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
|
/*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* This is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language. See
the file Tech.Notes for some information on the internals.
Written by: Philip Hazel <ph10@cam.ac.uk>
Copyright (c) 1997-2003 University of Cambridge
-----------------------------------------------------------------------------
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. This software is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
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.
4. If PCRE is embedded in any software that is released under the GNU
General Purpose Licence (GPL), then the terms of that licence shall
supersede any condition above with which it is incompatible.
-----------------------------------------------------------------------------
*/
/* Modified by Shawn Wagner for PennMUSH to fit in one file and remove
things we don't use, like a bunch of API functions and utf-8
support. If you want the full thing, see http://www.pcre.org. */
/* Modified by Alan Schwartz for PennMUSH to change the use of
* 'isblank' as a variable (reported to Philip Hazel for pcre 4.5) */
#include "config.h"
#include <limits.h>
#include <string.h>
#include <ctype.h>
#include <stdlib.h>
#include <stddef.h>
#include <setjmp.h>
#include "pcre.h"
#include "confmagic.h"
#undef min
#undef max
/* Bits of PCRE's conf.h */
#define NEWLINE '\n'
#define LINK_SIZE 2
#define MATCH_LIMIT 100000
#define NO_RECURSE
/* Bits of internal.h */
/* This header contains definitions that are shared between the different
modules, but which are not relevant to the outside. */
#define PCRE_DEFINITION /* Win32 __declspec(export) trigger for .dll */
#define EXPORT
/* PCRE keeps offsets in its compiled code as 2-byte quantities by default.
These are used, for example, to link from the start of a subpattern to its
alternatives and its end. The use of 2 bytes per offset limits the size of the
compiled regex to around 64K, which is big enough for almost everybody.
However, I received a request for an even bigger limit. For this reason, and
also to make the code easier to maintain, the storing and loading of offsets
from the byte string is now handled by the macros that are defined here.
The macros are controlled by the value of LINK_SIZE. This defaults to 2 in
the config.h file, but can be overridden by using -D on the command line. This
is automated on Unix systems via the "configure" command. */
#define PUT(a,n,d) \
(a[n] = (d) >> 8), \
(a[(n)+1] = (d) & 255)
#define GET(a,n) \
(((a)[n] << 8) | (a)[(n)+1])
#define MAX_PATTERN_SIZE (1 << 16)
/* Convenience macro defined in terms of the others */
#define PUTINC(a,n,d) PUT(a,n,d), a += LINK_SIZE
/* PCRE uses some other 2-byte quantities that do not change when the size of
offsets changes. There are used for repeat counts and for other things such as
capturing parenthesis numbers in back references. */
#define PUT2(a,n,d) \
a[n] = (d) >> 8; \
a[(n)+1] = (d) & 255
#define GET2(a,n) \
(((a)[n] << 8) | (a)[(n)+1])
#define PUT2INC(a,n,d) PUT2(a,n,d), a += 2
/* These are the public options that can change during matching. */
#define PCRE_IMS (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL)
/* Private options flags start at the most significant end of the four bytes,
but skip the top bit so we can use ints for convenience without getting tangled
with negative values. The public options defined in pcre.h start at the least
significant end. Make sure they don't overlap, though now that we have expanded
to four bytes there is plenty of space. */
#define PCRE_FIRSTSET 0x40000000 /* first_byte is set */
#define PCRE_REQCHSET 0x20000000 /* req_byte is set */
#define PCRE_STARTLINE 0x10000000 /* start after \n for multiline */
#define PCRE_ICHANGED 0x08000000 /* i option changes within regex */
/* Options for the "extra" block produced by pcre_study(). */
#define PCRE_STUDY_MAPPED 0x01 /* a map of starting chars exists */
/* Masks for identifying the public options which are permitted at compile
time, run time or study time, respectively. */
#define PUBLIC_OPTIONS \
(PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \
PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \
PCRE_NO_AUTO_CAPTURE|PCRE_NO_UTF8_CHECK)
#define PUBLIC_EXEC_OPTIONS \
(PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK)
#define PUBLIC_STUDY_OPTIONS 0 /* None defined */
/* Magic number to provide a small check against being handed junk. */
#define MAGIC_NUMBER 0x50435245UL /* 'PCRE' */
/* Negative values for the firstchar and reqchar variables */
#define REQ_UNSET (-2)
#define REQ_NONE (-1)
/* Flags added to firstbyte or reqbyte; a "non-literal" item is either a
variable-length repeat, or a anything other than literal characters. */
#define REQ_CASELESS 0x0100 /* indicates caselessness */
#define REQ_VARY 0x0200 /* reqbyte followed non-literal item */
/* Miscellaneous definitions */
typedef int BOOL;
#define FALSE 0
#define TRUE 1
/* Escape items that are just an encoding of a particular data value. Note that
ESC_n is defined as yet another macro, which is set in config.h to either \n
(the default) or \r (which some people want). */
#ifndef ESC_e
#define ESC_e 27
#endif
#ifndef ESC_f
#define ESC_f '\f'
#endif
#ifndef ESC_n
#define ESC_n NEWLINE
#endif
#ifndef ESC_r
#define ESC_r '\r'
#endif
/* We can't officially use ESC_t because it is a POSIX reserved identifier
(presumably because of all the others like size_t). */
#ifndef ESC_tee
#define ESC_tee '\t'
#endif
/* These are escaped items that aren't just an encoding of a particular data
value such as \n. They must have non-zero values, as check_escape() returns
their negation. Also, they must appear in the same order as in the opcode
definitions below, up to ESC_z. There's a dummy for OP_ANY because it
corresponds to "." rather than an escape sequence. The final one must be
ESC_REF as subsequent values are used for \1, \2, \3, etc. There is are two
tests in the code for an escape greater than ESC_b and less than ESC_Z to
detect the types that may be repeated. These are the types that consume a
character. If any new escapes are put in between that don't consume a
character, that code will have to change. */
enum { ESC_A = 1, ESC_G, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s, ESC_W,
ESC_w, ESC_dum1, ESC_C, ESC_Z, ESC_z, ESC_E, ESC_Q, ESC_REF
};
/* Flag bits and data types for the extended class (OP_XCLASS) for classes that
contain UTF-8 characters with values greater than 255. */
#define XCL_NOT 0x01 /* Flag: this is a negative class */
#define XCL_MAP 0x02 /* Flag: a 32-byte map is present */
#define XCL_END 0 /* Marks end of individual items */
#define XCL_SINGLE 1 /* Single item (one multibyte char) follows */
#define XCL_RANGE 2 /* A range (two multibyte chars) follows */
/* Opcode table: OP_BRA must be last, as all values >= it are used for brackets
that extract substrings. Starting from 1 (i.e. after OP_END), the values up to
OP_EOD must correspond in order to the list of escapes immediately above.
Note that whenever this list is updated, the two macro definitions that follow
must also be updated to match. */
enum {
OP_END, /* 0 End of pattern */
/* Values corresponding to backslashed metacharacters */
OP_SOD, /* 1 Start of data: \A */
OP_SOM, /* 2 Start of match (subject + offset): \G */
OP_NOT_WORD_BOUNDARY, /* 3 \B */
OP_WORD_BOUNDARY, /* 4 \b */
OP_NOT_DIGIT, /* 5 \D */
OP_DIGIT, /* 6 \d */
OP_NOT_WHITESPACE, /* 7 \S */
OP_WHITESPACE, /* 8 \s */
OP_NOT_WORDCHAR, /* 9 \W */
OP_WORDCHAR, /* 10 \w */
OP_ANY, /* 11 Match any character */
OP_ANYBYTE, /* 12 Match any byte (\C); different to OP_ANY for UTF-8 */
OP_EODN, /* 13 End of data or \n at end of data: \Z. */
OP_EOD, /* 14 End of data: \z */
OP_OPT, /* 15 Set runtime options */
OP_CIRC, /* 16 Start of line - varies with multiline switch */
OP_DOLL, /* 17 End of line - varies with multiline switch */
OP_CHARS, /* 18 Match string of characters */
OP_NOT, /* 19 Match anything but the following char */
OP_STAR, /* 20 The maximizing and minimizing versions of */
OP_MINSTAR, /* 21 all these opcodes must come in pairs, with */
OP_PLUS, /* 22 the minimizing one second. */
OP_MINPLUS, /* 23 This first set applies to single characters */
OP_QUERY, /* 24 */
OP_MINQUERY, /* 25 */
OP_UPTO, /* 26 From 0 to n matches */
OP_MINUPTO, /* 27 */
OP_EXACT, /* 28 Exactly n matches */
OP_NOTSTAR, /* 29 The maximizing and minimizing versions of */
OP_NOTMINSTAR, /* 30 all these opcodes must come in pairs, with */
OP_NOTPLUS, /* 31 the minimizing one second. */
OP_NOTMINPLUS, /* 32 This set applies to "not" single characters */
OP_NOTQUERY, /* 33 */
OP_NOTMINQUERY, /* 34 */
OP_NOTUPTO, /* 35 From 0 to n matches */
OP_NOTMINUPTO, /* 36 */
OP_NOTEXACT, /* 37 Exactly n matches */
OP_TYPESTAR, /* 38 The maximizing and minimizing versions of */
OP_TYPEMINSTAR, /* 39 all these opcodes must come in pairs, with */
OP_TYPEPLUS, /* 40 the minimizing one second. These codes must */
OP_TYPEMINPLUS, /* 41 be in exactly the same order as those above. */
OP_TYPEQUERY, /* 42 This set applies to character types such as \d */
OP_TYPEMINQUERY, /* 43 */
OP_TYPEUPTO, /* 44 From 0 to n matches */
OP_TYPEMINUPTO, /* 45 */
OP_TYPEEXACT, /* 46 Exactly n matches */
OP_CRSTAR, /* 47 The maximizing and minimizing versions of */
OP_CRMINSTAR, /* 48 all these opcodes must come in pairs, with */
OP_CRPLUS, /* 49 the minimizing one second. These codes must */
OP_CRMINPLUS, /* 50 be in exactly the same order as those above. */
OP_CRQUERY, /* 51 These are for character classes and back refs */
OP_CRMINQUERY, /* 52 */
OP_CRRANGE, /* 53 These are different to the three seta above. */
OP_CRMINRANGE, /* 54 */
OP_CLASS, /* 55 Match a character class, chars < 256 only */
OP_NCLASS, /* 56 Same, but the bitmap was created from a negative
class - the difference is relevant only when a UTF-8
character > 255 is encountered. */
OP_XCLASS, /* 57 Extended class for handling UTF-8 chars within the
class. This does both positive and negative. */
OP_REF, /* 58 Match a back reference */
OP_RECURSE, /* 59 Match a numbered subpattern (possibly recursive) */
OP_CALLOUT, /* 60 Call out to external function if provided */
OP_ALT, /* 61 Start of alternation */
OP_KET, /* 62 End of group that doesn't have an unbounded repeat */
OP_KETRMAX, /* 63 These two must remain together and in this */
OP_KETRMIN, /* 64 order. They are for groups the repeat for ever. */
/* The assertions must come before ONCE and COND */
OP_ASSERT, /* 65 Positive lookahead */
OP_ASSERT_NOT, /* 66 Negative lookahead */
OP_ASSERTBACK, /* 67 Positive lookbehind */
OP_ASSERTBACK_NOT, /* 68 Negative lookbehind */
OP_REVERSE, /* 69 Move pointer back - used in lookbehind assertions */
/* ONCE and COND must come after the assertions, with ONCE first, as there's
a test for >= ONCE for a subpattern that isn't an assertion. */
OP_ONCE, /* 70 Once matched, don't back up into the subpattern */
OP_COND, /* 71 Conditional group */
OP_CREF, /* 72 Used to hold an extraction string number (cond ref) */
OP_BRAZERO, /* 73 These two must remain together and in this */
OP_BRAMINZERO, /* 74 order. */
OP_BRANUMBER, /* 75 Used for extracting brackets whose number is greater
than can fit into an opcode. */
OP_BRA /* 76 This and greater values are used for brackets that
extract substrings up to a basic limit. After that,
use is made of OP_BRANUMBER. */
};
/* WARNING: There is an implicit assumption in study.c that all opcodes are
less than 128 in value. This makes handling UTF-8 character sequences easier.
*/
/* This macro defines textual names for all the opcodes. There are used only
for debugging, in pcre.c when DEBUG is defined, and also in pcretest.c. The
macro is referenced only in printint.c. */
#define OP_NAME_LIST \
"End", "\\A", "\\G", "\\B", "\\b", "\\D", "\\d", \
"\\S", "\\s", "\\W", "\\w", "Any", "Anybyte", "\\Z", "\\z", \
"Opt", "^", "$", "chars", "not", \
"*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
"*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
"*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
"*", "*?", "+", "+?", "?", "??", "{", "{", \
"class", "nclass", "xclass", "Ref", "Recurse", "Callout", \
"Alt", "Ket", "KetRmax", "KetRmin", "Assert", "Assert not", \
"AssertB", "AssertB not", "Reverse", "Once", "Cond", "Cond ref",\
"Brazero", "Braminzero", "Branumber", "Bra"
/* This macro defines the length of fixed length operations in the compiled
regex. The lengths are used when searching for specific things, and also in the
debugging printing of a compiled regex. We use a macro so that it can be
incorporated both into pcre.c and pcretest.c without being publicly exposed.
As things have been extended, some of these are no longer fixed lenths, but are
minima instead. For example, the length of a single-character repeat may vary
in UTF-8 mode. The code that uses this table must know about such things. */
#define OP_LENGTHS \
1, /* End */ \
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* \A, \G, \B, \B, \D, \d, \S, \s, \W, \w */ \
1, 1, 1, 1, 2, 1, 1, /* Any, Anybyte, \Z, \z, Opt, ^, $ */ \
2, /* Chars - the minimum length */ \
2, /* not */ \
/* Positive single-char repeats ** These are */ \
2, 2, 2, 2, 2, 2, /* *, *?, +, +?, ?, ?? ** minima in */ \
4, 4, 4, /* upto, minupto, exact ** UTF-8 mode */ \
/* Negative single-char repeats - only for chars < 256 */ \
2, 2, 2, 2, 2, 2, /* NOT *, *?, +, +?, ?, ?? */ \
4, 4, 4, /* NOT upto, minupto, exact */ \
/* Positive type repeats */ \
2, 2, 2, 2, 2, 2, /* Type *, *?, +, +?, ?, ?? */ \
4, 4, 4, /* Type upto, minupto, exact */ \
/* Character class & ref repeats */ \
1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */ \
5, 5, /* CRRANGE, CRMINRANGE */ \
33, /* CLASS */ \
33, /* NCLASS */ \
0, /* XCLASS - variable length */ \
3, /* REF */ \
1+LINK_SIZE, /* RECURSE */ \
2, /* CALLOUT */ \
1+LINK_SIZE, /* Alt */ \
1+LINK_SIZE, /* Ket */ \
1+LINK_SIZE, /* KetRmax */ \
1+LINK_SIZE, /* KetRmin */ \
1+LINK_SIZE, /* Assert */ \
1+LINK_SIZE, /* Assert not */ \
1+LINK_SIZE, /* Assert behind */ \
1+LINK_SIZE, /* Assert behind not */ \
1+LINK_SIZE, /* Reverse */ \
1+LINK_SIZE, /* Once */ \
1+LINK_SIZE, /* COND */ \
3, /* CREF */ \
1, 1, /* BRAZERO, BRAMINZERO */ \
3, /* BRANUMBER */ \
1+LINK_SIZE /* BRA */ \
/* The highest extraction number before we have to start using additional
bytes. (Originally PCRE didn't have support for extraction counts highter than
this number.) The value is limited by the number of opcodes left after OP_BRA,
i.e. 255 - OP_BRA. We actually set it a bit lower to leave room for additional
opcodes. */
#define EXTRACT_BASIC_MAX 150
/* A magic value for OP_CREF to indicate the "in recursion" condition. */
#define CREF_RECURSE 0xffff
/* The texts of compile-time error messages are defined as macros here so that
they can be accessed by the POSIX wrapper and converted into error codes. Yes,
I could have used error codes in the first place, but didn't feel like changing
just to accommodate the POSIX wrapper. */
#define ERR1 "\\ at end of pattern"
#define ERR2 "\\c at end of pattern"
#define ERR3 "unrecognized character follows \\"
#define ERR4 "numbers out of order in {} quantifier"
#define ERR5 "number too big in {} quantifier"
#define ERR6 "missing terminating ] for character class"
#define ERR7 "invalid escape sequence in character class"
#define ERR8 "range out of order in character class"
#define ERR9 "nothing to repeat"
#define ERR10 "operand of unlimited repeat could match the empty string"
#define ERR11 "internal error: unexpected repeat"
#define ERR12 "unrecognized character after (?"
#define ERR13 "POSIX named classes are supported only within a class"
#define ERR14 "missing )"
#define ERR15 "reference to non-existent subpattern"
#define ERR16 "erroffset passed as NULL"
#define ERR17 "unknown option bit(s) set"
#define ERR18 "missing ) after comment"
#define ERR19 "parentheses nested too deeply"
#define ERR20 "regular expression too large"
#define ERR21 "failed to get memory"
#define ERR22 "unmatched parentheses"
#define ERR23 "internal error: code overflow"
#define ERR24 "unrecognized character after (?<"
#define ERR25 "lookbehind assertion is not fixed length"
#define ERR26 "malformed number after (?("
#define ERR27 "conditional group contains more than two branches"
#define ERR28 "assertion expected after (?("
#define ERR29 "(?R or (?digits must be followed by )"
#define ERR30 "unknown POSIX class name"
#define ERR31 "POSIX collating elements are not supported"
#define ERR32 "this version of PCRE is not compiled with PCRE_UTF8 support"
#define ERR33 "spare error"
#define ERR34 "character value in \\x{...} sequence is too large"
#define ERR35 "invalid condition (?(0)"
#define ERR36 "\\C not allowed in lookbehind assertion"
#define ERR37 "PCRE does not support \\L, \\l, \\N, \\P, \\p, \\U, \\u, or \\X"
#define ERR38 "number after (?C is > 255"
#define ERR39 "closing ) for (?C expected"
#define ERR40 "recursive call could loop indefinitely"
#define ERR41 "unrecognized character after (?P"
#define ERR42 "syntax error after (?P"
#define ERR43 "two named groups have the same name"
#define ERR44 "invalid UTF-8 string"
/* All character handling must be done as unsigned characters. Otherwise there
are problems with top-bit-set characters and functions such as isspace().
However, we leave the interface to the outside world as char *, because that
should make things easier for callers. We define a short type for unsigned char
to save lots of typing. I tried "uchar", but it causes problems on Digital
Unix, where it is defined in sys/types, so use "uschar" instead. */
typedef unsigned char uschar;
/* The real format of the start of the pcre block; the index of names and the
code vector run on as long as necessary after the end. */
typedef struct real_pcre {
unsigned long int magic_number;
size_t size; /* Total that was malloced */
const unsigned char *tables; /* Pointer to tables */
unsigned long int options;
unsigned short int top_bracket;
unsigned short int top_backref;
unsigned short int first_byte;
unsigned short int req_byte;
unsigned short int name_entry_size; /* Size of any name items; 0 => none */
unsigned short int name_count; /* Number of name items */
} real_pcre;
/* The format of the block used to store data from pcre_study(). */
typedef struct pcre_study_data {
size_t size; /* Total that was malloced */
uschar options;
uschar start_bits[32];
} pcre_study_data;
/* Structure for passing "static" information around between the functions
doing the compiling, so that they are thread-safe. */
typedef struct compile_data {
const uschar *lcc; /* Points to lower casing table */
const uschar *fcc; /* Points to case-flipping table */
const uschar *cbits; /* Points to character type table */
const uschar *ctypes; /* Points to table of type maps */
const uschar *start_code; /* The start of the compiled code */
uschar *name_table; /* The name/number table */
int names_found; /* Number of entries so far */
int name_entry_size; /* Size of each entry */
int top_backref; /* Maximum back reference */
unsigned int backref_map; /* Bitmap of low back refs */
int req_varyopt; /* "After variable item" flag for reqbyte */
} compile_data;
/* Structure for maintaining a chain of pointers to the currently incomplete
branches, for testing for left recursion. */
typedef struct branch_chain {
struct branch_chain *outer;
uschar *current;
} branch_chain;
/* Structure for items in a linked list that represents an explicit recursive
call within the pattern. */
typedef struct recursion_info {
struct recursion_info *prevrec; /* Previous recursion record (or NULL) */
int group_num; /* Number of group that was called */
const uschar *after_call; /* "Return value": points after the call in the expr */
const uschar *save_start; /* Old value of md->start_match */
int *offset_save; /* Pointer to start of saved offsets */
int saved_max; /* Number of saved offsets */
} recursion_info;
/* When compiling in a mode that doesn't use recursive calls to match(),
a structure is used to remember local variables on the heap. It is defined in
pcre.c, close to the match() function, so that it is easy to keep it in step
with any changes of local variable. However, the pointer to the current frame
must be saved in some "static" place over a longjmp(). We declare the
structure here so that we can put a pointer in the match_data structure.
