UTF8_CTYPE_LOCALE) {
                Perl_ck_warner(aTHX_ packWARN(WARN_LOCALE),
                                                B_ON_NON_UTF8_LOCALE_IS_WRONG);
            }
            goto do_nboundu;
        }

        FBC_NBOUND(isWORDCHAR_LC, isWORDCHAR_LC_uvchr, isWORDCHAR_LC_utf8);
        break;

    case BOUND: /* regcomp.c makes sure that this only has the traditional \b
                   meaning */
        assert(FLAGS(c) == TRADITIONAL_BOUND);

        FBC_BOUND(isWORDCHAR, isWORDCHAR_uni, isWORDCHAR_utf8);
        break;

    case BOUNDA: /* regcomp.c makes sure that this only has the traditional \b
                   meaning */
        assert(FLAGS(c) == TRADITIONAL_BOUND);

        FBC_BOUND_A(isWORDCHAR_A);
        break;

    case NBOUND: /* regcomp.c makes sure that this only has the traditional \b
                   meaning */
        assert(FLAGS(c) == TRADITIONAL_BOUND);

        FBC_NBOUND(isWORDCHAR, isWORDCHAR_uni, isWORDCHAR_utf8);
        break;

    case NBOUNDA: /* regcomp.c makes sure that this only has the traditional \b
                   meaning */
        assert(FLAGS(c) == TRADITIONAL_BOUND);

        FBC_NBOUND_A(isWORDCHAR_A);
        break;

    case NBOUNDU:
        if ((bound_type) FLAGS(c) == TRADITIONAL_BOUND) {
            FBC_NBOUND(isWORDCHAR_L1, isWORDCHAR_uni, isWORDCHAR_utf8);
            break;
        }

      do_nboundu:

        to_complement = 1;
        /* FALLTHROUGH */

    case BOUNDU:
      do_boundu:
        switch((bound_type) FLAGS(c)) {
            case TRADITIONAL_BOUND:
                FBC_BOUND(isWORDCHAR_L1, isWORDCHAR_uni, isWORDCHAR_utf8);
                break;
            case GCB_BOUND:
                if (s == reginfo->strbeg) {
                    if (reginfo->intuit || regtry(reginfo, &s))
                    {
                        goto got_it;
                    }

                    /* Didn't match.  Try at the next position (if there is one) */
                    s += (utf8_target) ? UTF8SKIP(s) : 1;
                    if (UNLIKELY(s >= reginfo->strend)) {
                        break;
                    }
                }

                if (utf8_target) {
                    GCB_enum before = getGCB_VAL_UTF8(
                                               reghop3((U8*)s, -1,
                                                       (U8*)(reginfo->strbeg)),
                                               (U8*) reginfo->strend);
                    while (s < strend) {
                        GCB_enum after = getGCB_VAL_UTF8((U8*) s,
                                                        (U8*) reginfo->strend);
                        if (   (to_complement ^ isGCB(before, after))
                            && (reginfo->intuit || regtry(reginfo, &s)))
                        {
                            goto got_it;
                        }
                        before = after;
                        s += UTF8SKIP(s);
                    }
                }
                else {  /* Not utf8.  Everything is a GCB except between CR and
                           LF */
                    while (s < strend) {
                        if ((to_complement ^ (   UCHARAT(s - 1) != '\r'
                                              || UCHARAT(s) != '\n'))
                            && (reginfo->intuit || regtry(reginfo, &s)))
                        {
                            goto got_it;
                        }
                        s++;
                    }
                }

                /* And, since this is a bound, it can match after the final
                 * character in the string */
                if ((reginfo->intuit || regtry(reginfo, &s))) {
                    goto got_it;
                }
                break;

            case LB_BOUND:
                if (s == reginfo->strbeg) {
                    if (reginfo->intuit || regtry(reginfo, &s)) {
                        goto got_it;
                    }
                    s += (utf8_target) ? UTF8SKIP(s) : 1;
                    if (UNLIKELY(s >= reginfo->strend)) {
                        break;
                    }
                }

                if (utf8_target) {
                    LB_enum before = getLB_VAL_UTF8(reghop3((U8*)s,
                                                               -1,
                                                               (U8*)(reginfo->strbeg)),
                                                       (U8*) reginfo->strend);
                    while (s < strend) {
                        LB_enum after = getLB_VAL_UTF8((U8*) s, (U8*) reginfo->strend);
                        if (to_complement ^ isLB(before,
                                                 after,
                                                 (U8*) reginfo->strbeg,
                                                 (U8*) s,
                                                 (U8*) reginfo->strend,
                                                 utf8_target)
                            && (reginfo->intuit || regtry(reginfo, &s)))
                        {
                            goto got_it;
                        }
                        before = after;
                        s += UTF8SKIP(s);
                    }
                }
                else {  /* Not utf8. */
                    LB_enum before = getLB_VAL_CP((U8) *(s -1));
                    while (s < strend) {
                        LB_enum after = getLB_VAL_CP((U8) *s);
                        if (to_complement ^ isLB(before,
                                                 after,
                                                 (U8*) reginfo->strbeg,
                                                 (U8*) s,
                                                 (U8*) reginfo->strend,
                                                 utf8_target)
                            && (reginfo->intuit || regtry(reginfo, &s)))
                        {
                            goto got_it;
                        }
                        before = after;
                        s++;
                    }
                }

                if (reginfo->intuit || regtry(reginfo, &s)) {
                    goto got_it;
                }

                break;

            case SB_BOUND:
                if (s == reginfo->strbeg) {
                    if (reginfo->intuit || regtry(reginfo, &s)) {
                        goto got_it;
                    }
                    s += (utf8_target) ? UTF8SKIP(s) : 1;
                    if (UNLIKELY(s >= reginfo->strend)) {
                        break;
                    }
                }

                if (utf8_target) {
                    SB_enum before = getSB_VAL_UTF8(reghop3((U8*)s,
                                                        -1,
                                                        (U8*)(reginfo->strbeg)),
                                                      (U8*) reginfo->strend);
                    while (s < strend) {
                        SB_enum after = getSB_VAL_UTF8((U8*) s,
                                                         (U8*) reginfo->strend);
                        if ((to_complement ^ isSB(before,
                                                  after,
                                                  (U8*) reginfo->strbeg,
                                                  (U8*) s,
                                                  (U8*) reginfo->strend,
                                                  utf8_target))
                            && (reginfo->intuit || regtry(reginfo, &s)))
                        {
                            goto got_it;
                        }
                        before = after;
                        s += UTF8SKIP(s);
                    }
                }
                else {  /* Not utf8. */
                    SB_enum before = getSB_VAL_CP((U8) *(s -1));
                    while (s < strend) {
                        SB_enum after = getSB_VAL_CP((U8) *s);
                        if ((to_complement ^ isSB(before,
                                                  after,
                                                  (U8*) reginfo->strbeg,
                                                  (U8*) s,
                                                  (U8*) reginfo->strend,
                                                  utf8_target))
                            && (reginfo->intuit || regtry(reginfo, &s)))
                        {
                            goto got_it;
                        }
                        before = after;
                        s++;
                    }
                }

                /* Here are at the final position in the target string.  The SB
                 * value is always true here, so matches, depending on other
                 * constraints */
                if (reginfo->intuit || regtry(reginfo, &s)) {
                    goto got_it;
                }

                break;

            case WB_BOUND:
                if (s == reginfo->strbeg) {
                    if (reginfo->intuit || regtry(reginfo, &s)) {
                        goto got_it;
                    }
                    s += (utf8_target) ? UTF8SKIP(s) : 1;
                    if (UNLIKELY(s >= reginfo->strend)) {
                        break;
                    }
                }

                if (utf8_target) {
                    /* We are at a boundary between char_sub_0 and char_sub_1.
                     * We also keep track of the value for char_sub_-1 as we
                     * loop through the line.   Context may be needed to make a
                     * determination, and if so, this can save having to
                     * recalculate it */
                    WB_enum previous = WB_UNKNOWN;
                    WB_enum before = getWB_VAL_UTF8(
                                              reghop3((U8*)s,
                                                      -1,
                                                      (U8*)(reginfo->strbeg)),
                                              (U8*) reginfo->strend);
                    while (s < strend) {
                        WB_enum after = getWB_VAL_UTF8((U8*) s,
                                                        (U8*) reginfo->strend);
                        if ((to_complement ^ isWB(previous,
                                                  before,
                                                  after,
                                                  (U8*) reginfo->strbeg,
                                                  (U8*) s,
                                                  (U8*) reginfo->strend,
                                                  utf8_target))
                            && (reginfo->intuit || regtry(reginfo, &s)))
                        {
                            goto got_it;
                        }
                        previous = before;
                        before = after;
                        s += UTF8SKIP(s);
                    }
                }
                else {  /* Not utf8. */
                    WB_enum previous = WB_UNKNOWN;
                    WB_enum before = getWB_VAL_CP((U8) *(s -1));
                    while (s < strend) {
                        WB_enum after = getWB_VAL_CP((U8) *s);
                        if ((to_complement ^ isWB(previous,
                                                  before,
                                                  after,
                                                  (U8*) reginfo->strbeg,
                                                  (U8*) s,
                                                  (U8*) reginfo->strend,
                                                  utf8_target))
                            && (reginfo->intuit || regtry(reginfo, &s)))
                        {
                            goto got_it;
                        }
                        previous = before;
                        before = after;
                        s++;
                    }
                }

                if (reginfo->intuit || regtry(reginfo, &s)) {
                    goto got_it;
                }
        }
        break;

    case LNBREAK:
        REXEC_FBC_CSCAN(is_LNBREAK_utf8_safe(s, strend),
                        is_LNBREAK_latin1_safe(s, strend)
        );
        break;

    /* The argument to all the POSIX node types is the class number to pass to
     * _generic_isCC() to build a mask for searching in PL_charclass[] */

    case NPOSIXL:
        to_complement = 1;
        /* FALLTHROUGH */

    case POSIXL:
        _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
        REXEC_FBC_CSCAN(to_complement ^ cBOOL(isFOO_utf8_lc(FLAGS(c), (U8 *) s)),
                        to_complement ^ cBOOL(isFOO_lc(FLAGS(c), *s)));
        break;

    case NPOSIXD:
        to_complement = 1;
        /* FALLTHROUGH */

    case POSIXD:
        if (utf8_target) {
            goto posix_utf8;
        }
        goto posixa;

    case NPOSIXA:
        if (utf8_target) {
            /* The complement of something that matches only ASCII matches all
             * non-ASCII, plus everything in ASCII that isn't in the class. */
            REXEC_FBC_UTF8_CLASS_SCAN(! isASCII_utf8(s)
                                      || ! _generic_isCC_A(*s, FLAGS(c)));
            break;
        }

        to_complement = 1;
        /* FALLTHROUGH */

    case POSIXA:
      posixa:
        /* Don't need to worry about utf8, as it can match only a single
         * byte invariant character. */
        REXEC_FBC_CLASS_SCAN(
                        to_complement ^ cBOOL(_generic_isCC_A(*s, FLAGS(c))));
        break;

    case NPOSIXU:
        to_complement = 1;
        /* FALLTHROUGH */

    case POSIXU:
        if (! utf8_target) {
            REXEC_FBC_CLASS_SCAN(to_complement ^ cBOOL(_generic_isCC(*s,
                                                                    FLAGS(c))));
        }
        else {

          posix_utf8:
            classnum = (_char_class_number) FLAGS(c);
            if (classnum < _FIRST_NON_SWASH_CC) {
                while (s < strend) {

                    /* We avoid loading in the swash as long as possible, but
                     * should we have to, we jump to a separate loop.  This
                     * extra 'if' statement is what keeps this code from being
                     * just a call to REXEC_FBC_UTF8_CLASS_SCAN() */
                    if (UTF8_IS_ABOVE_LATIN1(*s)) {
                        goto found_above_latin1;
                    }
                    if ((UTF8_IS_INVARIANT(*s)
                         && to_complement ^ cBOOL(_generic_isCC((U8) *s,
                                                                classnum)))
                        || (UTF8_IS_DOWNGRADEABLE_START(*s)
                            && to_complement ^ cBOOL(
                                _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*s,
                                                                      *(s + 1)),
                                              classnum))))
                    {
                        if (tmp && (reginfo->intuit || regtry(reginfo, &s)))
                            goto got_it;
                        else {
                            tmp = doevery;
                        }
                    }
                    else {
                        tmp = 1;
                    }
                    s += UTF8SKIP(s);
                }
            }
            else switch (classnum) {    /* These classes are implemented as
                                           macros */
                case _CC_ENUM_SPACE:
                    REXEC_FBC_UTF8_CLASS_SCAN(
                                        to_complement ^ cBOOL(isSPACE_utf8(s)));
                    break;

                case _CC_ENUM_BLANK:
                    REXEC_FBC_UTF8_CLASS_SCAN(
                                        to_complement ^ cBOOL(isBLANK_utf8(s)));
                    break;

                case _CC_ENUM_XDIGIT:
                    REXEC_FBC_UTF8_CLASS_SCAN(
                                       to_complement ^ cBOOL(isXDIGIT_utf8(s)));
                    break;

                case _CC_ENUM_VERTSPACE:
                    REXEC_FBC_UTF8_CLASS_SCAN(
                                       to_complement ^ cBOOL(isVERTWS_utf8(s)));
                    break;

                case _CC_ENUM_CNTRL:
                    REXEC_FBC_UTF8_CLASS_SCAN(
                                        to_complement ^ cBOOL(isCNTRL_utf8(s)));
                    break;

                default:
                    Perl_croak(aTHX_ "panic: find_byclass() node %d='%s' has an unexpected character class '%d'", OP(c), PL_reg_name[OP(c)], classnum);
                    NOT_REACHED; /* NOTREACHED */
            }
        }
        break;

      found_above_latin1:   /* Here we have to load a swash to get the result
                               for the current code point */
        if (! PL_utf8_swash_ptrs[classnum]) {
            U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;
            PL_utf8_swash_ptrs[classnum] =
                    _core_swash_init("utf8",
                                     "",
                                     &PL_sv_undef, 1, 0,
                                     PL_XPosix_ptrs[classnum], &flags);
        }

        /* This is a copy of the loop above for swash classes, though using the
         * FBC macro instead of being expanded out.  Since we've loaded the
         * swash, we don't have to check for that each time through the loop */
        REXEC_FBC_UTF8_CLASS_SCAN(
                to_complement ^ cBOOL(_generic_utf8(
                                      classnum,
                                      s,
                                      swash_fetch(PL_utf8_swash_ptrs[classnum],
                                                  (U8 *) s, TRUE))));
        break;

    case AHOCORASICKC:
    case AHOCORASICK:
        {
            DECL_TRIE_TYPE(c);
            /* what trie are we using right now */
            reg_ac_data *aho = (reg_ac_data*)progi->data->data[ ARG( c ) ];
            reg_trie_data *trie = (reg_trie_data*)progi->data->data[ aho->trie ];
            HV *widecharmap = MUTABLE_HV(progi->data->data[ aho->trie + 1 ]);

            const char *last_start = strend - trie->minlen;
#ifdef DEBUGGING
            const char *real_start = s;
#endif
            STRLEN maxlen = trie->maxlen;
            SV *sv_points;
            U8 **points; /* map of where we were in the input string
                            when reading a given char. For ASCII this
                            is unnecessary overhead as the relationship
                            is always 1:1, but for Unicode, especially
                            case folded Unicode this is not true. */
            U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
            U8 *bitmap=NULL;


            GET_RE_DEBUG_FLAGS_DECL;

            /* We can't just allocate points here. We need to wrap it in
             * an SV so it gets freed properly if there is a croak while
             * running the match */
            ENTER;
            SAVETMPS;
            sv_points=newSV(maxlen * sizeof(U8 *));
            SvCUR_set(sv_points,
                maxlen * sizeof(U8 *));
            SvPOK_on(sv_points);
            sv_2mortal(sv_points);
            points=(U8**)SvPV_nolen(sv_points );
            if ( trie_type != trie_utf8_fold
                 && (trie->bitmap || OP(c)==AHOCORASICKC) )
            {
                if (trie->bitmap)
                    bitmap=(U8*)trie->bitmap;
                else
                    bitmap=(U8*)ANYOF_BITMAP(c);
            }
            /* this is the Aho-Corasick algorithm modified a touch
               to include special handling for long "unknown char" sequences.
               The basic idea being that we use AC as long as we are dealing
               with a possible matching char, when we encounter an unknown char
               (and we have not encountered an accepting state) we scan forward
               until we find a legal starting char.
               AC matching is basically that of trie matching, except that when
               we encounter a failing transition, we fall back to the current
               states "fail state", and try the current char again, a process
               we repeat until we reach the root state, state 1, or a legal
               transition. If we fail on the root state then we can either
               terminate if we have reached an accepting state previously, or
               restart the entire process from the beginning if we have not.

             */
            while (s <= last_start) {
                const U32 uniflags = UTF8_ALLOW_DEFAULT;
                U8 *uc = (U8*)s;
                U16 charid = 0;
                U32 base = 1;
                U32 state = 1;
                UV uvc = 0;
                STRLEN len = 0;
                STRLEN foldlen = 0;
                U8 *uscan = (U8*)NULL;
                U8 *leftmost = NULL;
#ifdef DEBUGGING
                U32 accepted_word= 0;
#endif
                U32 pointpos = 0;

                while ( state && uc <= (U8*)strend ) {
                    int failed=0;
                    U32 word = aho->states[ state ].wordnum;

                    if( state==1 ) {
                        if ( bitmap ) {
                            DEBUG_TRIE_EXECUTE_r(
                                if ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
                                    dump_exec_pos( (char *)uc, c, strend, real_start,
                                        (char *)uc, utf8_target, 0 );
                                    Perl_re_printf( aTHX_
                                        " Scanning for legal start char...\n");
                                }
                            );
                            if (utf8_target) {
                                while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
                                    uc += UTF8SKIP(uc);
                                }
                            } else {
                                while ( uc <= (U8*)last_start  && !BITMAP_TEST(bitmap,*uc) ) {
                                    uc++;
                                }
                            }
                            s= (char *)uc;
                        }
                        if (uc >(U8*)last_start) break;
                    }

                    if ( word ) {
                        U8 *lpos= points[ (pointpos - trie->wordinfo[word].len) % maxlen ];
                        if (!leftmost || lpos < leftmost) {
                            DEBUG_r(accepted_word=word);
                            leftmost= lpos;
                        }
                        if (base==0) break;

                    }
                    points[pointpos++ % maxlen]= uc;
                    if (foldlen || uc < (U8*)strend) {
                        REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc,
                                             (U8 *) strend, uscan, len, uvc,
                                             charid, foldlen, foldbuf,
                                             uniflags);
                        DEBUG_TRIE_EXECUTE_r({
                            dump_exec_pos( (char *)uc, c, strend,
                                        real_start, s, utf8_target, 0);
                            Perl_re_printf( aTHX_
                                " Charid:%3u CP:%4"UVxf" ",
                                 charid, uvc);
                        });
                    }
                    else {
                        len = 0;
                        charid = 0;
                    }


                    do {
#ifdef DEBUGGING
                        word = aho->states[ state ].wordnum;
#endif
                        base = aho->states[ state ].trans.base;

                        DEBUG_TRIE_EXECUTE_r({
                            if (failed)
                                dump_exec_pos( (char *)uc, c, strend, real_start,
                                    s,   utf8_target, 0 );
                            Perl_re_printf( aTHX_
                                "%sState: %4"UVxf", word=%"UVxf,
                                failed ? " Fail transition to " : "",
                                (UV)state, (UV)word);
                        });
                        if ( base ) {
                            U32 tmp;
                            I32 offset;
                            if (charid &&
                                 ( ((offset = base + charid
                                    - 1 - trie->uniquecharcount)) >= 0)
                                 && ((U32)offset < trie->lasttrans)
                                 && trie->trans[offset].check == state
                                 && (tmp=trie->trans[offset].next))
                            {
                                DEBUG_TRIE_EXECUTE_r(
                                    Perl_re_printf( aTHX_ " - legal\n"));
                                state = tmp;
                                break;
                            }
                            else {
                                DEBUG_TRIE_EXECUTE_r(
                                    Perl_re_printf( aTHX_ " - fail\n"));
                                failed = 1;
                                state = aho->fail[state];
                            }
                        }
                        else {
                            /* we must be accepting here */
                            DEBUG_TRIE_EXECUTE_r(
                                    Perl_re_printf( aTHX_ " - accepting\n"));
                            failed = 1;
                            break;
                        }
                    } while(state);
                    uc += len;
                    if (failed) {
                        if (leftmost)
                            break;
                        if (!state) state = 1;
                    }
                }
                if ( aho->states[ state ].wordnum ) {
                    U8 *lpos = points[ (pointpos - trie->wordinfo[aho->states[ state ].wordnum].len) % maxlen ];
                    if (!leftmost || lpos < leftmost) {
                        DEBUG_r(accepted_word=aho->states[ state ].wordnum);
                        leftmost = lpos;
                    }
                }
                if (leftmost) {
                    s = (char*)leftmost;
                    DEBUG_TRIE_EXECUTE_r({
                        Perl_re_printf( aTHX_  "Matches word #%"UVxf" at position %"IVdf". Trying full pattern...\n",
                            (UV)accepted_word, (IV)(s - real_start)
                        );
                    });
                    if (reginfo->intuit || regtry(reginfo, &s)) {
                        FREETMPS;
                        LEAVE;
                        goto got_it;
                    }
                    s = HOPc(s,1);
                    DEBUG_TRIE_EXECUTE_r({
                        Perl_re_printf( aTHX_ "Pattern failed. Looking for new start point...\n");
                    });
                } else {
                    DEBUG_TRIE_EXECUTE_r(
                        Perl_re_printf( aTHX_ "No match.\n"));
                    break;
                }
            }
            FREETMPS;
            LEAVE;
        }
        break;
    default:
        Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c));
    }
    return 0;
  got_it:
    return s;
}

/* set RX_SAVED_COPY, RX_SUBBEG etc.
 * flags have same meanings as with regexec_flags() */

static void
S_reg_set_capture_string(pTHX_ REGEXP * const rx,
                            char *strbeg,
                            char *strend,
                            SV *sv,
                            U32 flags,
                            bool utf8_target)
{
    struct regexp *const prog = ReANY(rx);

    if (flags & REXEC_COPY_STR) {
#ifdef PERL_ANY_COW
        if (SvCANCOW(sv)) {
            DEBUG_C(Perl_re_printf( aTHX_
                              "Copy on write: regexp capture, type %d\n",
                                    (int) SvTYPE(sv)));
            /* Create a new COW SV to share the match string and store
             * in saved_copy, unless the current COW SV in saved_copy
             * is valid and suitable for our purpose */
            if ((   prog->saved_copy
                 && SvIsCOW(prog->saved_copy)
                 && SvPOKp(prog->saved_copy)
                 && SvIsCOW(sv)
                 && SvPOKp(sv)
                 && SvPVX(sv) == SvPVX(prog->saved_copy)))
            {
                /* just reuse saved_copy SV */
                if (RXp_MATCH_COPIED(prog)) {
                    Safefree(prog->subbeg);
                    RXp_MATCH_COPIED_off(prog);
                }
            }
            else {
                /* create new COW SV to share string */
                RX_MATCH_COPY_FREE(rx);
                prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv);
            }
            prog->subbeg = (char *)SvPVX_const(prog->saved_copy);
            assert (SvPOKp(prog->saved_copy));
            prog->sublen  = strend - strbeg;
            prog->suboffset = 0;
            prog->subcoffset = 0;
        } else
#endif
        {
            SSize_t min = 0;
            SSize_t max = strend - strbeg;
            SSize_t sublen;

            if (    (flags & REXEC_COPY_SKIP_POST)
                && !(prog->extflags & RXf_PMf_KEEPCOPY) /* //p */
                && !(PL_sawampersand & SAWAMPERSAND_RIGHT)
            ) { /* don't copy $' part of string */
                U32 n = 0;
                max = -1;
                /* calculate the right-most part of the string covered
                 * by a capture. Due to lookahead, this may be to
                 * the right of $&, so we have to scan all captures */
                while (n <= prog->lastparen) {
                    if (prog->offs[n].end > max)
                        max = prog->offs[n].end;
                    n++;
                }
                if (max == -1)
                    max = (PL_sawampersand & SAWAMPERSAND_LEFT)
                            ? prog->offs[0].start
                            : 0;
                assert(max >= 0 && max <= strend - strbeg);
            }

            if (    (flags & REXEC_COPY_SKIP_PRE)
                && !(prog->extflags & RXf_PMf_KEEPCOPY) /* //p */
                && !(PL_sawampersand & SAWAMPERSAND_LEFT)
            ) { /* don't copy $` part of string */
                U32 n = 0;
                min = max;
                /* calculate the left-most part of the string covered
                 * by a capture. Due to lookbehind, this may be to
                 * the left of $&, so we have to scan all captures */
                while (min && n <= prog->lastparen) {
                    if (   prog->offs[n].start != -1
                        && prog->offs[n].start < min)
                    {
                        min = prog->offs[n].start;
                    }
                    n++;
                }
                if ((PL_sawampersand & SAWAMPERSAND_RIGHT)
                    && min >  prog->offs[0].end
                )
                    min = prog->offs[0].end;

            }

            assert(min >= 0 && min <= max && min <= strend - strbeg);
            sublen = max - min;

            if (RX_MATCH_COPIED(rx)) {
                if (sublen > prog->sublen)
                    prog->subbeg =
                            (char*)saferealloc(prog->subbeg, sublen+1);
            }
            else
                prog->subbeg = (char*)safemalloc(sublen+1);
            Copy(strbeg + min, prog->subbeg, sublen, char);
            prog->subbeg[sublen] = '\0';
            prog->suboffset = min;
            prog->sublen = sublen;
            RX_MATCH_COPIED_on(rx);
        }
        prog->subcoffset = prog->suboffset;
        if (prog->suboffset && utf8_target) {
            /* Convert byte offset to chars.
             * XXX ideally should only compute this if @-/@+
             * has been seen, a la PL_sawampersand ??? */

            /* If there's a direct correspondence between the
             * string which we're matching and the original SV,
             * then we can use the utf8 len cache associated with
             * the SV. In particular, it means that under //g,
             * sv_pos_b2u() will use the previously cached
             * position to speed up working out the new length of
             * subcoffset, rather than counting from the start of
             * the string each time. This stops
             *   $x = "\x{100}" x 1E6; 1 while $x =~ /(.)/g;
             * from going quadratic */
            if (SvPOKp(sv) && SvPVX(sv) == strbeg)
                prog->subcoffset = sv_pos_b2u_flags(sv, prog->subcoffset,
                                                SV_GMAGIC|SV_CONST_RETURN);
            else
                prog->subcoffset = utf8_length((U8*)strbeg,
                                    (U8*)(strbeg+prog->suboffset));
        }
    }
    else {
        RX_MATCH_COPY_FREE(rx);
        prog->subbeg = strbeg;
        prog->suboffset = 0;
        prog->subcoffset = 0;
        prog->sublen = strend - strbeg;
    }
}




/*
 - regexec_flags - match a regexp against a string
 */
I32
Perl_regexec_flags(pTHX_ REGEXP * const rx, char *stringarg, char *strend,
	      char *strbeg, SSize_t minend, SV *sv, void *data, U32 flags)
/* stringarg: the point in the string at which to begin matching */
/* strend:    pointer to null at end of string */
/* strbeg:    real beginning of string */
/* minend:    end of match must be >= minend bytes after stringarg. */
/* sv:        SV being matched: only used for utf8 flag, pos() etc; string
 *            itself is accessed via the pointers above */
/* data:      May be used for some additional optimizations.
              Currently unused. */
/* flags:     For optimizations. See REXEC_* in regexp.h */

{
    struct regexp *const prog = ReANY(rx);
    char *s;
    regnode *c;
    char *startpos;
    SSize_t minlen;		/* must match at least this many chars */
    SSize_t dontbother = 0;	/* how many characters not to try at end */
    const bool utf8_target = cBOOL(DO_UTF8(sv));
    I32 multiline;
    RXi_GET_DECL(prog,progi);
    regmatch_info reginfo_buf;  /* create some info to pass to regtry etc */
    regmatch_info *const reginfo = &reginfo_buf;
    regexp_paren_pair *swap = NULL;
    I32 oldsave;
    GET_RE_DEBUG_FLAGS_DECL;

    PERL_ARGS_ASSERT_REGEXEC_FLAGS;
    PERL_UNUSED_ARG(data);

    /* Be paranoid... */
    if (prog == NULL) {
	Perl_croak(aTHX_ "NULL regexp parameter");
    }

    DEBUG_EXECUTE_r(
        debug_start_match(rx, utf8_target, stringarg, strend,
        "Matching");
    );

    startpos = stringarg;

    /* set these early as they may be used by the HOP macros below */
    reginfo->strbeg = strbeg;
    reginfo->strend = strend;
    reginfo->is_utf8_target = cBOOL(utf8_target);

    if (prog->intflags & PREGf_GPOS_SEEN) {
        MAGIC *mg;

        /* set reginfo->ganch, the position where \G can match */

        reginfo->ganch =
            (flags & REXEC_IGNOREPOS)
            ? stringarg /* use start pos rather than pos() */
            : ((mg = mg_find_mglob(sv)) && mg->mg_len >= 0)
              /* Defined pos(): */
            ? strbeg + MgBYTEPOS(mg, sv, strbeg, strend-strbeg)
            : strbeg; /* pos() not defined; use start of string */

        DEBUG_GPOS_r(Perl_re_printf( aTHX_
            "GPOS ganch set to strbeg[%"IVdf"]\n", (IV)(reginfo->ganch - strbeg)));

        /* in the presence of \G, we may need to start looking earlier in
         * the string than the suggested start point of stringarg:
         * if prog->gofs is set, then that's a known, fixed minimum
         * offset, such as
         * /..\G/:   gofs = 2
         * /ab|c\G/: gofs = 1
         * or if the minimum offset isn't known, then we have to go back
         * to the start of the string, e.g. /w+\G/
         */

        if (prog->intflags & PREGf_ANCH_GPOS) {
            if (prog->gofs) {
                startpos = HOPBACKc(reginfo->ganch, prog->gofs);
                if (!startpos ||
                    ((flags & REXEC_FAIL_ON_UNDERFLOW) && startpos < stringarg))
                {
                    DEBUG_r(Perl_re_printf( aTHX_
                            "fail: ganch-gofs before earliest possible start\n"));
                    return 0;
                }
            }
            else
                startpos = reginfo->ganch;
        }
        else if (prog->gofs) {
            startpos = HOPBACKc(startpos, prog->gofs);
            if (!startpos)
                startpos = strbeg;
        }
        else if (prog->intflags & PREGf_GPOS_FLOAT)
            startpos = strbeg;
    }

    minlen = prog->minlen;
    if ((startpos + minlen) > strend || startpos < strbeg) {
        DEBUG_r(Perl_re_printf( aTHX_
                    "Regex match can't succeed, so not even tried\n"));
        return 0;
    }

    /* at the end of this function, we'll do a LEAVE_SCOPE(oldsave),
     * which will call destuctors to reset PL_regmatch_state, free higher
     * PL_regmatch_slabs, and clean up regmatch_info_aux and
     * regmatch_info_aux_eval */

    oldsave = PL_savestack_ix;

    s = startpos;

    if ((prog->extflags & RXf_USE_INTUIT)
        && !(flags & REXEC_CHECKED))
    {
	s = re_intuit_start(rx, sv, strbeg, startpos, strend,
                                    flags, NULL);
	if (!s)
	    return 0;

	if (prog->extflags & RXf_CHECK_ALL) {
            /* we can match based purely on the result of INTUIT.
             * Set up captures etc just for $& and $-[0]
             * (an intuit-only match wont have $1,$2,..) */
            assert(!prog->nparens);

            /* s/// doesn't like it if $& is earlier than where we asked it to
             * start searching (which can happen on something like /.\G/) */
            if (       (flags & REXEC_FAIL_ON_UNDERFLOW)
                    && (s < stringarg))
            {
                /* this should only be possible under \G */
                assert(prog->intflags & PREGf_GPOS_SEEN);
                DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
                    "matched, but failing for REXEC_FAIL_ON_UNDERFLOW\n"));
                goto phooey;
            }

            /* match via INTUIT shouldn't have any captures.
             * Let @-, @+, $^N know */
            prog->lastparen = prog->lastcloseparen = 0;
            RX_MATCH_UTF8_set(rx, utf8_target);
            prog->offs[0].start = s - strbeg;
            prog->offs[0].end = utf8_target
                ? (char*)utf8_hop((U8*)s, prog->minlenret) - strbeg
                : s - strbeg + prog->minlenret;
            if ( !(flags & REXEC_NOT_FIRST) )
                S_reg_set_capture_string(aTHX_ rx,
                                        strbeg, strend,
                                        sv, flags, utf8_target);

	    return 1;
        }
    }

    multiline = prog->extflags & RXf_PMf_MULTILINE;
    
    if (strend - s < (minlen+(prog->check_offset_min<0?prog->check_offset_min:0))) {
        DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
			      "String too short [regexec_flags]...\n"));
	goto phooey;
    }
    
    /* Check validity of program. */
    if (UCHARAT(progi->program) != REG_MAGIC) {
	Perl_croak(aTHX_ "corrupted regexp program");
    }

    RX_MATCH_TAINTED_off(rx);
    RX_MATCH_UTF8_set(rx, utf8_target);

    reginfo->prog = rx;	 /* Yes, sorry that this is confusing.  */
    reginfo->intuit = 0;
    reginfo->is_utf8_pat = cBOOL(RX_UTF8(rx));
    reginfo->warned = FALSE;
    reginfo->sv = sv;
    reginfo->poscache_maxiter = 0; /* not yet started a countdown */
    /* see how far we have to get to not match where we matched before */
    reginfo->till = stringarg + minend;

    if (prog->extflags & RXf_EVAL_SEEN && SvPADTMP(sv)) {
        /* SAVEFREESV, not sv_mortalcopy, as this SV must last until after
           S_cleanup_regmatch_info_aux has executed (registered by
           SAVEDESTRUCTOR_X below).  S_cleanup_regmatch_info_aux modifies
           magic belonging to this SV.
           Not newSVsv, either, as it does not COW.
        */
        reginfo->sv = newSV(0);
        SvSetSV_nosteal(reginfo->sv, sv);
        SAVEFREESV(reginfo->sv);
    }

    /* reserve next 2 or 3 slots in PL_regmatch_state:
     * slot N+0: may currently be in use: skip it
     * slot N+1: use for regmatch_info_aux struct
     * slot N+2: use for regmatch_info_aux_eval struct if we have (?{})'s
     * slot N+3: ready for use by regmatch()
     */

    {
        regmatch_state *old_regmatch_state;
        regmatch_slab  *old_regmatch_slab;
        int i, max = (prog->extflags & RXf_EVAL_SEEN) ? 2 : 1;

        /* on first ever match, allocate first slab */
        if (!PL_regmatch_slab) {
            Newx(PL_regmatch_slab, 1, regmatch_slab);
            PL_regmatch_slab->prev = NULL;
            PL_regmatch_slab->next = NULL;
            PL_regmatch_state = SLAB_FIRST(PL_regmatch_slab);
        }

        old_regmatch_state = PL_regmatch_state;
        old_regmatch_slab  = PL_regmatch_slab;

        for (i=0; i <= max; i++) {
            if (i == 1)
                reginfo->info_aux = &(PL_regmatch_state->u.info_aux);
            else if (i ==2)
                reginfo->info_aux_eval =
                reginfo->info_aux->info_aux_eval =
                            &(PL_regmatch_state->u.info_aux_eval);

            if (++PL_regmatch_state >  SLAB_LAST(PL_regmatch_slab))
                PL_regmatch_state = S_push_slab(aTHX);
        }

        /* note initial PL_regmatch_state position; at end of match we'll
         * pop back to there and free any higher slabs */

        reginfo->info_aux->old_regmatch_state = old_regmatch_state;
        reginfo->info_aux->old_regmatch_slab  = old_regmatch_slab;
        reginfo->info_aux->poscache = NULL;

        SAVEDESTRUCTOR_X(S_cleanup_regmatch_info_aux, reginfo->info_aux);

        if ((prog->extflags & RXf_EVAL_SEEN))
            S_setup_eval_state(aTHX_ reginfo);
        else
            reginfo->info_aux_eval = reginfo->info_aux->info_aux_eval = NULL;
    }

    /* If there is a "must appear" string, look for it. */

    if (PL_curpm && (PM_GETRE(PL_curpm) == rx)) {
        /* We have to be careful. If the previous successful match
           was from this regex we don't want a subsequent partially
           successful match to clobber the old results.
           So when we detect this possibility we add a swap buffer
           to the re, and switch the buffer each match. If we fail,
           we switch it back; otherwise we leave it swapped.
        */
        swap = prog->offs;
        /* do we need a save destructor here for eval dies? */
        Newxz(prog->offs, (prog->nparens + 1), regexp_paren_pair);
        DEBUG_BUFFERS_r(Perl_re_printf( aTHX_
	    "rex=0x%"UVxf" saving  offs: orig=0x%"UVxf" new=0x%"UVxf"\n",
	    PTR2UV(prog),
	    PTR2UV(swap),
	    PTR2UV(prog->offs)
	));
    }

    if (prog->recurse_locinput)
        Zero(prog->recurse_locinput,prog->nparens + 1, char *);

    /* Simplest case: anchored match need be tried only once, or with
     * MBOL, only at the beginning of each line.
     *
     * Note that /.*.../ sets PREGf_IMPLICIT|MBOL, while /.*.../s sets
     * PREGf_IMPLICIT|SBOL. The idea is that with /.*.../s, if it doesn't
     * match at the start of the string then it won't match anywhere else
     * either; while with /.*.../, if it doesn't match at the beginning,
     * the earliest it could match is at the start of the next line */

    if (prog->intflags & (PREGf_ANCH & ~PREGf_ANCH_GPOS)) {
        char *end;

	if (regtry(reginfo, &s))
	    goto got_it;

        if (!(prog->intflags & PREGf_ANCH_MBOL))
            goto phooey;

        /* didn't match at start, try at other newline positions */

        if (minlen)
            dontbother = minlen - 1;
        end = HOP3c(strend, -dontbother, strbeg) - 1;

        /* skip to next newline */

        while (s <= end) { /* note it could be possible to match at the end of the string */
            /* NB: newlines are the same in unicode as they are in latin */
            if (*s++ != '\n')
                continue;
            if (prog->check_substr || prog->check_utf8) {
            /* note that with PREGf_IMPLICIT, intuit can only fail
             * or return the start position, so it's of limited utility.
             * Nevertheless, I made the decision that the potential for
             * quick fail was still worth it - DAPM */
                s = re_intuit_start(rx, sv, strbeg, s, strend, flags, NULL);
                if (!s)
                    goto phooey;
            }
            if (regtry(reginfo, &s))
                goto got_it;
        }
        goto phooey;
    } /* end anchored search */

    if (prog->intflags & PREGf_ANCH_GPOS)
    {
        /* PREGf_ANCH_GPOS should never be true if PREGf_GPOS_SEEN is not true */
        assert(prog->intflags & PREGf_GPOS_SEEN);
        /* For anchored \G, the only position it can match from is
         * (ganch-gofs); we already set startpos to this above; if intuit
         * moved us on from there, we can't possibly succeed */
        assert(startpos == HOPBACKc(reginfo->ganch, prog->gofs));
	if (s == startpos && regtry(reginfo, &s))
	    goto got_it;
	goto phooey;
    }

    /* Messy cases:  unanchored match. */
    if ((prog->anchored_substr || prog->anchored_utf8) && prog->intflags & PREGf_SKIP) {
	/* we have /x+whatever/ */
	/* it must be a one character string (XXXX Except is_utf8_pat?) */
	char ch;
#ifdef DEBUGGING
	int did_match = 0;
#endif
	if (utf8_target) {
            if (! prog->anchored_utf8) {
                to_utf8_substr(prog);
            }
            ch = SvPVX_const(prog->anchored_utf8)[0];
	    REXEC_FBC_SCAN(
		if (*s == ch) {
		    DEBUG_EXECUTE_r( did_match = 1 );
		    if (regtry(reginfo, &s)) goto got_it;
		    s += UTF8SKIP(s);
		    while (s < strend && *s == ch)
			s += UTF8SKIP(s);
		}
	    );

	}
	else {
            if (! prog->anchored_substr) {
                if (! to_byte_substr(prog)) {
                    NON_UTF8_TARGET_BUT_UTF8_REQUIRED(phooey);
                }
            }
            ch = SvPVX_const(prog->anchored_substr)[0];
	    REXEC_FBC_SCAN(
		if (*s == ch) {
		    DEBUG_EXECUTE_r( did_match = 1 );
		    if (regtry(reginfo, &s)) goto got_it;
		    s++;
		    while (s < strend && *s == ch)
			s++;
		}
	    );
	}
	DEBUG_EXECUTE_r(if (!did_match)
                Perl_re_printf( aTHX_
                                  "Did not find anchored character...\n")
               );
    }
    else if (prog->anchored_substr != NULL
	      || prog->anchored_utf8 != NULL
	      || ((prog->float_substr != NULL || prog->float_utf8 != NULL)
		  && prog->float_max_offset < strend - s)) {
	SV *must;
	SSize_t back_max;
	SSize_t back_min;
	char *last;
	char *last1;		/* Last position checked before */
#ifdef DEBUGGING
	int did_match = 0;
#endif
	if (prog->anchored_substr || prog->anchored_utf8) {
	    if (utf8_target) {
                if (! prog->anchored_utf8) {
                    to_utf8_substr(prog);
                }
                must = prog->anchored_utf8;
            }
            else {
                if (! prog->anchored_substr) {
                    if (! to_byte_substr(prog)) {
                        NON_UTF8_TARGET_BUT_UTF8_REQUIRED(phooey);
                    }
                }
                must = prog->anchored_substr;
            }
	    back_max = back_min = prog->anchored_offset;
	} else {
	    if (utf8_target) {
                if (! prog->float_utf8) {
                    to_utf8_substr(prog);
                }
                must = prog->float_utf8;
            }
            else {
                if (! prog->float_substr) {
                    if (! to_byte_substr(prog)) {
                        NON_UTF8_TARGET_BUT_UTF8_REQUIRED(phooey);
                    }
                }
                must = prog->float_substr;
            }
	    back_max = prog->float_max_offset;
	    back_min = prog->float_min_offset;
	}
	    
        if (back_min<0) {
	    last = strend;
	} else {
            last = HOP3c(strend,	/* Cannot start after this */
        	  -(SSize_t)(CHR_SVLEN(must)
        		 - (SvTAIL(must) != 0) + back_min), strbeg);
        }
	if (s > reginfo->strbeg)
	    last1 = HOPc(s, -1);
	else
	    last1 = s - 1;	/* bogus */

	/* XXXX check_substr already used to find "s", can optimize if
	   check_substr==must. */
	dontbother = 0;
	strend = HOPc(strend, -dontbother);
	while ( (s <= last) &&
		(s = fbm_instr((unsigned char*)HOP4c(s, back_min, strbeg,  strend),
				  (unsigned char*)strend, must,
				  multiline ? FBMrf_MULTILINE : 0)) ) {
	    DEBUG_EXECUTE_r( did_match = 1 );
	    if (HOPc(s, -back_max) > last1) {
		last1 = HOPc(s, -back_min);
		s = HOPc(s, -back_max);
	    }
	    else {
		char * const t = (last1 >= reginfo->strbeg)
                                    ? HOPc(last1, 1) : last1 + 1;

		last1 = HOPc(s, -back_min);
		s = t;
	    }
	    if (utf8_target) {
		while (s <= last1) {
		    if (regtry(reginfo, &s))
			goto got_it;
                    if (s >= last1) {
                        s++; /* to break out of outer loop */
                        break;
                    }
                    s += UTF8SKIP(s);
		}
	    }
	    else {
		while (s <= last1) {
		    if (regtry(reginfo, &s))
			goto got_it;
		    s++;
		}
	    }
	}
	DEBUG_EXECUTE_r(if (!did_match) {
            RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
                SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
            Perl_re_printf( aTHX_  "Did not find %s substr %s%s...\n",
			      ((must == prog->anchored_substr || must == prog->anchored_utf8)
			       ? "anchored" : "floating"),
                quoted, RE_SV_TAIL(must));
        });		    
	goto phooey;
    }
    else if ( (c = progi->regstclass) ) {
	if (minlen) {
	    const OPCODE op = OP(progi->regstclass);
	    /* don't bother with what can't match */
	    if (PL_regkind[op] != EXACT && PL_regkind[op] != TRIE)
	        strend = HOPc(strend, -(minlen - 1));
	}
	DEBUG_EXECUTE_r({
	    SV * const prop = sv_newmortal();
            regprop(prog, prop, c, reginfo, NULL);
	    {
		RE_PV_QUOTED_DECL(quoted,utf8_target,PERL_DEBUG_PAD_ZERO(1),
		    s,strend-s,60);
                Perl_re_printf( aTHX_
		    "Matching stclass %.*s against %s (%d bytes)\n",
		    (int)SvCUR(prop), SvPVX_const(prop),
		     quoted, (int)(strend - s));
	    }
	});
        if (find_byclass(prog, c, s, strend, reginfo))
	    goto got_it;
        DEBUG_EXECUTE_r(Perl_re_printf( aTHX_  "Contradicts stclass... [regexec_flags]\n"));
    }
    else {
	dontbother = 0;
	if (prog->float_substr != NULL || prog->float_utf8 != NULL) {
	    /* Trim the end. */
	    char *last= NULL;
	    SV* float_real;
	    STRLEN len;
	    const char *little;

	    if (utf8_target) {
                if (! prog->float_utf8) {
                    to_utf8_substr(prog);
                }
                float_real = prog->float_utf8;
            }
            else {
                if (! prog->float_substr) {
                    if (! to_byte_substr(prog)) {
                        NON_UTF8_TARGET_BUT_UTF8_REQUIRED(phooey);
                    }
                }
                float_real = prog->float_substr;
            }

            little = SvPV_const(float_real, len);
	    if (SvTAIL(float_real)) {
                    /* This means that float_real contains an artificial \n on
                     * the end due to the presence of something like this:
                     * /foo$/ where we can match both "foo" and "foo\n" at the
                     * end of the string.  So we have to compare the end of the
                     * string first against the float_real without the \n and
                     * then against the full float_real with the string.  We
                     * have to watch out for cases where the string might be
                     * smaller than the float_real or the float_real without
                     * the \n. */
		    char *checkpos= strend - len;
		    DEBUG_OPTIMISE_r(
                        Perl_re_printf( aTHX_
			    "%sChecking for float_real.%s\n",
			    PL_colors[4], PL_colors[5]));
		    if (checkpos + 1 < strbeg) {
                        /* can't match, even if we remove the trailing \n
                         * string is too short to match */
			DEBUG_EXECUTE_r(
                            Perl_re_printf( aTHX_
				"%sString shorter than required trailing substring, cannot match.%s\n",
				PL_colors[4], PL_colors[5]));
			goto phooey;
		    } else if (memEQ(checkpos + 1, little, len - 1)) {
                        /* can match, the end of the string matches without the
                         * "\n" */
			last = checkpos + 1;
		    } else if (checkpos < strbeg) {
                        /* cant match, string is too short when the "\n" is
                         * included */
			DEBUG_EXECUTE_r(
                            Perl_re_printf( aTHX_
				"%sString does not contain required trailing substring, cannot match.%s\n",
				PL_colors[4], PL_colors[5]));
			goto phooey;
		    } else if (!multiline) {
                        /* non multiline match, so compare with the "\n" at the
                         * end of the string */
			if (memEQ(checkpos, little, len)) {
			    last= checkpos;
			} else {
			    DEBUG_EXECUTE_r(
                                Perl_re_printf( aTHX_
				    "%sString does not contain required trailing substring, cannot match.%s\n",
				    PL_colors[4], PL_colors[5]));
			    goto phooey;
			}
		    } else {
                        /* multiline match, so we have to search for a place
                         * where the full string is located */
			goto find_last;
		    }
	    } else {
		  find_last:
		    if (len)
			last = rninstr(s, strend, little, little + len);
		    else
			last = strend;	/* matching "$" */
	    }
	    if (!last) {
                /* at one point this block contained a comment which was
                 * probably incorrect, which said that this was a "should not
                 * happen" case.  Even if it was true when it was written I am
                 * pretty sure it is not anymore, so I have removed the comment
                 * and replaced it with this one. Yves */
		DEBUG_EXECUTE_r(
                    Perl_re_printf( aTHX_
			"%sString does not contain required substring, cannot match.%s\n",
                        PL_colors[4], PL_colors[5]
	            ));
		goto phooey;
	    }
	    dontbother = strend - last + prog->float_min_offset;
	}
	if (minlen && (dontbother < minlen))
	    dontbother = minlen - 1;
	strend -= dontbother; 		   /* this one's always in bytes! */
	/* We don't know much -- general case. */
	if (utf8_target) {
	    for (;;) {
		if (regtry(reginfo, &s))
		    goto got_it;
		if (s >= strend)
		    break;
		s += UTF8SKIP(s);
	    };
	}
	else {
	    do {
		if (regtry(reginfo, &s))
		    goto got_it;
	    } while (s++ < strend);
	}
    }