NOTE: This isn't used for a "normal" compilation of pcre. */
struct heapframe;
/* Structure for passing "static" information around between the functions
doing the matching, so that they are thread-safe. */
typedef struct match_data {
unsigned long int match_call_count; /* As it says */
unsigned long int match_limit; /* As it says */
int *offset_vector; /* Offset vector */
int offset_end; /* One past the end */
int offset_max; /* The maximum usable for return data */
const uschar *lcc; /* Points to lower casing table */
const uschar *ctypes; /* Points to table of type maps */
BOOL offset_overflow; /* Set if too many extractions */
BOOL notbol; /* NOTBOL flag */
BOOL noteol; /* NOTEOL flag */
BOOL utf8; /* UTF8 flag */
BOOL endonly; /* Dollar not before final \n */
BOOL notempty; /* Empty string match not wanted */
const uschar *start_code; /* For use when recursing */
const uschar *start_subject; /* Start of the subject string */
const uschar *end_subject; /* End of the subject string */
const uschar *start_match; /* Start of this match attempt */
const uschar *end_match_ptr; /* Subject position at end match */
int end_offset_top; /* Highwater mark at end of match */
int capture_last; /* Most recent capture number */
int start_offset; /* The start offset value */
recursion_info *recursive; /* Linked list of recursion data */
void *callout_data; /* To pass back to callouts */
struct heapframe *thisframe; /* Used only when compiling for no recursion */
} match_data;
/* Bit definitions for entries in the pcre_ctypes table. */
#define ctype_space 0x01
#define ctype_letter 0x02
#define ctype_digit 0x04
#define ctype_xdigit 0x08
#define ctype_word 0x10 /* alphameric or '_' */
#define ctype_meta 0x80 /* regexp meta char or zero (end pattern) */
/* Offsets for the bitmap tables in pcre_cbits. Each table contains a set
of bits for a class map. Some classes are built by combining these tables. */
#define cbit_space 0 /* [:space:] or \s */
#define cbit_xdigit 32 /* [:xdigit:] */
#define cbit_digit 64 /* [:digit:] or \d */
#define cbit_upper 96 /* [:upper:] */
#define cbit_lower 128 /* [:lower:] */
#define cbit_word 160 /* [:word:] or \w */
#define cbit_graph 192 /* [:graph:] */
#define cbit_print 224 /* [:print:] */
#define cbit_punct 256 /* [:punct:] */
#define cbit_cntrl 288 /* [:cntrl:] */
#define cbit_length 320 /* Length of the cbits table */
/* Offsets of the various tables from the base tables pointer, and
total length. */
#define lcc_offset 0
#define fcc_offset 256
#define cbits_offset 512
#define ctypes_offset (cbits_offset + cbit_length)
#define tables_length (ctypes_offset + 256)
/* End of internal.h */
/* chartables.c */
/*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* This file is automatically written by the dftables auxiliary
program. If you edit it by hand, you might like to edit the Makefile to
prevent its ever being regenerated.
This file is #included in the compilation of pcre.c to build the default
character tables which are used when no tables are passed to the compile
function. */
static unsigned char pcre_default_tables[] = {
/* This table is a lower casing table. */
0, 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, 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, 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,
/* This table is a case flipping table. */
0, 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, 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, 91, 92, 93, 94, 95,
96, 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, 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,
/* This table contains bit maps for various character classes.
Each map is 32 bytes long and the bits run from the least
significant end of each byte. The classes that have their own
maps are: space, xdigit, digit, upper, lower, word, graph
print, punct, and cntrl. Other classes are built from combinations. */
0x00, 0x3e, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xff, 0x03,
0x7e, 0x00, 0x00, 0x00, 0x7e, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xff, 0x03,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0xfe, 0xff, 0xff, 0x07, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0xfe, 0xff, 0xff, 0x07,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xff, 0x03,
0xfe, 0xff, 0xff, 0x87, 0xfe, 0xff, 0xff, 0x07,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0xfe, 0xff, 0xff, 0xff,
0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x7f,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0xff, 0xff, 0xff, 0xff,
0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x7f,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0xfe, 0xff, 0x00, 0xfc,
0x01, 0x00, 0x00, 0xf8, 0x01, 0x00, 0x00, 0x78,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0xff, 0xff, 0xff, 0xff, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
/* This table identifies various classes of character by individual bits:
0x01 white space character
0x02 letter
0x04 decimal digit
0x08 hexadecimal digit
0x10 alphanumeric or '_'
0x80 regular expression metacharacter or binary zero
*/
0x80, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0- 7 */
0x00, 0x01, 0x01, 0x00, 0x01, 0x01, 0x00, 0x00, /* 8- 15 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 16- 23 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 24- 31 */
0x01, 0x00, 0x00, 0x00, 0x80, 0x00, 0x00, 0x00, /* - ' */
0x80, 0x80, 0x80, 0x80, 0x00, 0x00, 0x80, 0x00, /* ( - / */
0x1c, 0x1c, 0x1c, 0x1c, 0x1c, 0x1c, 0x1c, 0x1c, /* 0 - 7 */
0x1c, 0x1c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80, /* 8 - ? */
0x00, 0x1a, 0x1a, 0x1a, 0x1a, 0x1a, 0x1a, 0x12, /* @ - G */
0x12, 0x12, 0x12, 0x12, 0x12, 0x12, 0x12, 0x12, /* H - O */
0x12, 0x12, 0x12, 0x12, 0x12, 0x12, 0x12, 0x12, /* P - W */
0x12, 0x12, 0x12, 0x80, 0x00, 0x00, 0x80, 0x10, /* X - _ */
0x00, 0x1a, 0x1a, 0x1a, 0x1a, 0x1a, 0x1a, 0x12, /* ` - g */
0x12, 0x12, 0x12, 0x12, 0x12, 0x12, 0x12, 0x12, /* h - o */
0x12, 0x12, 0x12, 0x12, 0x12, 0x12, 0x12, 0x12, /* p - w */
0x12, 0x12, 0x12, 0x80, 0x80, 0x00, 0x00, 0x00, /* x -127 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 128-135 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 136-143 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 144-151 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 152-159 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 160-167 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 168-175 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 176-183 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 184-191 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 192-199 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 200-207 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 208-215 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 216-223 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 224-231 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 232-239 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 240-247 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
}; /* 248-255 */
/* End of chartables.c */
/* get.c */
/* This module contains some convenience functions for extracting substrings
from the subject string after a regex match has succeeded. The original idea
for these functions came from Scott Wimer <scottw@cgibuilder.com>. */
/*************************************************
* Copy captured string to given buffer *
*************************************************/
/* This function copies a single captured substring into a given buffer.
Note that we use memcpy() rather than strncpy() in case there are binary zeros
in the string.
Arguments:
subject the subject string that was matched
ovector pointer to the offsets table
stringcount the number of substrings that were captured
(i.e. the yield of the pcre_exec call, unless
that was zero, in which case it should be 1/3
of the offset table size)
stringnumber the number of the required substring
buffer where to put the substring
size the size of the buffer
Returns: if successful:
the length of the copied string, not including the zero
that is put on the end; can be zero
if not successful:
PCRE_ERROR_NOMEMORY (-6) buffer too small
PCRE_ERROR_NOSUBSTRING (-7) no such captured substring
*/
int
pcre_copy_substring(const char *subject, int *ovector, int stringcount,
int stringnumber, char *buffer, int size)
{
int yield;
if (stringnumber < 0 || stringnumber >= stringcount)
return PCRE_ERROR_NOSUBSTRING;
stringnumber *= 2;
yield = ovector[stringnumber + 1] - ovector[stringnumber];
if (size < yield + 1)
return PCRE_ERROR_NOMEMORY;
memcpy(buffer, subject + ovector[stringnumber], yield);
buffer[yield] = 0;
return yield;
}
/* End of get.c */
/* maketables.c */
/*************************************************
* Create PCRE character tables *
*************************************************/
/* This function builds a set of character tables for use by PCRE and returns
a pointer to them. They are build using the ctype functions, and consequently
their contents will depend upon the current locale setting. When compiled as
part of the library, the store is obtained via pcre_malloc(), but when compiled
inside dftables, use malloc().
Arguments: none
Returns: pointer to the contiguous block of data
*/
const unsigned char *
pcre_maketables(void)
{
unsigned char *yield, *p;
int i;
#ifndef DFTABLES
yield = (unsigned char *) malloc(tables_length);
#else
yield = (unsigned char *) malloc(tables_length);
#endif
if (yield == NULL)
return NULL;
p = yield;
/* First comes the lower casing table */
for (i = 0; i < 256; i++)
*p++ = tolower(i);
/* Next the case-flipping table */
for (i = 0; i < 256; i++)
*p++ = islower(i) ? toupper(i) : tolower(i);
/* Then the character class tables. Don't try to be clever and save effort
on exclusive ones - in some locales things may be different. Note that the
table for "space" includes everything "isspace" gives, including VT in the
default locale. This makes it work for the POSIX class [:space:]. */
memset(p, 0, cbit_length);
for (i = 0; i < 256; i++) {
if (isdigit(i)) {
p[cbit_digit + i / 8] |= 1 << (i & 7);
p[cbit_word + i / 8] |= 1 << (i & 7);
}
if (isupper(i)) {
p[cbit_upper + i / 8] |= 1 << (i & 7);
p[cbit_word + i / 8] |= 1 << (i & 7);
}
if (islower(i)) {
p[cbit_lower + i / 8] |= 1 << (i & 7);
p[cbit_word + i / 8] |= 1 << (i & 7);
}
if (i == '_')
p[cbit_word + i / 8] |= 1 << (i & 7);
if (isspace(i))
p[cbit_space + i / 8] |= 1 << (i & 7);
if (isxdigit(i))
p[cbit_xdigit + i / 8] |= 1 << (i & 7);
if (isgraph(i))
p[cbit_graph + i / 8] |= 1 << (i & 7);
if (isprint(i))
p[cbit_print + i / 8] |= 1 << (i & 7);
if (ispunct(i))
p[cbit_punct + i / 8] |= 1 << (i & 7);
if (iscntrl(i))
p[cbit_cntrl + i / 8] |= 1 << (i & 7);
}
p += cbit_length;
/* Finally, the character type table. In this, we exclude VT from the white
space chars, because Perl doesn't recognize it as such for \s and for comments
within regexes. */
for (i = 0; i < 256; i++) {
int x = 0;
if (i != 0x0b && isspace(i))
x += ctype_space;
if (isalpha(i))
x += ctype_letter;
if (isdigit(i))
x += ctype_digit;
if (isxdigit(i))
x += ctype_xdigit;
if (isalnum(i) || i == '_')
x += ctype_word;
/* Note: strchr includes the terminating zero in the characters it considers.
In this instance, that is ok because we want binary zero to be flagged as a
meta-character, which in this sense is any character that terminates a run
of data characters. */
if (strchr("*+?{^.$|()[", i) != 0)
x += ctype_meta;
*p++ = x;
}
return yield;
}
/* End of maketables.c */
/* study.c */
/*************************************************
* Set a bit and maybe its alternate case *
*************************************************/
/* Given a character, set its bit in the table, and also the bit for the other
version of a letter if we are caseless.
Arguments:
start_bits points to the bit map
c is the character
caseless the caseless flag
cd the block with char table pointers
Returns: nothing
*/
static void
set_bit(uschar * start_bits, int c, BOOL caseless, compile_data * cd)
{
start_bits[c / 8] |= (1 << (c & 7));
if (caseless && (cd->ctypes[c] & ctype_letter) != 0)
start_bits[cd->fcc[c] / 8] |= (1 << (cd->fcc[c] & 7));
}
/*************************************************
* Create bitmap of starting chars *
*************************************************/
/* This function scans a compiled unanchored expression and attempts to build a
bitmap of the set of initial characters. If it can't, it returns FALSE. As time
goes by, we may be able to get more clever at doing this.
Arguments:
code points to an expression
start_bits points to a 32-byte table, initialized to 0
caseless the current state of the caseless flag
utf8 TRUE if in UTF-8 mode
cd the block with char table pointers
Returns: TRUE if table built, FALSE otherwise
*/
static BOOL
set_start_bits(const uschar * code, uschar * start_bits, BOOL caseless,
BOOL utf8, compile_data * cd)
{
register int c;
/* This next statement and the later reference to dummy are here in order to
trick the optimizer of the IBM C compiler for OS/2 into generating correct
code. Apparently IBM isn't going to fix the problem, and we would rather not
disable optimization (in this module it actually makes a big difference, and
the pcre module can use all the optimization it can get). */
volatile int dummy;
do {
const uschar *tcode = code + 1 + LINK_SIZE;
BOOL try_next = TRUE;
while (try_next) {
/* If a branch starts with a bracket or a positive lookahead assertion,
recurse to set bits from within them. That's all for this branch. */
if ((int) *tcode >= OP_BRA || *tcode == OP_ASSERT) {
if (!set_start_bits(tcode, start_bits, caseless, utf8, cd))
return FALSE;
try_next = FALSE;
}
else
switch (*tcode) {
default:
return FALSE;
/* Skip over callout */
case OP_CALLOUT:
tcode += 2;
break;
/* Skip over extended extraction bracket number */
case OP_BRANUMBER:
tcode += 3;
break;
/* Skip over lookbehind and negative lookahead assertions */
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
do
tcode += GET(tcode, 1);
while (*tcode == OP_ALT);
tcode += 1 + LINK_SIZE;
break;
/* Skip over an option setting, changing the caseless flag */
case OP_OPT:
caseless = (tcode[1] & PCRE_CASELESS) != 0;
tcode += 2;
break;
/* BRAZERO does the bracket, but carries on. */
case OP_BRAZERO:
case OP_BRAMINZERO:
if (!set_start_bits(++tcode, start_bits, caseless, utf8, cd))
return FALSE;
dummy = 1;
do
tcode += GET(tcode, 1);
while (*tcode == OP_ALT);
tcode += 1 + LINK_SIZE;
break;
/* Single-char * or ? sets the bit and tries the next item */
case OP_STAR:
case OP_MINSTAR:
case OP_QUERY:
case OP_MINQUERY:
set_bit(start_bits, tcode[1], caseless, cd);
tcode += 2;
break;
/* Single-char upto sets the bit and tries the next */
case OP_UPTO:
case OP_MINUPTO:
set_bit(start_bits, tcode[3], caseless, cd);
tcode += 4;
break;
/* At least one single char sets the bit and stops */
case OP_EXACT: /* Fall through */
tcode++;
case OP_CHARS: /* Fall through */
tcode++;
case OP_PLUS:
case OP_MINPLUS:
set_bit(start_bits, tcode[1], caseless, cd);
try_next = FALSE;
break;
/* Single character type sets the bits and stops */
case OP_NOT_DIGIT:
for (c = 0; c < 32; c++)
start_bits[c] |= ~cd->cbits[c + cbit_digit];
try_next = FALSE;
break;
case OP_DIGIT:
for (c = 0; c < 32; c++)
start_bits[c] |= cd->cbits[c + cbit_digit];
try_next = FALSE;
break;
case OP_NOT_WHITESPACE:
for (c = 0; c < 32; c++)
start_bits[c] |= ~cd->cbits[c + cbit_space];
try_next = FALSE;
break;
case OP_WHITESPACE:
for (c = 0; c < 32; c++)
start_bits[c] |= cd->cbits[c + cbit_space];
try_next = FALSE;
break;
case OP_NOT_WORDCHAR:
for (c = 0; c < 32; c++)
start_bits[c] |= ~cd->cbits[c + cbit_word];
try_next = FALSE;
break;
case OP_WORDCHAR:
for (c = 0; c < 32; c++)
start_bits[c] |= cd->cbits[c + cbit_word];
try_next = FALSE;
break;
/* One or more character type fudges the pointer and restarts, knowing
it will hit a single character type and stop there. */
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
tcode++;
break;
case OP_TYPEEXACT:
tcode += 3;
break;
/* Zero or more repeats of character types set the bits and then
try again. */
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
tcode += 2; /* Fall through */
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
switch (tcode[1]) {
case OP_ANY:
return FALSE;
case OP_NOT_DIGIT:
for (c = 0; c < 32; c++)
start_bits[c] |= ~cd->cbits[c + cbit_digit];
break;
case OP_DIGIT:
for (c = 0; c < 32; c++)
start_bits[c] |= cd->cbits[c + cbit_digit];
break;
case OP_NOT_WHITESPACE:
for (c = 0; c < 32; c++)
start_bits[c] |= ~cd->cbits[c + cbit_space];
break;
case OP_WHITESPACE:
for (c = 0; c < 32; c++)
start_bits[c] |= cd->cbits[c + cbit_space];
break;
case OP_NOT_WORDCHAR:
for (c = 0; c < 32; c++)
start_bits[c] |= ~cd->cbits[c + cbit_word];
break;
case OP_WORDCHAR:
for (c = 0; c < 32; c++)
start_bits[c] |= cd->cbits[c + cbit_word];
break;
}
tcode += 2;
break;
/* Character class where all the information is in a bit map: set the
bits and either carry on or not, according to the repeat count. If it was
a negative class, and we are operating with UTF-8 characters, any byte
with a value >= 0xc4 is a potentially valid starter because it starts a
character with a value > 255. */
case OP_NCLASS:
if (utf8) {
start_bits[24] |= 0xf0; /* Bits for 0xc4 - 0xc8 */
memset(start_bits + 25, 0xff, 7); /* Bits for 0xc9 - 0xff */
}
/* Fall through */
case OP_CLASS:
{
tcode++;
/* In UTF-8 mode, the bits in a bit map correspond to character
values, not to byte values. However, the bit map we are constructing is
for byte values. So we have to do a conversion for characters whose
value is > 127. In fact, there are only two possible starting bytes for
characters in the range 128 - 255. */
if (utf8) {
for (c = 0; c < 16; c++)
start_bits[c] |= tcode[c];
for (c = 128; c < 256; c++) {
if ((tcode[c / 8] && (1 << (c & 7))) != 0) {
int d = (c >> 6) | 0xc0; /* Set bit for this starter */
start_bits[d / 8] |= (1 << (d & 7)); /* and then skip on to the */
c = (c & 0xc0) + 0x40 - 1; /* next relevant character. */
}
}
}
/* In non-UTF-8 mode, the two bit maps are completely compatible. */
else {
for (c = 0; c < 32; c++)
start_bits[c] |= tcode[c];
}
/* Advance past the bit map, and act on what follows */
tcode += 32;
switch (*tcode) {
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRQUERY:
case OP_CRMINQUERY:
tcode++;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
if (((tcode[1] << 8) + tcode[2]) == 0)
tcode += 5;
else
try_next = FALSE;
break;
default:
try_next = FALSE;
break;
}
}
break; /* End of bitmap class handling */
} /* End of switch */
} /* End of try_next loop */
code += GET(code, 1); /* Advance to next branch */
}
while (*code == OP_ALT);
return TRUE;
}
/*************************************************
* Study a compiled expression *
*************************************************/
/* This function is handed a compiled expression that it must study to produce
information that will speed up the matching. It returns a pcre_extra block
which then gets handed back to pcre_exec().
Arguments:
re points to the compiled expression
options contains option bits
errorptr points to where to place error messages;
set NULL unless error
Returns: pointer to a pcre_extra block, with study_data filled in and the
appropriate flag set;
NULL on error or if no optimization possible
*/
EXPORT pcre_extra *
pcre_study(const pcre * external_re, int options, const char **errorptr)
{
uschar start_bits[32];
pcre_extra *extra;
pcre_study_data *study;
const real_pcre *re = (const real_pcre *) external_re;
uschar *code = (uschar *) re + sizeof(real_pcre) +
(re->name_count * re->name_entry_size);
compile_data compile_block;
*errorptr = NULL;
if (re == NULL || re->magic_number != MAGIC_NUMBER) {
*errorptr = "argument is not a compiled regular expression";
return NULL;
}
if ((options & ~PUBLIC_STUDY_OPTIONS) != 0) {
*errorptr = "unknown or incorrect option bit(s) set";
return NULL;
}
/* For an anchored pattern, or an unanchored pattern that has a first char, or
a multiline pattern that matches only at "line starts", no further processing
at present. */
if ((re->options & (PCRE_ANCHORED | PCRE_FIRSTSET | PCRE_STARTLINE)) != 0)
return NULL;
/* Set the character tables in the block which is passed around */
compile_block.lcc = re->tables + lcc_offset;
compile_block.fcc = re->tables + fcc_offset;
compile_block.cbits = re->tables + cbits_offset;
compile_block.ctypes = re->tables + ctypes_offset;
/* See if we can find a fixed set of initial characters for the pattern. */
memset(start_bits, 0, 32 * sizeof(uschar));
if (!set_start_bits(code, start_bits, (re->options & PCRE_CASELESS) != 0,
(re->options & PCRE_UTF8) != 0, &compile_block))
return NULL;
/* Get a pcre_extra block and a pcre_study_data block. The study data is put in
the latter, which is pointed to by the former, which may also get additional
data set later by the calling program. At the moment, the size of
pcre_study_data is fixed. We nevertheless save it in a field for returning via
the pcre_fullinfo() function so that if it becomes variable in the future, we
don't have to change that code. */
extra = (pcre_extra *) malloc(sizeof(pcre_extra) + sizeof(pcre_study_data));
if (extra == NULL) {
*errorptr = "failed to get memory";
return NULL;
}
study = (pcre_study_data *) ((char *) extra + sizeof(pcre_extra));
extra->flags = PCRE_EXTRA_STUDY_DATA;
extra->study_data = study;
study->size = sizeof(pcre_study_data);
study->options = PCRE_STUDY_MAPPED;
memcpy(study->start_bits, start_bits, sizeof(start_bits));
return extra;
}
/* End of study.c */
/* pcre.c */
#define DPRINTF(p) /*nothing */
/* Maximum number of items on the nested bracket stacks at compile time. This
applies to the nesting of all kinds of parentheses. It does not limit
un-nested, non-capturing parentheses. This number can be made bigger if
necessary - it is used to dimension one int and one unsigned char vector at
compile time. */
#define BRASTACK_SIZE 200
/* Maximum number of ints of offset to save on the stack for recursive calls.