    /* Failure. */
    goto phooey;

  got_it:
    /* s/// doesn't like it if $& is earlier than where we asked it to
     * start searching (which can happen on something like /.\G/) */
    if (       (flags & REXEC_FAIL_ON_UNDERFLOW)
            && (prog->offs[0].start < stringarg - strbeg))
    {
        /* this should only be possible under \G */
        assert(prog->intflags & PREGf_GPOS_SEEN);
        DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
            "matched, but failing for REXEC_FAIL_ON_UNDERFLOW\n"));
        goto phooey;
    }

    DEBUG_BUFFERS_r(
	if (swap)
            Perl_re_printf( aTHX_
		"rex=0x%"UVxf" freeing offs: 0x%"UVxf"\n",
		PTR2UV(prog),
		PTR2UV(swap)
	    );
    );
    Safefree(swap);

    /* clean up; this will trigger destructors that will free all slabs
     * above the current one, and cleanup the regmatch_info_aux
     * and regmatch_info_aux_eval sructs */

    LEAVE_SCOPE(oldsave);

    if (RXp_PAREN_NAMES(prog)) 
        (void)hv_iterinit(RXp_PAREN_NAMES(prog));

    /* make sure $`, $&, $', and $digit will work later */
    if ( !(flags & REXEC_NOT_FIRST) )
        S_reg_set_capture_string(aTHX_ rx,
                                    strbeg, reginfo->strend,
                                    sv, flags, utf8_target);

    return 1;

  phooey:
    DEBUG_EXECUTE_r(Perl_re_printf( aTHX_  "%sMatch failed%s\n",
			  PL_colors[4], PL_colors[5]));

    /* clean up; this will trigger destructors that will free all slabs
     * above the current one, and cleanup the regmatch_info_aux
     * and regmatch_info_aux_eval sructs */

    LEAVE_SCOPE(oldsave);

    if (swap) {
        /* we failed :-( roll it back */
        DEBUG_BUFFERS_r(Perl_re_printf( aTHX_
	    "rex=0x%"UVxf" rolling back offs: freeing=0x%"UVxf" restoring=0x%"UVxf"\n",
	    PTR2UV(prog),
	    PTR2UV(prog->offs),
	    PTR2UV(swap)
	));
        Safefree(prog->offs);
        prog->offs = swap;
    }
    return 0;
}


/* Set which rex is pointed to by PL_reg_curpm, handling ref counting.
 * Do inc before dec, in case old and new rex are the same */
#define SET_reg_curpm(Re2)                          \
    if (reginfo->info_aux_eval) {                   \
	(void)ReREFCNT_inc(Re2);		    \
	ReREFCNT_dec(PM_GETRE(PL_reg_curpm));	    \
	PM_SETRE((PL_reg_curpm), (Re2));	    \
    }


/*
 - regtry - try match at specific point
 */
STATIC bool			/* 0 failure, 1 success */
S_regtry(pTHX_ regmatch_info *reginfo, char **startposp)
{
    CHECKPOINT lastcp;
    REGEXP *const rx = reginfo->prog;
    regexp *const prog = ReANY(rx);
    SSize_t result;
#ifdef DEBUGGING
    U32 depth = 0; /* used by REGCP_SET */
#endif
    RXi_GET_DECL(prog,progi);
    GET_RE_DEBUG_FLAGS_DECL;

    PERL_ARGS_ASSERT_REGTRY;

    reginfo->cutpoint=NULL;

    prog->offs[0].start = *startposp - reginfo->strbeg;
    prog->lastparen = 0;
    prog->lastcloseparen = 0;

    /* XXXX What this code is doing here?!!!  There should be no need
       to do this again and again, prog->lastparen should take care of
       this!  --ilya*/

    /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code.
     * Actually, the code in regcppop() (which Ilya may be meaning by
     * prog->lastparen), is not needed at all by the test suite
     * (op/regexp, op/pat, op/split), but that code is needed otherwise
     * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/
     * Meanwhile, this code *is* needed for the
     * above-mentioned test suite tests to succeed.  The common theme
     * on those tests seems to be returning null fields from matches.
     * --jhi updated by dapm */
#if 1
    if (prog->nparens) {
	regexp_paren_pair *pp = prog->offs;
	I32 i;
	for (i = prog->nparens; i > (I32)prog->lastparen; i--) {
	    ++pp;
	    pp->start = -1;
	    pp->end = -1;
	}
    }
#endif
    REGCP_SET(lastcp);
    result = regmatch(reginfo, *startposp, progi->program + 1);
    if (result != -1) {
	prog->offs[0].end = result;
	return 1;
    }
    if (reginfo->cutpoint)
        *startposp= reginfo->cutpoint;
    REGCP_UNWIND(lastcp);
    return 0;
}


#define sayYES goto yes
#define sayNO goto no
#define sayNO_SILENT goto no_silent

/* we dont use STMT_START/END here because it leads to 
   "unreachable code" warnings, which are bogus, but distracting. */
#define CACHEsayNO \
    if (ST.cache_mask) \
       reginfo->info_aux->poscache[ST.cache_offset] |= ST.cache_mask; \
    sayNO

/* this is used to determine how far from the left messages like
   'failed...' are printed in regexec.c. It should be set such that
   messages are inline with the regop output that created them.
*/
#define REPORT_CODE_OFF 29
#define INDENT_CHARS(depth) ((int)(depth) % 20)
#ifdef DEBUGGING
int
Perl_re_exec_indentf(pTHX_ const char *fmt, U32 depth, ...)
{
    va_list ap;
    int result;
    PerlIO *f= Perl_debug_log;
    PERL_ARGS_ASSERT_RE_EXEC_INDENTF;
    va_start(ap, depth);
    PerlIO_printf(f, "%*s|%4"UVuf"| %*s", REPORT_CODE_OFF, "", (UV)depth, INDENT_CHARS(depth), "" );
    result = PerlIO_vprintf(f, fmt, ap);
    va_end(ap);
    return result;
}
#endif /* DEBUGGING */


#define CHRTEST_UNINIT -1001 /* c1/c2 haven't been calculated yet */
#define CHRTEST_VOID   -1000 /* the c1/c2 "next char" test should be skipped */
#define CHRTEST_NOT_A_CP_1 -999
#define CHRTEST_NOT_A_CP_2 -998

/* grab a new slab and return the first slot in it */

STATIC regmatch_state *
S_push_slab(pTHX)
{
#if PERL_VERSION < 9 && !defined(PERL_CORE)
    dMY_CXT;
#endif
    regmatch_slab *s = PL_regmatch_slab->next;
    if (!s) {
	Newx(s, 1, regmatch_slab);
	s->prev = PL_regmatch_slab;
	s->next = NULL;
	PL_regmatch_slab->next = s;
    }
    PL_regmatch_slab = s;
    return SLAB_FIRST(s);
}


/* push a new state then goto it */

#define PUSH_STATE_GOTO(state, node, input) \
    pushinput = input; \
    scan = node; \
    st->resume_state = state; \
    goto push_state;

/* push a new state with success backtracking, then goto it */

#define PUSH_YES_STATE_GOTO(state, node, input) \
    pushinput = input; \
    scan = node; \
    st->resume_state = state; \
    goto push_yes_state;




/*

regmatch() - main matching routine

This is basically one big switch statement in a loop. We execute an op,
set 'next' to point the next op, and continue. If we come to a point which
we may need to backtrack to on failure such as (A|B|C), we push a
backtrack state onto the backtrack stack. On failure, we pop the top
state, and re-enter the loop at the state indicated. If there are no more
states to pop, we return failure.

Sometimes we also need to backtrack on success; for example /A+/, where
after successfully matching one A, we need to go back and try to
match another one; similarly for lookahead assertions: if the assertion
completes successfully, we backtrack to the state just before the assertion
and then carry on.  In these cases, the pushed state is marked as
'backtrack on success too'. This marking is in fact done by a chain of
pointers, each pointing to the previous 'yes' state. On success, we pop to
the nearest yes state, discarding any intermediate failure-only states.
Sometimes a yes state is pushed just to force some cleanup code to be
called at the end of a successful match or submatch; e.g. (??{$re}) uses
it to free the inner regex.

Note that failure backtracking rewinds the cursor position, while
success backtracking leaves it alone.

A pattern is complete when the END op is executed, while a subpattern
such as (?=foo) is complete when the SUCCESS op is executed. Both of these
ops trigger the "pop to last yes state if any, otherwise return true"
behaviour.

A common convention in this function is to use A and B to refer to the two
subpatterns (or to the first nodes thereof) in patterns like /A*B/: so A is
the subpattern to be matched possibly multiple times, while B is the entire
rest of the pattern. Variable and state names reflect this convention.

The states in the main switch are the union of ops and failure/success of
substates associated with with that op.  For example, IFMATCH is the op
that does lookahead assertions /(?=A)B/ and so the IFMATCH state means
'execute IFMATCH'; while IFMATCH_A is a state saying that we have just
successfully matched A and IFMATCH_A_fail is a state saying that we have
just failed to match A. Resume states always come in pairs. The backtrack
state we push is marked as 'IFMATCH_A', but when that is popped, we resume
at IFMATCH_A or IFMATCH_A_fail, depending on whether we are backtracking
on success or failure.

The struct that holds a backtracking state is actually a big union, with
one variant for each major type of op. The variable st points to the
top-most backtrack struct. To make the code clearer, within each
block of code we #define ST to alias the relevant union.

Here's a concrete example of a (vastly oversimplified) IFMATCH
implementation:

    switch (state) {
    ....

#define ST st->u.ifmatch

    case IFMATCH: // we are executing the IFMATCH op, (?=A)B
	ST.foo = ...; // some state we wish to save
	...
	// push a yes backtrack state with a resume value of
	// IFMATCH_A/IFMATCH_A_fail, then continue execution at the
	// first node of A:
	PUSH_YES_STATE_GOTO(IFMATCH_A, A, newinput);
	// NOTREACHED

    case IFMATCH_A: // we have successfully executed A; now continue with B
	next = B;
	bar = ST.foo; // do something with the preserved value
	break;

    case IFMATCH_A_fail: // A failed, so the assertion failed
	...;   // do some housekeeping, then ...
	sayNO; // propagate the failure

#undef ST

    ...
    }

For any old-timers reading this who are familiar with the old recursive
approach, the code above is equivalent to:

    case IFMATCH: // we are executing the IFMATCH op, (?=A)B
    {
	int foo = ...
	...
	if (regmatch(A)) {
	    next = B;
	    bar = foo;
	    break;
	}
	...;   // do some housekeeping, then ...
	sayNO; // propagate the failure
    }

The topmost backtrack state, pointed to by st, is usually free. If you
want to claim it, populate any ST.foo fields in it with values you wish to
save, then do one of

	PUSH_STATE_GOTO(resume_state, node, newinput);
	PUSH_YES_STATE_GOTO(resume_state, node, newinput);

which sets that backtrack state's resume value to 'resume_state', pushes a
new free entry to the top of the backtrack stack, then goes to 'node'.
On backtracking, the free slot is popped, and the saved state becomes the
new free state. An ST.foo field in this new top state can be temporarily
accessed to retrieve values, but once the main loop is re-entered, it
becomes available for reuse.

Note that the depth of the backtrack stack constantly increases during the
left-to-right execution of the pattern, rather than going up and down with
the pattern nesting. For example the stack is at its maximum at Z at the
end of the pattern, rather than at X in the following:

    /(((X)+)+)+....(Y)+....Z/

The only exceptions to this are lookahead/behind assertions and the cut,
(?>A), which pop all the backtrack states associated with A before
continuing.
 
Backtrack state structs are allocated in slabs of about 4K in size.
PL_regmatch_state and st always point to the currently active state,
and PL_regmatch_slab points to the slab currently containing
PL_regmatch_state.  The first time regmatch() is called, the first slab is
allocated, and is never freed until interpreter destruction. When the slab
is full, a new one is allocated and chained to the end. At exit from
regmatch(), slabs allocated since entry are freed.

*/
 

#define DEBUG_STATE_pp(pp)                                  \
    DEBUG_STATE_r({                                         \
        DUMP_EXEC_POS(locinput, scan, utf8_target,depth);   \
        Perl_re_printf( aTHX_                                           \
            "%*s" pp " %s%s%s%s%s\n",                       \
            INDENT_CHARS(depth), "",                        \
            PL_reg_name[st->resume_state],                  \
            ((st==yes_state||st==mark_state) ? "[" : ""),   \
            ((st==yes_state) ? "Y" : ""),                   \
            ((st==mark_state) ? "M" : ""),                  \
            ((st==yes_state||st==mark_state) ? "]" : "")    \
        );                                                  \
    });


#define REG_NODE_NUM(x) ((x) ? (int)((x)-prog) : -1)

#ifdef DEBUGGING

STATIC void
S_debug_start_match(pTHX_ const REGEXP *prog, const bool utf8_target,
    const char *start, const char *end, const char *blurb)
{
    const bool utf8_pat = RX_UTF8(prog) ? 1 : 0;

    PERL_ARGS_ASSERT_DEBUG_START_MATCH;

    if (!PL_colorset)   
            reginitcolors();    
    {
        RE_PV_QUOTED_DECL(s0, utf8_pat, PERL_DEBUG_PAD_ZERO(0), 
            RX_PRECOMP_const(prog), RX_PRELEN(prog), 60);   
        
        RE_PV_QUOTED_DECL(s1, utf8_target, PERL_DEBUG_PAD_ZERO(1),
            start, end - start, 60); 
        
        Perl_re_printf( aTHX_
            "%s%s REx%s %s against %s\n", 
		       PL_colors[4], blurb, PL_colors[5], s0, s1); 
        
        if (utf8_target||utf8_pat)
            Perl_re_printf( aTHX_  "UTF-8 %s%s%s...\n",
                utf8_pat ? "pattern" : "",
                utf8_pat && utf8_target ? " and " : "",
                utf8_target ? "string" : ""
            ); 
    }
}

STATIC void
S_dump_exec_pos(pTHX_ const char *locinput, 
                      const regnode *scan, 
                      const char *loc_regeol, 
                      const char *loc_bostr, 
                      const char *loc_reg_starttry,
                      const bool utf8_target,
                      const U32 depth
                )
{
    const int docolor = *PL_colors[0] || *PL_colors[2] || *PL_colors[4];
    const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */
    int l = (loc_regeol - locinput) > taill ? taill : (loc_regeol - locinput);
    /* The part of the string before starttry has one color
       (pref0_len chars), between starttry and current
       position another one (pref_len - pref0_len chars),
       after the current position the third one.
       We assume that pref0_len <= pref_len, otherwise we
       decrease pref0_len.  */
    int pref_len = (locinput - loc_bostr) > (5 + taill) - l
	? (5 + taill) - l : locinput - loc_bostr;
    int pref0_len;

    PERL_ARGS_ASSERT_DUMP_EXEC_POS;

    while (utf8_target && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len)))
	pref_len++;
    pref0_len = pref_len  - (locinput - loc_reg_starttry);
    if (l + pref_len < (5 + taill) && l < loc_regeol - locinput)
	l = ( loc_regeol - locinput > (5 + taill) - pref_len
	      ? (5 + taill) - pref_len : loc_regeol - locinput);
    while (utf8_target && UTF8_IS_CONTINUATION(*(U8*)(locinput + l)))
	l--;
    if (pref0_len < 0)
	pref0_len = 0;
    if (pref0_len > pref_len)
	pref0_len = pref_len;
    {
	const int is_uni = utf8_target ? 1 : 0;

	RE_PV_COLOR_DECL(s0,len0,is_uni,PERL_DEBUG_PAD(0),
	    (locinput - pref_len),pref0_len, 60, 4, 5);
	
	RE_PV_COLOR_DECL(s1,len1,is_uni,PERL_DEBUG_PAD(1),
		    (locinput - pref_len + pref0_len),
		    pref_len - pref0_len, 60, 2, 3);
	
	RE_PV_COLOR_DECL(s2,len2,is_uni,PERL_DEBUG_PAD(2),
		    locinput, loc_regeol - locinput, 10, 0, 1);

	const STRLEN tlen=len0+len1+len2;
        Perl_re_printf( aTHX_
                    "%4"IVdf" <%.*s%.*s%s%.*s>%*s|%4u| ",
		    (IV)(locinput - loc_bostr),
		    len0, s0,
		    len1, s1,
		    (docolor ? "" : "> <"),
		    len2, s2,
		    (int)(tlen > 19 ? 0 :  19 - tlen),
                    "",
                    depth);
    }
}

#endif

/* reg_check_named_buff_matched()
 * Checks to see if a named buffer has matched. The data array of 
 * buffer numbers corresponding to the buffer is expected to reside
 * in the regexp->data->data array in the slot stored in the ARG() of
 * node involved. Note that this routine doesn't actually care about the
 * name, that information is not preserved from compilation to execution.
 * Returns the index of the leftmost defined buffer with the given name
 * or 0 if non of the buffers matched.
 */
STATIC I32
S_reg_check_named_buff_matched(const regexp *rex, const regnode *scan)
{
    I32 n;
    RXi_GET_DECL(rex,rexi);
    SV *sv_dat= MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
    I32 *nums=(I32*)SvPVX(sv_dat);

    PERL_ARGS_ASSERT_REG_CHECK_NAMED_BUFF_MATCHED;

    for ( n=0; n<SvIVX(sv_dat); n++ ) {
        if ((I32)rex->lastparen >= nums[n] &&
            rex->offs[nums[n]].end != -1)
        {
            return nums[n];
        }
    }
    return 0;
}


static bool
S_setup_EXACTISH_ST_c1_c2(pTHX_ const regnode * const text_node, int *c1p,
        U8* c1_utf8, int *c2p, U8* c2_utf8, regmatch_info *reginfo)
{
    /* This function determines if there are one or two characters that match
     * the first character of the passed-in EXACTish node <text_node>, and if
     * so, returns them in the passed-in pointers.
     *
     * If it determines that no possible character in the target string can
     * match, it returns FALSE; otherwise TRUE.  (The FALSE situation occurs if
     * the first character in <text_node> requires UTF-8 to represent, and the
     * target string isn't in UTF-8.)
     *
     * If there are more than two characters that could match the beginning of
     * <text_node>, or if more context is required to determine a match or not,
     * it sets both *<c1p> and *<c2p> to CHRTEST_VOID.
     *
     * The motiviation behind this function is to allow the caller to set up
     * tight loops for matching.  If <text_node> is of type EXACT, there is
     * only one possible character that can match its first character, and so
     * the situation is quite simple.  But things get much more complicated if
     * folding is involved.  It may be that the first character of an EXACTFish
     * node doesn't participate in any possible fold, e.g., punctuation, so it
     * can be matched only by itself.  The vast majority of characters that are
     * in folds match just two things, their lower and upper-case equivalents.
     * But not all are like that; some have multiple possible matches, or match
     * sequences of more than one character.  This function sorts all that out.
     *
     * Consider the patterns A*B or A*?B where A and B are arbitrary.  In a
     * loop of trying to match A*, we know we can't exit where the thing
     * following it isn't a B.  And something can't be a B unless it is the
     * beginning of B.  By putting a quick test for that beginning in a tight
     * loop, we can rule out things that can't possibly be B without having to
     * break out of the loop, thus avoiding work.  Similarly, if A is a single
     * character, we can make a tight loop matching A*, using the outputs of
     * this function.
     *
     * If the target string to match isn't in UTF-8, and there aren't
     * complications which require CHRTEST_VOID, *<c1p> and *<c2p> are set to
     * the one or two possible octets (which are characters in this situation)
     * that can match.  In all cases, if there is only one character that can
     * match, *<c1p> and *<c2p> will be identical.
     *
     * If the target string is in UTF-8, the buffers pointed to by <c1_utf8>
     * and <c2_utf8> will contain the one or two UTF-8 sequences of bytes that
     * can match the beginning of <text_node>.  They should be declared with at
     * least length UTF8_MAXBYTES+1.  (If the target string isn't in UTF-8, it is
     * undefined what these contain.)  If one or both of the buffers are
     * invariant under UTF-8, *<c1p>, and *<c2p> will also be set to the
     * corresponding invariant.  If variant, the corresponding *<c1p> and/or
     * *<c2p> will be set to a negative number(s) that shouldn't match any code
     * point (unless inappropriately coerced to unsigned).   *<c1p> will equal
     * *<c2p> if and only if <c1_utf8> and <c2_utf8> are the same. */

    const bool utf8_target = reginfo->is_utf8_target;

    UV c1 = (UV)CHRTEST_NOT_A_CP_1;
    UV c2 = (UV)CHRTEST_NOT_A_CP_2;
    bool use_chrtest_void = FALSE;
    const bool is_utf8_pat = reginfo->is_utf8_pat;

    /* Used when we have both utf8 input and utf8 output, to avoid converting
     * to/from code points */
    bool utf8_has_been_setup = FALSE;

    dVAR;

    U8 *pat = (U8*)STRING(text_node);
    U8 folded[UTF8_MAX_FOLD_CHAR_EXPAND * UTF8_MAXBYTES_CASE + 1] = { '\0' };

    if (OP(text_node) == EXACT || OP(text_node) == EXACTL) {

        /* In an exact node, only one thing can be matched, that first
         * character.  If both the pat and the target are UTF-8, we can just
         * copy the input to the output, avoiding finding the code point of
         * that character */
        if (!is_utf8_pat) {
            c2 = c1 = *pat;
        }
        else if (utf8_target) {
            Copy(pat, c1_utf8, UTF8SKIP(pat), U8);
            Copy(pat, c2_utf8, UTF8SKIP(pat), U8);
            utf8_has_been_setup = TRUE;
        }
        else {
            c2 = c1 = valid_utf8_to_uvchr(pat, NULL);
        }
    }
    else { /* an EXACTFish node */
        U8 *pat_end = pat + STR_LEN(text_node);

        /* An EXACTFL node has at least some characters unfolded, because what
         * they match is not known until now.  So, now is the time to fold
         * the first few of them, as many as are needed to determine 'c1' and
         * 'c2' later in the routine.  If the pattern isn't UTF-8, we only need
         * to fold if in a UTF-8 locale, and then only the Sharp S; everything
         * else is 1-1 and isn't assumed to be folded.  In a UTF-8 pattern, we
         * need to fold as many characters as a single character can fold to,
         * so that later we can check if the first ones are such a multi-char
         * fold.  But, in such a pattern only locale-problematic characters
         * aren't folded, so we can skip this completely if the first character
         * in the node isn't one of the tricky ones */
        if (OP(text_node) == EXACTFL) {

            if (! is_utf8_pat) {
                if (IN_UTF8_CTYPE_LOCALE && *pat == LATIN_SMALL_LETTER_SHARP_S)
                {
                    folded[0] = folded[1] = 's';
                    pat = folded;
                    pat_end = folded + 2;
                }
            }
            else if (is_PROBLEMATIC_LOCALE_FOLDEDS_START_utf8(pat)) {
                U8 *s = pat;
                U8 *d = folded;
                int i;

                for (i = 0; i < UTF8_MAX_FOLD_CHAR_EXPAND && s < pat_end; i++) {
                    if (isASCII(*s)) {
                        *(d++) = (U8) toFOLD_LC(*s);
                        s++;
                    }
                    else {
                        STRLEN len;
                        _to_utf8_fold_flags(s,
                                            d,
                                            &len,
                                            FOLD_FLAGS_FULL | FOLD_FLAGS_LOCALE);
                        d += len;
                        s += UTF8SKIP(s);
                    }
                }

                pat = folded;
                pat_end = d;
            }
        }

        if ((is_utf8_pat && is_MULTI_CHAR_FOLD_utf8_safe(pat, pat_end))
             || (!is_utf8_pat && is_MULTI_CHAR_FOLD_latin1_safe(pat, pat_end)))
        {
            /* Multi-character folds require more context to sort out.  Also
             * PL_utf8_foldclosures used below doesn't handle them, so have to
             * be handled outside this routine */
            use_chrtest_void = TRUE;
        }
        else { /* an EXACTFish node which doesn't begin with a multi-char fold */
            c1 = is_utf8_pat ? valid_utf8_to_uvchr(pat, NULL) : *pat;
            if (c1 > 255) {
                /* Load the folds hash, if not already done */
                SV** listp;
                if (! PL_utf8_foldclosures) {
                    _load_PL_utf8_foldclosures();
                }