If the offset vector is bigger, malloc is used. This should be a multiple of 3,
because the offset vector is always a multiple of 3 long. */
#define REC_STACK_SAVE_MAX 30
/* The number of bytes in a literal character string above which we can't add
any more is set at 250 in order to allow for UTF-8 characters. (In theory it
could be 255 when UTF-8 support is excluded, but that means that some of the
test output would be different, which just complicates things.) */
#define MAXLIT 250
/* The maximum remaining length of subject we are prepared to search for a
req_byte match. */
#define REQ_BYTE_MAX 1000
/* Table of sizes for the fixed-length opcodes. It's defined in a macro so that
the definition is next to the definition of the opcodes in internal.h. */
static const uschar OP_lengths[] = { OP_LENGTHS };
/* Min and max values for the common repeats; for the maxima, 0 => infinity */
static const char rep_min[] = { 0, 0, 1, 1, 0, 0 };
static const char rep_max[] = { 0, 0, 0, 0, 1, 1 };
/* Table for handling escaped characters in the range '0'-'z'. Positive returns
are simple data values; negative values are for special things like \d and so
on. Zero means further processing is needed (for things like \x), or the escape
is invalid. */
static const short int escapes[] = {
0, 0, 0, 0, 0, 0, 0, 0, /* 0 - 7 */
0, 0, ':', ';', '<', '=', '>', '?', /* 8 - ? */
'@', -ESC_A, -ESC_B, -ESC_C, -ESC_D, -ESC_E, 0, -ESC_G, /* @ - G */
0, 0, 0, 0, 0, 0, 0, 0, /* H - O */
0, -ESC_Q, 0, -ESC_S, 0, 0, 0, -ESC_W, /* P - W */
0, 0, -ESC_Z, '[', '\\', ']', '^', '_', /* X - _ */
'`', 7, -ESC_b, 0, -ESC_d, ESC_e, ESC_f, 0, /* ` - g */
0, 0, 0, 0, 0, 0, ESC_n, 0, /* h - o */
0, 0, ESC_r, -ESC_s, ESC_tee, 0, 0, -ESC_w, /* p - w */
0, 0, -ESC_z /* x - z */
};
/* Tables of names of POSIX character classes and their lengths. The list is
terminated by a zero length entry. The first three must be alpha, upper, lower,
as this is assumed for handling case independence. */
static const char *const posix_names[] = {
"alpha", "lower", "upper",
"alnum", "ascii", "blank", "cntrl", "digit", "graph",
"print", "punct", "space", "word", "xdigit"
};
static const uschar posix_name_lengths[] = {
5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0
};
/* Table of class bit maps for each POSIX class; up to three may be combined
to form the class. The table for [:blank:] is dynamically modified to remove
the vertical space characters. */
static const int posix_class_maps[] = {
cbit_lower, cbit_upper, -1, /* alpha */
cbit_lower, -1, -1, /* lower */
cbit_upper, -1, -1, /* upper */
cbit_digit, cbit_lower, cbit_upper, /* alnum */
cbit_print, cbit_cntrl, -1, /* ascii */
cbit_space, -1, -1, /* blank - a GNU extension */
cbit_cntrl, -1, -1, /* cntrl */
cbit_digit, -1, -1, /* digit */
cbit_graph, -1, -1, /* graph */
cbit_print, -1, -1, /* print */
cbit_punct, -1, -1, /* punct */
cbit_space, -1, -1, /* space */
cbit_word, -1, -1, /* word - a Perl extension */
cbit_xdigit, -1, -1 /* xdigit */
};
/* Table to identify digits and hex digits. This is used when compiling
patterns. Note that the tables in chartables are dependent on the locale, and
may mark arbitrary characters as digits - but the PCRE compiling code expects
to handle only 0-9, a-z, and A-Z as digits when compiling. That is why we have
a private table here. It costs 256 bytes, but it is a lot faster than doing
character value tests (at least in some simple cases I timed), and in some
applications one wants PCRE to compile efficiently as well as match
efficiently.
For convenience, we use the same bit definitions as in chartables:
0x04 decimal digit
0x08 hexadecimal digit
Then we can use ctype_digit and ctype_xdigit in the code. */
static const unsigned char digitab[] = {
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0- 7 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 8- 15 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 16- 23 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 24- 31 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* - ' */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* ( - / */
0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, /* 0 - 7 */
0x0c, 0x0c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 8 - ? */
0x00, 0x08, 0x08, 0x08, 0x08, 0x08, 0x08, 0x00, /* @ - G */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* H - O */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* P - W */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* X - _ */
0x00, 0x08, 0x08, 0x08, 0x08, 0x08, 0x08, 0x00, /* ` - g */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* h - o */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* p - w */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* x -127 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 128-135 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 136-143 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 144-151 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 152-159 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 160-167 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 168-175 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 176-183 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 184-191 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 192-199 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 200-207 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 208-215 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 216-223 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 224-231 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 232-239 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 240-247 */
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
}; /* 248-255 */
/* Definition to allow mutual recursion */
static BOOL
compile_regex(int, int, int *, uschar **, const uschar **, const char **,
BOOL, int, int *, int *, branch_chain *, compile_data *);
/* Structure for building a chain of data that actually lives on the
stack, for holding the values of the subject pointer at the start of each
subpattern, so as to detect when an empty string has been matched by a
subpattern - to break infinite loops. When NO_RECURSE is set, these blocks
are on the heap, not on the stack. */
typedef struct eptrblock {
struct eptrblock *epb_prev;
const uschar *epb_saved_eptr;
} eptrblock;
/* Flag bits for the match() function */
#define match_condassert 0x01 /* Called to check a condition assertion */
#define match_isgroup 0x02 /* Set if start of bracketed group */
/* Non-error returns from the match() function. Error returns are externally
defined PCRE_ERROR_xxx codes, which are all negative. */
#define MATCH_MATCH 1
#define MATCH_NOMATCH 0
/*************************************************
* Global variables *
*************************************************/
/* PCRE is thread-clean and doesn't use any global variables in the normal
sense. However, it calls memory allocation and free functions via the four
indirections below, and it can optionally do callouts. These values can be
changed by the caller, but are shared between all threads. However, when
compiling for Virtual Pascal, things are done differently (see pcre.in). */
#ifndef VPCOMPAT
#ifdef __cplusplus
extern "C" void *(*pcre_malloc) (size_t) = malloc;
extern "C" void (*pcre_free) (void *) = free;
extern "C" void *(*pcre_stack_malloc) (size_t) = malloc;
extern "C" void (*pcre_stack_free) (void *) = free;
extern "C" int (*pcre_callout) (pcre_callout_block *) = NULL;
#else
void *(*pcre_malloc) (size_t) = malloc;
void (*pcre_free) (void *) = free;
void *(*pcre_stack_malloc) (size_t) = malloc;
void (*pcre_stack_free) (void *) = free;
int (*pcre_callout) (pcre_callout_block *) = NULL;
#endif
#endif
/*************************************************
* Macros and tables for character handling *
*************************************************/
/* When UTF-8 encoding is being used, a character is no longer just a single
byte. The macros for character handling generate simple sequences when used in
byte-mode, and more complicated ones for UTF-8 characters. */
#define GETCHAR(c, eptr) c = *eptr;
#define GETCHARINC(c, eptr) c = *eptr++;
#define GETCHARINCTEST(c, eptr) c = *eptr++;
#define GETCHARLEN(c, eptr, len) c = *eptr;
#define BACKCHAR(eptr)
/*************************************************
* Handle escapes *
*************************************************/
/* This function is called when a \ has been encountered. It either returns a
positive value for a simple escape such as \n, or a negative value which
encodes one of the more complicated things such as \d. When UTF-8 is enabled,
a positive value greater than 255 may be returned. On entry, ptr is pointing at
the \. On exit, it is on the final character of the escape sequence.
Arguments:
ptrptr points to the pattern position pointer
errorptr points to the pointer to the error message
bracount number of previous extracting brackets
options the options bits
isclass TRUE if inside a character class
Returns: zero or positive => a data character
negative => a special escape sequence
on error, errorptr is set
*/
static int
check_escape(const uschar ** ptrptr, const char **errorptr, int bracount,
int options, BOOL isclass)
{
const uschar *ptr = *ptrptr;
int c, i;
/* If backslash is at the end of the pattern, it's an error. */
c = *(++ptr);
if (c == 0)
*errorptr = ERR1;
/* Non-alphamerics are literals. For digits or letters, do an initial lookup in
a table. A non-zero result is something that can be returned immediately.
Otherwise further processing may be required. */
else if (c < '0' || c > 'z') {
} /* Not alphameric */
else if ((i = escapes[c - '0']) != 0)
c = i;
/* Escapes that need further processing, or are illegal. */
else {
const uschar *oldptr;
switch (c) {
/* A number of Perl escapes are not handled by PCRE. We give an explicit
error. */
case 'l':
case 'L':
case 'N':
case 'p':
case 'P':
case 'u':
case 'U':
case 'X':
*errorptr = ERR37;
break;
/* The handling of escape sequences consisting of a string of digits
starting with one that is not zero is not straightforward. By experiment,
the way Perl works seems to be as follows:
Outside a character class, the digits are read as a decimal number. If the
number is less than 10, or if there are that many previous extracting
left brackets, then it is a back reference. Otherwise, up to three octal
digits are read to form an escaped byte. Thus \123 is likely to be octal
123 (cf \0123, which is octal 012 followed by the literal 3). If the octal
value is greater than 377, the least significant 8 bits are taken. Inside a
character class, \ followed by a digit is always an octal number. */
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
case '8':
case '9':
if (!isclass) {
oldptr = ptr;
c -= '0';
while ((digitab[ptr[1]] & ctype_digit) != 0)
c = c * 10 + *(++ptr) - '0';
if (c < 10 || c <= bracount) {
c = -(ESC_REF + c);
break;
}
ptr = oldptr; /* Put the pointer back and fall through */
}
/* Handle an octal number following \. If the first digit is 8 or 9, Perl
generates a binary zero byte and treats the digit as a following literal.
Thus we have to pull back the pointer by one. */
if ((c = *ptr) >= '8') {
ptr--;
c = 0;
break;
}
/* \0 always starts an octal number, but we may drop through to here with a
larger first octal digit. */
case '0':
c -= '0';
while (i++ < 2 && ptr[1] >= '0' && ptr[1] <= '7')
c = c * 8 + *(++ptr) - '0';
c &= 255; /* Take least significant 8 bits */
break;
/* \x is complicated when UTF-8 is enabled. \x{ddd} is a character number
which can be greater than 0xff, but only if the ddd are hex digits. */
case 'x':
/* Read just a single hex char */
c = 0;
while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0) {
int cc; /* Some compilers don't like ++ */
cc = *(++ptr); /* in initializers */
if (cc >= 'a')
cc -= 32; /* Convert to upper case */
c = c * 16 + cc - ((cc < 'A') ? '0' : ('A' - 10));
}
break;
/* Other special escapes not starting with a digit are straightforward */
case 'c':
c = *(++ptr);
if (c == 0) {
*errorptr = ERR2;
return 0;
}
/* A letter is upper-cased; then the 0x40 bit is flipped. This coding
is ASCII-specific, but then the whole concept of \cx is ASCII-specific.
(However, an EBCDIC equivalent has now been added.) */
if (c >= 'a' && c <= 'z')
c -= 32;
c ^= 0x40;
break;
/* PCRE_EXTRA enables extensions to Perl in the matter of escapes. Any
other alphameric following \ is an error if PCRE_EXTRA was set; otherwise,
for Perl compatibility, it is a literal. This code looks a bit odd, but
there used to be some cases other than the default, and there may be again
in future, so I haven't "optimized" it. */
default:
if ((options & PCRE_EXTRA) != 0)
switch (c) {
default:
*errorptr = ERR3;
break;
}
break;
}
}
*ptrptr = ptr;
return c;
}
/*************************************************
* Check for counted repeat *
*************************************************/
/* This function is called when a '{' is encountered in a place where it might
start a quantifier. It looks ahead to see if it really is a quantifier or not.
It is only a quantifier if it is one of the forms {ddd} {ddd,} or {ddd,ddd}
where the ddds are digits.
Arguments:
p pointer to the first char after '{'
Returns: TRUE or FALSE
*/
static BOOL
is_counted_repeat(const uschar * p)
{
if ((digitab[*p++] & ctype_digit) == 0)
return FALSE;
while ((digitab[*p] & ctype_digit) != 0)
p++;
if (*p == '}')
return TRUE;
if (*p++ != ',')
return FALSE;
if (*p == '}')
return TRUE;
if ((digitab[*p++] & ctype_digit) == 0)
return FALSE;
while ((digitab[*p] & ctype_digit) != 0)
p++;
return (*p == '}');
}
/*************************************************
* Read repeat counts *
*************************************************/
/* Read an item of the form {n,m} and return the values. This is called only
after is_counted_repeat() has confirmed that a repeat-count quantifier exists,
so the syntax is guaranteed to be correct, but we need to check the values.
Arguments:
p pointer to first char after '{'
minp pointer to int for min
maxp pointer to int for max
returned as -1 if no max
errorptr points to pointer to error message
Returns: pointer to '}' on success;
current ptr on error, with errorptr set
*/
static const uschar *
read_repeat_counts(const uschar * p, int *minp, int *maxp,
const char **errorptr)
{
int min = 0;
int max = -1;
while ((digitab[*p] & ctype_digit) != 0)
min = min * 10 + *p++ - '0';
if (*p == '}')
max = min;
else {
if (*(++p) != '}') {
max = 0;
while ((digitab[*p] & ctype_digit) != 0)
max = max * 10 + *p++ - '0';
if (max < min) {
*errorptr = ERR4;
return p;
}
}
}
/* Do paranoid checks, then fill in the required variables, and pass back the
pointer to the terminating '}'. */
if (min > 65535 || max > 65535)
*errorptr = ERR5;
else {
*minp = min;
*maxp = max;
}
return p;
}
/*************************************************
* Find first significant op code *
*************************************************/
/* This is called by several functions that scan a compiled expression looking
for a fixed first character, or an anchoring op code etc. It skips over things
that do not influence this. For some calls, a change of option is important.
Arguments:
code pointer to the start of the group
options pointer to external options
optbit the option bit whose changing is significant, or
zero if none are
Returns: pointer to the first significant opcode
*/
static const uschar *
first_significant_code(const uschar * code, int *options, int optbit)
{
for (;;) {
switch ((int) *code) {
case OP_OPT:
if (optbit > 0 && ((int) code[1] & optbit) != (*options & optbit))
*options = (int) code[1];
code += 2;
break;
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
do
code += GET(code, 1);
while (*code == OP_ALT);
/* Fall through */
case OP_CALLOUT:
case OP_CREF:
case OP_BRANUMBER:
case OP_WORD_BOUNDARY:
case OP_NOT_WORD_BOUNDARY:
code += OP_lengths[*code];
break;
default:
return code;
}
}
/* Control never reaches here */
}
/*************************************************
* Find the fixed length of a pattern *
*************************************************/
/* Scan a pattern and compute the fixed length of subject that will match it,
if the length is fixed. This is needed for dealing with backward assertions.
In UTF8 mode, the result is in characters rather than bytes.
Arguments:
code points to the start of the pattern (the bracket)
options the compiling options
Returns: the fixed length, or -1 if there is no fixed length,
or -2 if \C was encountered
*/
static int
find_fixedlength(uschar * code, int options)
{
int length = -1;
register int branchlength = 0;
register uschar *cc = code + 1 + LINK_SIZE;
/* Scan along the opcodes for this branch. If we get to the end of the
branch, check the length against that of the other branches. */
for (;;) {
int d;
register int op = *cc;
if (op >= OP_BRA)
op = OP_BRA;
switch (op) {
case OP_BRA:
case OP_ONCE:
case OP_COND:
d = find_fixedlength(cc, options);
if (d < 0)
return d;
branchlength += d;
do
cc += GET(cc, 1);
while (*cc == OP_ALT);
cc += 1 + LINK_SIZE;
break;
/* Reached end of a branch; if it's a ket it is the end of a nested
call. If it's ALT it is an alternation in a nested call. If it is
END it's the end of the outer call. All can be handled by the same code. */
case OP_ALT:
case OP_KET:
case OP_KETRMAX:
case OP_KETRMIN:
case OP_END:
if (length < 0)
length = branchlength;
else if (length != branchlength)
return -1;
if (*cc != OP_ALT)
return length;
cc += 1 + LINK_SIZE;
branchlength = 0;
break;
/* Skip over assertive subpatterns */
case OP_ASSERT:
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
do
cc += GET(cc, 1);
while (*cc == OP_ALT);
/* Fall through */
/* Skip over things that don't match chars */
case OP_REVERSE:
case OP_BRANUMBER:
case OP_CREF:
case OP_OPT:
case OP_CALLOUT:
case OP_SOD:
case OP_SOM:
case OP_EOD:
case OP_EODN:
case OP_CIRC:
case OP_DOLL:
case OP_NOT_WORD_BOUNDARY:
case OP_WORD_BOUNDARY:
cc += OP_lengths[*cc];
break;
/* Handle char strings. In UTF-8 mode we must count characters, not bytes.
This requires a scan of the string, unfortunately. We assume valid UTF-8
strings, so all we do is reduce the length by one for every byte whose bits
are 10xxxxxx. */
case OP_CHARS:
branchlength += *(++cc);
cc += *cc + 1;
break;
/* Handle exact repetitions. The count is already in characters, but we
need to skip over a multibyte character in UTF8 mode. */
case OP_EXACT:
branchlength += GET2(cc, 1);
cc += 4;
break;
case OP_TYPEEXACT:
branchlength += GET2(cc, 1);
cc += 4;
break;
/* Handle single-char matchers */
case OP_NOT_DIGIT:
case OP_DIGIT:
case OP_NOT_WHITESPACE:
case OP_WHITESPACE:
case OP_NOT_WORDCHAR:
case OP_WORDCHAR:
case OP_ANY:
branchlength++;
cc++;
break;
/* The single-byte matcher isn't allowed */
case OP_ANYBYTE:
return -2;
/* Check a class for variable quantification */
case OP_CLASS:
case OP_NCLASS:
cc += 33;
switch (*cc) {
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRQUERY:
case OP_CRMINQUERY:
return -1;
case OP_CRRANGE:
case OP_CRMINRANGE:
if (GET2(cc, 1) != GET2(cc, 3))
return -1;
branchlength += GET2(cc, 1);
cc += 5;
break;
default:
branchlength++;
}
break;
/* Anything else is variable length */
default:
return -1;
}
}
/* Control never gets here */
}
/*************************************************
* Scan compiled regex for numbered bracket *
*************************************************/
/* This little function scans through a compiled pattern until it finds a
capturing bracket with the given number.
Arguments:
code points to start of expression
utf8 TRUE in UTF-8 mode
number the required bracket number
Returns: pointer to the opcode for the bracket, or NULL if not found
*/
static const uschar *
find_bracket(const uschar * code, BOOL utf8, int number)
{
utf8 = utf8; /* Stop pedantic compilers complaining */
for (;;) {
register int c = *code;
if (c == OP_END)
return NULL;
else if (c == OP_CHARS)
code += code[1] + OP_lengths[c];
else if (c > OP_BRA) {
int n = c - OP_BRA;
if (n > EXTRACT_BASIC_MAX)
n = GET2(code, 2 + LINK_SIZE);
if (n == number)
return (uschar *) code;
code += OP_lengths[OP_BRA];
} else {
code += OP_lengths[c];
}
}
}
/*************************************************
* Scan compiled regex for recursion reference *
*************************************************/
/* This little function scans through a compiled pattern until it finds an
instance of OP_RECURSE.
Arguments:
code points to start of expression
utf8 TRUE in UTF-8 mode
Returns: pointer to the opcode for OP_RECURSE, or NULL if not found
*/
static const uschar *
find_recurse(const uschar * code, BOOL utf8)
{
utf8 = utf8; /* Stop pedantic compilers complaining */
for (;;) {
register int c = *code;
if (c == OP_END)
return NULL;
else if (c == OP_RECURSE)
return code;
else if (c == OP_CHARS)
code += code[1] + OP_lengths[c];
else if (c > OP_BRA) {
code += OP_lengths[OP_BRA];
} else {
code += OP_lengths[c];
}
}
}
/*************************************************
* Scan compiled branch for non-emptiness *
*************************************************/
/* This function scans through a branch of a compiled pattern to see whether it
can match the empty string or not. It is called only from could_be_empty()
below. Note that first_significant_code() skips over assertions. If we hit an
unclosed bracket, we return "empty" - this means we've struck an inner bracket
whose current branch will already have been scanned.
Arguments:
code points to start of search
endcode points to where to stop
utf8 TRUE if in UTF8 mode
Returns: TRUE if what is matched could be empty
*/
static BOOL
could_be_empty_branch(const uschar * code, const uschar * endcode, BOOL utf8)
{
register int c;
for (code = first_significant_code(code + 1 + LINK_SIZE, NULL, 0);
code < endcode;
code = first_significant_code(code + OP_lengths[c], NULL, 0)) {
const uschar *ccode;
c = *code;
if (c >= OP_BRA) {
BOOL empty_branch;
if (GET(code, 1) == 0)
return TRUE; /* Hit unclosed bracket */
/* Scan a closed bracket */
empty_branch = FALSE;
do {
if (!empty_branch && could_be_empty_branch(code, endcode, utf8))
empty_branch = TRUE;
code += GET(code, 1);
}
while (*code == OP_ALT);
if (!empty_branch)
return FALSE; /* All branches are non-empty */
code += 1 + LINK_SIZE;
c = *code;
}
else
switch (c) {
/* Check for quantifiers after a class */
case OP_CLASS:
case OP_NCLASS:
ccode = code + 33;
switch (*ccode) {
case OP_CRSTAR: /* These could be empty; continue */
case OP_CRMINSTAR:
case OP_CRQUERY:
case OP_CRMINQUERY:
break;
default: /* Non-repeat => class must match */
case OP_CRPLUS: /* These repeats aren't empty */
case OP_CRMINPLUS:
return FALSE;
case OP_CRRANGE:
case OP_CRMINRANGE:
if (GET2(ccode, 1) > 0)
return FALSE; /* Minimum > 0 */
break;
}
break;
/* Opcodes that must match a character */
case OP_NOT_DIGIT:
case OP_DIGIT:
case OP_NOT_WHITESPACE:
case OP_WHITESPACE:
case OP_NOT_WORDCHAR:
case OP_WORDCHAR:
case OP_ANY:
case OP_ANYBYTE:
case OP_CHARS:
case OP_NOT:
case OP_PLUS:
case OP_MINPLUS:
case OP_EXACT:
case OP_NOTPLUS:
case OP_NOTMINPLUS:
case OP_NOTEXACT:
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEEXACT:
return FALSE;
/* End of branch */
case OP_KET:
case OP_KETRMAX:
case OP_KETRMIN:
case OP_ALT:
return TRUE;
/* In UTF-8 mode, STAR, MINSTAR, QUERY, MINQUERY, UPTO, and MINUPTO may be
followed by a multibyte character */
}
}
return TRUE;
}
/*************************************************
* Scan compiled regex for non-emptiness *
*************************************************/
/* This function is called to check for left recursive calls. We want to check
the current branch of the current pattern to see if it could match the empty
string. If it could, we must look outwards for branches at other levels,
stopping when we pass beyond the bracket which is the subject of the recursion.
Arguments:
code points to start of the recursion
endcode points to where to stop (current RECURSE item)
bcptr points to the chain of current (unclosed) branch starts
utf8 TRUE if in UTF-8 mode
Returns: TRUE if what is matched could be empty
*/
static BOOL
could_be_empty(const uschar * code, const uschar * endcode,
branch_chain * bcptr, BOOL utf8)
{
while (bcptr != NULL && bcptr->current >= code) {
if (!could_be_empty_branch(bcptr->current, endcode, utf8))
return FALSE;
bcptr = bcptr->outer;
}
return TRUE;
}
/*************************************************
* Check for POSIX class syntax *
*************************************************/
/* This function is called when the sequence "[:" or "[." or "[=" is
encountered in a character class. It checks whether this is followed by an
optional ^ and then a sequence of letters, terminated by a matching ":]" or
".]" or "=]".
Argument:
ptr pointer to the initial [
endptr where to return the end pointer
cd pointer to compile data
Returns: TRUE or FALSE
*/
static BOOL
check_posix_syntax(const uschar * ptr, const uschar ** endptr,
compile_data * cd)
{
int terminator; /* Don't combine these lines; the Solaris cc */
terminator = *(++ptr); /* compiler warns about "non-constant" initializer. */
if (*(++ptr) == '^')
ptr++;
while ((cd->ctypes[*ptr] & ctype_letter) != 0)
ptr++;
if (*ptr == terminator && ptr[1] == ']') {
*endptr = ptr;
return TRUE;
}
return FALSE;
}
/*************************************************
* Check POSIX class name *
*************************************************/
/* This function is called to check the name given in a POSIX-style class entry
such as [:alnum:].
Arguments:
ptr points to the first letter
len the length of the name
Returns: a value representing the name, or -1 if unknown
*/
static int
check_posix_name(const uschar * ptr, int len)
{
register int yield = 0;
while (posix_name_lengths[yield] != 0) {
if (len == posix_name_lengths[yield] &&
strncmp((const char *) ptr, posix_names[yield], len) == 0)
return yield;
yield++;
}
return -1;
}
/*************************************************
* Adjust OP_RECURSE items in repeated group *
*************************************************/
/* OP_RECURSE items contain an offset from the start of the regex to the group
that is referenced. This means that groups can be replicated for fixed
repetition simply by copying (because the recursion is allowed to refer to
earlier groups that are outside the current group). However, when a group is
optional (i.e. the minimum quantifier is zero), OP_BRAZERO is inserted before
it, after it has been compiled. This means that any OP_RECURSE items within it
that refer to the group itself or any contained groups have to have their
offsets adjusted. That is the job of this function. Before it is called, the
partially compiled regex must be temporarily terminated with OP_END.
Arguments:
group points to the start of the group
adjust the amount by which the group is to be moved
utf8 TRUE in UTF-8 mode
cd contains pointers to tables etc.
Returns: nothing
*/
static void
adjust_recurse(uschar * group, int adjust, BOOL utf8, compile_data * cd)
{
uschar *ptr = group;
while ((ptr = (uschar *) find_recurse(ptr, utf8)) != NULL) {
int offset = GET(ptr, 1);
if (cd->start_code + offset >= group)
PUT(ptr, 1, offset + adjust);
ptr += 1 + LINK_SIZE;
}
}
/*************************************************
* Compile one branch *
*************************************************/
/* Scan the pattern, compiling it into the code vector. If the options are
changed during the branch, the pointer is used to change the external options
bits.
Arguments:
optionsptr pointer to the option bits
brackets points to number of extracting brackets used
code points to the pointer to the current code point
ptrptr points to the current pattern pointer
errorptr points to pointer to error message
firstbyteptr set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)
reqbyteptr set to the last literal character required, else < 0
bcptr points to current branch chain
cd contains pointers to tables etc.
Returns: TRUE on success
FALSE, with *errorptr set on error
*/
static BOOL
compile_branch(int *optionsptr, int *brackets, uschar ** codeptr,
const uschar ** ptrptr, const char **errorptr, int *firstbyteptr,
int *reqbyteptr, branch_chain * bcptr, compile_data * cd)
{
int repeat_type, op_type;
int repeat_min = 0, repeat_max = 0; /* To please picky compilers */
int bravalue = 0;
int length;
int greedy_default, greedy_non_default;
int firstbyte, reqbyte;
int zeroreqbyte, zerofirstbyte;
int req_caseopt, reqvary, tempreqvary;
int condcount = 0;
int options = *optionsptr;
register int c;
register uschar *code = *codeptr;
uschar *tempcode;
BOOL inescq = FALSE;
BOOL groupsetfirstbyte = FALSE;
const uschar *ptr = *ptrptr;
const uschar *tempptr;
uschar *previous = NULL;
uschar class[32];
BOOL utf8 = FALSE;
/* Set up the default and non-default settings for greediness */
greedy_default = ((options & PCRE_UNGREEDY) != 0);
greedy_non_default = greedy_default ^ 1;
/* Initialize no first char, no required char. REQ_UNSET means "no char
matching encountered yet". It gets changed to REQ_NONE if we hit something that
matches a non-fixed char first char; reqbyte just remains unset if we never
find one.
When we hit a repeat whose minimum is zero, we may have to adjust these values
to take the zero repeat into account. This is implemented by setting them to
zerofirstbyte and zeroreqbyte when such a repeat is encountered. The individual
item types that can be repeated set these backoff variables appropriately. */
firstbyte = reqbyte = zerofirstbyte = zeroreqbyte = REQ_UNSET;
/* The variable req_caseopt contains either the REQ_CASELESS value or zero,
according to the current setting of the caseless flag. REQ_CASELESS is a bit
value > 255. It is added into the firstbyte or reqbyte variables to record the
case status of the value. */
req_caseopt = ((options & PCRE_CASELESS) != 0) ? REQ_CASELESS : 0;
/* Switch on next character until the end of the branch */
for (;; ptr++) {
BOOL negate_class;
BOOL possessive_quantifier;
int class_charcount;
int class_lastchar;
int newoptions;
int recno;
int skipbytes;
int subreqbyte;
int subfirstbyte;
c = *ptr;
if (inescq && c != 0)
goto NORMAL_CHAR;
if ((options & PCRE_EXTENDED) != 0) {
if ((cd->ctypes[c] & ctype_space) != 0)
continue;
if (c == '#') {
/* The space before the ; is to avoid a warning on a silly compiler
on the Macintosh. */
while ((c = *(++ptr)) != 0 && c != NEWLINE) ;
if (c != 0)
continue; /* Else fall through to handle end of string */
}
}
switch (c) {
/* The branch terminates at end of string, |, or ). */
case 0:
case '|':
case ')':
*firstbyteptr = firstbyte;
*reqbyteptr = reqbyte;
*codeptr = code;
*ptrptr = ptr;
return TRUE;
/* Handle single-character metacharacters. In multiline mode, ^ disables
the setting of any following char as a first character. */
case '^':
if ((options & PCRE_MULTILINE) != 0) {
if (firstbyte == REQ_UNSET)
firstbyte = REQ_NONE;
}
previous = NULL;
*code++ = OP_CIRC;
break;
case '$':
previous = NULL;
*code++ = OP_DOLL;
break;
/* There can never be a first char if '.' is first, whatever happens about
repeats. The value of reqbyte doesn't change either. */
case '.':
if (firstbyte == REQ_UNSET)
firstbyte = REQ_NONE;
zerofirstbyte = firstbyte;
zeroreqbyte = reqbyte;
previous = code;
*code++ = OP_ANY;
break;
/* Character classes. If the included characters are all < 255 in value, we
build a 32-byte bitmap of the permitted characters, except in the special
case where there is only one such character. For negated classes, we build
the map as usual, then invert it at the end. However, we use a different
opcode so that data characters > 255 can be handled correctly.
If the class contains characters outside the 0-255 range, a different
opcode is compiled. It may optionally have a bit map for characters < 256,
but those above are are explicitly listed afterwards. A flag byte tells
whether the bitmap is present, and whether this is a negated class or not.
*/
case '[':
previous = code;
/* PCRE supports POSIX class stuff inside a class. Perl gives an error if
they are encountered at the top level, so we'll do that too. */
if ((ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&
check_posix_syntax(ptr, &tempptr, cd)) {
*errorptr = (ptr[1] == ':') ? ERR13 : ERR31;
goto FAILED;
}
/* If the first character is '^', set the negation flag and skip it. */
if ((c = *(++ptr)) == '^') {
negate_class = TRUE;
c = *(++ptr);
} else {
negate_class = FALSE;
}
/* Keep a count of chars with values < 256 so that we can optimize the case
of just a single character (as long as it's < 256). For higher valued UTF-8
characters, we don't yet do any optimization. */
class_charcount = 0;
class_lastchar = -1;
/* Initialize the 32-char bit map to all zeros. We have to build the
map in a temporary bit of store, in case the class contains only 1
character (< 256), because in that case the compiled code doesn't use the
bit map. */
memset(class, 0, 32 * sizeof(uschar));
/* Process characters until ] is reached. By writing this as a "do" it
means that an initial ] is taken as a data character. The first pass
through the regex checked the overall syntax, so we don't need to be very
strict here. At the start of the loop, c contains the first byte of the
character. */
do {
/* Inside \Q...\E everything is literal except \E */
if (inescq) {
if (c == '\\' && ptr[1] == 'E') {
inescq = FALSE;
ptr++;
continue;
} else
goto LONE_SINGLE_CHARACTER;
}
/* Handle POSIX class names. Perl allows a negation extension of the
form [:^name:]. A square bracket that doesn't match the syntax is
treated as a literal. We also recognize the POSIX constructions
[.ch.] and [=ch=] ("collating elements") and fault them, as Perl
5.6 and 5.8 do. */
if (c == '[' &&
(ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&
check_posix_syntax(ptr, &tempptr, cd)) {
BOOL local_negate = FALSE;
int posix_class, i;
register const uschar *cbits = cd->cbits;
if (ptr[1] != ':') {
*errorptr = ERR31;
goto FAILED;
}
ptr += 2;
if (*ptr == '^') {
local_negate = TRUE;
ptr++;
}
posix_class = check_posix_name(ptr, tempptr - ptr);
if (posix_class < 0) {
*errorptr = ERR30;
goto FAILED;
}
/* If matching is caseless, upper and lower are converted to
alpha. This relies on the fact that the class table starts with
alpha, lower, upper as the first 3 entries. */
if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
posix_class = 0;
/* Or into the map we are building up to 3 of the static class
tables, or their negations. The [:blank:] class sets up the same
chars as the [:space:] class (all white space). We remove the vertical
white space chars afterwards. */
posix_class *= 3;
for (i = 0; i < 3; i++) {
BOOL blankclass = strncmp((char *) ptr, "blank", 5) == 0;
int taboffset = posix_class_maps[posix_class + i];
if (taboffset < 0)
break;
if (local_negate) {
for (c = 0; c < 32; c++)
class[c] |= ~cbits[c + taboffset];
if (blankclass)
class[1] |= 0x3c;
} else {
for (c = 0; c < 32; c++)
class[c] |= cbits[c + taboffset];
if (blankclass)
class[1] &= ~0x3c;
}
}
ptr = tempptr + 1;
class_charcount = 10; /* Set > 1; assumes more than 1 per class */
continue; /* End of POSIX syntax handling */
}
/* Backslash may introduce a single character, or it may introduce one
of the specials, which just set a flag. Escaped items are checked for
validity in the pre-compiling pass. The sequence \b is a special case.
Inside a class (and only there) it is treated as backspace. Elsewhere
it marks a word boundary. Other escapes have preset maps ready to
or into the one we are building. We assume they have more than one
character in them, so set class_charcount bigger than one. */
if (c == '\\') {
c = check_escape(&ptr, errorptr, *brackets, options, TRUE);
if (-c == ESC_b)
c = '\b'; /* \b is backslash in a class */
if (-c == ESC_Q) { /* Handle start of quoted string */
if (ptr[1] == '\\' && ptr[2] == 'E') {
ptr += 2; /* avoid empty string */
} else
inescq = TRUE;
continue;
}
else if (c < 0) {
register const uschar *cbits = cd->cbits;
class_charcount = 10; /* Greater than 1 is what matters */
switch (-c) {
case ESC_d:
for (c = 0; c < 32; c++)
class[c] |= cbits[c + cbit_digit];
continue;
case ESC_D:
for (c = 0; c < 32; c++)
class[c] |= ~cbits[c + cbit_digit];
continue;
case ESC_w:
for (c = 0; c < 32; c++)
class[c] |= cbits[c + cbit_word];
continue;
case ESC_W:
for (c = 0; c < 32; c++)
class[c] |= ~cbits[c + cbit_word];
continue;
case ESC_s:
for (c = 0; c < 32; c++)
class[c] |= cbits[c + cbit_space];
class[1] &= ~0x08; /* Perl 5.004 onwards omits VT from \s */
continue;
case ESC_S:
for (c = 0; c < 32; c++)
class[c] |= ~cbits[c + cbit_space];
class[1] |= 0x08; /* Perl 5.004 onwards omits VT from \s */
continue;
/* Unrecognized escapes are faulted if PCRE is running in its
strict mode. By default, for compatibility with Perl, they are
treated as literals. */
default:
if ((options & PCRE_EXTRA) != 0) {
*errorptr = ERR7;
goto FAILED;
}
c = *ptr; /* The final character */
}
}
/* Fall through if we have a single character (c >= 0). This may be
> 256 in UTF-8 mode. */
}
/* End of backslash handling */
/* A single character may be followed by '-' to form a range. However,
Perl does not permit ']' to be the end of the range. A '-' character
here is treated as a literal. */
if (ptr[1] == '-' && ptr[2] != ']') {
int d;
ptr += 2;
d = *ptr;
/* The second part of a range can be a single-character escape, but
not any of the other escapes. Perl 5.6 treats a hyphen as a literal
in such circumstances. */
if (d == '\\') {
const uschar *oldptr = ptr;
d = check_escape(&ptr, errorptr, *brackets, options, TRUE);
/* \b is backslash; any other special means the '-' was literal */
if (d < 0) {
if (d == -ESC_b)
d = '\b';
else {
ptr = oldptr - 2;
goto LONE_SINGLE_CHARACTER; /* A few lines below */
}
}
}
/* Check that the two values are in the correct order */
if (d < c) {
*errorptr = ERR8;
goto FAILED;
}
/* If d is greater than 255, we can't just use the bit map, so set up
for the UTF-8 supporting class type. If we are not caseless, we can
just set up a single range. If we are caseless, the characters < 256
are handled with a bitmap, in order to get the case-insensitive
handling. */
/* We use the bit map if the range is entirely < 255, or if part of it
is < 255 and matching is caseless. */
for (; c <= d; c++) {
class[c / 8] |= (1 << (c & 7));
if ((options & PCRE_CASELESS) != 0) {
int uc = cd->fcc[c]; /* flip case */
class[uc / 8] |= (1 << (uc & 7));
}
class_charcount++; /* in case a one-char range */
class_lastchar = c;
}
continue; /* Go get the next char in the class */
}
/* Handle a lone single character - we can get here for a normal
non-escape char, or after \ that introduces a single character. */
LONE_SINGLE_CHARACTER:
/* Handle a multibyte character */
/* Handle a single-byte character */
{
class[c / 8] |= (1 << (c & 7));
if ((options & PCRE_CASELESS) != 0) {
c = cd->fcc[c]; /* flip case */
class[c / 8] |= (1 << (c & 7));
}
class_charcount++;
class_lastchar = c;
}
}
/* Loop until ']' reached; the check for end of string happens inside the
loop. This "while" is the end of the "do" above. */
while ((c = *(++ptr)) != ']' || inescq);
/* If class_charcount is 1, we saw precisely one character with a value <
256. In UTF-8 mode, we can optimize if there were no characters >= 256 and
the one character is < 128. In non-UTF-8 mode we can always optimize.
The optimization throws away the bit map. We turn the item into a
1-character OP_CHARS if it's positive, or OP_NOT if it's negative. Note
that OP_NOT does not support multibyte characters. In the positive case, it
can cause firstbyte to be set. Otherwise, there can be no first char if
this item is first, whatever repeat count may follow. In the case of
reqbyte, save the previous value for reinstating. */
if (class_charcount == 1) {
zeroreqbyte = reqbyte;
if (negate_class) {
if (firstbyte == REQ_UNSET)
firstbyte = REQ_NONE;
zerofirstbyte = firstbyte;
*code++ = OP_NOT;
} else {
if (firstbyte == REQ_UNSET) {
zerofirstbyte = REQ_NONE;
firstbyte = class_lastchar | req_caseopt;
} else {
zerofirstbyte = firstbyte;
reqbyte = class_lastchar | req_caseopt | cd->req_varyopt;
}
*code++ = OP_CHARS;
*code++ = 1;
}
*code++ = class_lastchar;
break; /* End of class handling */
}
/* End of 1-byte optimization */
/* Otherwise, if this is the first thing in the branch, there can be no
first char setting, whatever the repeat count. Any reqbyte setting must
remain unchanged after any kind of repeat. */
if (firstbyte == REQ_UNSET)
firstbyte = REQ_NONE;
zerofirstbyte = firstbyte;
zeroreqbyte = reqbyte;
/* If there are characters with values > 255, we have to compile an
extended class, with its own opcode. If there are no characters < 256,
we can omit the bitmap. */
/* If there are no characters > 255, negate the 32-byte map if necessary,
and copy it into the code vector. If this is the first thing in the branch,
there can be no first char setting, whatever the repeat count. Any reqbyte
setting must remain unchanged after any kind of repeat. */
if (negate_class) {
*code++ = OP_NCLASS;
for (c = 0; c < 32; c++)
code[c] = ~class[c];
} else {
*code++ = OP_CLASS;
memcpy(code, class, 32);
}
code += 32;
break;
/* Various kinds of repeat */
case '{':
if (!is_counted_repeat(ptr + 1))
goto NORMAL_CHAR;
ptr = read_repeat_counts(ptr + 1, &repeat_min, &repeat_max, errorptr);
if (*errorptr != NULL)
goto FAILED;
goto REPEAT;
case '*':
repeat_min = 0;
repeat_max = -1;
goto REPEAT;
case '+':
repeat_min = 1;
repeat_max = -1;
goto REPEAT;
case '?':
repeat_min = 0;
repeat_max = 1;
REPEAT:
if (previous == NULL) {
*errorptr = ERR9;
goto FAILED;
}
if (repeat_min == 0) {
firstbyte = zerofirstbyte; /* Adjust for zero repeat */
reqbyte = zeroreqbyte; /* Ditto */
}
/* Remember whether this is a variable length repeat */
reqvary = (repeat_min == repeat_max) ? 0 : REQ_VARY;
op_type = 0; /* Default single-char op codes */
possessive_quantifier = FALSE; /* Default not possessive quantifier */
/* Save start of previous item, in case we have to move it up to make space
for an inserted OP_ONCE for the additional '+' extension. */
tempcode = previous;
/* If the next character is '+', we have a possessive quantifier. This
implies greediness, whatever the setting of the PCRE_UNGREEDY option.