                /* The fold closures data structure is a hash with the keys
                 * being the UTF-8 of every character that is folded to, like
                 * 'k', and the values each an array of all code points that
                 * fold to its key.  e.g. [ 'k', 'K', KELVIN_SIGN ].
                 * Multi-character folds are not included */
                if ((! (listp = hv_fetch(PL_utf8_foldclosures,
                                        (char *) pat,
                                        UTF8SKIP(pat),
                                        FALSE))))
                {
                    /* Not found in the hash, therefore there are no folds
                    * containing it, so there is only a single character that
                    * could match */
                    c2 = c1;
                }
                else {  /* Does participate in folds */
                    AV* list = (AV*) *listp;
                    if (av_tindex_nomg(list) != 1) {

                        /* If there aren't exactly two folds to this, it is
                         * outside the scope of this function */
                        use_chrtest_void = TRUE;
                    }
                    else {  /* There are two.  Get them */
                        SV** c_p = av_fetch(list, 0, FALSE);
                        if (c_p == NULL) {
                            Perl_croak(aTHX_ "panic: invalid PL_utf8_foldclosures structure");
                        }
                        c1 = SvUV(*c_p);

                        c_p = av_fetch(list, 1, FALSE);
                        if (c_p == NULL) {
                            Perl_croak(aTHX_ "panic: invalid PL_utf8_foldclosures structure");
                        }
                        c2 = SvUV(*c_p);

                        /* Folds that cross the 255/256 boundary are forbidden
                         * if EXACTFL (and isnt a UTF8 locale), or EXACTFA and
                         * one is ASCIII.  Since the pattern character is above
                         * 255, and its only other match is below 256, the only
                         * legal match will be to itself.  We have thrown away
                         * the original, so have to compute which is the one
                         * above 255. */
                        if ((c1 < 256) != (c2 < 256)) {
                            if ((OP(text_node) == EXACTFL
                                 && ! IN_UTF8_CTYPE_LOCALE)
                                || ((OP(text_node) == EXACTFA
                                    || OP(text_node) == EXACTFA_NO_TRIE)
                                    && (isASCII(c1) || isASCII(c2))))
                            {
                                if (c1 < 256) {
                                    c1 = c2;
                                }
                                else {
                                    c2 = c1;
                                }
                            }
                        }
                    }
                }
            }
            else /* Here, c1 is <= 255 */
                if (utf8_target
                    && HAS_NONLATIN1_FOLD_CLOSURE(c1)
                    && ( ! (OP(text_node) == EXACTFL && ! IN_UTF8_CTYPE_LOCALE))
                    && ((OP(text_node) != EXACTFA
                        && OP(text_node) != EXACTFA_NO_TRIE)
                        || ! isASCII(c1)))
            {
                /* Here, there could be something above Latin1 in the target
                 * which folds to this character in the pattern.  All such
                 * cases except LATIN SMALL LETTER Y WITH DIAERESIS have more
                 * than two characters involved in their folds, so are outside
                 * the scope of this function */
                if (UNLIKELY(c1 == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS)) {
                    c2 = LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS;
                }
                else {
                    use_chrtest_void = TRUE;
                }
            }
            else { /* Here nothing above Latin1 can fold to the pattern
                      character */
                switch (OP(text_node)) {

                    case EXACTFL:   /* /l rules */
                        c2 = PL_fold_locale[c1];
                        break;

                    case EXACTF:   /* This node only generated for non-utf8
                                    patterns */
                        assert(! is_utf8_pat);
                        if (! utf8_target) {    /* /d rules */
                            c2 = PL_fold[c1];
                            break;
                        }
                        /* FALLTHROUGH */
                        /* /u rules for all these.  This happens to work for
                        * EXACTFA as nothing in Latin1 folds to ASCII */
                    case EXACTFA_NO_TRIE:   /* This node only generated for
                                            non-utf8 patterns */
                        assert(! is_utf8_pat);
                        /* FALLTHROUGH */
                    case EXACTFA:
                    case EXACTFU_SS:
                    case EXACTFU:
                        c2 = PL_fold_latin1[c1];
                        break;

                    default:
                        Perl_croak(aTHX_ "panic: Unexpected op %u", OP(text_node));
                        NOT_REACHED; /* NOTREACHED */
                }
            }
        }
    }

    /* Here have figured things out.  Set up the returns */
    if (use_chrtest_void) {
        *c2p = *c1p = CHRTEST_VOID;
    }
    else if (utf8_target) {
        if (! utf8_has_been_setup) {    /* Don't have the utf8; must get it */
            uvchr_to_utf8(c1_utf8, c1);
            uvchr_to_utf8(c2_utf8, c2);
        }

        /* Invariants are stored in both the utf8 and byte outputs; Use
         * negative numbers otherwise for the byte ones.  Make sure that the
         * byte ones are the same iff the utf8 ones are the same */
        *c1p = (UTF8_IS_INVARIANT(*c1_utf8)) ? *c1_utf8 : CHRTEST_NOT_A_CP_1;
        *c2p = (UTF8_IS_INVARIANT(*c2_utf8))
                ? *c2_utf8
                : (c1 == c2)
                  ? CHRTEST_NOT_A_CP_1
                  : CHRTEST_NOT_A_CP_2;
    }
    else if (c1 > 255) {
       if (c2 > 255) {  /* both possibilities are above what a non-utf8 string
                           can represent */
           return FALSE;
       }

       *c1p = *c2p = c2;    /* c2 is the only representable value */
    }
    else {  /* c1 is representable; see about c2 */
       *c1p = c1;
       *c2p = (c2 < 256) ? c2 : c1;
    }

    return TRUE;
}

PERL_STATIC_INLINE bool
S_isGCB(const GCB_enum before, const GCB_enum after)
{
    /* returns a boolean indicating if there is a Grapheme Cluster Boundary
     * between the inputs.  See http://www.unicode.org/reports/tr29/ */

    return GCB_table[before][after];
}

/* Combining marks attach to most classes that precede them, but this defines
 * the exceptions (from TR14) */
#define LB_CM_ATTACHES_TO(prev) ( ! (   prev == LB_EDGE                 \
                                     || prev == LB_Mandatory_Break      \
                                     || prev == LB_Carriage_Return      \
                                     || prev == LB_Line_Feed            \
                                     || prev == LB_Next_Line            \
                                     || prev == LB_Space                \
                                     || prev == LB_ZWSpace))

STATIC bool
S_isLB(pTHX_ LB_enum before,
             LB_enum after,
             const U8 * const strbeg,
             const U8 * const curpos,
             const U8 * const strend,
             const bool utf8_target)
{
    U8 * temp_pos = (U8 *) curpos;
    LB_enum prev = before;

    /* Is the boundary between 'before' and 'after' line-breakable?
     * Most of this is just a table lookup of a generated table from Unicode
     * rules.  But some rules require context to decide, and so have to be
     * implemented in code */

    PERL_ARGS_ASSERT_ISLB;

    /* Rule numbers in the comments below are as of Unicode 8.0 */

  redo:
    before = prev;
    switch (LB_table[before][after]) {
        case LB_BREAKABLE:
            return TRUE;

        case LB_NOBREAK:
        case LB_NOBREAK_EVEN_WITH_SP_BETWEEN:
            return FALSE;

        case LB_SP_foo + LB_BREAKABLE:
        case LB_SP_foo + LB_NOBREAK:
        case LB_SP_foo + LB_NOBREAK_EVEN_WITH_SP_BETWEEN:

            /* When we have something following a SP, we have to look at the
             * context in order to know what to do.
             *
             * SP SP should not reach here because LB7: Do not break before
             * spaces.  (For two spaces in a row there is nothing that
             * overrides that) */
            assert(after != LB_Space);

            /* Here we have a space followed by a non-space.  Mostly this is a
             * case of LB18: "Break after spaces".  But there are complications
             * as the handling of spaces is somewhat tricky.  They are in a
             * number of rules, which have to be applied in priority order, but
             * something earlier in the string can cause a rule to be skipped
             * and a lower priority rule invoked.  A prime example is LB7 which
             * says don't break before a space.  But rule LB8 (lower priority)
             * says that the first break opportunity after a ZW is after any
             * span of spaces immediately after it.  If a ZW comes before a SP
             * in the input, rule LB8 applies, and not LB7.  Other such rules
             * involve combining marks which are rules 9 and 10, but they may
             * override higher priority rules if they come earlier in the
             * string.  Since we're doing random access into the middle of the
             * string, we have to look for rules that should get applied based
             * on both string position and priority.  Combining marks do not
             * attach to either ZW nor SP, so we don't have to consider them
             * until later.
             *
             * To check for LB8, we have to find the first non-space character
             * before this span of spaces */
            do {
                prev = backup_one_LB(strbeg, &temp_pos, utf8_target);
            }
            while (prev == LB_Space);

            /* LB8 Break before any character following a zero-width space,
             * even if one or more spaces intervene.
             *      ZW SP* ÷
             * So if we have a ZW just before this span, and to get here this
             * is the final space in the span. */
            if (prev == LB_ZWSpace) {
                return TRUE;
            }

            /* Here, not ZW SP+.  There are several rules that have higher
             * priority than LB18 and can be resolved now, as they don't depend
             * on anything earlier in the string (except ZW, which we have
             * already handled).  One of these rules is LB11 Do not break
             * before Word joiner, but we have specially encoded that in the
             * lookup table so it is caught by the single test below which
             * catches the other ones. */
            if (LB_table[LB_Space][after] - LB_SP_foo
                                            == LB_NOBREAK_EVEN_WITH_SP_BETWEEN)
            {
                return FALSE;
            }

            /* If we get here, we have to XXX consider combining marks. */
            if (prev == LB_Combining_Mark) {

                /* What happens with these depends on the character they
                 * follow.  */
                do {
                    prev = backup_one_LB(strbeg, &temp_pos, utf8_target);
                }
                while (prev == LB_Combining_Mark);

                /* Most times these attach to and inherit the characteristics
                 * of that character, but not always, and when not, they are to
                 * be treated as AL by rule LB10. */
                if (! LB_CM_ATTACHES_TO(prev)) {
                    prev = LB_Alphabetic;
                }
            }

            /* Here, we have the character preceding the span of spaces all set
             * up.  We follow LB18: "Break after spaces" unless the table shows
             * that is overriden */
            return LB_table[prev][after] != LB_NOBREAK_EVEN_WITH_SP_BETWEEN;

        case LB_CM_foo:

            /* We don't know how to treat the CM except by looking at the first
             * non-CM character preceding it */
            do {
                prev = backup_one_LB(strbeg, &temp_pos, utf8_target);
            }
            while (prev == LB_Combining_Mark);

            /* Here, 'prev' is that first earlier non-CM character.  If the CM
             * attatches to it, then it inherits the behavior of 'prev'.  If it
             * doesn't attach, it is to be treated as an AL */
            if (! LB_CM_ATTACHES_TO(prev)) {
                prev = LB_Alphabetic;
            }

            goto redo;

        case LB_HY_or_BA_then_foo + LB_BREAKABLE:
        case LB_HY_or_BA_then_foo + LB_NOBREAK:

            /* LB21a Don't break after Hebrew + Hyphen.
             * HL (HY | BA) × */

            if (backup_one_LB(strbeg, &temp_pos, utf8_target)
                                                          == LB_Hebrew_Letter)
            {
                return FALSE;
            }

            return LB_table[prev][after] - LB_HY_or_BA_then_foo == LB_BREAKABLE;

        case LB_PR_or_PO_then_OP_or_HY + LB_BREAKABLE:
        case LB_PR_or_PO_then_OP_or_HY + LB_NOBREAK:

            /* LB25a (PR | PO) × ( OP | HY )? NU */
            if (advance_one_LB(&temp_pos, strend, utf8_target) == LB_Numeric) {
                return FALSE;
            }

            return LB_table[prev][after] - LB_PR_or_PO_then_OP_or_HY
                                                                == LB_BREAKABLE;

        case LB_SY_or_IS_then_various + LB_BREAKABLE:
        case LB_SY_or_IS_then_various + LB_NOBREAK:
        {
            /* LB25d NU (SY | IS)* × (NU | SY | IS | CL | CP ) */

            LB_enum temp = prev;
            do {
                temp = backup_one_LB(strbeg, &temp_pos, utf8_target);
            }
            while (temp == LB_Break_Symbols || temp == LB_Infix_Numeric);
            if (temp == LB_Numeric) {
                return FALSE;
            }

            return LB_table[prev][after] - LB_SY_or_IS_then_various
                                                               == LB_BREAKABLE;
        }

        case LB_various_then_PO_or_PR + LB_BREAKABLE:
        case LB_various_then_PO_or_PR + LB_NOBREAK:
        {
            /* LB25e NU (SY | IS)* (CL | CP)? × (PO | PR) */

            LB_enum temp = prev;
            if (temp == LB_Close_Punctuation || temp == LB_Close_Parenthesis)
            {
                temp = backup_one_LB(strbeg, &temp_pos, utf8_target);
            }
            while (temp == LB_Break_Symbols || temp == LB_Infix_Numeric) {
                temp = backup_one_LB(strbeg, &temp_pos, utf8_target);
            }
            if (temp == LB_Numeric) {
                return FALSE;
            }
            return LB_various_then_PO_or_PR;
        }

        default:
            break;
    }

#ifdef DEBUGGING
    Perl_re_printf( aTHX_  "Unhandled LB pair: LB_table[%d, %d] = %d\n",
                                  before, after, LB_table[before][after]);
    assert(0);
#endif
    return TRUE;
}

STATIC LB_enum
S_advance_one_LB(pTHX_ U8 ** curpos, const U8 * const strend, const bool utf8_target)
{
    LB_enum lb;

    PERL_ARGS_ASSERT_ADVANCE_ONE_LB;

    if (*curpos >= strend) {
        return LB_EDGE;
    }

    if (utf8_target) {
        *curpos += UTF8SKIP(*curpos);
        if (*curpos >= strend) {
            return LB_EDGE;
        }
        lb = getLB_VAL_UTF8(*curpos, strend);
    }
    else {
        (*curpos)++;
        if (*curpos >= strend) {
            return LB_EDGE;
        }
        lb = getLB_VAL_CP(**curpos);
    }

    return lb;
}

STATIC LB_enum
S_backup_one_LB(pTHX_ const U8 * const strbeg, U8 ** curpos, const bool utf8_target)
{
    LB_enum lb;

    PERL_ARGS_ASSERT_BACKUP_ONE_LB;

    if (*curpos < strbeg) {
        return LB_EDGE;
    }

    if (utf8_target) {
        U8 * prev_char_pos = reghopmaybe3(*curpos, -1, strbeg);
        U8 * prev_prev_char_pos;

        if (! prev_char_pos) {
            return LB_EDGE;
        }

        if ((prev_prev_char_pos = reghopmaybe3((U8 *) prev_char_pos, -1, strbeg))) {
            lb = getLB_VAL_UTF8(prev_prev_char_pos, prev_char_pos);
            *curpos = prev_char_pos;
            prev_char_pos = prev_prev_char_pos;
        }
        else {
            *curpos = (U8 *) strbeg;
            return LB_EDGE;
        }
    }
    else {
        if (*curpos - 2 < strbeg) {
            *curpos = (U8 *) strbeg;
            return LB_EDGE;
        }
        (*curpos)--;
        lb = getLB_VAL_CP(*(*curpos - 1));
    }

    return lb;
}

STATIC bool
S_isSB(pTHX_ SB_enum before,
             SB_enum after,
             const U8 * const strbeg,
             const U8 * const curpos,
             const U8 * const strend,
             const bool utf8_target)
{
    /* returns a boolean indicating if there is a Sentence Boundary Break
     * between the inputs.  See http://www.unicode.org/reports/tr29/ */

    U8 * lpos = (U8 *) curpos;
    bool has_para_sep = FALSE;
    bool has_sp = FALSE;

    PERL_ARGS_ASSERT_ISSB;

    /* Break at the start and end of text.
        SB1.  sot  ÷
        SB2.  ÷  eot
      But unstated in Unicode is don't break if the text is empty */
    if (before == SB_EDGE || after == SB_EDGE) {
        return before != after;
    }

    /* SB 3: Do not break within CRLF. */
    if (before == SB_CR && after == SB_LF) {
        return FALSE;
    }

    /* Break after paragraph separators.  CR and LF are considered
     * so because Unicode views text as like word processing text where there
     * are no newlines except between paragraphs, and the word processor takes
     * care of wrapping without there being hard line-breaks in the text *./
       SB4.  Sep | CR | LF  ÷ */
    if (before == SB_Sep || before == SB_CR || before == SB_LF) {
        return TRUE;
    }

    /* Ignore Format and Extend characters, except after sot, Sep, CR, or LF.
     * (See Section 6.2, Replacing Ignore Rules.)
        SB5.  X (Extend | Format)*  →  X */
    if (after == SB_Extend || after == SB_Format) {

        /* Implied is that the these characters attach to everything
         * immediately prior to them except for those separator-type
         * characters.  And the rules earlier have already handled the case
         * when one of those immediately precedes the extend char */
        return FALSE;
    }

    if (before == SB_Extend || before == SB_Format) {
        U8 * temp_pos = lpos;
        const SB_enum backup = backup_one_SB(strbeg, &temp_pos, utf8_target);
        if (   backup != SB_EDGE
            && backup != SB_Sep
            && backup != SB_CR
            && backup != SB_LF)
        {
            before = backup;
            lpos = temp_pos;
        }

        /* Here, both 'before' and 'backup' are these types; implied is that we
         * don't break between them */
        if (backup == SB_Extend || backup == SB_Format) {
            return FALSE;
        }
    }

    /* Do not break after ambiguous terminators like period, if they are
     * immediately followed by a number or lowercase letter, if they are
     * between uppercase letters, if the first following letter (optionally
     * after certain punctuation) is lowercase, or if they are followed by
     * "continuation" punctuation such as comma, colon, or semicolon. For
     * example, a period may be an abbreviation or numeric period, and thus may
     * not mark the end of a sentence.

     * SB6. ATerm  ×  Numeric */
    if (before == SB_ATerm && after == SB_Numeric) {
        return FALSE;
    }

    /* SB7.  (Upper | Lower) ATerm  ×  Upper */
    if (before == SB_ATerm && after == SB_Upper) {
        U8 * temp_pos = lpos;
        SB_enum backup = backup_one_SB(strbeg, &temp_pos, utf8_target);
        if (backup == SB_Upper || backup == SB_Lower) {
            return FALSE;
        }
    }

    /* The remaining rules that aren't the final one, all require an STerm or
     * an ATerm after having backed up over some Close* Sp*, and in one case an
     * optional Paragraph separator, although one rule doesn't have any Sp's in it.
     * So do that backup now, setting flags if either Sp or a paragraph
     * separator are found */

    if (before == SB_Sep || before == SB_CR || before == SB_LF) {
        has_para_sep = TRUE;
        before = backup_one_SB(strbeg, &lpos, utf8_target);
    }

    if (before == SB_Sp) {
        has_sp = TRUE;
        do {
            before = backup_one_SB(strbeg, &lpos, utf8_target);
        }
        while (before == SB_Sp);
    }

    while (before == SB_Close) {
        before = backup_one_SB(strbeg, &lpos, utf8_target);
    }

    /* The next few rules apply only when the backed-up-to is an ATerm, and in
     * most cases an STerm */
    if (before == SB_STerm || before == SB_ATerm) {

        /* So, here the lhs matches
         *      (STerm | ATerm) Close* Sp* (Sep | CR | LF)?
         * and we have set flags if we found an Sp, or the optional Sep,CR,LF.
         * The rules that apply here are:
         *
         * SB8    ATerm Close* Sp*  ×  ( ¬(OLetter | Upper | Lower | Sep | CR
                                           | LF | STerm | ATerm) )* Lower
           SB8a  (STerm | ATerm) Close* Sp*  ×  (SContinue | STerm | ATerm)
           SB9   (STerm | ATerm) Close*  ×  (Close | Sp | Sep | CR | LF)
           SB10  (STerm | ATerm) Close* Sp*  ×  (Sp | Sep | CR | LF)
           SB11  (STerm | ATerm) Close* Sp* (Sep | CR | LF)?  ÷
         */

        /* And all but SB11 forbid having seen a paragraph separator */
        if (! has_para_sep) {
            if (before == SB_ATerm) {          /* SB8 */
                U8 * rpos = (U8 *) curpos;
                SB_enum later = after;

                while (    later != SB_OLetter
                        && later != SB_Upper
                        && later != SB_Lower
                        && later != SB_Sep
                        && later != SB_CR
                        && later != SB_LF
                        && later != SB_STerm
                        && later != SB_ATerm
                        && later != SB_EDGE)
                {
                    later = advance_one_SB(&rpos, strend, utf8_target);
                }
                if (later == SB_Lower) {
                    return FALSE;
                }
            }

            if (   after == SB_SContinue    /* SB8a */
                || after == SB_STerm
                || after == SB_ATerm)
            {
                return FALSE;
            }

            if (! has_sp) {     /* SB9 applies only if there was no Sp* */
                if (   after == SB_Close
                    || after == SB_Sp
                    || after == SB_Sep
                    || after == SB_CR
                    || after == SB_LF)
                {
                    return FALSE;
                }
            }

            /* SB10.  This and SB9 could probably be combined some way, but khw
             * has decided to follow the Unicode rule book precisely for
             * simplified maintenance */
            if (   after == SB_Sp
                || after == SB_Sep
                || after == SB_CR
                || after == SB_LF)
            {
                return FALSE;
            }
        }

        /* SB11.  */
        return TRUE;
    }

    /* Otherwise, do not break.
    SB12.  Any  ×  Any */

    return FALSE;
}

STATIC SB_enum
S_advance_one_SB(pTHX_ U8 ** curpos, const U8 * const strend, const bool utf8_target)
{
    SB_enum sb;

    PERL_ARGS_ASSERT_ADVANCE_ONE_SB;

    if (*curpos >= strend) {
        return SB_EDGE;
    }

    if (utf8_target) {
        do {
            *curpos += UTF8SKIP(*curpos);
            if (*curpos >= strend) {
                return SB_EDGE;
            }
            sb = getSB_VAL_UTF8(*curpos, strend);
        } while (sb == SB_Extend || sb == SB_Format);
    }
    else {
        do {
            (*curpos)++;
            if (*curpos >= strend) {
                return SB_EDGE;
            }
            sb = getSB_VAL_CP(**curpos);
        } while (sb == SB_Extend || sb == SB_Format);
    }

    return sb;
}

STATIC SB_enum
S_backup_one_SB(pTHX_ const U8 * const strbeg, U8 ** curpos, const bool utf8_target)
{
    SB_enum sb;

    PERL_ARGS_ASSERT_BACKUP_ONE_SB;

    if (*curpos < strbeg) {
        return SB_EDGE;
    }

    if (utf8_target) {
        U8 * prev_char_pos = reghopmaybe3(*curpos, -1, strbeg);
        if (! prev_char_pos) {
            return SB_EDGE;
        }

        /* Back up over Extend and Format.  curpos is always just to the right
         * of the characater whose value we are getting */
        do {
            U8 * prev_prev_char_pos;
            if ((prev_prev_char_pos = reghopmaybe3((U8 *) prev_char_pos, -1,
                                                                      strbeg)))
            {
                sb = getSB_VAL_UTF8(prev_prev_char_pos, prev_char_pos);
                *curpos = prev_char_pos;
                prev_char_pos = prev_prev_char_pos;
            }
            else {
                *curpos = (U8 *) strbeg;
                return SB_EDGE;
            }
        } while (sb == SB_Extend || sb == SB_Format);
    }
    else {
        do {
            if (*curpos - 2 < strbeg) {
                *curpos = (U8 *) strbeg;
                return SB_EDGE;
            }
            (*curpos)--;
            sb = getSB_VAL_CP(*(*curpos - 1));
        } while (sb == SB_Extend || sb == SB_Format);
    }

    return sb;
}

STATIC bool
S_isWB(pTHX_ WB_enum previous,
             WB_enum before,
             WB_enum after,
             const U8 * const strbeg,
             const U8 * const curpos,
             const U8 * const strend,
             const bool utf8_target)
{
    /*  Return a boolean as to if the boundary between 'before' and 'after' is
     *  a Unicode word break, using their published algorithm, but tailored for
     *  Perl by treating spans of white space as one unit.  Context may be
     *  needed to make this determination.  If the value for the character
     *  before 'before' is known, it is passed as 'previous'; otherwise that
     *  should be set to WB_UNKNOWN.  The other input parameters give the
     *  boundaries and current position in the matching of the string.  That
     *  is, 'curpos' marks the position where the character whose wb value is
     *  'after' begins.  See http://www.unicode.org/reports/tr29/ */