If the next character is '?' this is a minimizing repeat, by default,
but if PCRE_UNGREEDY is set, it works the other way round. We change the
repeat type to the non-default. */
if (ptr[1] == '+') {
repeat_type = 0; /* Force greedy */
possessive_quantifier = TRUE;
ptr++;
} else if (ptr[1] == '?') {
repeat_type = greedy_non_default;
ptr++;
} else
repeat_type = greedy_default;
/* If previous was a recursion, we need to wrap it inside brackets so that
it can be replicated if necessary. */
if (*previous == OP_RECURSE) {
memmove(previous + 1 + LINK_SIZE, previous, 1 + LINK_SIZE);
code += 1 + LINK_SIZE;
*previous = OP_BRA;
PUT(previous, 1, code - previous);
*code = OP_KET;
PUT(code, 1, code - previous);
code += 1 + LINK_SIZE;
}
/* If previous was a string of characters, chop off the last one and use it
as the subject of the repeat. If there was only one character, we can
abolish the previous item altogether. If a one-char item has a minumum of
more than one, ensure that it is set in reqbyte - it might not be if a
sequence such as x{3} is the first thing in a branch because the x will
have gone into firstbyte instead. */
if (*previous == OP_CHARS) {
/* Deal with UTF-8 characters that take up more than one byte. It's
easier to write this out separately than try to macrify it. Use c to
hold the length of the character in bytes, plus 0x80 to flag that it's a
length rather than a small character. */
/* Handle the case of a single byte - either with no UTF8 support, or
with UTF-8 disabled, or for a UTF-8 character < 128. */
{
c = *(--code);
if (code == previous + 2) { /* There was only one character */
code = previous; /* Abolish the previous item */
if (repeat_min > 1)
reqbyte = c | req_caseopt | cd->req_varyopt;
} else {
previous[1]--; /* adjust length */
tempcode = code; /* Adjust position to be moved for '+' */
}
}
goto OUTPUT_SINGLE_REPEAT; /* Code shared with single character types */
}
/* If previous was a single negated character ([^a] or similar), we use
one of the special opcodes, replacing it. The code is shared with single-
character repeats by setting opt_type to add a suitable offset into
repeat_type. OP_NOT is currently used only for single-byte chars. */
else if (*previous == OP_NOT) {
op_type = OP_NOTSTAR - OP_STAR; /* Use "not" opcodes */
c = previous[1];
code = previous;
goto OUTPUT_SINGLE_REPEAT;
}
/* If previous was a character type match (\d or similar), abolish it and
create a suitable repeat item. The code is shared with single-character
repeats by setting op_type to add a suitable offset into repeat_type. */
else if (*previous < OP_EODN) {
op_type = OP_TYPESTAR - OP_STAR; /* Use type opcodes */
c = *previous;
code = previous;
OUTPUT_SINGLE_REPEAT:
/* If the maximum is zero then the minimum must also be zero; Perl allows
this case, so we do too - by simply omitting the item altogether. */
if (repeat_max == 0)
goto END_REPEAT;
/* Combine the op_type with the repeat_type */
repeat_type += op_type;
/* A minimum of zero is handled either as the special case * or ?, or as
an UPTO, with the maximum given. */
if (repeat_min == 0) {
if (repeat_max == -1)
*code++ = OP_STAR + repeat_type;
else if (repeat_max == 1)
*code++ = OP_QUERY + repeat_type;
else {
*code++ = OP_UPTO + repeat_type;
PUT2INC(code, 0, repeat_max);
}
}
/* The case {1,} is handled as the special case + */
else if (repeat_min == 1 && repeat_max == -1)
*code++ = OP_PLUS + repeat_type;
/* The case {n,n} is just an EXACT, while the general case {n,m} is
handled as an EXACT followed by an UPTO. An EXACT of 1 is optimized. */
else {
if (repeat_min != 1) {
*code++ = OP_EXACT + op_type; /* NB EXACT doesn't have repeat_type */
PUT2INC(code, 0, repeat_min);
}
/* If the mininum is 1 and the previous item was a character string,
we either have to put back the item that got cancelled if the string
length was 1, or add the character back onto the end of a longer
string. For a character type nothing need be done; it will just get
put back naturally. Note that the final character is always going to
get added below, so we leave code ready for its insertion. */
else if (*previous == OP_CHARS) {
if (code == previous)
code += 2;
else
/* In UTF-8 mode, a multibyte char has its length in c, with the 0x80
bit set as a flag. The length will always be between 2 and 6. */
previous[1]++;
}
/* For a single negated character we also have to put back the
item that got cancelled. At present this applies only to single byte
characters in any mode. */
else if (*previous == OP_NOT)
code++;
/* If the maximum is unlimited, insert an OP_STAR. Before doing so,
we have to insert the character for the previous code. In UTF-8 mode,
long characters have their length in c, with the 0x80 bit as a flag. */
if (repeat_max < 0) {
*code++ = c;
*code++ = OP_STAR + repeat_type;
}
/* Else insert an UPTO if the max is greater than the min, again
preceded by the character, for the previously inserted code. */
else if (repeat_max != repeat_min) {
*code++ = c;
repeat_max -= repeat_min;
*code++ = OP_UPTO + repeat_type;
PUT2INC(code, 0, repeat_max);
}
}
/* The character or character type itself comes last in all cases. */
*code++ = c;
}
/* If previous was a character class or a back reference, we put the repeat
stuff after it, but just skip the item if the repeat was {0,0}. */
else if (*previous == OP_CLASS ||
*previous == OP_NCLASS || *previous == OP_REF) {
if (repeat_max == 0) {
code = previous;
goto END_REPEAT;
}
if (repeat_min == 0 && repeat_max == -1)
*code++ = OP_CRSTAR + repeat_type;
else if (repeat_min == 1 && repeat_max == -1)
*code++ = OP_CRPLUS + repeat_type;
else if (repeat_min == 0 && repeat_max == 1)
*code++ = OP_CRQUERY + repeat_type;
else {
*code++ = OP_CRRANGE + repeat_type;
PUT2INC(code, 0, repeat_min);
if (repeat_max == -1)
repeat_max = 0; /* 2-byte encoding for max */
PUT2INC(code, 0, repeat_max);
}
}
/* If previous was a bracket group, we may have to replicate it in certain
cases. */
else if (*previous >= OP_BRA || *previous == OP_ONCE ||
*previous == OP_COND) {
register int i;
int ketoffset = 0;
int len = code - previous;
uschar *bralink = NULL;
/* If the maximum repeat count is unlimited, find the end of the bracket
by scanning through from the start, and compute the offset back to it
from the current code pointer. There may be an OP_OPT setting following
the final KET, so we can't find the end just by going back from the code
pointer. */
if (repeat_max == -1) {
register uschar *ket = previous;
do
ket += GET(ket, 1);
while (*ket != OP_KET);
ketoffset = code - ket;
}
/* The case of a zero minimum is special because of the need to stick
OP_BRAZERO in front of it, and because the group appears once in the
data, whereas in other cases it appears the minimum number of times. For
this reason, it is simplest to treat this case separately, as otherwise
the code gets far too messy. There are several special subcases when the
minimum is zero. */
if (repeat_min == 0) {
/* If the maximum is also zero, we just omit the group from the output
altogether. */
if (repeat_max == 0) {
code = previous;
goto END_REPEAT;
}
/* If the maximum is 1 or unlimited, we just have to stick in the
BRAZERO and do no more at this point. However, we do need to adjust
any OP_RECURSE calls inside the group that refer to the group itself or
any internal group, because the offset is from the start of the whole
regex. Temporarily terminate the pattern while doing this. */
if (repeat_max <= 1) {
*code = OP_END;
adjust_recurse(previous, 1, utf8, cd);
memmove(previous + 1, previous, len);
code++;
*previous++ = OP_BRAZERO + repeat_type;
}
/* If the maximum is greater than 1 and limited, we have to replicate
in a nested fashion, sticking OP_BRAZERO before each set of brackets.
The first one has to be handled carefully because it's the original
copy, which has to be moved up. The remainder can be handled by code
that is common with the non-zero minimum case below. We have to
adjust the value or repeat_max, since one less copy is required. Once
again, we may have to adjust any OP_RECURSE calls inside the group. */
else {
int offset;
*code = OP_END;
adjust_recurse(previous, 2 + LINK_SIZE, utf8, cd);
memmove(previous + 2 + LINK_SIZE, previous, len);
code += 2 + LINK_SIZE;
*previous++ = OP_BRAZERO + repeat_type;
*previous++ = OP_BRA;
/* We chain together the bracket offset fields that have to be
filled in later when the ends of the brackets are reached. */
offset = (bralink == NULL) ? 0 : previous - bralink;
bralink = previous;
PUTINC(previous, 0, offset);
}
repeat_max--;
}
/* If the minimum is greater than zero, replicate the group as many
times as necessary, and adjust the maximum to the number of subsequent
copies that we need. If we set a first char from the group, and didn't
set a required char, copy the latter from the former. */
else {
if (repeat_min > 1) {
if (groupsetfirstbyte && reqbyte < 0)
reqbyte = firstbyte;
for (i = 1; i < repeat_min; i++) {
memcpy(code, previous, len);
code += len;
}
}
if (repeat_max > 0)
repeat_max -= repeat_min;
}
/* This code is common to both the zero and non-zero minimum cases. If
the maximum is limited, it replicates the group in a nested fashion,
remembering the bracket starts on a stack. In the case of a zero minimum,
the first one was set up above. In all cases the repeat_max now specifies
the number of additional copies needed. */
if (repeat_max >= 0) {
for (i = repeat_max - 1; i >= 0; i--) {
*code++ = OP_BRAZERO + repeat_type;
/* All but the final copy start a new nesting, maintaining the
chain of brackets outstanding. */
if (i != 0) {
int offset;
*code++ = OP_BRA;
offset = (bralink == NULL) ? 0 : code - bralink;
bralink = code;
PUTINC(code, 0, offset);
}
memcpy(code, previous, len);
code += len;
}
/* Now chain through the pending brackets, and fill in their length
fields (which are holding the chain links pro tem). */
while (bralink != NULL) {
int oldlinkoffset;
int offset = code - bralink + 1;
uschar *bra = code - offset;
oldlinkoffset = GET(bra, 1);
bralink = (oldlinkoffset == 0) ? NULL : bralink - oldlinkoffset;
*code++ = OP_KET;
PUTINC(code, 0, offset);
PUT(bra, 1, offset);
}
}
/* If the maximum is unlimited, set a repeater in the final copy. We
can't just offset backwards from the current code point, because we
don't know if there's been an options resetting after the ket. The
correct offset was computed above. */
else
code[-ketoffset] = OP_KETRMAX + repeat_type;
}
/* Else there's some kind of shambles */
else {
*errorptr = ERR11;
goto FAILED;
}
/* If the character following a repeat is '+', we wrap the entire repeated
item inside OP_ONCE brackets. This is just syntactic sugar, taken from
Sun's Java package. The repeated item starts at tempcode, not at previous,
which might be the first part of a string whose (former) last char we
repeated. However, we don't support '+' after a greediness '?'. */
if (possessive_quantifier) {
int len = code - tempcode;
memmove(tempcode + 1 + LINK_SIZE, tempcode, len);
code += 1 + LINK_SIZE;
len += 1 + LINK_SIZE;
tempcode[0] = OP_ONCE;
*code++ = OP_KET;
PUTINC(code, 0, len);
PUT(tempcode, 1, len);
}
/* In all case we no longer have a previous item. We also set the
"follows varying string" flag for subsequently encountered reqbytes if
it isn't already set and we have just passed a varying length item. */
END_REPEAT:
previous = NULL;
cd->req_varyopt |= reqvary;
break;
/* Start of nested bracket sub-expression, or comment or lookahead or
lookbehind or option setting or condition. First deal with special things
that can come after a bracket; all are introduced by ?, and the appearance
of any of them means that this is not a referencing group. They were
checked for validity in the first pass over the string, so we don't have to
check for syntax errors here. */
case '(':
newoptions = options;
skipbytes = 0;
if (*(++ptr) == '?') {
int set, unset;
int *optset;
switch (*(++ptr)) {
case '#': /* Comment; skip to ket */
ptr++;
while (*ptr != ')')
ptr++;
continue;
case ':': /* Non-extracting bracket */
bravalue = OP_BRA;
ptr++;
break;
case '(':
bravalue = OP_COND; /* Conditional group */
/* Condition to test for recursion */
if (ptr[1] == 'R') {
code[1 + LINK_SIZE] = OP_CREF;
PUT2(code, 2 + LINK_SIZE, CREF_RECURSE);
skipbytes = 3;
ptr += 3;
}
/* Condition to test for a numbered subpattern match. We know that
if a digit follows ( then there will just be digits until ) because
the syntax was checked in the first pass. */
else if ((digitab[ptr[1]] && ctype_digit) != 0) {
int condref; /* Don't amalgamate; some compilers */
condref = *(++ptr) - '0'; /* grumble at autoincrement in declaration */
while (*(++ptr) != ')')
condref = condref * 10 + *ptr - '0';
if (condref == 0) {
*errorptr = ERR35;
goto FAILED;
}
ptr++;
code[1 + LINK_SIZE] = OP_CREF;
PUT2(code, 2 + LINK_SIZE, condref);
skipbytes = 3;
}
/* For conditions that are assertions, we just fall through, having
set bravalue above. */
break;
case '=': /* Positive lookahead */
bravalue = OP_ASSERT;
ptr++;
break;
case '!': /* Negative lookahead */
bravalue = OP_ASSERT_NOT;
ptr++;
break;
case '<': /* Lookbehinds */
switch (*(++ptr)) {
case '=': /* Positive lookbehind */
bravalue = OP_ASSERTBACK;
ptr++;
break;
case '!': /* Negative lookbehind */
bravalue = OP_ASSERTBACK_NOT;
ptr++;
break;
}
break;
case '>': /* One-time brackets */
bravalue = OP_ONCE;
ptr++;
break;
case 'C': /* Callout - may be followed by digits */
*code++ = OP_CALLOUT;
{
int n = 0;
while ((digitab[*(++ptr)] & ctype_digit) != 0)
n = n * 10 + *ptr - '0';
if (n > 255) {
*errorptr = ERR38;
goto FAILED;
}
*code++ = n;
}
previous = NULL;
continue;
case 'P': /* Named subpattern handling */
if (*(++ptr) == '<') { /* Definition */
int i, namelen;
uschar *slot = cd->name_table;
const uschar *name; /* Don't amalgamate; some compilers */
name = ++ptr; /* grumble at autoincrement in declaration */
while (*ptr++ != '>') ;
namelen = ptr - name - 1;
for (i = 0; i < cd->names_found; i++) {
int crc = memcmp(name, slot + 2, namelen);
if (crc == 0) {
if (slot[2 + namelen] == 0) {
*errorptr = ERR43;
goto FAILED;
}
crc = -1; /* Current name is substring */
}
if (crc < 0) {
memmove(slot + cd->name_entry_size, slot,
(cd->names_found - i) * cd->name_entry_size);
break;
}
slot += cd->name_entry_size;
}
PUT2(slot, 0, *brackets + 1);
memcpy(slot + 2, name, namelen);
slot[2 + namelen] = 0;
cd->names_found++;
goto NUMBERED_GROUP;
}
if (*ptr == '=' || *ptr == '>') { /* Reference or recursion */
int i, namelen;
int type = *ptr++;
const uschar *name = ptr;
uschar *slot = cd->name_table;
while (*ptr != ')')
ptr++;
namelen = ptr - name;
for (i = 0; i < cd->names_found; i++) {
if (strncmp((char *) name, (char *) slot + 2, namelen) == 0)
break;
slot += cd->name_entry_size;
}
if (i >= cd->names_found) {
*errorptr = ERR15;
goto FAILED;
}
recno = GET2(slot, 0);
if (type == '>')
goto HANDLE_RECURSION; /* A few lines below */
/* Back reference */
previous = code;
*code++ = OP_REF;
PUT2INC(code, 0, recno);
cd->backref_map |= (recno < 32) ? (1 << recno) : 1;
if (recno > cd->top_backref)
cd->top_backref = recno;
continue;
}
/* Should never happen */
break;
case 'R': /* Pattern recursion */
ptr++; /* Same as (?0) */
/* Fall through */
/* Recursion or "subroutine" call */
case '0':
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
case '8':
case '9':
{
const uschar *called;
recno = 0;
while ((digitab[*ptr] & ctype_digit) != 0)
recno = recno * 10 + *ptr++ - '0';
/* Come here from code above that handles a named recursion */
HANDLE_RECURSION:
previous = code;
/* Find the bracket that is being referenced. Temporarily end the
regex in case it doesn't exist. */
*code = OP_END;
called = (recno == 0) ?
cd->start_code : find_bracket(cd->start_code, utf8, recno);
if (called == NULL) {
*errorptr = ERR15;
goto FAILED;
}
/* If the subpattern is still open, this is a recursive call. We
check to see if this is a left recursion that could loop for ever,
and diagnose that case. */
if (GET(called, 1) == 0
&& could_be_empty(called, code, bcptr, utf8)) {
*errorptr = ERR40;
goto FAILED;
}
/* Insert the recursion/subroutine item */
*code = OP_RECURSE;
PUT(code, 1, called - cd->start_code);
code += 1 + LINK_SIZE;
}
continue;
/* Character after (? not specially recognized */
default: /* Option setting */
set = unset = 0;
optset = &set;
while (*ptr != ')' && *ptr != ':') {
switch (*ptr++) {
case '-':
optset = &unset;
break;
case 'i':
*optset |= PCRE_CASELESS;
break;
case 'm':
*optset |= PCRE_MULTILINE;
break;
case 's':
*optset |= PCRE_DOTALL;
break;
case 'x':
*optset |= PCRE_EXTENDED;
break;
case 'U':
*optset |= PCRE_UNGREEDY;
break;
case 'X':
*optset |= PCRE_EXTRA;
break;
}
}
/* Set up the changed option bits, but don't change anything yet. */
newoptions = (options | set) & (~unset);
/* If the options ended with ')' this is not the start of a nested
group with option changes, so the options change at this level. Compile
code to change the ims options if this setting actually changes any of
them. We also pass the new setting back so that it can be put at the
start of any following branches, and when this group ends (if we are in
a group), a resetting item can be compiled.
Note that if this item is right at the start of the pattern, the
options will have been abstracted and made global, so there will be no
change to compile. */
if (*ptr == ')') {
if ((options & PCRE_IMS) != (newoptions & PCRE_IMS)) {
*code++ = OP_OPT;
*code++ = newoptions & PCRE_IMS;
}
/* Change options at this level, and pass them back for use
in subsequent branches. Reset the greedy defaults and the case
value for firstbyte and reqbyte. */
*optionsptr = options = newoptions;
greedy_default = ((newoptions & PCRE_UNGREEDY) != 0);
greedy_non_default = greedy_default ^ 1;
req_caseopt = ((options & PCRE_CASELESS) != 0) ? REQ_CASELESS : 0;
previous = NULL; /* This item can't be repeated */
continue; /* It is complete */
}
/* If the options ended with ':' we are heading into a nested group
with possible change of options. Such groups are non-capturing and are
not assertions of any kind. All we need to do is skip over the ':';
the newoptions value is handled below. */
bravalue = OP_BRA;
ptr++;
}
}
/* If PCRE_NO_AUTO_CAPTURE is set, all unadorned brackets become
non-capturing and behave like (?:...) brackets */
else if ((options & PCRE_NO_AUTO_CAPTURE) != 0) {
bravalue = OP_BRA;
}
/* Else we have a referencing group; adjust the opcode. If the bracket
number is greater than EXTRACT_BASIC_MAX, we set the opcode one higher, and
arrange for the true number to follow later, in an OP_BRANUMBER item. */
else {
NUMBERED_GROUP:
if (++(*brackets) > EXTRACT_BASIC_MAX) {
bravalue = OP_BRA + EXTRACT_BASIC_MAX + 1;
code[1 + LINK_SIZE] = OP_BRANUMBER;
PUT2(code, 2 + LINK_SIZE, *brackets);
skipbytes = 3;
} else
bravalue = OP_BRA + *brackets;
}
/* Process nested bracketed re. Assertions may not be repeated, but other
kinds can be. We copy code into a non-register variable in order to be able
to pass its address because some compilers complain otherwise. Pass in a
new setting for the ims options if they have changed. */
previous = (bravalue >= OP_ONCE) ? code : NULL;
*code = bravalue;
tempcode = code;
tempreqvary = cd->req_varyopt; /* Save value before bracket */
if (!compile_regex(newoptions, /* The complete new option state */
options & PCRE_IMS, /* The previous ims option state */
brackets, /* Extracting bracket count */
&tempcode, /* Where to put code (updated) */
&ptr, /* Input pointer (updated) */
errorptr, /* Where to put an error message */
(bravalue == OP_ASSERTBACK || bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */
skipbytes, /* Skip over OP_COND/OP_BRANUMBER */
&subfirstbyte, /* For possible first char */
&subreqbyte, /* For possible last char */
bcptr, /* Current branch chain */
cd)) /* Tables block */
goto FAILED;
/* At the end of compiling, code is still pointing to the start of the
group, while tempcode has been updated to point past the end of the group
and any option resetting that may follow it. The pattern pointer (ptr)
is on the bracket. */
/* If this is a conditional bracket, check that there are no more than
two branches in the group. */
else if (bravalue == OP_COND) {
uschar *tc = code;
condcount = 0;
do {
condcount++;
tc += GET(tc, 1);
}
while (*tc != OP_KET);
if (condcount > 2) {
*errorptr = ERR27;
goto FAILED;
}
/* If there is just one branch, we must not make use of its firstbyte or
reqbyte, because this is equivalent to an empty second branch. */
if (condcount == 1)
subfirstbyte = subreqbyte = REQ_NONE;
}
/* Handle updating of the required and first characters. Update for normal
brackets of all kinds, and conditions with two branches (see code above).
If the bracket is followed by a quantifier with zero repeat, we have to
back off. Hence the definition of zeroreqbyte and zerofirstbyte outside the
main loop so that they can be accessed for the back off. */
zeroreqbyte = reqbyte;
zerofirstbyte = firstbyte;
groupsetfirstbyte = FALSE;
if (bravalue >= OP_BRA || bravalue == OP_ONCE || bravalue == OP_COND) {
/* If we have not yet set a firstbyte in this branch, take it from the
subpattern, remembering that it was set here so that a repeat of more
than one can replicate it as reqbyte if necessary. If the subpattern has
no firstbyte, set "none" for the whole branch. In both cases, a zero
repeat forces firstbyte to "none". */
if (firstbyte == REQ_UNSET) {
if (subfirstbyte >= 0) {
firstbyte = subfirstbyte;
groupsetfirstbyte = TRUE;
} else
firstbyte = REQ_NONE;
zerofirstbyte = REQ_NONE;
}
/* If firstbyte was previously set, convert the subpattern's firstbyte
into reqbyte if there wasn't one, using the vary flag that was in
existence beforehand. */
else if (subfirstbyte >= 0 && subreqbyte < 0)
subreqbyte = subfirstbyte | tempreqvary;
/* If the subpattern set a required byte (or set a first byte that isn't
really the first byte - see above), set it. */
if (subreqbyte >= 0)
reqbyte = subreqbyte;
}
/* For a forward assertion, we take the reqbyte, if set. This can be
helpful if the pattern that follows the assertion doesn't set a different
char. For example, it's useful for /(?=abcde).+/. We can't set firstbyte
for an assertion, however because it leads to incorrect effect for patterns
such as /(?=a)a.+/ when the "real" "a" would then become a reqbyte instead
of a firstbyte. This is overcome by a scan at the end if there's no
firstbyte, looking for an asserted first char. */
else if (bravalue == OP_ASSERT && subreqbyte >= 0)
reqbyte = subreqbyte;
/* Now update the main code pointer to the end of the group. */
code = tempcode;
/* Error if hit end of pattern */
if (*ptr != ')') {
*errorptr = ERR14;
goto FAILED;
}
break;
/* Check \ for being a real metacharacter; if not, fall through and handle
it as a data character at the start of a string. Escape items are checked
for validity in the pre-compiling pass. */
case '\\':
tempptr = ptr;
c = check_escape(&ptr, errorptr, *brackets, options, FALSE);
/* Handle metacharacters introduced by \. For ones like \d, the ESC_ values
are arranged to be the negation of the corresponding OP_values. For the
back references, the values are ESC_REF plus the reference number. Only
back references and those types that consume a character may be repeated.
We can test for values between ESC_b and ESC_Z for the latter; this may
have to change if any new ones are ever created. */
if (c < 0) {
if (-c == ESC_Q) { /* Handle start of quoted string */
if (ptr[1] == '\\' && ptr[2] == 'E')
ptr += 2; /* avoid empty string */
else
inescq = TRUE;
continue;
}
/* For metasequences that actually match a character, we disable the
setting of a first character if it hasn't already been set. */
if (firstbyte == REQ_UNSET && -c > ESC_b && -c < ESC_Z)
firstbyte = REQ_NONE;
/* Set values to reset to if this is followed by a zero repeat. */
zerofirstbyte = firstbyte;
zeroreqbyte = reqbyte;
/* Back references are handled specially */
if (-c >= ESC_REF) {
int number = -c - ESC_REF;
previous = code;
*code++ = OP_REF;
PUT2INC(code, 0, number);
} else {
previous = (-c > ESC_b && -c < ESC_Z) ? code : NULL;
*code++ = -c;
}
continue;
}
/* Data character: reset and fall through */
ptr = tempptr;
c = '\\';
/* Handle a run of data characters until a metacharacter is encountered.