    U8 * before_pos = (U8 *) curpos;
    U8 * after_pos = (U8 *) curpos;
    WB_enum prev = before;
    WB_enum next;

    PERL_ARGS_ASSERT_ISWB;

    /* Rule numbers in the comments below are as of Unicode 8.0 */

  redo:
    before = prev;
    switch (WB_table[before][after]) {
        case WB_BREAKABLE:
            return TRUE;

        case WB_NOBREAK:
            return FALSE;

        case WB_hs_then_hs:     /* 2 horizontal spaces in a row */
            next = advance_one_WB(&after_pos, strend, utf8_target,
                                 FALSE /* Don't skip Extend nor Format */ );
            /* A space immediately preceeding an Extend or Format is attached
             * to by them, and hence gets separated from previous spaces.
             * Otherwise don't break between horizontal white space */
            return next == WB_Extend || next == WB_Format;

        /* WB4 Ignore Format and Extend characters, except when they appear at
         * the beginning of a region of text.  This code currently isn't
         * general purpose, but it works as the rules are currently and likely
         * to be laid out.  The reason it works is that when 'they appear at
         * the beginning of a region of text', the rule is to break before
         * them, just like any other character.  Therefore, the default rule
         * applies and we don't have to look in more depth.  Should this ever
         * change, we would have to have 2 'case' statements, like in the
         * rules below, and backup a single character (not spacing over the
         * extend ones) and then see if that is one of the region-end
         * characters and go from there */
        case WB_Ex_or_FO_then_foo:
            prev = backup_one_WB(&previous, strbeg, &before_pos, utf8_target);
            goto redo;

        case WB_DQ_then_HL + WB_BREAKABLE:
        case WB_DQ_then_HL + WB_NOBREAK:

            /* WB7c  Hebrew_Letter Double_Quote  ×  Hebrew_Letter */

            if (backup_one_WB(&previous, strbeg, &before_pos, utf8_target)
                                                            == WB_Hebrew_Letter)
            {
                return FALSE;
            }

             return WB_table[before][after] - WB_DQ_then_HL == WB_BREAKABLE;

        case WB_HL_then_DQ + WB_BREAKABLE:
        case WB_HL_then_DQ + WB_NOBREAK:

            /* WB7b  Hebrew_Letter  ×  Double_Quote Hebrew_Letter */

            if (advance_one_WB(&after_pos, strend, utf8_target,
                                       TRUE /* Do skip Extend and Format */ )
                                                            == WB_Hebrew_Letter)
            {
                return FALSE;
            }

            return WB_table[before][after] - WB_HL_then_DQ == WB_BREAKABLE;

        case WB_LE_or_HL_then_MB_or_ML_or_SQ + WB_NOBREAK:
        case WB_LE_or_HL_then_MB_or_ML_or_SQ + WB_BREAKABLE:

            /* WB6  (ALetter | Hebrew_Letter)  ×  (MidLetter | MidNumLet
             *       | Single_Quote) (ALetter | Hebrew_Letter) */

            next = advance_one_WB(&after_pos, strend, utf8_target,
                                       TRUE /* Do skip Extend and Format */ );

            if (next == WB_ALetter || next == WB_Hebrew_Letter)
            {
                return FALSE;
            }

            return WB_table[before][after]
                            - WB_LE_or_HL_then_MB_or_ML_or_SQ == WB_BREAKABLE;

        case WB_MB_or_ML_or_SQ_then_LE_or_HL + WB_NOBREAK:
        case WB_MB_or_ML_or_SQ_then_LE_or_HL + WB_BREAKABLE:

            /* WB7  (ALetter | Hebrew_Letter) (MidLetter | MidNumLet
             *       | Single_Quote)  ×  (ALetter | Hebrew_Letter) */

            prev = backup_one_WB(&previous, strbeg, &before_pos, utf8_target);
            if (prev == WB_ALetter || prev == WB_Hebrew_Letter)
            {
                return FALSE;
            }

            return WB_table[before][after]
                            - WB_MB_or_ML_or_SQ_then_LE_or_HL == WB_BREAKABLE;

        case WB_MB_or_MN_or_SQ_then_NU + WB_NOBREAK:
        case WB_MB_or_MN_or_SQ_then_NU + WB_BREAKABLE:

            /* WB11  Numeric (MidNum | (MidNumLet | Single_Quote))  ×  Numeric
             * */

            if (backup_one_WB(&previous, strbeg, &before_pos, utf8_target)
                                                            == WB_Numeric)
            {
                return FALSE;
            }

            return WB_table[before][after]
                                - WB_MB_or_MN_or_SQ_then_NU == WB_BREAKABLE;

        case WB_NU_then_MB_or_MN_or_SQ + WB_NOBREAK:
        case WB_NU_then_MB_or_MN_or_SQ + WB_BREAKABLE:

            /* WB12  Numeric  ×  (MidNum | MidNumLet | Single_Quote) Numeric */

            if (advance_one_WB(&after_pos, strend, utf8_target,
                                       TRUE /* Do skip Extend and Format */ )
                                                            == WB_Numeric)
            {
                return FALSE;
            }

            return WB_table[before][after]
                                - WB_NU_then_MB_or_MN_or_SQ == WB_BREAKABLE;

        default:
            break;
    }

#ifdef DEBUGGING
    Perl_re_printf( aTHX_  "Unhandled WB pair: WB_table[%d, %d] = %d\n",
                                  before, after, WB_table[before][after]);
    assert(0);
#endif
    return TRUE;
}

STATIC WB_enum
S_advance_one_WB(pTHX_ U8 ** curpos,
                       const U8 * const strend,
                       const bool utf8_target,
                       const bool skip_Extend_Format)
{
    WB_enum wb;

    PERL_ARGS_ASSERT_ADVANCE_ONE_WB;

    if (*curpos >= strend) {
        return WB_EDGE;
    }

    if (utf8_target) {

        /* Advance over Extend and Format */
        do {
            *curpos += UTF8SKIP(*curpos);
            if (*curpos >= strend) {
                return WB_EDGE;
            }
            wb = getWB_VAL_UTF8(*curpos, strend);
        } while (    skip_Extend_Format
                 && (wb == WB_Extend || wb == WB_Format));
    }
    else {
        do {
            (*curpos)++;
            if (*curpos >= strend) {
                return WB_EDGE;
            }
            wb = getWB_VAL_CP(**curpos);
        } while (    skip_Extend_Format
                 && (wb == WB_Extend || wb == WB_Format));
    }

    return wb;
}

STATIC WB_enum
S_backup_one_WB(pTHX_ WB_enum * previous, const U8 * const strbeg, U8 ** curpos, const bool utf8_target)
{
    WB_enum wb;

    PERL_ARGS_ASSERT_BACKUP_ONE_WB;

    /* If we know what the previous character's break value is, don't have
        * to look it up */
    if (*previous != WB_UNKNOWN) {
        wb = *previous;

        /* But we need to move backwards by one */
        if (utf8_target) {
            *curpos = reghopmaybe3(*curpos, -1, strbeg);
            if (! *curpos) {
                *previous = WB_EDGE;
                *curpos = (U8 *) strbeg;
            }
            else {
                *previous = WB_UNKNOWN;
            }
        }
        else {
            (*curpos)--;
            *previous = (*curpos <= strbeg) ? WB_EDGE : WB_UNKNOWN;
        }

        /* And we always back up over these two types */
        if (wb != WB_Extend && wb != WB_Format) {
            return wb;
        }
    }

    if (*curpos < strbeg) {
        return WB_EDGE;
    }

    if (utf8_target) {
        U8 * prev_char_pos = reghopmaybe3(*curpos, -1, strbeg);
        if (! prev_char_pos) {
            return WB_EDGE;
        }

        /* Back up over Extend and Format.  curpos is always just to the right
         * of the characater whose value we are getting */
        do {
            U8 * prev_prev_char_pos;
            if ((prev_prev_char_pos = reghopmaybe3((U8 *) prev_char_pos,
                                                   -1,
                                                   strbeg)))
            {
                wb = getWB_VAL_UTF8(prev_prev_char_pos, prev_char_pos);
                *curpos = prev_char_pos;
                prev_char_pos = prev_prev_char_pos;
            }
            else {
                *curpos = (U8 *) strbeg;
                return WB_EDGE;
            }
        } while (wb == WB_Extend || wb == WB_Format);
    }
    else {
        do {
            if (*curpos - 2 < strbeg) {
                *curpos = (U8 *) strbeg;
                return WB_EDGE;
            }
            (*curpos)--;
            wb = getWB_VAL_CP(*(*curpos - 1));
        } while (wb == WB_Extend || wb == WB_Format);
    }

    return wb;
}

#define EVAL_CLOSE_PAREN_IS(st,expr)                        \
(                                                           \
    (   ( st )                                         ) && \
    (   ( st )->u.eval.close_paren                     ) && \
    ( ( ( st )->u.eval.close_paren ) == ( (expr) + 1 ) )    \
)

#define EVAL_CLOSE_PAREN_IS_TRUE(st,expr)                   \
(                                                           \
    (   ( st )                                         ) && \
    (   ( st )->u.eval.close_paren                     ) && \
    (   ( expr )                                       ) && \
    ( ( ( st )->u.eval.close_paren ) == ( (expr) + 1 ) )    \
)


#define EVAL_CLOSE_PAREN_SET(st,expr) \
    (st)->u.eval.close_paren = ( (expr) + 1 )

#define EVAL_CLOSE_PAREN_CLEAR(st) \
    (st)->u.eval.close_paren = 0

/* returns -1 on failure, $+[0] on success */
STATIC SSize_t
S_regmatch(pTHX_ regmatch_info *reginfo, char *startpos, regnode *prog)
{

#if PERL_VERSION < 9 && !defined(PERL_CORE)
    dMY_CXT;
#endif
    dVAR;
    const bool utf8_target = reginfo->is_utf8_target;
    const U32 uniflags = UTF8_ALLOW_DEFAULT;
    REGEXP *rex_sv = reginfo->prog;
    regexp *rex = ReANY(rex_sv);
    RXi_GET_DECL(rex,rexi);
    /* the current state. This is a cached copy of PL_regmatch_state */
    regmatch_state *st;
    /* cache heavy used fields of st in registers */
    regnode *scan;
    regnode *next;
    U32 n = 0;	/* general value; init to avoid compiler warning */
    SSize_t ln = 0; /* len or last;  init to avoid compiler warning */
    char *locinput = startpos;
    char *pushinput; /* where to continue after a PUSH */
    I32 nextchr;   /* is always set to UCHARAT(locinput), or -1 at EOS */

    bool result = 0;	    /* return value of S_regmatch */
    int depth = 0;	    /* depth of backtrack stack */
    U32 nochange_depth = 0; /* depth of GOSUB recursion with nochange */
    const U32 max_nochange_depth =
        (3 * rex->nparens > MAX_RECURSE_EVAL_NOCHANGE_DEPTH) ?
        3 * rex->nparens : MAX_RECURSE_EVAL_NOCHANGE_DEPTH;
    regmatch_state *yes_state = NULL; /* state to pop to on success of
							    subpattern */
    /* mark_state piggy backs on the yes_state logic so that when we unwind 
       the stack on success we can update the mark_state as we go */
    regmatch_state *mark_state = NULL; /* last mark state we have seen */
    regmatch_state *cur_eval = NULL; /* most recent EVAL_AB state */
    struct regmatch_state  *cur_curlyx = NULL; /* most recent curlyx */
    U32 state_num;
    bool no_final = 0;      /* prevent failure from backtracking? */
    bool do_cutgroup = 0;   /* no_final only until next branch/trie entry */
    char *startpoint = locinput;
    SV *popmark = NULL;     /* are we looking for a mark? */
    SV *sv_commit = NULL;   /* last mark name seen in failure */
    SV *sv_yes_mark = NULL; /* last mark name we have seen 
                               during a successful match */
    U32 lastopen = 0;       /* last open we saw */
    bool has_cutgroup = RX_HAS_CUTGROUP(rex) ? 1 : 0;   
    SV* const oreplsv = GvSVn(PL_replgv);
    /* these three flags are set by various ops to signal information to
     * the very next op. They have a useful lifetime of exactly one loop
     * iteration, and are not preserved or restored by state pushes/pops
     */
    bool sw = 0;	    /* the condition value in (?(cond)a|b) */
    bool minmod = 0;	    /* the next "{n,m}" is a "{n,m}?" */
    int logical = 0;	    /* the following EVAL is:
				0: (?{...})
				1: (?(?{...})X|Y)
				2: (??{...})
			       or the following IFMATCH/UNLESSM is:
			        false: plain (?=foo)
				true:  used as a condition: (?(?=foo))
			    */
    PAD* last_pad = NULL;
    dMULTICALL;
    U8 gimme = G_SCALAR;
    CV *caller_cv = NULL;	/* who called us */
    CV *last_pushed_cv = NULL;	/* most recently called (?{}) CV */
    CHECKPOINT runops_cp;	/* savestack position before executing EVAL */
    U32 maxopenparen = 0;       /* max '(' index seen so far */
    int to_complement;  /* Invert the result? */
    _char_class_number classnum;
    bool is_utf8_pat = reginfo->is_utf8_pat;
    bool match = FALSE;

/* Solaris Studio 12.3 messes up fetching PL_charclass['\n'] */
#if (defined(__SUNPRO_C) && (__SUNPRO_C == 0x5120) && defined(__x86_64) && defined(USE_64_BIT_ALL))
#  define SOLARIS_BAD_OPTIMIZER
    const U32 *pl_charclass_dup = PL_charclass;
#  define PL_charclass pl_charclass_dup
#endif

#ifdef DEBUGGING
    GET_RE_DEBUG_FLAGS_DECL;
#endif

    /* protect against undef(*^R) */
    SAVEFREESV(SvREFCNT_inc_simple_NN(oreplsv));

    /* shut up 'may be used uninitialized' compiler warnings for dMULTICALL */
    multicall_oldcatch = 0;
    PERL_UNUSED_VAR(multicall_cop);

    PERL_ARGS_ASSERT_REGMATCH;

    DEBUG_OPTIMISE_r( DEBUG_EXECUTE_r({
            Perl_re_printf( aTHX_ "regmatch start\n");
    }));

    st = PL_regmatch_state;

    /* Note that nextchr is a byte even in UTF */
    SET_nextchr;
    scan = prog;
    while (scan != NULL) {


	next = scan + NEXT_OFF(scan);
	if (next == scan)
	    next = NULL;
	state_num = OP(scan);

      reenter_switch:
        DEBUG_EXECUTE_r(
            if (state_num <= REGNODE_MAX) {
                SV * const prop = sv_newmortal();
                regnode *rnext = regnext(scan);

                DUMP_EXEC_POS( locinput, scan, utf8_target, depth );
                regprop(rex, prop, scan, reginfo, NULL);
                Perl_re_printf( aTHX_
                    "%*s%"IVdf":%s(%"IVdf")\n",
                    INDENT_CHARS(depth), "",
                    (IV)(scan - rexi->program),
                    SvPVX_const(prop),
                    (PL_regkind[OP(scan)] == END || !rnext) ?
                        0 : (IV)(rnext - rexi->program));
            }
        );

        to_complement = 0;

        SET_nextchr;
        assert(nextchr < 256 && (nextchr >= 0 || nextchr == NEXTCHR_EOS));

	switch (state_num) {
	case SBOL: /*  /^../ and /\A../  */
	    if (locinput == reginfo->strbeg)
		break;
	    sayNO;

	case MBOL: /*  /^../m  */
	    if (locinput == reginfo->strbeg ||
		(!NEXTCHR_IS_EOS && locinput[-1] == '\n'))
	    {
		break;
	    }
	    sayNO;

	case GPOS: /*  \G  */
	    if (locinput == reginfo->ganch)
		break;
	    sayNO;

	case KEEPS: /*   \K  */
	    /* update the startpoint */
	    st->u.keeper.val = rex->offs[0].start;
	    rex->offs[0].start = locinput - reginfo->strbeg;
	    PUSH_STATE_GOTO(KEEPS_next, next, locinput);
	    NOT_REACHED; /* NOTREACHED */

	case KEEPS_next_fail:
	    /* rollback the start point change */
	    rex->offs[0].start = st->u.keeper.val;
	    sayNO_SILENT;
	    NOT_REACHED; /* NOTREACHED */

	case MEOL: /* /..$/m  */
	    if (!NEXTCHR_IS_EOS && nextchr != '\n')
		sayNO;
	    break;

	case SEOL: /* /..$/  */
	    if (!NEXTCHR_IS_EOS && nextchr != '\n')
		sayNO;
	    if (reginfo->strend - locinput > 1)
		sayNO;
	    break;

	case EOS: /*  \z  */
	    if (!NEXTCHR_IS_EOS)
		sayNO;
	    break;

	case SANY: /*  /./s  */
	    if (NEXTCHR_IS_EOS)
		sayNO;
            goto increment_locinput;

	case REG_ANY: /*  /./  */
	    if ((NEXTCHR_IS_EOS) || nextchr == '\n')
		sayNO;
            goto increment_locinput;


#undef  ST
#define ST st->u.trie
        case TRIEC: /* (ab|cd) with known charclass */
            /* In this case the charclass data is available inline so
               we can fail fast without a lot of extra overhead. 
             */
            if(!NEXTCHR_IS_EOS && !ANYOF_BITMAP_TEST(scan, nextchr)) {
                DEBUG_EXECUTE_r(
                    Perl_re_exec_indentf( aTHX_  "%sfailed to match trie start class...%s\n",
                              depth, PL_colors[4], PL_colors[5])
                );
                sayNO_SILENT;
                NOT_REACHED; /* NOTREACHED */
            }
            /* FALLTHROUGH */
	case TRIE:  /* (ab|cd)  */
	    /* the basic plan of execution of the trie is:
	     * At the beginning, run though all the states, and
	     * find the longest-matching word. Also remember the position
	     * of the shortest matching word. For example, this pattern:
	     *    1  2 3 4    5
	     *    ab|a|x|abcd|abc
	     * when matched against the string "abcde", will generate
	     * accept states for all words except 3, with the longest
	     * matching word being 4, and the shortest being 2 (with
	     * the position being after char 1 of the string).
	     *
	     * Then for each matching word, in word order (i.e. 1,2,4,5),
	     * we run the remainder of the pattern; on each try setting
	     * the current position to the character following the word,
	     * returning to try the next word on failure.
	     *
	     * We avoid having to build a list of words at runtime by
	     * using a compile-time structure, wordinfo[].prev, which
	     * gives, for each word, the previous accepting word (if any).
	     * In the case above it would contain the mappings 1->2, 2->0,
	     * 3->0, 4->5, 5->1.  We can use this table to generate, from
	     * the longest word (4 above), a list of all words, by
	     * following the list of prev pointers; this gives us the
	     * unordered list 4,5,1,2. Then given the current word we have
	     * just tried, we can go through the list and find the
	     * next-biggest word to try (so if we just failed on word 2,
	     * the next in the list is 4).
	     *
	     * Since at runtime we don't record the matching position in
	     * the string for each word, we have to work that out for
	     * each word we're about to process. The wordinfo table holds
	     * the character length of each word; given that we recorded
	     * at the start: the position of the shortest word and its
	     * length in chars, we just need to move the pointer the
	     * difference between the two char lengths. Depending on
	     * Unicode status and folding, that's cheap or expensive.
	     *
	     * This algorithm is optimised for the case where are only a
	     * small number of accept states, i.e. 0,1, or maybe 2.
	     * With lots of accepts states, and having to try all of them,
	     * it becomes quadratic on number of accept states to find all
	     * the next words.
	     */

	    {
                /* what type of TRIE am I? (utf8 makes this contextual) */
                DECL_TRIE_TYPE(scan);

                /* what trie are we using right now */
		reg_trie_data * const trie
        	    = (reg_trie_data*)rexi->data->data[ ARG( scan ) ];
		HV * widecharmap = MUTABLE_HV(rexi->data->data[ ARG( scan ) + 1 ]);
                U32 state = trie->startstate;

                if (scan->flags == EXACTL || scan->flags == EXACTFLU8) {
                    _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
                    if (utf8_target
                        && UTF8_IS_ABOVE_LATIN1(nextchr)
                        && scan->flags == EXACTL)
                    {
                        /* We only output for EXACTL, as we let the folder
                         * output this message for EXACTFLU8 to avoid
                         * duplication */
                        _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(locinput,
                                                               reginfo->strend);
                    }
                }
                if (   trie->bitmap
                    && (NEXTCHR_IS_EOS || !TRIE_BITMAP_TEST(trie, nextchr)))
                {
        	    if (trie->states[ state ].wordnum) {
        	         DEBUG_EXECUTE_r(
                            Perl_re_exec_indentf( aTHX_  "%smatched empty string...%s\n",
                                          depth, PL_colors[4], PL_colors[5])
                        );
			if (!trie->jump)
			    break;
        	    } else {
        	        DEBUG_EXECUTE_r(
                            Perl_re_exec_indentf( aTHX_  "%sfailed to match trie start class...%s\n",
                                          depth, PL_colors[4], PL_colors[5])
                        );
        	        sayNO_SILENT;
        	   }
                }

            { 
		U8 *uc = ( U8* )locinput;

		STRLEN len = 0;
		STRLEN foldlen = 0;
		U8 *uscan = (U8*)NULL;
		U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
		U32 charcount = 0; /* how many input chars we have matched */
		U32 accepted = 0; /* have we seen any accepting states? */

		ST.jump = trie->jump;
		ST.me = scan;
		ST.firstpos = NULL;
		ST.longfold = FALSE; /* char longer if folded => it's harder */
		ST.nextword = 0;

		/* fully traverse the TRIE; note the position of the
		   shortest accept state and the wordnum of the longest
		   accept state */

		while ( state && uc <= (U8*)(reginfo->strend) ) {
                    U32 base = trie->states[ state ].trans.base;
                    UV uvc = 0;
                    U16 charid = 0;
		    U16 wordnum;
                    wordnum = trie->states[ state ].wordnum;

		    if (wordnum) { /* it's an accept state */
			if (!accepted) {
			    accepted = 1;
			    /* record first match position */
			    if (ST.longfold) {
				ST.firstpos = (U8*)locinput;
				ST.firstchars = 0;
			    }
			    else {
				ST.firstpos = uc;
				ST.firstchars = charcount;
			    }
			}
			if (!ST.nextword || wordnum < ST.nextword)
			    ST.nextword = wordnum;
			ST.topword = wordnum;
		    }

		    DEBUG_TRIE_EXECUTE_r({
                                DUMP_EXEC_POS( (char *)uc, scan, utf8_target, depth );
                                Perl_re_exec_indentf( aTHX_
                                    "%sState: %4"UVxf" Accepted: %c ",
                                    depth, PL_colors[4],
			            (UV)state, (accepted ? 'Y' : 'N'));
		    });

		    /* read a char and goto next state */
		    if ( base && (foldlen || uc < (U8*)(reginfo->strend))) {
			I32 offset;
			REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc,
                                             (U8 *) reginfo->strend, uscan,
                                             len, uvc, charid, foldlen,
                                             foldbuf, uniflags);
			charcount++;
			if (foldlen>0)
			    ST.longfold = TRUE;
			if (charid &&
			     ( ((offset =
			      base + charid - 1 - trie->uniquecharcount)) >= 0)

			     && ((U32)offset < trie->lasttrans)
			     && trie->trans[offset].check == state)
			{
			    state = trie->trans[offset].next;
			}
			else {
			    state = 0;
			}
			uc += len;

		    }
		    else {
			state = 0;
		    }
		    DEBUG_TRIE_EXECUTE_r(
                        Perl_re_printf( aTHX_
		            "Charid:%3x CP:%4"UVxf" After State: %4"UVxf"%s\n",
		            charid, uvc, (UV)state, PL_colors[5] );
		    );
		}
		if (!accepted)
		   sayNO;