The first character is guaranteed not to be whitespace or # when the
extended flag is set. */
NORMAL_CHAR:
default:
previous = code;
*code = OP_CHARS;
code += 2;
length = 0;
do {
/* If in \Q...\E, check for the end; if not, we always have a literal */
if (inescq) {
if (c == '\\' && ptr[1] == 'E') {
inescq = FALSE;
ptr++;
} else {
*code++ = c;
length++;
}
continue;
}
/* Skip white space and comments for /x patterns */
if ((options & PCRE_EXTENDED) != 0) {
if ((cd->ctypes[c] & ctype_space) != 0)
continue;
if (c == '#') {
/* The space before the ; is to avoid a warning on a silly compiler
on the Macintosh. */
while ((c = *(++ptr)) != 0 && c != NEWLINE) ;
if (c == 0)
break;
continue;
}
}
/* Backslash may introduce a data char or a metacharacter. Escaped items
are checked for validity in the pre-compiling pass. Stop the string
before a metaitem. */
if (c == '\\') {
tempptr = ptr;
c = check_escape(&ptr, errorptr, *brackets, options, FALSE);
if (c < 0) {
ptr = tempptr;
break;
}
/* If a character is > 127 in UTF-8 mode, we have to turn it into
two or more bytes in the UTF-8 encoding. */
}
/* Ordinary character or single-char escape */
*code++ = c;
length++;
}
/* This "while" is the end of the "do" above. */
while (length < MAXLIT && (cd->ctypes[c = *(++ptr)] & ctype_meta) == 0);
/* Update the first and last requirements. These are always bytes, even in
UTF-8 mode. However, there is a special case to be considered when there
are only one or two characters. Because this gets messy in UTF-8 mode, the
code is kept separate. When we get here "length" contains the number of
bytes. */
/* This is the code for non-UTF-8 operation, either without UTF-8 support,
or when UTF-8 is not enabled. */
{
/* firstbyte was not previously set; take it from this string */
if (firstbyte == REQ_UNSET) {
if (length == 1) {
zerofirstbyte = REQ_NONE;
firstbyte = previous[2] | req_caseopt;
zeroreqbyte = reqbyte;
} else {
zerofirstbyte = firstbyte = previous[2] | req_caseopt;
zeroreqbyte = (length > 2) ?
(code[-2] | req_caseopt | cd->req_varyopt) : reqbyte;
reqbyte = code[-1] | req_caseopt | cd->req_varyopt;
}
}
/* firstbyte was previously set */
else {
zerofirstbyte = firstbyte;
zeroreqbyte = (length == 1) ? reqbyte :
code[-2] | req_caseopt | cd->req_varyopt;
reqbyte = code[-1] | req_caseopt | cd->req_varyopt;
}
}
/* Set the length in the data vector, and advance to the next state. */
previous[1] = length;
if (length < MAXLIT)
ptr--;
break;
}
} /* end of big loop */
/* Control never reaches here by falling through, only by a goto for all the
error states. Pass back the position in the pattern so that it can be displayed
to the user for diagnosing the error. */
FAILED:
*ptrptr = ptr;
return FALSE;
}
/*************************************************
* Compile sequence of alternatives *
*************************************************/
/* On entry, ptr is pointing past the bracket character, but on return
it points to the closing bracket, or vertical bar, or end of string.
The code variable is pointing at the byte into which the BRA operator has been
stored. If the ims options are changed at the start (for a (?ims: group) or
during any branch, we need to insert an OP_OPT item at the start of every
following branch to ensure they get set correctly at run time, and also pass
the new options into every subsequent branch compile.
Argument:
options option bits, including any changes for this subpattern
oldims previous settings of ims option bits
brackets -> int containing the number of extracting brackets used
codeptr -> the address of the current code pointer
ptrptr -> the address of the current pattern pointer
errorptr -> pointer to error message
lookbehind TRUE if this is a lookbehind assertion
skipbytes skip this many bytes at start (for OP_COND, OP_BRANUMBER)
firstbyteptr place to put the first required character, or a negative number
reqbyteptr place to put the last required character, or a negative number
bcptr pointer to the chain of currently open branches
cd points to the data block with tables pointers etc.
Returns: TRUE on success
*/
static BOOL
compile_regex(int options, int oldims, int *brackets, uschar ** codeptr,
const uschar ** ptrptr, const char **errorptr, BOOL lookbehind,
int skipbytes, int *firstbyteptr, int *reqbyteptr,
branch_chain * bcptr, compile_data * cd)
{
const uschar *ptr = *ptrptr;
uschar *code = *codeptr;
uschar *last_branch = code;
uschar *start_bracket = code;
uschar *reverse_count = NULL;
int firstbyte, reqbyte;
int branchfirstbyte, branchreqbyte;
branch_chain bc;
bc.outer = bcptr;
bc.current = code;
firstbyte = reqbyte = REQ_UNSET;
/* Offset is set zero to mark that this bracket is still open */
PUT(code, 1, 0);
code += 1 + LINK_SIZE + skipbytes;
/* Loop for each alternative branch */
for (;;) {
/* Handle a change of ims options at the start of the branch */
if ((options & PCRE_IMS) != oldims) {
*code++ = OP_OPT;
*code++ = options & PCRE_IMS;
}
/* Set up dummy OP_REVERSE if lookbehind assertion */
if (lookbehind) {
*code++ = OP_REVERSE;
reverse_count = code;
PUTINC(code, 0, 0);
}
/* Now compile the branch */
if (!compile_branch(&options, brackets, &code, &ptr, errorptr,
&branchfirstbyte, &branchreqbyte, &bc, cd)) {
*ptrptr = ptr;
return FALSE;
}
/* If this is the first branch, the firstbyte and reqbyte values for the
branch become the values for the regex. */
if (*last_branch != OP_ALT) {
firstbyte = branchfirstbyte;
reqbyte = branchreqbyte;
}
/* If this is not the first branch, the first char and reqbyte have to
match the values from all the previous branches, except that if the previous
value for reqbyte didn't have REQ_VARY set, it can still match, and we set
REQ_VARY for the regex. */
else {
/* If we previously had a firstbyte, but it doesn't match the new branch,
we have to abandon the firstbyte for the regex, but if there was previously
no reqbyte, it takes on the value of the old firstbyte. */
if (firstbyte >= 0 && firstbyte != branchfirstbyte) {
if (reqbyte < 0)
reqbyte = firstbyte;
firstbyte = REQ_NONE;
}
/* If we (now or from before) have no firstbyte, a firstbyte from the
branch becomes a reqbyte if there isn't a branch reqbyte. */
if (firstbyte < 0 && branchfirstbyte >= 0 && branchreqbyte < 0)
branchreqbyte = branchfirstbyte;
/* Now ensure that the reqbytes match */
if ((reqbyte & ~REQ_VARY) != (branchreqbyte & ~REQ_VARY))
reqbyte = REQ_NONE;
else
reqbyte |= branchreqbyte; /* To "or" REQ_VARY */
}
/* If lookbehind, check that this branch matches a fixed-length string,
and put the length into the OP_REVERSE item. Temporarily mark the end of
the branch with OP_END. */
if (lookbehind) {
int length;
*code = OP_END;
length = find_fixedlength(last_branch, options);
DPRINTF(("fixed length = %d\n", length));
if (length < 0) {
*errorptr = (length == -2) ? ERR36 : ERR25;
*ptrptr = ptr;
return FALSE;
}
PUT(reverse_count, 0, length);
}
/* Reached end of expression, either ')' or end of pattern. Go back through
the alternative branches and reverse the chain of offsets, with the field in
the BRA item now becoming an offset to the first alternative. If there are
no alternatives, it points to the end of the group. The length in the
terminating ket is always the length of the whole bracketed item. If any of
the ims options were changed inside the group, compile a resetting op-code
following, except at the very end of the pattern. Return leaving the pointer
at the terminating char. */
if (*ptr != '|') {
int length = code - last_branch;
do {
int prev_length = GET(last_branch, 1);
PUT(last_branch, 1, length);
length = prev_length;
last_branch -= length;
}
while (length > 0);
/* Fill in the ket */
*code = OP_KET;
PUT(code, 1, code - start_bracket);
code += 1 + LINK_SIZE;
/* Resetting option if needed */
if ((options & PCRE_IMS) != oldims && *ptr == ')') {
*code++ = OP_OPT;
*code++ = oldims;
}
/* Set values to pass back */
*codeptr = code;
*ptrptr = ptr;
*firstbyteptr = firstbyte;
*reqbyteptr = reqbyte;
return TRUE;
}
/* Another branch follows; insert an "or" node. Its length field points back
to the previous branch while the bracket remains open. At the end the chain
is reversed. It's done like this so that the start of the bracket has a
zero offset until it is closed, making it possible to detect recursion. */
*code = OP_ALT;
PUT(code, 1, code - last_branch);
bc.current = last_branch = code;
code += 1 + LINK_SIZE;
ptr++;
}
/* Control never reaches here */
}
/*************************************************
* Check for anchored expression *
*************************************************/
/* Try to find out if this is an anchored regular expression. Consider each
alternative branch. If they all start with OP_SOD or OP_CIRC, or with a bracket
all of whose alternatives start with OP_SOD or OP_CIRC (recurse ad lib), then
it's anchored. However, if this is a multiline pattern, then only OP_SOD
counts, since OP_CIRC can match in the middle.
We can also consider a regex to be anchored if OP_SOM starts all its branches.
This is the code for \G, which means "match at start of match position, taking
into account the match offset".
A branch is also implicitly anchored if it starts with .* and DOTALL is set,
because that will try the rest of the pattern at all possible matching points,
so there is no point trying again.... er ....
.... except when the .* appears inside capturing parentheses, and there is a
subsequent back reference to those parentheses. We haven't enough information
to catch that case precisely.
At first, the best we could do was to detect when .* was in capturing brackets
and the highest back reference was greater than or equal to that level.
However, by keeping a bitmap of the first 31 back references, we can catch some
of the more common cases more precisely.
Arguments:
code points to start of expression (the bracket)
options points to the options setting
bracket_map a bitmap of which brackets we are inside while testing; this
handles up to substring 31; after that we just have to take
the less precise approach
backref_map the back reference bitmap
Returns: TRUE or FALSE
*/
static BOOL
is_anchored(register const uschar * code, int *options,
unsigned int bracket_map, unsigned int backref_map)
{
do {
const uschar *scode =
first_significant_code(code + 1 + LINK_SIZE, options, PCRE_MULTILINE);
register int op = *scode;
/* Capturing brackets */
if (op > OP_BRA) {
int new_map;
op -= OP_BRA;
if (op > EXTRACT_BASIC_MAX)
op = GET2(scode, 2 + LINK_SIZE);
new_map = bracket_map | ((op < 32) ? (1 << op) : 1);
if (!is_anchored(scode, options, new_map, backref_map))
return FALSE;
}
/* Other brackets */
else if (op == OP_BRA || op == OP_ASSERT || op == OP_ONCE || op == OP_COND) {
if (!is_anchored(scode, options, bracket_map, backref_map))
return FALSE;
}
/* .* is not anchored unless DOTALL is set and it isn't in brackets that
are or may be referenced. */
else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR) &&
(*options & PCRE_DOTALL) != 0) {
if (scode[1] != OP_ANY || (bracket_map & backref_map) != 0)
return FALSE;
}
/* Check for explicit anchoring */
else if (op != OP_SOD && op != OP_SOM &&
((*options & PCRE_MULTILINE) != 0 || op != OP_CIRC))
return FALSE;
code += GET(code, 1);
}
while (*code == OP_ALT); /* Loop for each alternative */
return TRUE;
}
/*************************************************
* Check for starting with ^ or .* *
*************************************************/
/* This is called to find out if every branch starts with ^ or .* so that
"first char" processing can be done to speed things up in multiline
matching and for non-DOTALL patterns that start with .* (which must start at
the beginning or after \n). As in the case of is_anchored() (see above), we
have to take account of back references to capturing brackets that contain .*
because in that case we can't make the assumption.
Arguments:
code points to start of expression (the bracket)
bracket_map a bitmap of which brackets we are inside while testing; this
handles up to substring 31; after that we just have to take
the less precise approach
backref_map the back reference bitmap
Returns: TRUE or FALSE
*/
static BOOL
is_startline(const uschar * code, unsigned int bracket_map,
unsigned int backref_map)
{
do {
const uschar *scode = first_significant_code(code + 1 + LINK_SIZE, NULL, 0);
register int op = *scode;
/* Capturing brackets */
if (op > OP_BRA) {
int new_map;
op -= OP_BRA;
if (op > EXTRACT_BASIC_MAX)
op = GET2(scode, 2 + LINK_SIZE);
new_map = bracket_map | ((op < 32) ? (1 << op) : 1);
if (!is_startline(scode, new_map, backref_map))
return FALSE;
}
/* Other brackets */
else if (op == OP_BRA || op == OP_ASSERT || op == OP_ONCE || op == OP_COND) {
if (!is_startline(scode, bracket_map, backref_map))
return FALSE;
}
/* .* is not anchored unless DOTALL is set and it isn't in brackets that
may be referenced. */
else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR) {
if (scode[1] != OP_ANY || (bracket_map & backref_map) != 0)
return FALSE;
}
/* Check for explicit circumflex */
else if (op != OP_CIRC)
return FALSE;
code += GET(code, 1);
}
while (*code == OP_ALT); /* Loop for each alternative */
return TRUE;
}
/*************************************************
* Check for asserted fixed first char *
*************************************************/
/* During compilation, the "first char" settings from forward assertions are
discarded, because they can cause conflicts with actual literals that follow.
However, if we end up without a first char setting for an unanchored pattern,
it is worth scanning the regex to see if there is an initial asserted first
char. If all branches start with the same asserted char, or with a bracket all
of whose alternatives start with the same asserted char (recurse ad lib), then
we return that char, otherwise -1.
Arguments:
code points to start of expression (the bracket)
options pointer to the options (used to check casing changes)
inassert TRUE if in an assertion
Returns: -1 or the fixed first char
*/
static int
find_firstassertedchar(const uschar * code, int *options, BOOL inassert)
{
register int c = -1;
do {
int d;
const uschar *scode =
first_significant_code(code + 1 + LINK_SIZE, options, PCRE_CASELESS);
register int op = *scode;
if (op >= OP_BRA)
op = OP_BRA;
switch (op) {
default:
return -1;
case OP_BRA:
case OP_ASSERT:
case OP_ONCE:
case OP_COND:
if ((d = find_firstassertedchar(scode, options, op == OP_ASSERT)) < 0)
return -1;
if (c < 0)
c = d;
else if (c != d)
return -1;
break;
case OP_EXACT: /* Fall through */
scode++;
case OP_CHARS: /* Fall through */
scode++;
case OP_PLUS:
case OP_MINPLUS:
if (!inassert)
return -1;
if (c < 0) {
c = scode[1];
if ((*options & PCRE_CASELESS) != 0)
c |= REQ_CASELESS;
} else if (c != scode[1])
return -1;
break;
}
code += GET(code, 1);
}
while (*code == OP_ALT);
return c;
}
/*************************************************
* Compile a Regular Expression *
*************************************************/
/* This function takes a string and returns a pointer to a block of store
holding a compiled version of the expression.
Arguments:
pattern the regular expression
options various option bits
errorptr pointer to pointer to error text
erroroffset ptr offset in pattern where error was detected
tables pointer to character tables or NULL
Returns: pointer to compiled data block, or NULL on error,
with errorptr and erroroffset set
*/
EXPORT pcre *
pcre_compile(const char *pattern, int options, const char **errorptr,
int *erroroffset, const unsigned char *tables)
{
real_pcre *re;
int length = 1 + LINK_SIZE; /* For initial BRA plus length */
int runlength;
int c, firstbyte, reqbyte;
int bracount = 0;
int branch_extra = 0;
int branch_newextra;
int item_count = -1;
int name_count = 0;
int max_name_size = 0;
BOOL inescq = FALSE;
unsigned int brastackptr = 0;
size_t size;
uschar *code;
const uschar *codestart;
const uschar *ptr;
compile_data compile_block;
int brastack[BRASTACK_SIZE];
uschar bralenstack[BRASTACK_SIZE];
/* We can't pass back an error message if errorptr is NULL; I guess the best we
can do is just return NULL. */
if (errorptr == NULL)
return NULL;
*errorptr = NULL;
/* However, we can give a message for this error */
if (erroroffset == NULL) {
*errorptr = ERR16;
return NULL;
}
*erroroffset = 0;
/* Can't support UTF8 unless PCRE has been compiled to include the code. */
if ((options & PCRE_UTF8) != 0) {
*errorptr = ERR32;
return NULL;
}
if ((options & ~PUBLIC_OPTIONS) != 0) {
*errorptr = ERR17;
return NULL;
}
/* Set up pointers to the individual character tables */
if (tables == NULL)
tables = pcre_default_tables;
compile_block.lcc = tables + lcc_offset;
compile_block.fcc = tables + fcc_offset;
compile_block.cbits = tables + cbits_offset;
compile_block.ctypes = tables + ctypes_offset;
/* Maximum back reference and backref bitmap. This is updated for numeric
references during the first pass, but for named references during the actual
compile pass. The bitmap records up to 31 back references to help in deciding
whether (.*) can be treated as anchored or not. */
compile_block.top_backref = 0;
compile_block.backref_map = 0;
/* Reflect pattern for debugging output */
DPRINTF(("------------------------------------------------------------------\n"));
DPRINTF(("%s\n", pattern));
/* The first thing to do is to make a pass over the pattern to compute the
amount of store required to hold the compiled code. This does not have to be
perfect as long as errors are overestimates. At the same time we can detect any
flag settings right at the start, and extract them. Make an attempt to correct
for any counted white space if an "extended" flag setting appears late in the
pattern. We can't be so clever for #-comments. */
ptr = (const uschar *) (pattern - 1);
while ((c = *(++ptr)) != 0) {
int min, max;
int class_optcount;
int bracket_length;
int duplength;
/* If we are inside a \Q...\E sequence, all chars are literal */
if (inescq)
goto NORMAL_CHAR;
/* Otherwise, first check for ignored whitespace and comments */
if ((options & PCRE_EXTENDED) != 0) {
if ((compile_block.ctypes[c] & ctype_space) != 0)
continue;
if (c == '#') {
/* The space before the ; is to avoid a warning on a silly compiler
on the Macintosh. */
while ((c = *(++ptr)) != 0 && c != NEWLINE) ;
if (c == 0)
break;
continue;
}
}
item_count++; /* Is zero for the first non-comment item */
switch (c) {
/* A backslashed item may be an escaped "normal" character or a
character type. For a "normal" character, put the pointers and
character back so that tests for whitespace etc. in the input
are done correctly. */
case '\\':
{
const uschar *save_ptr = ptr;
c = check_escape(&ptr, errorptr, bracount, options, FALSE);
if (*errorptr != NULL)
goto PCRE_ERROR_RETURN;
if (c >= 0) {
ptr = save_ptr;
c = '\\';
goto NORMAL_CHAR;
}
}
/* If \Q, enter "literal" mode */
if (-c == ESC_Q) {
inescq = TRUE;
continue;
}
/* Other escapes need one byte, and are of length one for repeats */
length++;
/* A back reference needs an additional 2 bytes, plus either one or 5
bytes for a repeat. We also need to keep the value of the highest
back reference. */
if (c <= -ESC_REF) {
int refnum = -c - ESC_REF;
compile_block.backref_map |= (refnum < 32) ? (1 << refnum) : 1;
if (refnum > compile_block.top_backref)
compile_block.top_backref = refnum;
length += 2; /* For single back reference */
if (ptr[1] == '{' && is_counted_repeat(ptr + 2)) {
ptr = read_repeat_counts(ptr + 2, &min, &max, errorptr);
if (*errorptr != NULL)
goto PCRE_ERROR_RETURN;
if ((min == 0 && (max == 1 || max == -1)) || (min == 1 && max == -1))
length++;
else
length += 5;
if (ptr[1] == '?')
ptr++;
}
}
continue;
case '^': /* Single-byte metacharacters */
case '.':
case '$':
length++;
continue;
case '*': /* These repeats won't be after brackets; */
case '+': /* those are handled separately */
case '?':
length++;
goto POSESSIVE; /* A few lines below */
/* This covers the cases of braced repeats after a single char, metachar,
class, or back reference. */
case '{':
if (!is_counted_repeat(ptr + 1))
goto NORMAL_CHAR;
ptr = read_repeat_counts(ptr + 1, &min, &max, errorptr);
if (*errorptr != NULL)
goto PCRE_ERROR_RETURN;
/* These special cases just insert one extra opcode */
if ((min == 0 && (max == 1 || max == -1)) || (min == 1 && max == -1))
length++;
/* These cases might insert additional copies of a preceding character. */
else {
/* Not UTF-8 mode: all characters are one byte */
{
if (min != 1) {
length--; /* Uncount the original char or metachar */
if (min > 0)
length += 4;
}
length += (max > 0) ? 4 : 2;
}
}
if (ptr[1] == '?')
ptr++; /* Needs no extra length */
POSESSIVE: /* Test for possessive quantifier */
if (ptr[1] == '+') {
ptr++;
length += 2 + 2 * LINK_SIZE; /* Allow for atomic brackets */
}
continue;
/* An alternation contains an offset to the next branch or ket. If any ims
options changed in the previous branch(es), and/or if we are in a
lookbehind assertion, extra space will be needed at the start of the
branch. This is handled by branch_extra. */
case '|':
length += 1 + LINK_SIZE + branch_extra;
continue;
/* A character class uses 33 characters provided that all the character
values are less than 256. Otherwise, it uses a bit map for low valued
characters, and individual items for others. Don't worry about character
types that aren't allowed in classes - they'll get picked up during the
compile. A character class that contains only one single-byte character
uses 2 or 3 bytes, depending on whether it is negated or not. Notice this
where we can. (In UTF-8 mode we can do this only for chars < 128.) */
case '[':
class_optcount = 0;
if (*(++ptr) == '^')
ptr++;
/* Written as a "do" so that an initial ']' is taken as data */
if (*ptr != 0)
do {
/* Inside \Q...\E everything is literal except \E */
if (inescq) {
if (*ptr != '\\' || ptr[1] != 'E')
goto NON_SPECIAL_CHARACTER;
inescq = FALSE;
ptr += 1;
continue;
}
/* Outside \Q...\E, check for escapes */
if (*ptr == '\\') {
int ch = check_escape(&ptr, errorptr, bracount, options, TRUE);
if (*errorptr != NULL)
goto PCRE_ERROR_RETURN;
/* \b is backspace inside a class */
if (-ch == ESC_b)
ch = '\b';
/* \Q enters quoting mode */
if (-ch == ESC_Q) {
inescq = TRUE;
continue;
}
/* Handle escapes that turn into characters */
if (ch >= 0) {
class_optcount++; /* for possible optimization */
} else
class_optcount = 10; /* \d, \s etc; make sure > 1 */
}
/* Check the syntax for POSIX stuff. The bits we actually handle are
checked during the real compile phase. */
else if (*ptr == '[' && check_posix_syntax(ptr, &ptr, &compile_block)) {
ptr++;
class_optcount = 10; /* Make sure > 1 */
}
/* Anything else just increments the possible optimization count. If
there are wide characters, we are going to have to use an XCLASS. */
else {
NON_SPECIAL_CHARACTER:
class_optcount++;
}
}
while (*(++ptr) != 0 && (inescq || *ptr != ']')); /* Concludes "do" above */
if (*ptr == 0) { /* Missing terminating ']' */
*errorptr = ERR6;
goto PCRE_ERROR_RETURN;
}
/* We can optimize when there was only one optimizable character. Repeats
for positive and negated single one-byte chars are handled by the general
code. Here, we handle repeats for the class opcodes. */
if (class_optcount == 1)
length += 3;
else {
length += 33;
/* A repeat needs either 1 or 5 bytes. If it is a possessive quantifier,
we also need extra for wrapping the whole thing in a sub-pattern. */
if (*ptr != 0 && ptr[1] == '{' && is_counted_repeat(ptr + 2)) {
ptr = read_repeat_counts(ptr + 2, &min, &max, errorptr);
if (*errorptr != NULL)
goto PCRE_ERROR_RETURN;
if ((min == 0 && (max == 1 || max == -1)) || (min == 1 && max == -1))
length++;
else
length += 5;
if (ptr[1] == '+') {
ptr++;
length += 2 + 2 * LINK_SIZE;
} else if (ptr[1] == '?')
ptr++;
}
}
continue;
/* Brackets may be genuine groups or special things */
case '(':
branch_newextra = 0;
bracket_length = 1 + LINK_SIZE;
/* Handle special forms of bracket, which all start (? */
if (ptr[1] == '?') {
int set, unset;
int *optset;
switch (c = ptr[2]) {
/* Skip over comments entirely */
case '#':
ptr += 3;
while (*ptr != 0 && *ptr != ')')
ptr++;
if (*ptr == 0) {
*errorptr = ERR18;
goto PCRE_ERROR_RETURN;
}
continue;
/* Non-referencing groups and lookaheads just move the pointer on, and
then behave like a non-special bracket, except that they don't increment
the count of extracting brackets. Ditto for the "once only" bracket,
which is in Perl from version 5.005. */
case ':':
case '=':
case '!':
case '>':
ptr += 2;
break;
/* (?R) specifies a recursive call to the regex, which is an extension
to provide the facility which can be obtained by (?p{perl-code}) in
Perl 5.6. In Perl 5.8 this has become (??{perl-code}).