		/* calculate total number of accept states */
		{
		    U16 w = ST.topword;
		    accepted = 0;
		    while (w) {
			w = trie->wordinfo[w].prev;
			accepted++;
		    }
		    ST.accepted = accepted;
		}

		DEBUG_EXECUTE_r(
                    Perl_re_exec_indentf( aTHX_  "%sgot %"IVdf" possible matches%s\n",
                        depth,
			PL_colors[4], (IV)ST.accepted, PL_colors[5] );
		);
		goto trie_first_try; /* jump into the fail handler */
	    }}
	    NOT_REACHED; /* NOTREACHED */

	case TRIE_next_fail: /* we failed - try next alternative */
        {
            U8 *uc;
            if ( ST.jump) {
                REGCP_UNWIND(ST.cp);
                UNWIND_PAREN(ST.lastparen, ST.lastcloseparen);
	    }
	    if (!--ST.accepted) {
	        DEBUG_EXECUTE_r({
                    Perl_re_exec_indentf( aTHX_  "%sTRIE failed...%s\n",
                        depth,
			PL_colors[4],
			PL_colors[5] );
		});
		sayNO_SILENT;
	    }
	    {
		/* Find next-highest word to process.  Note that this code
		 * is O(N^2) per trie run (O(N) per branch), so keep tight */
		U16 min = 0;
		U16 word;
		U16 const nextword = ST.nextword;
		reg_trie_wordinfo * const wordinfo
		    = ((reg_trie_data*)rexi->data->data[ARG(ST.me)])->wordinfo;
		for (word=ST.topword; word; word=wordinfo[word].prev) {
		    if (word > nextword && (!min || word < min))
			min = word;
		}
		ST.nextword = min;
	    }

          trie_first_try:
            if (do_cutgroup) {
                do_cutgroup = 0;
                no_final = 0;
            }

            if ( ST.jump) {
                ST.lastparen = rex->lastparen;
                ST.lastcloseparen = rex->lastcloseparen;
	        REGCP_SET(ST.cp);
            }

	    /* find start char of end of current word */
	    {
		U32 chars; /* how many chars to skip */
		reg_trie_data * const trie
		    = (reg_trie_data*)rexi->data->data[ARG(ST.me)];

		assert((trie->wordinfo[ST.nextword].len - trie->prefixlen)
			    >=  ST.firstchars);
		chars = (trie->wordinfo[ST.nextword].len - trie->prefixlen)
			    - ST.firstchars;
		uc = ST.firstpos;

		if (ST.longfold) {
		    /* the hard option - fold each char in turn and find
		     * its folded length (which may be different */
		    U8 foldbuf[UTF8_MAXBYTES_CASE + 1];
		    STRLEN foldlen;
		    STRLEN len;
		    UV uvc;
		    U8 *uscan;

		    while (chars) {
			if (utf8_target) {
			    uvc = utf8n_to_uvchr((U8*)uc, UTF8_MAXLEN, &len,
						    uniflags);
			    uc += len;
			}
			else {
			    uvc = *uc;
			    uc++;
			}
			uvc = to_uni_fold(uvc, foldbuf, &foldlen);
			uscan = foldbuf;
			while (foldlen) {
			    if (!--chars)
				break;
			    uvc = utf8n_to_uvchr(uscan, foldlen, &len,
                                                 uniflags);
			    uscan += len;
			    foldlen -= len;
			}
		    }
		}
		else {
		    if (utf8_target)
			while (chars--)
			    uc += UTF8SKIP(uc);
		    else
			uc += chars;
		}
	    }

	    scan = ST.me + ((ST.jump && ST.jump[ST.nextword])
			    ? ST.jump[ST.nextword]
			    : NEXT_OFF(ST.me));

	    DEBUG_EXECUTE_r({
                Perl_re_exec_indentf( aTHX_  "%sTRIE matched word #%d, continuing%s\n",
                    depth,
		    PL_colors[4],
		    ST.nextword,
		    PL_colors[5]
		    );
	    });

	    if (ST.accepted > 1 || has_cutgroup) {
		PUSH_STATE_GOTO(TRIE_next, scan, (char*)uc);
		NOT_REACHED; /* NOTREACHED */
	    }
	    /* only one choice left - just continue */
	    DEBUG_EXECUTE_r({
		AV *const trie_words
		    = MUTABLE_AV(rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET]);
		SV ** const tmp = trie_words
                        ? av_fetch(trie_words, ST.nextword - 1, 0) : NULL;
		SV *sv= tmp ? sv_newmortal() : NULL;

                Perl_re_exec_indentf( aTHX_  "%sonly one match left, short-circuiting: #%d <%s>%s\n",
                    depth, PL_colors[4],
		    ST.nextword,
		    tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0,
			    PL_colors[0], PL_colors[1],
			    (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0)|PERL_PV_ESCAPE_NONASCII
			) 
		    : "not compiled under -Dr",
		    PL_colors[5] );
	    });

	    locinput = (char*)uc;
	    continue; /* execute rest of RE */
            /* NOTREACHED */
        }
#undef  ST

	case EXACTL:             /*  /abc/l       */
            _CHECK_AND_WARN_PROBLEMATIC_LOCALE;

            /* Complete checking would involve going through every character
             * matched by the string to see if any is above latin1.  But the
             * comparision otherwise might very well be a fast assembly
             * language routine, and I (khw) don't think slowing things down
             * just to check for this warning is worth it.  So this just checks
             * the first character */
            if (utf8_target && UTF8_IS_ABOVE_LATIN1(*locinput)) {
                _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(locinput, reginfo->strend);
            }
            /* FALLTHROUGH */
	case EXACT: {            /*  /abc/        */
	    char *s = STRING(scan);
	    ln = STR_LEN(scan);
	    if (utf8_target != is_utf8_pat) {
		/* The target and the pattern have differing utf8ness. */
		char *l = locinput;
		const char * const e = s + ln;

		if (utf8_target) {
                    /* The target is utf8, the pattern is not utf8.
                     * Above-Latin1 code points can't match the pattern;
                     * invariants match exactly, and the other Latin1 ones need
                     * to be downgraded to a single byte in order to do the
                     * comparison.  (If we could be confident that the target
                     * is not malformed, this could be refactored to have fewer
                     * tests by just assuming that if the first bytes match, it
                     * is an invariant, but there are tests in the test suite
                     * dealing with (??{...}) which violate this) */
		    while (s < e) {
			if (l >= reginfo->strend
                            || UTF8_IS_ABOVE_LATIN1(* (U8*) l))
                        {
                            sayNO;
                        }
                        if (UTF8_IS_INVARIANT(*(U8*)l)) {
			    if (*l != *s) {
                                sayNO;
                            }
                            l++;
                        }
                        else {
                            if (EIGHT_BIT_UTF8_TO_NATIVE(*l, *(l+1)) != * (U8*) s)
                            {
                                sayNO;
                            }
                            l += 2;
                        }
			s++;
		    }
		}
		else {
		    /* The target is not utf8, the pattern is utf8. */
		    while (s < e) {
                        if (l >= reginfo->strend
                            || UTF8_IS_ABOVE_LATIN1(* (U8*) s))
                        {
                            sayNO;
                        }
                        if (UTF8_IS_INVARIANT(*(U8*)s)) {
			    if (*s != *l) {
                                sayNO;
                            }
                            s++;
                        }
                        else {
                            if (EIGHT_BIT_UTF8_TO_NATIVE(*s, *(s+1)) != * (U8*) l)
                            {
                                sayNO;
                            }
                            s += 2;
                        }
			l++;
		    }
		}
		locinput = l;
	    }
            else {
                /* The target and the pattern have the same utf8ness. */
                /* Inline the first character, for speed. */
                if (reginfo->strend - locinput < ln
                    || UCHARAT(s) != nextchr
                    || (ln > 1 && memNE(s, locinput, ln)))
                {
                    sayNO;
                }
                locinput += ln;
            }
	    break;
	    }

	case EXACTFL: {          /*  /abc/il      */
	    re_fold_t folder;
	    const U8 * fold_array;
	    const char * s;
	    U32 fold_utf8_flags;

            _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
            folder = foldEQ_locale;
            fold_array = PL_fold_locale;
	    fold_utf8_flags = FOLDEQ_LOCALE;
	    goto do_exactf;

        case EXACTFLU8:           /*  /abc/il; but all 'abc' are above 255, so
                                      is effectively /u; hence to match, target
                                      must be UTF-8. */
            if (! utf8_target) {
                sayNO;
            }
            fold_utf8_flags =  FOLDEQ_LOCALE | FOLDEQ_S1_ALREADY_FOLDED
                                             | FOLDEQ_S1_FOLDS_SANE;
	    folder = foldEQ_latin1;
	    fold_array = PL_fold_latin1;
	    goto do_exactf;

	case EXACTFU_SS:         /*  /\x{df}/iu   */
	case EXACTFU:            /*  /abc/iu      */
	    folder = foldEQ_latin1;
	    fold_array = PL_fold_latin1;
	    fold_utf8_flags = is_utf8_pat ? FOLDEQ_S1_ALREADY_FOLDED : 0;
	    goto do_exactf;

        case EXACTFA_NO_TRIE:   /* This node only generated for non-utf8
                                   patterns */
            assert(! is_utf8_pat);
            /* FALLTHROUGH */
	case EXACTFA:            /*  /abc/iaa     */
	    folder = foldEQ_latin1;
	    fold_array = PL_fold_latin1;
	    fold_utf8_flags = FOLDEQ_UTF8_NOMIX_ASCII;
	    goto do_exactf;

        case EXACTF:             /*  /abc/i    This node only generated for
                                               non-utf8 patterns */
            assert(! is_utf8_pat);
	    folder = foldEQ;
	    fold_array = PL_fold;
	    fold_utf8_flags = 0;

	  do_exactf:
	    s = STRING(scan);
	    ln = STR_LEN(scan);

	    if (utf8_target
                || is_utf8_pat
                || state_num == EXACTFU_SS
                || (state_num == EXACTFL && IN_UTF8_CTYPE_LOCALE))
            {
	      /* Either target or the pattern are utf8, or has the issue where
	       * the fold lengths may differ. */
		const char * const l = locinput;
		char *e = reginfo->strend;

		if (! foldEQ_utf8_flags(s, 0,  ln, is_utf8_pat,
			                l, &e, 0,  utf8_target, fold_utf8_flags))
		{
		    sayNO;
		}
		locinput = e;
		break;
	    }

	    /* Neither the target nor the pattern are utf8 */
	    if (UCHARAT(s) != nextchr
                && !NEXTCHR_IS_EOS
		&& UCHARAT(s) != fold_array[nextchr])
	    {
		sayNO;
	    }
	    if (reginfo->strend - locinput < ln)
		sayNO;
	    if (ln > 1 && ! folder(s, locinput, ln))
		sayNO;
	    locinput += ln;
	    break;
	}

	case NBOUNDL: /*  /\B/l  */
            to_complement = 1;
            /* FALLTHROUGH */

	case BOUNDL:  /*  /\b/l  */
        {
            bool b1, b2;
            _CHECK_AND_WARN_PROBLEMATIC_LOCALE;

            if (FLAGS(scan) != TRADITIONAL_BOUND) {
                if (! IN_UTF8_CTYPE_LOCALE) {
                    Perl_ck_warner(aTHX_ packWARN(WARN_LOCALE),
                                                B_ON_NON_UTF8_LOCALE_IS_WRONG);
                }
                goto boundu;
            }

	    if (utf8_target) {
		if (locinput == reginfo->strbeg)
		    b1 = isWORDCHAR_LC('\n');
		else {
                    b1 = isWORDCHAR_LC_utf8(reghop3((U8*)locinput, -1,
                                                        (U8*)(reginfo->strbeg)));
		}
                b2 = (NEXTCHR_IS_EOS)
                    ? isWORDCHAR_LC('\n')
                    : isWORDCHAR_LC_utf8((U8*)locinput);
	    }
	    else { /* Here the string isn't utf8 */
		b1 = (locinput == reginfo->strbeg)
                     ? isWORDCHAR_LC('\n')
                     : isWORDCHAR_LC(UCHARAT(locinput - 1));
                b2 = (NEXTCHR_IS_EOS)
                    ? isWORDCHAR_LC('\n')
                    : isWORDCHAR_LC(nextchr);
	    }
            if (to_complement ^ (b1 == b2)) {
                sayNO;
            }
	    break;
        }

	case NBOUND:  /*  /\B/   */
            to_complement = 1;
            /* FALLTHROUGH */

	case BOUND:   /*  /\b/   */
	    if (utf8_target) {
                goto bound_utf8;
            }
            goto bound_ascii_match_only;

	case NBOUNDA: /*  /\B/a  */
            to_complement = 1;
            /* FALLTHROUGH */

	case BOUNDA:  /*  /\b/a  */
        {
            bool b1, b2;

          bound_ascii_match_only:
            /* Here the string isn't utf8, or is utf8 and only ascii characters
             * are to match \w.  In the latter case looking at the byte just
             * prior to the current one may be just the final byte of a
             * multi-byte character.  This is ok.  There are two cases:
             * 1) it is a single byte character, and then the test is doing
             *    just what it's supposed to.
             * 2) it is a multi-byte character, in which case the final byte is
             *    never mistakable for ASCII, and so the test will say it is
             *    not a word character, which is the correct answer. */
            b1 = (locinput == reginfo->strbeg)
                 ? isWORDCHAR_A('\n')
                 : isWORDCHAR_A(UCHARAT(locinput - 1));
            b2 = (NEXTCHR_IS_EOS)
                ? isWORDCHAR_A('\n')
                : isWORDCHAR_A(nextchr);
            if (to_complement ^ (b1 == b2)) {
                sayNO;
            }
	    break;
        }

	case NBOUNDU: /*  /\B/u  */
            to_complement = 1;
            /* FALLTHROUGH */

	case BOUNDU:  /*  /\b/u  */

          boundu:
            if (UNLIKELY(reginfo->strbeg >= reginfo->strend)) {
                match = FALSE;
            }
            else if (utf8_target) {
              bound_utf8:
                switch((bound_type) FLAGS(scan)) {
                    case TRADITIONAL_BOUND:
                    {
                        bool b1, b2;
                        b1 = (locinput == reginfo->strbeg)
                             ? 0 /* isWORDCHAR_L1('\n') */
                             : isWORDCHAR_utf8(reghop3((U8*)locinput, -1,
                                                                (U8*)(reginfo->strbeg)));
                        b2 = (NEXTCHR_IS_EOS)
                            ? 0 /* isWORDCHAR_L1('\n') */
                            : isWORDCHAR_utf8((U8*)locinput);
                        match = cBOOL(b1 != b2);
                        break;
                    }
                    case GCB_BOUND:
                        if (locinput == reginfo->strbeg || NEXTCHR_IS_EOS) {
                            match = TRUE; /* GCB always matches at begin and
                                             end */
                        }
                        else {
                            /* Find the gcb values of previous and current
                             * chars, then see if is a break point */
                            match = isGCB(getGCB_VAL_UTF8(
                                                reghop3((U8*)locinput,
                                                        -1,
                                                        (U8*)(reginfo->strbeg)),
                                                (U8*) reginfo->strend),
                                          getGCB_VAL_UTF8((U8*) locinput,
                                                        (U8*) reginfo->strend));
                        }
                        break;

                    case LB_BOUND:
                        if (locinput == reginfo->strbeg) {
                            match = FALSE;
                        }
                        else if (NEXTCHR_IS_EOS) {
                            match = TRUE;
                        }
                        else {
                            match = isLB(getLB_VAL_UTF8(
                                                reghop3((U8*)locinput,
                                                        -1,
                                                        (U8*)(reginfo->strbeg)),
                                                (U8*) reginfo->strend),
                                          getLB_VAL_UTF8((U8*) locinput,
                                                        (U8*) reginfo->strend),
                                          (U8*) reginfo->strbeg,
                                          (U8*) locinput,
                                          (U8*) reginfo->strend,
                                          utf8_target);
                        }
                        break;

                    case SB_BOUND: /* Always matches at begin and end */
                        if (locinput == reginfo->strbeg || NEXTCHR_IS_EOS) {
                            match = TRUE;
                        }
                        else {
                            match = isSB(getSB_VAL_UTF8(
                                                reghop3((U8*)locinput,
                                                        -1,
                                                        (U8*)(reginfo->strbeg)),
                                                (U8*) reginfo->strend),
                                          getSB_VAL_UTF8((U8*) locinput,
                                                        (U8*) reginfo->strend),
                                          (U8*) reginfo->strbeg,
                                          (U8*) locinput,
                                          (U8*) reginfo->strend,
                                          utf8_target);
                        }
                        break;

                    case WB_BOUND:
                        if (locinput == reginfo->strbeg || NEXTCHR_IS_EOS) {
                            match = TRUE;
                        }
                        else {
                            match = isWB(WB_UNKNOWN,
                                         getWB_VAL_UTF8(
                                                reghop3((U8*)locinput,
                                                        -1,
                                                        (U8*)(reginfo->strbeg)),
                                                (U8*) reginfo->strend),
                                          getWB_VAL_UTF8((U8*) locinput,
                                                        (U8*) reginfo->strend),
                                          (U8*) reginfo->strbeg,
                                          (U8*) locinput,
                                          (U8*) reginfo->strend,
                                          utf8_target);
                        }
                        break;
                }
	    }
	    else {  /* Not utf8 target */
                switch((bound_type) FLAGS(scan)) {
                    case TRADITIONAL_BOUND:
                    {
                        bool b1, b2;
                        b1 = (locinput == reginfo->strbeg)
                            ? 0 /* isWORDCHAR_L1('\n') */
                            : isWORDCHAR_L1(UCHARAT(locinput - 1));
                        b2 = (NEXTCHR_IS_EOS)
                            ? 0 /* isWORDCHAR_L1('\n') */
                            : isWORDCHAR_L1(nextchr);
                        match = cBOOL(b1 != b2);
                        break;
                    }

                    case GCB_BOUND:
                        if (locinput == reginfo->strbeg || NEXTCHR_IS_EOS) {
                            match = TRUE; /* GCB always matches at begin and
                                             end */
                        }
                        else {  /* Only CR-LF combo isn't a GCB in 0-255
                                   range */
                            match =    UCHARAT(locinput - 1) != '\r'
                                    || UCHARAT(locinput) != '\n';
                        }
                        break;

                    case LB_BOUND:
                        if (locinput == reginfo->strbeg) {
                            match = FALSE;
                        }
                        else if (NEXTCHR_IS_EOS) {
                            match = TRUE;
                        }
                        else {
                            match = isLB(getLB_VAL_CP(UCHARAT(locinput -1)),
                                         getLB_VAL_CP(UCHARAT(locinput)),
                                         (U8*) reginfo->strbeg,
                                         (U8*) locinput,
                                         (U8*) reginfo->strend,
                                         utf8_target);
                        }
                        break;

                    case SB_BOUND: /* Always matches at begin and end */
                        if (locinput == reginfo->strbeg || NEXTCHR_IS_EOS) {
                            match = TRUE;
                        }
                        else {
                            match = isSB(getSB_VAL_CP(UCHARAT(locinput -1)),
                                         getSB_VAL_CP(UCHARAT(locinput)),
                                         (U8*) reginfo->strbeg,
                                         (U8*) locinput,
                                         (U8*) reginfo->strend,
                                         utf8_target);
                        }
                        break;

                    case WB_BOUND:
                        if (locinput == reginfo->strbeg || NEXTCHR_IS_EOS) {
                            match = TRUE;
                        }
                        else {
                            match = isWB(WB_UNKNOWN,
                                         getWB_VAL_CP(UCHARAT(locinput -1)),
                                         getWB_VAL_CP(UCHARAT(locinput)),
                                         (U8*) reginfo->strbeg,
                                         (U8*) locinput,
                                         (U8*) reginfo->strend,
                                         utf8_target);
                        }
                        break;
                }
	    }

            if (to_complement ^ ! match) {
                sayNO;
            }
	    break;

	case ANYOFL:  /*  /[abc]/l      */
            _CHECK_AND_WARN_PROBLEMATIC_LOCALE;

            if (ANYOFL_UTF8_LOCALE_REQD(FLAGS(scan)) && ! IN_UTF8_CTYPE_LOCALE)
            {
              Perl_ck_warner(aTHX_ packWARN(WARN_LOCALE), utf8_locale_required);
            }
            /* FALLTHROUGH */
	case ANYOFD:  /*   /[abc]/d       */
	case ANYOF:  /*   /[abc]/       */
            if (NEXTCHR_IS_EOS)
                sayNO;
	    if (utf8_target && ! UTF8_IS_INVARIANT(*locinput)) {
	        if (!reginclass(rex, scan, (U8*)locinput, (U8*)reginfo->strend,
                                                                   utf8_target))
		    sayNO;
		locinput += UTF8SKIP(locinput);
	    }
	    else {
		if (!REGINCLASS(rex, scan, (U8*)locinput, utf8_target))
		    sayNO;
		locinput++;
	    }
	    break;

        /* The argument (FLAGS) to all the POSIX node types is the class number
         * */

        case NPOSIXL:   /* \W or [:^punct:] etc. under /l */
            to_complement = 1;
            /* FALLTHROUGH */

        case POSIXL:    /* \w or [:punct:] etc. under /l */
            _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
            if (NEXTCHR_IS_EOS)
                sayNO;

            /* Use isFOO_lc() for characters within Latin1.  (Note that
             * UTF8_IS_INVARIANT works even on non-UTF-8 strings, or else
             * wouldn't be invariant) */
            if (UTF8_IS_INVARIANT(nextchr) || ! utf8_target) {
                if (! (to_complement ^ cBOOL(isFOO_lc(FLAGS(scan), (U8) nextchr)))) {
                    sayNO;
                }
            }
            else if (UTF8_IS_DOWNGRADEABLE_START(nextchr)) {
                if (! (to_complement ^ cBOOL(isFOO_lc(FLAGS(scan),
                                               EIGHT_BIT_UTF8_TO_NATIVE(nextchr,
                                               *(locinput + 1))))))
                {
                    sayNO;
                }
            }
            else { /* Here, must be an above Latin-1 code point */
                _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(locinput, reginfo->strend);
                goto utf8_posix_above_latin1;
            }

            /* Here, must be utf8 */
            locinput += UTF8SKIP(locinput);
            break;

        case NPOSIXD:   /* \W or [:^punct:] etc. under /d */
            to_complement = 1;
            /* FALLTHROUGH */

        case POSIXD:    /* \w or [:punct:] etc. under /d */
            if (utf8_target) {
                goto utf8_posix;
            }
            goto posixa;

        case NPOSIXA:   /* \W or [:^punct:] etc. under /a */

            if (NEXTCHR_IS_EOS) {
                sayNO;
            }

            /* All UTF-8 variants match */
            if (! UTF8_IS_INVARIANT(nextchr)) {
                goto increment_locinput;
            }

            to_complement = 1;
            goto join_nposixa;

        case POSIXA:    /* \w or [:punct:] etc. under /a */

          posixa:
            /* We get here through POSIXD, NPOSIXD, and NPOSIXA when not in
             * UTF-8, and also from NPOSIXA even in UTF-8 when the current
             * character is a single byte */

            if (NEXTCHR_IS_EOS) {
                sayNO;
            }

          join_nposixa:

            if (! (to_complement ^ cBOOL(_generic_isCC_A(nextchr,
                                                                FLAGS(scan)))))
            {
                sayNO;
            }

            /* Here we are either not in utf8, or we matched a utf8-invariant,
             * so the next char is the next byte */
            locinput++;
            break;

        case NPOSIXU:   /* \W or [:^punct:] etc. under /u */
            to_complement = 1;
            /* FALLTHROUGH */

        case POSIXU:    /* \w or [:punct:] etc. under /u */
          utf8_posix:
            if (NEXTCHR_IS_EOS) {
                sayNO;
            }

            /* Use _generic_isCC() for characters within Latin1.  (Note that
             * UTF8_IS_INVARIANT works even on non-UTF-8 strings, or else
             * wouldn't be invariant) */
            if (UTF8_IS_INVARIANT(nextchr) || ! utf8_target) {
                if (! (to_complement ^ cBOOL(_generic_isCC(nextchr,
                                                           FLAGS(scan)))))
                {
                    sayNO;
                }
                locinput++;
            }
            else if (UTF8_IS_DOWNGRADEABLE_START(nextchr)) {
                if (! (to_complement
                       ^ cBOOL(_generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(nextchr,
                                                               *(locinput + 1)),
                                             FLAGS(scan)))))
                {
                    sayNO;
                }
                locinput += 2;
            }
            else {  /* Handle above Latin-1 code points */
              utf8_posix_above_latin1:
                classnum = (_char_class_number) FLAGS(scan);
                if (classnum < _FIRST_NON_SWASH_CC) {