From PCRE 4.00, items such as (?3) specify subroutine-like "calls" to
the appropriate numbered brackets. This includes both recursive and
non-recursive calls. (?R) is now synonymous with (?0). */
case 'R':
ptr++;
case '0':
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
case '8':
case '9':
ptr += 2;
if (c != 'R')
while ((digitab[*(++ptr)] & ctype_digit) != 0) ;
if (*ptr != ')') {
*errorptr = ERR29;
goto PCRE_ERROR_RETURN;
}
length += 1 + LINK_SIZE;
/* If this item is quantified, it will get wrapped inside brackets so
as to use the code for quantified brackets. We jump down and use the
code that handles this for real brackets. */
if (ptr[1] == '+' || ptr[1] == '*' || ptr[1] == '?' || ptr[1] == '{') {
length += 2 + 2 * LINK_SIZE; /* to make bracketed */
duplength = 5 + 3 * LINK_SIZE;
goto HANDLE_QUANTIFIED_BRACKETS;
}
continue;
/* (?C) is an extension which provides "callout" - to provide a bit of
the functionality of the Perl (?{...}) feature. An optional number may
follow (default is zero). */
case 'C':
ptr += 2;
while ((digitab[*(++ptr)] & ctype_digit) != 0) ;
if (*ptr != ')') {
*errorptr = ERR39;
goto PCRE_ERROR_RETURN;
}
length += 2;
continue;
/* Named subpatterns are an extension copied from Python */
case 'P':
ptr += 3;
if (*ptr == '<') {
const uschar *p; /* Don't amalgamate; some compilers */
p = ++ptr; /* grumble at autoincrement in declaration */
while ((compile_block.ctypes[*ptr] & ctype_word) != 0)
ptr++;
if (*ptr != '>') {
*errorptr = ERR42;
goto PCRE_ERROR_RETURN;
}
name_count++;
if (ptr - p > max_name_size)
max_name_size = (ptr - p);
break;
}
if (*ptr == '=' || *ptr == '>') {
while ((compile_block.ctypes[*(++ptr)] & ctype_word) != 0) ;
if (*ptr != ')') {
*errorptr = ERR42;
goto PCRE_ERROR_RETURN;
}
break;
}
/* Unknown character after (?P */
*errorptr = ERR41;
goto PCRE_ERROR_RETURN;
/* Lookbehinds are in Perl from version 5.005 */
case '<':
ptr += 3;
if (*ptr == '=' || *ptr == '!') {
branch_newextra = 1 + LINK_SIZE;
length += 1 + LINK_SIZE; /* For the first branch */
break;
}
*errorptr = ERR24;
goto PCRE_ERROR_RETURN;
/* Conditionals are in Perl from version 5.005. The bracket must either
be followed by a number (for bracket reference) or by an assertion
group, or (a PCRE extension) by 'R' for a recursion test. */
case '(':
if (ptr[3] == 'R' && ptr[4] == ')') {
ptr += 4;
length += 3;
} else if ((digitab[ptr[3]] & ctype_digit) != 0) {
ptr += 4;
length += 3;
while ((digitab[*ptr] & ctype_digit) != 0)
ptr++;
if (*ptr != ')') {
*errorptr = ERR26;
goto PCRE_ERROR_RETURN;
}
} else { /* An assertion must follow */
ptr++; /* Can treat like ':' as far as spacing is concerned */
if (ptr[2] != '?' ||
(ptr[3] != '=' && ptr[3] != '!' && ptr[3] != '<')) {
ptr += 2; /* To get right offset in message */
*errorptr = ERR28;
goto PCRE_ERROR_RETURN;
}
}
break;
/* Else loop checking valid options until ) is met. Anything else is an
error. If we are without any brackets, i.e. at top level, the settings
act as if specified in the options, so massage the options immediately.
This is for backward compatibility with Perl 5.004. */
default:
set = unset = 0;
optset = &set;
ptr += 2;
for (;; ptr++) {
c = *ptr;
switch (c) {
case 'i':
*optset |= PCRE_CASELESS;
continue;
case 'm':
*optset |= PCRE_MULTILINE;
continue;
case 's':
*optset |= PCRE_DOTALL;
continue;
case 'x':
*optset |= PCRE_EXTENDED;
continue;
case 'X':
*optset |= PCRE_EXTRA;
continue;
case 'U':
*optset |= PCRE_UNGREEDY;
continue;
case '-':
optset = &unset;
continue;
/* A termination by ')' indicates an options-setting-only item; if
this is at the very start of the pattern (indicated by item_count
being zero), we use it to set the global options. This is helpful
when analyzing the pattern for first characters, etc. Otherwise
nothing is done here and it is handled during the compiling
process.
[Historical note: Up to Perl 5.8, options settings at top level
were always global settings, wherever they appeared in the pattern.
That is, they were equivalent to an external setting. From 5.8
onwards, they apply only to what follows (which is what you might
expect).] */
case ')':
if (item_count == 0) {
options = (options | set) & (~unset);
set = unset = 0; /* To save length */
item_count--; /* To allow for several */
}
/* Fall through */
/* A termination by ':' indicates the start of a nested group with
the given options set. This is again handled at compile time, but
we must allow for compiled space if any of the ims options are
set. We also have to allow for resetting space at the end of
the group, which is why 4 is added to the length and not just 2.
If there are several changes of options within the same group, this
will lead to an over-estimate on the length, but this shouldn't
matter very much. We also have to allow for resetting options at
the start of any alternations, which we do by setting
branch_newextra to 2. Finally, we record whether the case-dependent
flag ever changes within the regex. This is used by the "required
character" code. */
case ':':
if (((set | unset) & PCRE_IMS) != 0) {
length += 4;
branch_newextra = 2;
if (((set | unset) & PCRE_CASELESS) != 0)
options |= PCRE_ICHANGED;
}
goto END_OPTIONS;
/* Unrecognized option character */
default:
*errorptr = ERR12;
goto PCRE_ERROR_RETURN;
}
}
/* If we hit a closing bracket, that's it - this is a freestanding
option-setting. We need to ensure that branch_extra is updated if
necessary. The only values branch_newextra can have here are 0 or 2.
If the value is 2, then branch_extra must either be 2 or 5, depending
on whether this is a lookbehind group or not. */
END_OPTIONS:
if (c == ')') {
if (branch_newextra == 2 &&
(branch_extra == 0 || branch_extra == 1 + LINK_SIZE))
branch_extra += branch_newextra;
continue;
}
/* If options were terminated by ':' control comes here. Fall through
to handle the group below. */
}
}
/* Extracting brackets must be counted so we can process escapes in a
Perlish way. If the number exceeds EXTRACT_BASIC_MAX we are going to
need an additional 3 bytes of store per extracting bracket. However, if
PCRE_NO_AUTO)CAPTURE is set, unadorned brackets become non-capturing, so we
must leave the count alone (it will aways be zero). */
else if ((options & PCRE_NO_AUTO_CAPTURE) == 0) {
bracount++;
if (bracount > EXTRACT_BASIC_MAX)
bracket_length += 3;
}
/* Save length for computing whole length at end if there's a repeat that
requires duplication of the group. Also save the current value of
branch_extra, and start the new group with the new value. If non-zero, this
will either be 2 for a (?imsx: group, or 3 for a lookbehind assertion. */
if (brastackptr >= sizeof(brastack) / sizeof(int)) {
*errorptr = ERR19;
goto PCRE_ERROR_RETURN;
}
bralenstack[brastackptr] = branch_extra;
branch_extra = branch_newextra;
brastack[brastackptr++] = length;
length += bracket_length;
continue;
/* Handle ket. Look for subsequent max/min; for certain sets of values we
have to replicate this bracket up to that many times. If brastackptr is
0 this is an unmatched bracket which will generate an error, but take care
not to try to access brastack[-1] when computing the length and restoring
the branch_extra value. */
case ')':
length += 1 + LINK_SIZE;
if (brastackptr > 0) {
duplength = length - brastack[--brastackptr];
branch_extra = bralenstack[brastackptr];
} else
duplength = 0;
/* The following code is also used when a recursion such as (?3) is
followed by a quantifier, because in that case, it has to be wrapped inside
brackets so that the quantifier works. The value of duplength must be
set before arrival. */
HANDLE_QUANTIFIED_BRACKETS:
/* Leave ptr at the final char; for read_repeat_counts this happens
automatically; for the others we need an increment. */
if ((c = ptr[1]) == '{' && is_counted_repeat(ptr + 2)) {
ptr = read_repeat_counts(ptr + 2, &min, &max, errorptr);
if (*errorptr != NULL)
goto PCRE_ERROR_RETURN;
} else if (c == '*') {
min = 0;
max = -1;
ptr++;
} else if (c == '+') {
min = 1;
max = -1;
ptr++;
} else if (c == '?') {
min = 0;
max = 1;
ptr++;
} else {
min = 1;
max = 1;
}
/* If the minimum is zero, we have to allow for an OP_BRAZERO before the
group, and if the maximum is greater than zero, we have to replicate
maxval-1 times; each replication acquires an OP_BRAZERO plus a nesting
bracket set. */
if (min == 0) {
length++;
if (max > 0)
length += (max - 1) * (duplength + 3 + 2 * LINK_SIZE);
}
/* When the minimum is greater than zero, we have to replicate up to
minval-1 times, with no additions required in the copies. Then, if there
is a limited maximum we have to replicate up to maxval-1 times allowing
for a BRAZERO item before each optional copy and nesting brackets for all
but one of the optional copies. */
else {
length += (min - 1) * duplength;
if (max > min) /* Need this test as max=-1 means no limit */
length += (max - min) * (duplength + 3 + 2 * LINK_SIZE)
- (2 + 2 * LINK_SIZE);
}
/* Allow space for once brackets for "possessive quantifier" */
if (ptr[1] == '+') {
ptr++;
length += 2 + 2 * LINK_SIZE;
}
continue;
/* Non-special character. For a run of such characters the length required
is the number of characters + 2, except that the maximum run length is
MAXLIT. We won't get a skipped space or a non-data escape or the start of a
# comment as the first character, so the length can't be zero. */
NORMAL_CHAR:
default:
length += 2;
runlength = 0;
do {
/* If in a \Q...\E sequence, check for end; otherwise it's a literal */
if (inescq) {
if (c == '\\' && ptr[1] == 'E') {
inescq = FALSE;
ptr++;
} else
runlength++;
continue;
}
/* Skip whitespace and comments for /x */
if ((options & PCRE_EXTENDED) != 0) {
if ((compile_block.ctypes[c] & ctype_space) != 0)
continue;
if (c == '#') {
/* The space before the ; is to avoid a warning on a silly compiler
on the Macintosh. */
while ((c = *(++ptr)) != 0 && c != NEWLINE) ;
continue;
}
}
/* Backslash may introduce a data char or a metacharacter; stop the
string before the latter. */
if (c == '\\') {
const uschar *saveptr = ptr;
c = check_escape(&ptr, errorptr, bracount, options, FALSE);
if (*errorptr != NULL)
goto PCRE_ERROR_RETURN;
if (c < 0) {
ptr = saveptr;
break;
}
/* In UTF-8 mode, add on the number of additional bytes needed to
encode this character, and save the total length in case this is a
final char that is repeated. */
}
/* Ordinary character or single-char escape */
runlength++;
}
/* This "while" is the end of the "do" above. */
while (runlength < MAXLIT &&
(compile_block.ctypes[c = *(++ptr)] & ctype_meta) == 0);
/* If we hit a meta-character, back off to point to it */
if (runlength < MAXLIT)
ptr--;
/* If the last char in the string is a UTF-8 multibyte character, we must
set lastcharlength correctly. If it was specified as an escape, this will
already have been done above. However, we also have to support in-line
UTF-8 characters, so check backwards from where we are. */
length += runlength;
continue;
}
}
length += 2 + LINK_SIZE; /* For final KET and END */
if (length > MAX_PATTERN_SIZE) {
*errorptr = ERR20;
return NULL;
}
/* Compute the size of data block needed and get it, either from malloc or
externally provided function. */
size = length + sizeof(real_pcre) + name_count * (max_name_size + 3);
re = (real_pcre *) malloc(size);
if (re == NULL) {
*errorptr = ERR21;
return NULL;
}
/* Put in the magic number, and save the size, options, and table pointer */
re->magic_number = MAGIC_NUMBER;
re->size = size;
re->options = options;
re->tables = tables;
re->name_entry_size = max_name_size + 3;
re->name_count = name_count;
/* The starting points of the name/number translation table and of the code are
passed around in the compile data block. */
compile_block.names_found = 0;
compile_block.name_entry_size = max_name_size + 3;
compile_block.name_table = (uschar *) re + sizeof(real_pcre);
codestart = compile_block.name_table + re->name_entry_size * re->name_count;
compile_block.start_code = codestart;
compile_block.req_varyopt = 0;
/* Set up a starting, non-extracting bracket, then compile the expression. On
error, *errorptr will be set non-NULL, so we don't need to look at the result
of the function here. */
ptr = (const uschar *) pattern;
code = (uschar *) codestart;
*code = OP_BRA;
bracount = 0;
(void) compile_regex(options, options & PCRE_IMS, &bracount, &code, &ptr,
errorptr, FALSE, 0, &firstbyte, &reqbyte, NULL,
&compile_block);
re->top_bracket = bracount;
re->top_backref = compile_block.top_backref;
/* If not reached end of pattern on success, there's an excess bracket. */
if (*errorptr == NULL && *ptr != 0)
*errorptr = ERR22;
/* Fill in the terminating state and check for disastrous overflow, but
if debugging, leave the test till after things are printed out. */
*code++ = OP_END;
if (code - codestart > length)
*errorptr = ERR23;
/* Give an error if there's back reference to a non-existent capturing
subpattern. */
if (re->top_backref > re->top_bracket)
*errorptr = ERR15;
/* Failed to compile, or error while post-processing */
if (*errorptr != NULL) {
free(re);
PCRE_ERROR_RETURN:
*erroroffset = ptr - (const uschar *) pattern;
return NULL;
}
/* If the anchored option was not passed, set the flag if we can determine that
the pattern is anchored by virtue of ^ characters or \A or anything else (such
as starting with .* when DOTALL is set).
Otherwise, if we know what the first character has to be, save it, because that
speeds up unanchored matches no end. If not, see if we can set the
PCRE_STARTLINE flag. This is helpful for multiline matches when all branches
start with ^. and also when all branches start with .* for non-DOTALL matches.
*/
if ((options & PCRE_ANCHORED) == 0) {
int temp_options = options;
if (is_anchored(codestart, &temp_options, 0, compile_block.backref_map))
re->options |= PCRE_ANCHORED;
else {
if (firstbyte < 0)
firstbyte = find_firstassertedchar(codestart, &temp_options, FALSE);
if (firstbyte >= 0) { /* Remove caseless flag for non-caseable chars */
int ch = firstbyte & 255;
re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&
compile_block.fcc[ch] == ch) ? ch : firstbyte;
re->options |= PCRE_FIRSTSET;
} else if (is_startline(codestart, 0, compile_block.backref_map))
re->options |= PCRE_STARTLINE;
}
}
/* For an anchored pattern, we use the "required byte" only if it follows a
variable length item in the regex. Remove the caseless flag for non-caseable
chars. */
if (reqbyte >= 0 &&
((re->options & PCRE_ANCHORED) == 0 || (reqbyte & REQ_VARY) != 0)) {
int ch = reqbyte & 255;
re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&
compile_block.fcc[ch] ==
ch) ? (reqbyte & ~REQ_CASELESS) : reqbyte;
re->options |= PCRE_REQCHSET;
}
/* Print out the compiled data for debugging */
return (pcre *) re;
}
/*************************************************
* Match a back-reference *
*************************************************/
/* If a back reference hasn't been set, the length that is passed is greater
than the number of characters left in the string, so the match fails.
Arguments:
offset index into the offset vector
eptr points into the subject
length length to be matched
md points to match data block
ims the ims flags
Returns: TRUE if matched
*/
static BOOL
match_ref(int offset, register const uschar * eptr, int length, match_data * md,
unsigned long int ims)
{
const uschar *p = md->start_subject + md->offset_vector[offset];
/* Always fail if not enough characters left */
if (length > md->end_subject - eptr)
return FALSE;
/* Separate the caselesss case for speed */
if ((ims & PCRE_CASELESS) != 0) {
while (length-- > 0)
if (md->lcc[*p++] != md->lcc[*eptr++])
return FALSE;
} else {
while (length-- > 0)
if (*p++ != *eptr++)
return FALSE;
}
return TRUE;
}
/***************************************************************************
****************************************************************************
RECURSION IN THE match() FUNCTION
The match() function is highly recursive. Some regular expressions can cause
it to recurse thousands of times. I was writing for Unix, so I just let it
call itself recursively. This uses the stack for saving everything that has
to be saved for a recursive call. On Unix, the stack can be large, and this
works fine.
It turns out that on non-Unix systems there are problems with programs that
use a lot of stack. (This despite the fact that every last chip has oodles
of memory these days, and techniques for extending the stack have been known
for decades.) So....
There is a fudge, triggered by defining NO_RECURSE, which avoids recursive
calls by keeping local variables that need to be preserved in blocks of memory
obtained from malloc instead instead of on the stack. Macros are used to
achieve this so that the actual code doesn't look very different to what it
always used to.
****************************************************************************
***************************************************************************/
/* These versions of the macros use the stack, as normal */
#ifndef NO_RECURSE
#define REGISTER register
#define RMATCH(rx,ra,rb,rc,rd,re,rf,rg) rx = match(ra,rb,rc,rd,re,rf,rg)
#define RRETURN(ra) return ra
#else
/* These versions of the macros manage a private stack on the heap. Note
that the rd argument of RMATCH isn't actually used. It's the md argument of
match(), which never actually changes. */
#define REGISTER
#define RMATCH(rx,ra,rb,rc,rd,re,rf,rg)\
{\
heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\
if (setjmp(frame->Xwhere) == 0)\
{\
newframe->Xeptr = ra;\
newframe->Xecode = rb;\
newframe->Xoffset_top = rc;\
newframe->Xims = re;\
newframe->Xeptrb = rf;\
newframe->Xflags = rg;\
newframe->Xprevframe = frame;\
frame = newframe;\
DPRINTF(("restarting from line %d\n", __LINE__));\
goto HEAP_RECURSE;\
}\
else\
{\
DPRINTF(("longjumped back to line %d\n", __LINE__));\
frame = md->thisframe;\
rx = frame->Xresult;\
}\
}
#define RRETURN(ra)\
{\
heapframe *newframe = frame;\
frame = newframe->Xprevframe;\
(pcre_stack_free)(newframe);\
if (frame != NULL)\
{\
frame->Xresult = ra;\
md->thisframe = frame;\
longjmp(frame->Xwhere, 1);\
}\
return ra;\
}
/* Structure for remembering the local variables in a private frame */
typedef struct heapframe {
struct heapframe *Xprevframe;
/* Function arguments that may change */
const uschar *Xeptr;
const uschar *Xecode;
int Xoffset_top;
long int Xims;
eptrblock *Xeptrb;
int Xflags;
/* Function local variables */
const uschar *Xcallpat;
const uschar *Xcharptr;
const uschar *Xdata;
const uschar *Xlastptr;
const uschar *Xnext;
const uschar *Xpp;
const uschar *Xprev;
const uschar *Xsaved_eptr;
recursion_info Xnew_recursive;
BOOL Xcur_is_word;
BOOL Xcondition;
BOOL Xminimize;
BOOL Xprev_is_word;
unsigned long int Xoriginal_ims;
int Xctype;
int Xfc;
int Xfi;
int Xlength;
int Xmax;
int Xmin;
int Xnumber;
int Xoffset;
int Xop;
int Xsave_capture_last;
int Xsave_offset1, Xsave_offset2, Xsave_offset3;
int Xstacksave[REC_STACK_SAVE_MAX];
eptrblock Xnewptrb;
/* Place to pass back result, and where to jump back to */
int Xresult;
jmp_buf Xwhere;
} heapframe;
#endif
/***************************************************************************
***************************************************************************/
/*************************************************
* Match from current position *
*************************************************/
/* On entry ecode points to the first opcode, and eptr to the first character
in the subject string, while eptrb holds the value of eptr at the start of the
last bracketed group - used for breaking infinite loops matching zero-length
strings. This function is called recursively in many circumstances. Whenever it
returns a negative (error) response, the outer incarnation must also return the
same response.