                    /* Here, uses a swash to find such code points.  Load if if
                     * not done already */
                    if (! PL_utf8_swash_ptrs[classnum]) {
                        U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;
                        PL_utf8_swash_ptrs[classnum]
                                = _core_swash_init("utf8",
                                        "",
                                        &PL_sv_undef, 1, 0,
                                        PL_XPosix_ptrs[classnum], &flags);
                    }
                    if (! (to_complement
                           ^ cBOOL(swash_fetch(PL_utf8_swash_ptrs[classnum],
                                               (U8 *) locinput, TRUE))))
                    {
                        sayNO;
                    }
                }
                else {  /* Here, uses macros to find above Latin-1 code points */
                    switch (classnum) {
                        case _CC_ENUM_SPACE:
                            if (! (to_complement
                                        ^ cBOOL(is_XPERLSPACE_high(locinput))))
                            {
                                sayNO;
                            }
                            break;
                        case _CC_ENUM_BLANK:
                            if (! (to_complement
                                            ^ cBOOL(is_HORIZWS_high(locinput))))
                            {
                                sayNO;
                            }
                            break;
                        case _CC_ENUM_XDIGIT:
                            if (! (to_complement
                                            ^ cBOOL(is_XDIGIT_high(locinput))))
                            {
                                sayNO;
                            }
                            break;
                        case _CC_ENUM_VERTSPACE:
                            if (! (to_complement
                                            ^ cBOOL(is_VERTWS_high(locinput))))
                            {
                                sayNO;
                            }
                            break;
                        default:    /* The rest, e.g. [:cntrl:], can't match
                                       above Latin1 */
                            if (! to_complement) {
                                sayNO;
                            }
                            break;
                    }
                }
                locinput += UTF8SKIP(locinput);
            }
            break;

	case CLUMP: /* Match \X: logical Unicode character.  This is defined as
		       a Unicode extended Grapheme Cluster */
	    if (NEXTCHR_IS_EOS)
		sayNO;
	    if  (! utf8_target) {

		/* Match either CR LF  or '.', as all the other possibilities
		 * require utf8 */
		locinput++;	    /* Match the . or CR */
		if (nextchr == '\r' /* And if it was CR, and the next is LF,
				       match the LF */
		    && locinput < reginfo->strend
		    && UCHARAT(locinput) == '\n')
                {
                    locinput++;
                }
	    }
	    else {

                /* Get the gcb type for the current character */
                GCB_enum prev_gcb = getGCB_VAL_UTF8((U8*) locinput,
                                                       (U8*) reginfo->strend);

                /* Then scan through the input until we get to the first
                 * character whose type is supposed to be a gcb with the
                 * current character.  (There is always a break at the
                 * end-of-input) */
                locinput += UTF8SKIP(locinput);
                while (locinput < reginfo->strend) {
                    GCB_enum cur_gcb = getGCB_VAL_UTF8((U8*) locinput,
                                                         (U8*) reginfo->strend);
                    if (isGCB(prev_gcb, cur_gcb)) {
                        break;
                    }

                    prev_gcb = cur_gcb;
                    locinput += UTF8SKIP(locinput);
                }


	    }
	    break;
            
	case NREFFL:  /*  /\g{name}/il  */
	{   /* The capture buffer cases.  The ones beginning with N for the
	       named buffers just convert to the equivalent numbered and
	       pretend they were called as the corresponding numbered buffer
	       op.  */
	    /* don't initialize these in the declaration, it makes C++
	       unhappy */
	    const char *s;
	    char type;
	    re_fold_t folder;
	    const U8 *fold_array;
	    UV utf8_fold_flags;

            _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
	    folder = foldEQ_locale;
	    fold_array = PL_fold_locale;
	    type = REFFL;
	    utf8_fold_flags = FOLDEQ_LOCALE;
	    goto do_nref;

	case NREFFA:  /*  /\g{name}/iaa  */
	    folder = foldEQ_latin1;
	    fold_array = PL_fold_latin1;
	    type = REFFA;
	    utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII;
	    goto do_nref;

	case NREFFU:  /*  /\g{name}/iu  */
	    folder = foldEQ_latin1;
	    fold_array = PL_fold_latin1;
	    type = REFFU;
	    utf8_fold_flags = 0;
	    goto do_nref;

	case NREFF:  /*  /\g{name}/i  */
	    folder = foldEQ;
	    fold_array = PL_fold;
	    type = REFF;
	    utf8_fold_flags = 0;
	    goto do_nref;

	case NREF:  /*  /\g{name}/   */
	    type = REF;
	    folder = NULL;
	    fold_array = NULL;
	    utf8_fold_flags = 0;
	  do_nref:

	    /* For the named back references, find the corresponding buffer
	     * number */
	    n = reg_check_named_buff_matched(rex,scan);

            if ( ! n ) {
                sayNO;
	    }
	    goto do_nref_ref_common;

	case REFFL:  /*  /\1/il  */
            _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
	    folder = foldEQ_locale;
	    fold_array = PL_fold_locale;
	    utf8_fold_flags = FOLDEQ_LOCALE;
	    goto do_ref;

	case REFFA:  /*  /\1/iaa  */
	    folder = foldEQ_latin1;
	    fold_array = PL_fold_latin1;
	    utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII;
	    goto do_ref;

	case REFFU:  /*  /\1/iu  */
	    folder = foldEQ_latin1;
	    fold_array = PL_fold_latin1;
	    utf8_fold_flags = 0;
	    goto do_ref;

	case REFF:  /*  /\1/i  */
	    folder = foldEQ;
	    fold_array = PL_fold;
	    utf8_fold_flags = 0;
	    goto do_ref;

        case REF:  /*  /\1/    */
	    folder = NULL;
	    fold_array = NULL;
	    utf8_fold_flags = 0;

	  do_ref:
	    type = OP(scan);
	    n = ARG(scan);  /* which paren pair */

	  do_nref_ref_common:
	    ln = rex->offs[n].start;
	    reginfo->poscache_iter = reginfo->poscache_maxiter; /* Void cache */
	    if (rex->lastparen < n || ln == -1)
		sayNO;			/* Do not match unless seen CLOSEn. */
	    if (ln == rex->offs[n].end)
		break;

	    s = reginfo->strbeg + ln;
	    if (type != REF	/* REF can do byte comparison */
		&& (utf8_target || type == REFFU || type == REFFL))
	    {
		char * limit = reginfo->strend;

		/* This call case insensitively compares the entire buffer
		    * at s, with the current input starting at locinput, but
                    * not going off the end given by reginfo->strend, and
                    * returns in <limit> upon success, how much of the
                    * current input was matched */
		if (! foldEQ_utf8_flags(s, NULL, rex->offs[n].end - ln, utf8_target,
				    locinput, &limit, 0, utf8_target, utf8_fold_flags))
		{
		    sayNO;
		}
		locinput = limit;
		break;
	    }

	    /* Not utf8:  Inline the first character, for speed. */
	    if (!NEXTCHR_IS_EOS &&
                UCHARAT(s) != nextchr &&
		(type == REF ||
		 UCHARAT(s) != fold_array[nextchr]))
		sayNO;
	    ln = rex->offs[n].end - ln;
	    if (locinput + ln > reginfo->strend)
		sayNO;
	    if (ln > 1 && (type == REF
			   ? memNE(s, locinput, ln)
			   : ! folder(s, locinput, ln)))
		sayNO;
	    locinput += ln;
	    break;
	}

	case NOTHING: /* null op; e.g. the 'nothing' following
                       * the '*' in m{(a+|b)*}' */
	    break;
	case TAIL: /* placeholder while compiling (A|B|C) */
	    break;

#undef  ST
#define ST st->u.eval
#define CUR_EVAL cur_eval->u.eval

	{
	    SV *ret;
	    REGEXP *re_sv;
            regexp *re;
            regexp_internal *rei;
            regnode *startpoint;
            U32 arg;

	case GOSUB: /*    /(...(?1))/   /(...(?&foo))/   */
            arg= (U32)ARG(scan);
            if (cur_eval && cur_eval->locinput == locinput) {
                if ( ++nochange_depth > max_nochange_depth )
                    Perl_croak(aTHX_ 
                        "Pattern subroutine nesting without pos change"
                        " exceeded limit in regex");
            } else {
                nochange_depth = 0;
            }
	    re_sv = rex_sv;
            re = rex;
            rei = rexi;
            startpoint = scan + ARG2L(scan);
            EVAL_CLOSE_PAREN_SET( st, arg );
            /* Detect infinite recursion
             *
             * A pattern like /(?R)foo/ or /(?<x>(?&y)foo)(?<y>(?&x)bar)/
             * or "a"=~/(.(?2))((?<=(?=(?1)).))/ could recurse forever.
             * So we track the position in the string we are at each time
             * we recurse and if we try to enter the same routine twice from
             * the same position we throw an error.
             */
            if ( rex->recurse_locinput[arg] == locinput ) {
                /* FIXME: we should show the regop that is failing as part
                 * of the error message. */
                Perl_croak(aTHX_ "Infinite recursion in regex");
            } else {
                ST.prev_recurse_locinput= rex->recurse_locinput[arg];
                rex->recurse_locinput[arg]= locinput;

                DEBUG_r({
                    GET_RE_DEBUG_FLAGS_DECL;
                    DEBUG_STACK_r({
                        Perl_re_exec_indentf( aTHX_
                            "entering GOSUB, prev_recurse_locinput=%p recurse_locinput[%d]=%p\n",
                            depth, ST.prev_recurse_locinput, arg, rex->recurse_locinput[arg]
                        );
                    });
                });
            }

            /* Save all the positions seen so far. */
            ST.cp = regcppush(rex, 0, maxopenparen);
            REGCP_SET(ST.lastcp);

            /* and then jump to the code we share with EVAL */
            goto eval_recurse_doit;
            /* NOTREACHED */

        case EVAL:  /*   /(?{A})B/   /(??{A})B/  and /(?(?{A})X|Y)B/   */        
            if (cur_eval && cur_eval->locinput==locinput) {
		if ( ++nochange_depth > max_nochange_depth )
                    Perl_croak(aTHX_ "EVAL without pos change exceeded limit in regex");
            } else {
                nochange_depth = 0;
            }    
	    {
		/* execute the code in the {...} */

		dSP;
		IV before;
		OP * const oop = PL_op;
		COP * const ocurcop = PL_curcop;
		OP *nop;
		CV *newcv;

		/* save *all* paren positions */
		regcppush(rex, 0, maxopenparen);
		REGCP_SET(runops_cp);

		if (!caller_cv)
		    caller_cv = find_runcv(NULL);

		n = ARG(scan);

		if (rexi->data->what[n] == 'r') { /* code from an external qr */
                    newcv = (ReANY(
                                    (REGEXP*)(rexi->data->data[n])
                            ))->qr_anoncv;
		    nop = (OP*)rexi->data->data[n+1];
		}
		else if (rexi->data->what[n] == 'l') { /* literal code */
		    newcv = caller_cv;
		    nop = (OP*)rexi->data->data[n];
		    assert(CvDEPTH(newcv));
		}
		else {
		    /* literal with own CV */
		    assert(rexi->data->what[n] == 'L');
		    newcv = rex->qr_anoncv;
		    nop = (OP*)rexi->data->data[n];
		}

		/* normally if we're about to execute code from the same
		 * CV that we used previously, we just use the existing
		 * CX stack entry. However, its possible that in the
		 * meantime we may have backtracked, popped from the save
		 * stack, and undone the SAVECOMPPAD(s) associated with
		 * PUSH_MULTICALL; in which case PL_comppad no longer
		 * points to newcv's pad. */
		if (newcv != last_pushed_cv || PL_comppad != last_pad)
		{
                    U8 flags = (CXp_SUB_RE |
                                ((newcv == caller_cv) ? CXp_SUB_RE_FAKE : 0));
		    if (last_pushed_cv) {
                        /* PUSH/POP_MULTICALL save and restore the
                         * caller's PL_comppad; if we call multiple subs
                         * using the same CX block, we have to save and
                         * unwind the varying PL_comppad's ourselves,
                         * especially restoring the right PL_comppad on
                         * backtrack - so save it on the save stack */
                        SAVECOMPPAD();
			CHANGE_MULTICALL_FLAGS(newcv, flags);
		    }
		    else {
			PUSH_MULTICALL_FLAGS(newcv, flags);
		    }
		    last_pushed_cv = newcv;
		}
		else {
                    /* these assignments are just to silence compiler
                     * warnings */
		    multicall_cop = NULL;
		}
		last_pad = PL_comppad;

		/* the initial nextstate you would normally execute
		 * at the start of an eval (which would cause error
		 * messages to come from the eval), may be optimised
		 * away from the execution path in the regex code blocks;
		 * so manually set PL_curcop to it initially */
		{
		    OP *o = cUNOPx(nop)->op_first;
		    assert(o->op_type == OP_NULL);
		    if (o->op_targ == OP_SCOPE) {
			o = cUNOPo->op_first;
		    }
		    else {
			assert(o->op_targ == OP_LEAVE);
			o = cUNOPo->op_first;
			assert(o->op_type == OP_ENTER);
			o = OpSIBLING(o);
		    }

		    if (o->op_type != OP_STUB) {
			assert(    o->op_type == OP_NEXTSTATE
				|| o->op_type == OP_DBSTATE
				|| (o->op_type == OP_NULL
				    &&  (  o->op_targ == OP_NEXTSTATE
					|| o->op_targ == OP_DBSTATE
					)
				    )
			);
			PL_curcop = (COP*)o;
		    }
		}
		nop = nop->op_next;

                DEBUG_STATE_r( Perl_re_printf( aTHX_
		    "  re EVAL PL_op=0x%"UVxf"\n", PTR2UV(nop)) );

		rex->offs[0].end = locinput - reginfo->strbeg;
                if (reginfo->info_aux_eval->pos_magic)
                    MgBYTEPOS_set(reginfo->info_aux_eval->pos_magic,
                                  reginfo->sv, reginfo->strbeg,
                                  locinput - reginfo->strbeg);

                if (sv_yes_mark) {
                    SV *sv_mrk = get_sv("REGMARK", 1);
                    sv_setsv(sv_mrk, sv_yes_mark);
                }

		/* we don't use MULTICALL here as we want to call the
		 * first op of the block of interest, rather than the
		 * first op of the sub. Also, we don't want to free
                 * the savestack frame */
		before = (IV)(SP-PL_stack_base);
		PL_op = nop;
		CALLRUNOPS(aTHX);			/* Scalar context. */
		SPAGAIN;
		if ((IV)(SP-PL_stack_base) == before)
		    ret = &PL_sv_undef;   /* protect against empty (?{}) blocks. */
		else {
		    ret = POPs;
		    PUTBACK;
		}

		/* before restoring everything, evaluate the returned
		 * value, so that 'uninit' warnings don't use the wrong
		 * PL_op or pad. Also need to process any magic vars
		 * (e.g. $1) *before* parentheses are restored */

		PL_op = NULL;

                re_sv = NULL;
		if (logical == 0)        /*   (?{})/   */
		    sv_setsv(save_scalar(PL_replgv), ret); /* $^R */
		else if (logical == 1) { /*   /(?(?{...})X|Y)/    */
		    sw = cBOOL(SvTRUE(ret));
		    logical = 0;
		}
		else {                   /*  /(??{})  */
		    /*  if its overloaded, let the regex compiler handle
		     *  it; otherwise extract regex, or stringify  */
		    if (SvGMAGICAL(ret))
			ret = sv_mortalcopy(ret);
		    if (!SvAMAGIC(ret)) {
			SV *sv = ret;
			if (SvROK(sv))
			    sv = SvRV(sv);
			if (SvTYPE(sv) == SVt_REGEXP)
			    re_sv = (REGEXP*) sv;
			else if (SvSMAGICAL(ret)) {
			    MAGIC *mg = mg_find(ret, PERL_MAGIC_qr);
			    if (mg)
				re_sv = (REGEXP *) mg->mg_obj;
			}

			/* force any undef warnings here */
			if (!re_sv && !SvPOK(ret) && !SvNIOK(ret)) {
			    ret = sv_mortalcopy(ret);
			    (void) SvPV_force_nolen(ret);
			}
		    }

		}

		/* *** Note that at this point we don't restore
		 * PL_comppad, (or pop the CxSUB) on the assumption it may
		 * be used again soon. This is safe as long as nothing
		 * in the regexp code uses the pad ! */
		PL_op = oop;
		PL_curcop = ocurcop;
		S_regcp_restore(aTHX_ rex, runops_cp, &maxopenparen);
		PL_curpm = PL_reg_curpm;

		if (logical != 2)
		    break;
	    }

		/* only /(??{})/  from now on */
		logical = 0;
		{
		    /* extract RE object from returned value; compiling if
		     * necessary */

		    if (re_sv) {
			re_sv = reg_temp_copy(NULL, re_sv);
		    }
		    else {
			U32 pm_flags = 0;

			if (SvUTF8(ret) && IN_BYTES) {
			    /* In use 'bytes': make a copy of the octet
			     * sequence, but without the flag on */
			    STRLEN len;
			    const char *const p = SvPV(ret, len);
			    ret = newSVpvn_flags(p, len, SVs_TEMP);
			}
			if (rex->intflags & PREGf_USE_RE_EVAL)
			    pm_flags |= PMf_USE_RE_EVAL;

			/* if we got here, it should be an engine which
			 * supports compiling code blocks and stuff */
			assert(rex->engine && rex->engine->op_comp);
                        assert(!(scan->flags & ~RXf_PMf_COMPILETIME));
			re_sv = rex->engine->op_comp(aTHX_ &ret, 1, NULL,
				    rex->engine, NULL, NULL,
                                    /* copy /msixn etc to inner pattern */
                                    ARG2L(scan),
                                    pm_flags);

			if (!(SvFLAGS(ret)
			      & (SVs_TEMP | SVs_GMG | SVf_ROK))
			 && (!SvPADTMP(ret) || SvREADONLY(ret))) {
			    /* This isn't a first class regexp. Instead, it's
			       caching a regexp onto an existing, Perl visible
			       scalar.  */
			    sv_magic(ret, MUTABLE_SV(re_sv), PERL_MAGIC_qr, 0, 0);
			}
		    }
		    SAVEFREESV(re_sv);
		    re = ReANY(re_sv);
		}
                RXp_MATCH_COPIED_off(re);
                re->subbeg = rex->subbeg;
                re->sublen = rex->sublen;
                re->suboffset = rex->suboffset;
                re->subcoffset = rex->subcoffset;
                re->lastparen = 0;
                re->lastcloseparen = 0;
		rei = RXi_GET(re);
                DEBUG_EXECUTE_r(
                    debug_start_match(re_sv, utf8_target, locinput,
                                    reginfo->strend, "Matching embedded");
		);		
		startpoint = rei->program + 1;
                EVAL_CLOSE_PAREN_CLEAR(st); /* ST.close_paren = 0;
                                             * close_paren only for GOSUB */
                ST.prev_recurse_locinput= NULL; /* only used for GOSUB */
                /* Save all the seen positions so far. */
                ST.cp = regcppush(rex, 0, maxopenparen);
                REGCP_SET(ST.lastcp);
                /* and set maxopenparen to 0, since we are starting a "fresh" match */
                maxopenparen = 0;
                /* run the pattern returned from (??{...}) */

              eval_recurse_doit: /* Share code with GOSUB below this line
                            * At this point we expect the stack context to be
                            * set up correctly */

                /* invalidate the S-L poscache. We're now executing a
                 * different set of WHILEM ops (and their associated
                 * indexes) against the same string, so the bits in the
                 * cache are meaningless. Setting maxiter to zero forces
                 * the cache to be invalidated and zeroed before reuse.
		 * XXX This is too dramatic a measure. Ideally we should
                 * save the old cache and restore when running the outer
                 * pattern again */
		reginfo->poscache_maxiter = 0;

                /* the new regexp might have a different is_utf8_pat than we do */
                is_utf8_pat = reginfo->is_utf8_pat = cBOOL(RX_UTF8(re_sv));

		ST.prev_rex = rex_sv;
		ST.prev_curlyx = cur_curlyx;
		rex_sv = re_sv;
		SET_reg_curpm(rex_sv);
		rex = re;
		rexi = rei;
		cur_curlyx = NULL;
		ST.B = next;
		ST.prev_eval = cur_eval;
		cur_eval = st;
		/* now continue from first node in postoned RE */
		PUSH_YES_STATE_GOTO(EVAL_AB, startpoint, locinput);
		NOT_REACHED; /* NOTREACHED */
	}

	case EVAL_AB: /* cleanup after a successful (??{A})B */
            /* note: this is called twice; first after popping B, then A */
            DEBUG_STACK_r({
                Perl_re_exec_indentf( aTHX_  "EVAL_AB cur_eval=%p prev_eval=%p\n",
                    depth, cur_eval, ST.prev_eval);
            });

#define SET_RECURSE_LOCINPUT(STR,VAL)\
            if ( cur_eval && CUR_EVAL.close_paren ) {\
                DEBUG_STACK_r({ \
                    Perl_re_exec_indentf( aTHX_  STR " GOSUB%d ce=%p recurse_locinput=%p\n",\
                        depth,    \
                        CUR_EVAL.close_paren - 1,\
                        cur_eval, \
                        VAL);     \
                });               \
                rex->recurse_locinput[CUR_EVAL.close_paren - 1] = VAL;\
            }

            SET_RECURSE_LOCINPUT("EVAL_AB[before]", CUR_EVAL.prev_recurse_locinput);

	    rex_sv = ST.prev_rex;
            is_utf8_pat = reginfo->is_utf8_pat = cBOOL(RX_UTF8(rex_sv));
	    SET_reg_curpm(rex_sv);
	    rex = ReANY(rex_sv);
	    rexi = RXi_GET(rex);
            {
                /* preserve $^R across LEAVE's. See Bug 121070. */
                SV *save_sv= GvSV(PL_replgv);
                SvREFCNT_inc(save_sv);
                regcpblow(ST.cp); /* LEAVE in disguise */
                sv_setsv(GvSV(PL_replgv), save_sv);
                SvREFCNT_dec(save_sv);
            }
	    cur_eval = ST.prev_eval;
	    cur_curlyx = ST.prev_curlyx;

	    /* Invalidate cache. See "invalidate" comment above. */
	    reginfo->poscache_maxiter = 0;
            if ( nochange_depth )
	        nochange_depth--;

            SET_RECURSE_LOCINPUT("EVAL_AB[after]", cur_eval->locinput);
	    sayYES;


	case EVAL_AB_fail: /* unsuccessfully ran A or B in (??{A})B */
	    /* note: this is called twice; first after popping B, then A */
            DEBUG_STACK_r({
                Perl_re_exec_indentf( aTHX_  "EVAL_AB_fail cur_eval=%p prev_eval=%p\n",
                    depth, cur_eval, ST.prev_eval);
            });

            SET_RECURSE_LOCINPUT("EVAL_AB_fail[before]", CUR_EVAL.prev_recurse_locinput);

	    rex_sv = ST.prev_rex;
            is_utf8_pat = reginfo->is_utf8_pat = cBOOL(RX_UTF8(rex_sv));
	    SET_reg_curpm(rex_sv);
	    rex = ReANY(rex_sv);
	    rexi = RXi_GET(rex); 

	    REGCP_UNWIND(ST.lastcp);
	    regcppop(rex, &maxopenparen);
	    cur_eval = ST.prev_eval;
	    cur_curlyx = ST.prev_curlyx;

	    /* Invalidate cache. See "invalidate" comment above. */
	    reginfo->poscache_maxiter = 0;
	    if ( nochange_depth )
	        nochange_depth--;

            SET_RECURSE_LOCINPUT("EVAL_AB_fail[after]", cur_eval->locinput);
            sayNO_SILENT;
#undef ST

	case OPEN: /*  (  */
	    n = ARG(scan);  /* which paren pair */
	    rex->offs[n].start_tmp = locinput - reginfo->strbeg;
	    if (n > maxopenparen)
		maxopenparen = n;
            DEBUG_BUFFERS_r(Perl_re_printf( aTHX_
		"rex=0x%"UVxf" offs=0x%"UVxf": \\%"UVuf": set %"IVdf" tmp; maxopenparen=%"UVuf"\n",
		PTR2UV(rex),
		PTR2UV(rex->offs),
		(UV)n,
		(IV)rex->offs[n].start_tmp,
		(UV)maxopenparen
	    ));
            lastopen = n;
	    break;

/* XXX really need to log other places start/end are set too */
#define CLOSE_CAPTURE                                                      \
    rex->offs[n].start = rex->offs[n].start_tmp;                           \
    rex->offs[n].end = locinput - reginfo->strbeg;                         \
    DEBUG_BUFFERS_r(Perl_re_printf( aTHX_                                              \
        "rex=0x%"UVxf" offs=0x%"UVxf": \\%"UVuf": set %"IVdf"..%"IVdf"\n", \
        PTR2UV(rex),                                                       \
        PTR2UV(rex->offs),                                                 \
        (UV)n,                                                             \
        (IV)rex->offs[n].start,                                            \
        (IV)rex->offs[n].end                                               \
    ))

	case CLOSE:  /*  )  */
	    n = ARG(scan);  /* which paren pair */
	    CLOSE_CAPTURE;
	    if (n > rex->lastparen)
		rex->lastparen = n;
	    rex->lastcloseparen = n;
            if ( EVAL_CLOSE_PAREN_IS( cur_eval, n ) )
	        goto fake_end;

	    break;

        case ACCEPT:  /*  (*ACCEPT)  */
            if (scan->flags)
                sv_yes_mark = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
            if (ARG2L(scan)){
                regnode *cursor;
                for (cursor=scan;
                     cursor && OP(cursor)!=END; 
                     cursor=regnext(cursor)) 
                {
                    if ( OP(cursor)==CLOSE ){
                        n = ARG(cursor);
                        if ( n <= lastopen ) {
			    CLOSE_CAPTURE;
                            if (n > rex->lastparen)
                                rex->lastparen = n;
                            rex->lastcloseparen = n;
                            if ( n == ARG(scan) || EVAL_CLOSE_PAREN_IS(cur_eval, n) )
                                break;
                        }
                    }
                }
            }
	    goto fake_end;
	    /* NOTREACHED */

	case GROUPP:  /*  (?(1))  */
	    n = ARG(scan);  /* which paren pair */
	    sw = cBOOL(rex->lastparen >= n && rex->offs[n].end != -1);
	    break;

	case NGROUPP:  /*  (?(<name>))  */
	    /* reg_check_named_buff_matched returns 0 for no match */
	    sw = cBOOL(0 < reg_check_named_buff_matched(rex,scan));
	    break;

        case INSUBP:   /*  (?(R))  */
            n = ARG(scan);
            /* this does not need to use EVAL_CLOSE_PAREN macros, as the arg
             * of SCAN is already set up as matches a eval.close_paren */
            sw = cur_eval && (n == 0 || CUR_EVAL.close_paren == n);
            break;

        case DEFINEP:  /*  (?(DEFINE))  */
            sw = 0;
            break;

	case IFTHEN:   /*  (?(cond)A|B)  */
	    reginfo->poscache_iter = reginfo->poscache_maxiter; /* Void cache */
	    if (sw)
		next = NEXTOPER(NEXTOPER(scan));
	    else {
		next = scan + ARG(scan);
		if (OP(next) == IFTHEN) /* Fake one. */
		    next = NEXTOPER(NEXTOPER(next));
	    }
	    break;

	case LOGICAL:  /* modifier for EVAL and IFMATCH */
	    logical = scan->flags;
	    break;

/*******************************************************************

The CURLYX/WHILEM pair of ops handle the most generic case of the /A*B/
pattern, where A and B are subpatterns. (For simple A, CURLYM or
STAR/PLUS/CURLY/CURLYN are used instead.)

A*B is compiled as <CURLYX><A><WHILEM><B>

On entry to the subpattern, CURLYX is called. This pushes a CURLYX
state, which contains the current count, initialised to -1. It also sets
cur_curlyx to point to this state, with any previous value saved in the
state block.

CURLYX then jumps straight to the WHILEM op, rather than executing A,
since the pattern may possibly match zero times (i.e. it's a while {} loop
rather than a do {} while loop).

Each entry to WHILEM represents a successful match of A. The count in the
CURLYX block is incremented, another WHILEM state is pushed, and execution
passes to A or B depending on greediness and the current count.

For example, if matching against the string a1a2a3b (where the aN are
substrings that match /A/), then the match progresses as follows: (the
pushed states are interspersed with the bits of strings matched so far):

    <CURLYX cnt=-1>
    <CURLYX cnt=0><WHILEM>
    <CURLYX cnt=1><WHILEM> a1 <WHILEM>
    <CURLYX cnt=2><WHILEM> a1 <WHILEM> a2 <WHILEM>
    <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM>
    <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM> b

(Contrast this with something like CURLYM, which maintains only a single
backtrack state:

    <CURLYM cnt=0> a1
    a1 <CURLYM cnt=1> a2
    a1 a2 <CURLYM cnt=2> a3
    a1 a2 a3 <CURLYM cnt=3> b
)

Each WHILEM state block marks a point to backtrack to upon partial failure
of A or B, and also contains some minor state data related to that
iteration.  The CURLYX block, pointed to by cur_curlyx, contains the
overall state, such as the count, and pointers to the A and B ops.

This is complicated slightly by nested CURLYX/WHILEM's. Since cur_curlyx
must always point to the *current* CURLYX block, the rules are:

When executing CURLYX, save the old cur_curlyx in the CURLYX state block,
and set cur_curlyx to point the new block.

When popping the CURLYX block after a successful or unsuccessful match,
restore the previous cur_curlyx.

When WHILEM is about to execute B, save the current cur_curlyx, and set it
to the outer one saved in the CURLYX block.

When popping the WHILEM block after a successful or unsuccessful B match,
restore the previous cur_curlyx.

Here's an example for the pattern (AI* BI)*BO
I and O refer to inner and outer, C and W refer to CURLYX and WHILEM:

cur_
curlyx backtrack stack
------ ---------------
NULL   
CO     <CO prev=NULL> <WO>
CI     <CO prev=NULL> <WO> <CI prev=CO> <WI> ai 
CO     <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi 
NULL   <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi <WO prev=CO> bo

At this point the pattern succeeds, and we work back down the stack to
clean up, restoring as we go:

CO     <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi 
CI     <CO prev=NULL> <WO> <CI prev=CO> <WI> ai 
CO     <CO prev=NULL> <WO>
NULL   

*******************************************************************/

#define ST st->u.curlyx

	case CURLYX:    /* start of /A*B/  (for complex A) */
	{
	    /* No need to save/restore up to this paren */
	    I32 parenfloor = scan->flags;
	    
	    assert(next); /* keep Coverity happy */
	    if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */
		next += ARG(next);

	    /* XXXX Probably it is better to teach regpush to support
	       parenfloor > maxopenparen ... */
	    if (parenfloor > (I32)rex->lastparen)
		parenfloor = rex->lastparen; /* Pessimization... */

	    ST.prev_curlyx= cur_curlyx;
	    cur_curlyx = st;
	    ST.cp = PL_savestack_ix;

	    /* these fields contain the state of the current curly.
	     * they are accessed by subsequent WHILEMs */
	    ST.parenfloor = parenfloor;
	    ST.me = scan;
	    ST.B = next;
	    ST.minmod = minmod;
	    minmod = 0;
	    ST.count = -1;	/* this will be updated by WHILEM */
	    ST.lastloc = NULL;  /* this will be updated by WHILEM */

	    PUSH_YES_STATE_GOTO(CURLYX_end, PREVOPER(next), locinput);
	    NOT_REACHED; /* NOTREACHED */
	}

	case CURLYX_end: /* just finished matching all of A*B */
	    cur_curlyx = ST.prev_curlyx;
	    sayYES;
	    NOT_REACHED; /* NOTREACHED */

	case CURLYX_end_fail: /* just failed to match all of A*B */
	    regcpblow(ST.cp);
	    cur_curlyx = ST.prev_curlyx;
	    sayNO;
	    NOT_REACHED; /* NOTREACHED */


#undef ST
#define ST st->u.whilem

	case WHILEM:     /* just matched an A in /A*B/  (for complex A) */
	{
	    /* see the discussion above about CURLYX/WHILEM */
	    I32 n;
	    int min, max;
	    regnode *A;

	    assert(cur_curlyx); /* keep Coverity happy */

	    min = ARG1(cur_curlyx->u.curlyx.me);
	    max = ARG2(cur_curlyx->u.curlyx.me);
	    A = NEXTOPER(cur_curlyx->u.curlyx.me) + EXTRA_STEP_2ARGS;
	    n = ++cur_curlyx->u.curlyx.count; /* how many A's matched */
	    ST.save_lastloc = cur_curlyx->u.curlyx.lastloc;
	    ST.cache_offset = 0;
	    ST.cache_mask = 0;
	    

            DEBUG_EXECUTE_r( Perl_re_exec_indentf( aTHX_  "whilem: matched %ld out of %d..%d\n",
                  depth, (long)n, min, max)
	    );

	    /* First just match a string of min A's. */

	    if (n < min) {
		ST.cp = regcppush(rex, cur_curlyx->u.curlyx.parenfloor,
                                    maxopenparen);
		cur_curlyx->u.curlyx.lastloc = locinput;
		REGCP_SET(ST.lastcp);

		PUSH_STATE_GOTO(WHILEM_A_pre, A, locinput);
		NOT_REACHED; /* NOTREACHED */
	    }

	    /* If degenerate A matches "", assume A done. */

	    if (locinput == cur_curlyx->u.curlyx.lastloc) {
                DEBUG_EXECUTE_r( Perl_re_exec_indentf( aTHX_  "whilem: empty match detected, trying continuation...\n",
                   depth)
		);
		goto do_whilem_B_max;
	    }

	    /* super-linear cache processing.
             *
             * The idea here is that for certain types of CURLYX/WHILEM -
             * principally those whose upper bound is infinity (and
             * excluding regexes that have things like \1 and other very
             * non-regular expresssiony things), then if a pattern like
             * /....A*.../ fails and we backtrack to the WHILEM, then we
             * make a note that this particular WHILEM op was at string
             * position 47 (say) when the rest of pattern failed. Then, if
             * we ever find ourselves back at that WHILEM, and at string
             * position 47 again, we can just fail immediately rather than
             * running the rest of the pattern again.
             *
             * This is very handy when patterns start to go
             * 'super-linear', like in (a+)*(a+)*(a+)*, where you end up
             * with a combinatorial explosion of backtracking.
             *
             * The cache is implemented as a bit array, with one bit per
             * string byte position per WHILEM op (up to 16) - so its
             * between 0.25 and 2x the string size.
             *
             * To avoid allocating a poscache buffer every time, we do an
             * initially countdown; only after we have  executed a WHILEM
             * op (string-length x #WHILEMs) times do we allocate the
             * cache.
             *
             * The top 4 bits of scan->flags byte say how many different
             * relevant CURLLYX/WHILEM op pairs there are, while the
             * bottom 4-bits is the identifying index number of this
             * WHILEM.
             */

	    if (scan->flags) {

		if (!reginfo->poscache_maxiter) {
		    /* start the countdown: Postpone detection until we
		     * know the match is not *that* much linear. */
		    reginfo->poscache_maxiter
                        =    (reginfo->strend - reginfo->strbeg + 1)
                           * (scan->flags>>4);
		    /* possible overflow for long strings and many CURLYX's */
		    if (reginfo->poscache_maxiter < 0)
			reginfo->poscache_maxiter = I32_MAX;
		    reginfo->poscache_iter = reginfo->poscache_maxiter;
		}

		if (reginfo->poscache_iter-- == 0) {
		    /* initialise cache */
		    const SSize_t size = (reginfo->poscache_maxiter + 7)/8;
                    regmatch_info_aux *const aux = reginfo->info_aux;
		    if (aux->poscache) {
			if ((SSize_t)reginfo->poscache_size < size) {
			    Renew(aux->poscache, size, char);
			    reginfo->poscache_size = size;
			}
			Zero(aux->poscache, size, char);
		    }
		    else {
			reginfo->poscache_size = size;
			Newxz(aux->poscache, size, char);
		    }
                    DEBUG_EXECUTE_r( Perl_re_printf( aTHX_
      "%swhilem: Detected a super-linear match, switching on caching%s...\n",
			      PL_colors[4], PL_colors[5])
		    );
		}

		if (reginfo->poscache_iter < 0) {
		    /* have we already failed at this position? */
		    SSize_t offset, mask;

                    reginfo->poscache_iter = -1; /* stop eventual underflow */
		    offset  = (scan->flags & 0xf) - 1
                                +   (locinput - reginfo->strbeg)
                                  * (scan->flags>>4);
		    mask    = 1 << (offset % 8);
		    offset /= 8;
		    if (reginfo->info_aux->poscache[offset] & mask) {
                        DEBUG_EXECUTE_r( Perl_re_exec_indentf( aTHX_  "whilem: (cache) already tried at this position...\n",
                            depth)
			);
			sayNO; /* cache records failure */
		    }
		    ST.cache_offset = offset;
		    ST.cache_mask   = mask;
		}
	    }

	    /* Prefer B over A for minimal matching. */

	    if (cur_curlyx->u.curlyx.minmod) {
		ST.save_curlyx = cur_curlyx;
		cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx;
		ST.cp = regcppush(rex, ST.save_curlyx->u.curlyx.parenfloor,
                            maxopenparen);
		REGCP_SET(ST.lastcp);
		PUSH_YES_STATE_GOTO(WHILEM_B_min, ST.save_curlyx->u.curlyx.B,
                                    locinput);
		NOT_REACHED; /* NOTREACHED */
	    }

	    /* Prefer A over B for maximal matching. */

	    if (n < max) { /* More greed allowed? */
		ST.cp = regcppush(rex, cur_curlyx->u.curlyx.parenfloor,
                            maxopenparen);
		cur_curlyx->u.curlyx.lastloc = locinput;
		REGCP_SET(ST.lastcp);
		PUSH_STATE_GOTO(WHILEM_A_max, A, locinput);
		NOT_REACHED; /* NOTREACHED */
	    }
	    goto do_whilem_B_max;
	}
	NOT_REACHED; /* NOTREACHED */

	case WHILEM_B_min: /* just matched B in a minimal match */
	case WHILEM_B_max: /* just matched B in a maximal match */
	    cur_curlyx = ST.save_curlyx;
	    sayYES;
	    NOT_REACHED; /* NOTREACHED */

	case WHILEM_B_max_fail: /* just failed to match B in a maximal match */
	    cur_curlyx = ST.save_curlyx;
	    cur_curlyx->u.curlyx.lastloc = ST.save_lastloc;
	    cur_curlyx->u.curlyx.count--;
	    CACHEsayNO;
	    NOT_REACHED; /* NOTREACHED */

	case WHILEM_A_min_fail: /* just failed to match A in a minimal match */
	    /* FALLTHROUGH */
	case WHILEM_A_pre_fail: /* just failed to match even minimal A */
	    REGCP_UNWIND(ST.lastcp);
	    regcppop(rex, &maxopenparen);
	    cur_curlyx->u.curlyx.lastloc = ST.save_lastloc;
	    cur_curlyx->u.curlyx.count--;
	    CACHEsayNO;
	    NOT_REACHED; /* NOTREACHED */

	case WHILEM_A_max_fail: /* just failed to match A in a maximal match */
	    REGCP_UNWIND(ST.lastcp);
	    regcppop(rex, &maxopenparen); /* Restore some previous $<digit>s? */
            DEBUG_EXECUTE_r(Perl_re_exec_indentf( aTHX_  "whilem: failed, trying continuation...\n",
                depth)
	    );
	  do_whilem_B_max:
	    if (cur_curlyx->u.curlyx.count >= REG_INFTY
		&& ckWARN(WARN_REGEXP)
		&& !reginfo->warned)
	    {
                reginfo->warned	= TRUE;
		Perl_warner(aTHX_ packWARN(WARN_REGEXP),
		     "Complex regular subexpression recursion limit (%d) "
		     "exceeded",
		     REG_INFTY - 1);
	    }

	    /* now try B */
	    ST.save_curlyx = cur_curlyx;
	    cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx;
	    PUSH_YES_STATE_GOTO(WHILEM_B_max, ST.save_curlyx->u.curlyx.B,
                                locinput);
	    NOT_REACHED; /* NOTREACHED */

	case WHILEM_B_min_fail: /* just failed to match B in a minimal match */
	    cur_curlyx = ST.save_curlyx;
	    REGCP_UNWIND(ST.lastcp);
	    regcppop(rex, &maxopenparen);

	    if (cur_curlyx->u.curlyx.count >= /*max*/ARG2(cur_curlyx->u.curlyx.me)) {
		/* Maximum greed exceeded */
		if (cur_curlyx->u.curlyx.count >= REG_INFTY
		    && ckWARN(WARN_REGEXP)
                    && !reginfo->warned)
		{
                    reginfo->warned	= TRUE;
		    Perl_warner(aTHX_ packWARN(WARN_REGEXP),
			"Complex regular subexpression recursion "
			"limit (%d) exceeded",
			REG_INFTY - 1);
		}
		cur_curlyx->u.curlyx.count--;
		CACHEsayNO;
	    }

            DEBUG_EXECUTE_r(Perl_re_exec_indentf( aTHX_  "trying longer...\n", depth)
	    );
	    /* Try grabbing another A and see if it helps. */
	    cur_curlyx->u.curlyx.lastloc = locinput;
	    ST.cp = regcppush(rex, cur_curlyx->u.curlyx.parenfloor,
                            maxopenparen);
	    REGCP_SET(ST.lastcp);
	    PUSH_STATE_GOTO(WHILEM_A_min,
		/*A*/ NEXTOPER(ST.save_curlyx->u.curlyx.me) + EXTRA_STEP_2ARGS,
                locinput);
	    NOT_REACHED; /* NOTREACHED */

#undef  ST
#define ST st->u.branch

	case BRANCHJ:	    /*  /(...|A|...)/ with long next pointer */
	    next = scan + ARG(scan);
	    if (next == scan)
		next = NULL;
	    scan = NEXTOPER(scan);
	    /* FALLTHROUGH */

	case BRANCH:	    /*  /(...|A|...)/ */
	    scan = NEXTOPER(scan); /* scan now points to inner node */
	    ST.lastparen = rex->lastparen;
	    ST.lastcloseparen = rex->lastcloseparen;
	    ST.next_branch = next;
	    REGCP_SET(ST.cp);

	    /* Now go into the branch */
	    if (has_cutgroup) {
	        PUSH_YES_STATE_GOTO(BRANCH_next, scan, locinput);
	    } else {
	        PUSH_STATE_GOTO(BRANCH_next, scan, locinput);
	    }
	    NOT_REACHED; /* NOTREACHED */

        case CUTGROUP:  /*  /(*THEN)/  */
            sv_yes_mark = st->u.mark.mark_name = scan->flags
                ? MUTABLE_SV(rexi->data->data[ ARG( scan ) ])
                : NULL;
            PUSH_STATE_GOTO(CUTGROUP_next, next, locinput);
            NOT_REACHED; /* NOTREACHED */

        case CUTGROUP_next_fail:
            do_cutgroup = 1;
            no_final = 1;
            if (st->u.mark.mark_name)
                sv_commit = st->u.mark.mark_name;
            sayNO;	    
            NOT_REACHED; /* NOTREACHED */

        case BRANCH_next:
            sayYES;
            NOT_REACHED; /* NOTREACHED */

	case BRANCH_next_fail: /* that branch failed; try the next, if any */
	    if (do_cutgroup) {
	        do_cutgroup = 0;
	        no_final = 0;
	    }
	    REGCP_UNWIND(ST.cp);
            UNWIND_PAREN(ST.lastparen, ST.lastcloseparen);
	    scan = ST.next_branch;
	    /* no more branches? */
	    if (!scan || (OP(scan) != BRANCH && OP(scan) != BRANCHJ)) {
	        DEBUG_EXECUTE_r({
                    Perl_re_exec_indentf( aTHX_  "%sBRANCH failed...%s\n",
                        depth,
			PL_colors[4],
			PL_colors[5] );
		});
		sayNO_SILENT;
            }
	    continue; /* execute next BRANCH[J] op */
            /* NOTREACHED */
    
	case MINMOD: /* next op will be non-greedy, e.g. A*?  */
	    minmod = 1;
	    break;

#undef  ST
#define ST st->u.curlym

	case CURLYM:	/* /A{m,n}B/ where A is fixed-length */

	    /* This is an optimisation of CURLYX that enables us to push
	     * only a single backtracking state, no matter how many matches
	     * there are in {m,n}. It relies on the pattern being constant
	     * length, with no parens to influence future backrefs
	     */

	    ST.me = scan;
	    scan = NEXTOPER(scan) + NODE_STEP_REGNODE;

	    ST.lastparen      = rex->lastparen;
	    ST.lastcloseparen = rex->lastcloseparen;

	    /* if paren positive, emulate an OPEN/CLOSE around A */
	    if (ST.me->flags) {
		U32 paren = ST.me->flags;
		if (paren > maxopenparen)
		    maxopenparen = paren;
		scan += NEXT_OFF(scan); /* Skip former OPEN. */
	    }
	    ST.A = scan;
	    ST.B = next;
	    ST.alen = 0;
	    ST.count = 0;
	    ST.minmod = minmod;
	    minmod = 0;
	    ST.c1 = CHRTEST_UNINIT;
	    REGCP_SET(ST.cp);

	    if (!(ST.minmod ? ARG1(ST.me) : ARG2(ST.me))) /* min/max */
		goto curlym_do_B;

	  curlym_do_A: /* execute the A in /A{m,n}B/  */
	    PUSH_YES_STATE_GOTO(CURLYM_A, ST.A, locinput); /* match A */
	    NOT_REACHED; /* NOTREACHED */

	case CURLYM_A: /* we've just matched an A */
	    ST.count++;
	    /* after first match, determine A's length: u.curlym.alen */
	    if (ST.count == 1) {
		if (reginfo->is_utf8_target) {
		    char *s = st->locinput;
		    while (s < locinput) {
			ST.alen++;
			s += UTF8SKIP(s);
		    }
		}
		else {
		    ST.alen = locinput - st->locinput;
		}
		if (ST.alen == 0)
		    ST.count = ST.minmod ? ARG1(ST.me) : ARG2(ST.me);
	    }
	    DEBUG_EXECUTE_r(
                Perl_re_exec_indentf( aTHX_  "CURLYM now matched %"IVdf" times, len=%"IVdf"...\n",
                          depth, (IV) ST.count, (IV)ST.alen)
	    );

            if (EVAL_CLOSE_PAREN_IS_TRUE(cur_eval,(U32)ST.me->flags))
	        goto fake_end;
	        
	    {
		I32 max = (ST.minmod ? ARG1(ST.me) : ARG2(ST.me));
		if ( max == REG_INFTY || ST.count < max )
		    goto curlym_do_A; /* try to match another A */
	    }
	    goto curlym_do_B; /* try to match B */

	case CURLYM_A_fail: /* just failed to match an A */
	    REGCP_UNWIND(ST.cp);


	    if (ST.minmod || ST.count < ARG1(ST.me) /* min*/ 
                || EVAL_CLOSE_PAREN_IS_TRUE(cur_eval,(U32)ST.me->flags))
		sayNO;

	  curlym_do_B: /* execute the B in /A{m,n}B/  */
	    if (ST.c1 == CHRTEST_UNINIT) {
		/* calculate c1 and c2 for possible match of 1st char
		 * following curly */
		ST.c1 = ST.c2 = CHRTEST_VOID;
                assert(ST.B);
		if (HAS_TEXT(ST.B) || JUMPABLE(ST.B)) {
		    regnode *text_node = ST.B;
		    if (! HAS_TEXT(text_node))
			FIND_NEXT_IMPT(text_node);
	            /* this used to be 
	                
	                (HAS_TEXT(text_node) && PL_regkind[OP(text_node)] == EXACT)
	                
	            	But the former is redundant in light of the latter.
	            	
	            	if this changes back then the macro for 
	            	IS_TEXT and friends need to change.
	             */
		    if (PL_regkind[OP(text_node)] == EXACT) {
                        if (! S_setup_EXACTISH_ST_c1_c2(aTHX_
                           text_node, &ST.c1, ST.c1_utf8, &ST.c2, ST.c2_utf8,
                           reginfo))
                        {
                            sayNO;
                        }
		    }