Performance note: It might be tempting to extract commonly used fields from the
md structure (e.g. utf8, end_subject) into individual variables to improve
performance. Tests using gcc on a SPARC disproved this; in the first case, it
made performance worse.
Arguments:
eptr pointer in subject
ecode position in code
offset_top current top pointer
md pointer to "static" info for the match
ims current /i, /m, and /s options
eptrb pointer to chain of blocks containing eptr at start of
brackets - for testing for empty matches
flags can contain
match_condassert - this is an assertion condition
match_isgroup - this is the start of a bracketed group
Returns: MATCH_MATCH if matched ) these values are >= 0
MATCH_NOMATCH if failed to match )
a negative PCRE_ERROR_xxx value if aborted by an error condition
(e.g. stopped by recursion limit)
*/
static int
match(REGISTER const uschar * eptr, REGISTER const uschar * ecode,
int offset_top, match_data * md, unsigned long int ims, eptrblock * eptrb,
int flags)
{
/* These variables do not need to be preserved over recursion in this function,
so they can be ordinary variables in all cases. Mark them with "register"
because they are used a lot in loops. */
register int rrc; /* Returns from recursive calls */
register int i; /* Used for loops not involving calls to RMATCH() */
register int c; /* Character values not kept over RMATCH() calls */
/* When recursion is not being used, all "local" variables that have to be
preserved over calls to RMATCH() are part of a "frame" which is obtained from
heap storage. Set up the top-level frame here; others are obtained from the
heap whenever RMATCH() does a "recursion". See the macro definitions above. */
#ifdef NO_RECURSE
heapframe *frame = (pcre_stack_malloc) (sizeof(heapframe));
frame->Xprevframe = NULL; /* Marks the top level */
/* Copy in the original argument variables */
frame->Xeptr = eptr;
frame->Xecode = ecode;
frame->Xoffset_top = offset_top;
frame->Xims = ims;
frame->Xeptrb = eptrb;
frame->Xflags = flags;
/* This is where control jumps back to to effect "recursion" */
HEAP_RECURSE:
/* Macros make the argument variables come from the current frame */
#define eptr frame->Xeptr
#define ecode frame->Xecode
#define offset_top frame->Xoffset_top
#define ims frame->Xims
#define eptrb frame->Xeptrb
#define flags frame->Xflags
/* Ditto for the local variables */
#define callpat frame->Xcallpat
#define charptr frame->Xcharptr
#define data frame->Xdata
#define lastptr frame->Xlastptr
#define next frame->Xnext
#define pp frame->Xpp
#define prev frame->Xprev
#define saved_eptr frame->Xsaved_eptr
#define new_recursive frame->Xnew_recursive
#define cur_is_word frame->Xcur_is_word
#define condition frame->Xcondition
#define minimize frame->Xminimize
#define prev_is_word frame->Xprev_is_word
#define original_ims frame->Xoriginal_ims
#define ctype frame->Xctype
#define fc frame->Xfc
#define fi frame->Xfi
#define length frame->Xlength
#define max frame->Xmax
#define min frame->Xmin
#define number frame->Xnumber
#define offset frame->Xoffset
#define op frame->Xop
#define save_capture_last frame->Xsave_capture_last
#define save_offset1 frame->Xsave_offset1
#define save_offset2 frame->Xsave_offset2
#define save_offset3 frame->Xsave_offset3
#define stacksave frame->Xstacksave
#define newptrb frame->Xnewptrb
/* When recursion is being used, local variables are allocated on the stack and
get preserved during recursion in the normal way. In this environment, fi and
i, and fc and c, can be the same variables. */
#else
#define fi i
#define fc c
const uschar *callpat; /* Many of these variables are used ony */
const uschar *charptr; /* small blocks of the code. My normal */
const uschar *data; /* style of coding would have declared */
const uschar *lastptr; /* them within each of those blocks. */
const uschar *next; /* However, in order to accommodate the */
const uschar *pp; /* version of this code that uses an */
const uschar *prev; /* external "stack" implemented on the */
const uschar *saved_eptr; /* heap, it is easier to declare them */
/* all here, so the declarations can */
recursion_info new_recursive; /* be cut out in a block. The only */
/* declarations within blocks below are */
BOOL cur_is_word; /* for variables that do not have to */
BOOL condition; /* be preserved over a recursive call */
BOOL minimize; /* to RMATCH(). */
BOOL prev_is_word;
unsigned long int original_ims;
int ctype;
int length;
int max;
int min;
int number;
int offset;
int op;
int save_capture_last;
int save_offset1, save_offset2, save_offset3;
int stacksave[REC_STACK_SAVE_MAX];
eptrblock newptrb;
#endif
/* OK, now we can get on with the real code of the function. Recursion is
specified by the macros RMATCH and RRETURN. When NO_RECURSE is *not* defined,
these just turn into a recursive call to match() and a "return", respectively.
However, RMATCH isn't like a function call because it's quite a complicated
macro. It has to be used in one particular way. This shouldn't, however, impact
performance when true recursion is being used. */
if (md->match_call_count++ >= md->match_limit)
RRETURN(PCRE_ERROR_MATCHLIMIT);
original_ims = ims; /* Save for resetting on ')' */
/* At the start of a bracketed group, add the current subject pointer to the
stack of such pointers, to be re-instated at the end of the group when we hit
the closing ket. When match() is called in other circumstances, we don't add to
this stack. */
if ((flags & match_isgroup) != 0) {
newptrb.epb_prev = eptrb;
newptrb.epb_saved_eptr = eptr;
eptrb = &newptrb;
}
/* Now start processing the operations. */
for (;;) {
op = *ecode;
minimize = FALSE;
/* Opening capturing bracket. If there is space in the offset vector, save
the current subject position in the working slot at the top of the vector. We
mustn't change the current values of the data slot, because they may be set
from a previous iteration of this group, and be referred to by a reference
inside the group.
If the bracket fails to match, we need to restore this value and also the
values of the final offsets, in case they were set by a previous iteration of
the same bracket.
If there isn't enough space in the offset vector, treat this as if it were a
non-capturing bracket. Don't worry about setting the flag for the error case
here; that is handled in the code for KET. */
if (op > OP_BRA) {
number = op - OP_BRA;
/* For extended extraction brackets (large number), we have to fish out the
number from a dummy opcode at the start. */
if (number > EXTRACT_BASIC_MAX)
number = GET2(ecode, 2 + LINK_SIZE);
offset = number << 1;
if (offset < md->offset_max) {
save_offset1 = md->offset_vector[offset];
save_offset2 = md->offset_vector[offset + 1];
save_offset3 = md->offset_vector[md->offset_end - number];
save_capture_last = md->capture_last;
DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2,
save_offset3));
md->offset_vector[md->offset_end - number] = eptr - md->start_subject;
do {
RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb,
match_isgroup);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
md->capture_last = save_capture_last;
ecode += GET(ecode, 1);
}
while (*ecode == OP_ALT);
DPRINTF(("bracket %d failed\n", number));
md->offset_vector[offset] = save_offset1;
md->offset_vector[offset + 1] = save_offset2;
md->offset_vector[md->offset_end - number] = save_offset3;
RRETURN(MATCH_NOMATCH);
}
/* Insufficient room for saving captured contents */
else
op = OP_BRA;
}
/* Other types of node can be handled by a switch */
switch (op) {
case OP_BRA: /* Non-capturing bracket: optimized */
DPRINTF(("start bracket 0\n"));
do {
RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb,
match_isgroup);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
ecode += GET(ecode, 1);
}
while (*ecode == OP_ALT);
DPRINTF(("bracket 0 failed\n"));
RRETURN(MATCH_NOMATCH);
/* Conditional group: compilation checked that there are no more than
two branches. If the condition is false, skipping the first branch takes us
past the end if there is only one branch, but that's OK because that is
exactly what going to the ket would do. */
case OP_COND:
if (ecode[LINK_SIZE + 1] == OP_CREF) { /* Condition extract or recurse test */
offset = GET2(ecode, LINK_SIZE + 2) << 1; /* Doubled ref number */
condition = (offset == CREF_RECURSE * 2) ?
(md->recursive != NULL) :
(offset < offset_top && md->offset_vector[offset] >= 0);
RMATCH(rrc, eptr, ecode + (condition ?
(LINK_SIZE + 4) : (LINK_SIZE + 1 +
GET(ecode, 1))),
offset_top, md, ims, eptrb, match_isgroup);
RRETURN(rrc);
}
/* The condition is an assertion. Call match() to evaluate it - setting
the final argument TRUE causes it to stop at the end of an assertion. */
else {
RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,
match_condassert | match_isgroup);
if (rrc == MATCH_MATCH) {
ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
while (*ecode == OP_ALT)
ecode += GET(ecode, 1);
} else if (rrc != MATCH_NOMATCH) {
RRETURN(rrc); /* Need braces because of following else */
} else
ecode += GET(ecode, 1);
RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb,
match_isgroup);
RRETURN(rrc);
}
/* Control never reaches here */
/* Skip over conditional reference or large extraction number data if
encountered. */
case OP_CREF:
case OP_BRANUMBER:
ecode += 3;
break;
/* End of the pattern. If we are in a recursion, we should restore the
offsets appropriately and continue from after the call. */
case OP_END:
if (md->recursive != NULL && md->recursive->group_num == 0) {
recursion_info *rec = md->recursive;
DPRINTF(("Hit the end in a (?0) recursion\n"));
md->recursive = rec->prevrec;
memmove(md->offset_vector, rec->offset_save,
rec->saved_max * sizeof(int));
md->start_match = rec->save_start;
ims = original_ims;
ecode = rec->after_call;
break;
}
/* Otherwise, if PCRE_NOTEMPTY is set, fail if we have matched an empty
string - backtracking will then try other alternatives, if any. */
if (md->notempty && eptr == md->start_match)
RRETURN(MATCH_NOMATCH);
md->end_match_ptr = eptr; /* Record where we ended */
md->end_offset_top = offset_top; /* and how many extracts were taken */
RRETURN(MATCH_MATCH);
/* Change option settings */
case OP_OPT:
ims = ecode[1];
ecode += 2;
DPRINTF(("ims set to %02lx\n", ims));
break;
/* Assertion brackets. Check the alternative branches in turn - the
matching won't pass the KET for an assertion. If any one branch matches,
the assertion is true. Lookbehind assertions have an OP_REVERSE item at the
start of each branch to move the current point backwards, so the code at
this level is identical to the lookahead case. */
case OP_ASSERT:
case OP_ASSERTBACK:
do {
RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,
match_isgroup);
if (rrc == MATCH_MATCH)
break;
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
ecode += GET(ecode, 1);
}
while (*ecode == OP_ALT);
if (*ecode == OP_KET)
RRETURN(MATCH_NOMATCH);
/* If checking an assertion for a condition, return MATCH_MATCH. */
if ((flags & match_condassert) != 0)
RRETURN(MATCH_MATCH);
/* Continue from after the assertion, updating the offsets high water
mark, since extracts may have been taken during the assertion. */
do
ecode += GET(ecode, 1);
while (*ecode == OP_ALT);
ecode += 1 + LINK_SIZE;
offset_top = md->end_offset_top;
continue;
/* Negative assertion: all branches must fail to match */
case OP_ASSERT_NOT:
case OP_ASSERTBACK_NOT:
do {
RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,
match_isgroup);
if (rrc == MATCH_MATCH)
RRETURN(MATCH_NOMATCH);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
ecode += GET(ecode, 1);
}
while (*ecode == OP_ALT);
if ((flags & match_condassert) != 0)
RRETURN(MATCH_MATCH);
ecode += 1 + LINK_SIZE;
continue;
/* Move the subject pointer back. This occurs only at the start of
each branch of a lookbehind assertion. If we are too close to the start to
move back, this match function fails. When working with UTF-8 we move
back a number of characters, not bytes. */
case OP_REVERSE:
/* No UTF-8 support, or not in UTF-8 mode: count is byte count */
{
eptr -= GET(ecode, 1);
if (eptr < md->start_subject)
RRETURN(MATCH_NOMATCH);
}
/* Skip to next op code */
ecode += 1 + LINK_SIZE;
break;
/* The callout item calls an external function, if one is provided, passing
details of the match so far. This is mainly for debugging, though the
function is able to force a failure. */
case OP_CALLOUT:
if (pcre_callout != NULL) {
pcre_callout_block cb;
cb.version = 0; /* Version 0 of the callout block */
cb.callout_number = ecode[1];
cb.offset_vector = md->offset_vector;
cb.subject = (const char *) md->start_subject;
cb.subject_length = md->end_subject - md->start_subject;
cb.start_match = md->start_match - md->start_subject;
cb.current_position = eptr - md->start_subject;
cb.capture_top = offset_top / 2;
cb.capture_last = md->capture_last;
cb.callout_data = md->callout_data;
if ((rrc = (*pcre_callout) (&cb)) > 0)
RRETURN(MATCH_NOMATCH);
if (rrc < 0)
RRETURN(rrc);
}
ecode += 2;
break;
/* Recursion either matches the current regex, or some subexpression. The
offset data is the offset to the starting bracket from the start of the
whole pattern. (This is so that it works from duplicated subpatterns.)
If there are any capturing brackets started but not finished, we have to
save their starting points and reinstate them after the recursion. However,
we don't know how many such there are (offset_top records the completed
total) so we just have to save all the potential data. There may be up to
65535 such values, which is too large to put on the stack, but using malloc
for small numbers seems expensive. As a compromise, the stack is used when
there are no more than REC_STACK_SAVE_MAX values to store; otherwise malloc
is used. A problem is what to do if the malloc fails ... there is no way of
returning to the top level with an error. Save the top REC_STACK_SAVE_MAX
values on the stack, and accept that the rest may be wrong.
There are also other values that have to be saved. We use a chained
sequence of blocks that actually live on the stack. Thanks to Robin Houston
for the original version of this logic. */
case OP_RECURSE:
{
callpat = md->start_code + GET(ecode, 1);
new_recursive.group_num = *callpat - OP_BRA;
/* For extended extraction brackets (large number), we have to fish out
the number from a dummy opcode at the start. */
if (new_recursive.group_num > EXTRACT_BASIC_MAX)
new_recursive.group_num = GET2(callpat, 2 + LINK_SIZE);
/* Add to "recursing stack" */
new_recursive.prevrec = md->recursive;
md->recursive = &new_recursive;
/* Find where to continue from afterwards */
ecode += 1 + LINK_SIZE;
new_recursive.after_call = ecode;
/* Now save the offset data. */
new_recursive.saved_max = md->offset_end;
if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
new_recursive.offset_save = stacksave;
else {
new_recursive.offset_save =
(int *) malloc(new_recursive.saved_max * sizeof(int));
if (new_recursive.offset_save == NULL)
RRETURN(PCRE_ERROR_NOMEMORY);
}
memcpy(new_recursive.offset_save, md->offset_vector,
new_recursive.saved_max * sizeof(int));
new_recursive.save_start = md->start_match;
md->start_match = eptr;
/* OK, now we can do the recursion. For each top-level alternative we
restore the offset and recursion data. */
DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
do {
RMATCH(rrc, eptr, callpat + 1 + LINK_SIZE, offset_top, md, ims,
eptrb, match_isgroup);
if (rrc == MATCH_MATCH) {
md->recursive = new_recursive.prevrec;
if (new_recursive.offset_save != stacksave)
free(new_recursive.offset_save);
RRETURN(MATCH_MATCH);
} else if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
md->recursive = &new_recursive;
memcpy(md->offset_vector, new_recursive.offset_save,
new_recursive.saved_max * sizeof(int));
callpat += GET(callpat, 1);
}
while (*callpat == OP_ALT);
DPRINTF(("Recursion didn't match\n"));
md->recursive = new_recursive.prevrec;
if (new_recursive.offset_save != stacksave)
free(new_recursive.offset_save);
RRETURN(MATCH_NOMATCH);
}
/* Control never reaches here */
/* "Once" brackets are like assertion brackets except that after a match,
the point in the subject string is not moved back. Thus there can never be
a move back into the brackets. Friedl calls these "atomic" subpatterns.
Check the alternative branches in turn - the matching won't pass the KET
for this kind of subpattern. If any one branch matches, we carry on as at
the end of a normal bracket, leaving the subject pointer. */
case OP_ONCE:
{
prev = ecode;
saved_eptr = eptr;
do {
RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims,
eptrb, match_isgroup);
if (rrc == MATCH_MATCH)
break;
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
ecode += GET(ecode, 1);
}
while (*ecode == OP_ALT);
/* If hit the end of the group (which could be repeated), fail */
if (*ecode != OP_ONCE && *ecode != OP_ALT)
RRETURN(MATCH_NOMATCH);
/* Continue as from after the assertion, updating the offsets high water
mark, since extracts may have been taken. */
do
ecode += GET(ecode, 1);
while (*ecode == OP_ALT);
offset_top = md->end_offset_top;
eptr = md->end_match_ptr;
/* For a non-repeating ket, just continue at this level. This also
happens for a repeating ket if no characters were matched in the group.
This is the forcible breaking of infinite loops as implemented in Perl
5.005. If there is an options reset, it will get obeyed in the normal
course of events. */
if (*ecode == OP_KET || eptr == saved_eptr) {
ecode += 1 + LINK_SIZE;
break;
}
/* The repeating kets try the rest of the pattern or restart from the
preceding bracket, in the appropriate order. We need to reset any options
that changed within the bracket before re-running it, so check the next
opcode. */
if (ecode[1 + LINK_SIZE] == OP_OPT) {
ims = (ims & ~PCRE_IMS) | ecode[4];
DPRINTF(("ims set to %02lx at group repeat\n", ims));
}
if (*ecode == OP_KETRMIN) {
RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb,
0);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
} else { /* OP_KETRMAX */
RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb,
0);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
}
}
RRETURN(MATCH_NOMATCH);
/* An alternation is the end of a branch; scan along to find the end of the
bracketed group and go to there. */
case OP_ALT:
do
ecode += GET(ecode, 1);
while (*ecode == OP_ALT);
break;
/* BRAZERO and BRAMINZERO occur just before a bracket group, indicating
that it may occur zero times. It may repeat infinitely, or not at all -
i.e. it could be ()* or ()? in the pattern. Brackets with fixed upper
repeat limits are compiled as a number of copies, with the optional ones
preceded by BRAZERO or BRAMINZERO. */
case OP_BRAZERO:
{
next = ecode + 1;
RMATCH(rrc, eptr, next, offset_top, md, ims, eptrb, match_isgroup);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
do
next += GET(next, 1);
while (*next == OP_ALT);
ecode = next + 1 + LINK_SIZE;
}
break;
case OP_BRAMINZERO:
{
next = ecode + 1;
do
next += GET(next, 1);
while (*next == OP_ALT);
RMATCH(rrc, eptr, next + 1 + LINK_SIZE, offset_top, md, ims, eptrb,
match_isgroup);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
ecode++;
}
break;
/* End of a group, repeated or non-repeating. If we are at the end of
an assertion "group", stop matching and return MATCH_MATCH, but record the
current high water mark for use by positive assertions. Do this also
for the "once" (not-backup up) groups. */
case OP_KET:
case OP_KETRMIN:
case OP_KETRMAX:
{
prev = ecode - GET(ecode, 1);
saved_eptr = eptrb->epb_saved_eptr;
/* Back up the stack of bracket start pointers. */
eptrb = eptrb->epb_prev;
if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
*prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||
*prev == OP_ONCE) {
md->end_match_ptr = eptr; /* For ONCE */
md->end_offset_top = offset_top;
RRETURN(MATCH_MATCH);
}
/* In all other cases except a conditional group we have to check the
group number back at the start and if necessary complete handling an
extraction by setting the offsets and bumping the high water mark. */
if (*prev != OP_COND) {
number = *prev - OP_BRA;
/* For extended extraction brackets (large number), we have to fish out
the number from a dummy opcode at the start. */
if (number > EXTRACT_BASIC_MAX)
number = GET2(prev, 2 + LINK_SIZE);
offset = number << 1;
/* Test for a numbered group. This includes groups called as a result
of recursion. Note that whole-pattern recursion is coded as a recurse
into group 0, so it won't be picked up here. Instead, we catch it when
the OP_END is reached. */
if (number > 0) {
md->capture_last = number;
if (offset >= md->offset_max)
md->offset_overflow = TRUE;
else {
md->offset_vector[offset] =
md->offset_vector[md->offset_end - number];
md->offset_vector[offset + 1] = eptr - md->start_subject;
if (offset_top <= offset)
offset_top = offset + 2;
}
/* Handle a recursively called group. Restore the offsets
appropriately and continue from after the call. */
if (md->recursive != NULL && md->recursive->group_num == number) {
recursion_info *rec = md->recursive;
DPRINTF(("Recursion (%d) succeeded - continuing\n", number));
md->recursive = rec->prevrec;
md->start_match = rec->save_start;
memcpy(md->offset_vector, rec->offset_save,
rec->saved_max * sizeof(int));
ecode = rec->after_call;
ims = original_ims;
break;
}
}
}
/* Reset the value of the ims flags, in case they got changed during
the group. */
ims = original_ims;
DPRINTF(("ims reset to %02lx\n", ims));
/* For a non-repeating ket, just continue at this level. This also
happens for a repeating ket if no characters were matched in the group.
This is the forcible breaking of infinite loops as implemented in Perl
5.005. If there is an options reset, it will get obeyed in the normal
course of events. */
if (*ecode == OP_KET || eptr == saved_eptr) {
ecode += 1 + LINK_SIZE;
break;
}
/* The repeating kets try the rest of the pattern or restart from the
preceding bracket, in the appropriate order. */
if (*ecode == OP_KETRMIN) {
RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb,
0);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
} else { /* OP_KETRMAX */
RMATCH(rrc, eptr, prev, offset_top, md, ims, eptrb, match_isgroup);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
RMATCH(rrc, eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb,
0);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
}
}
RRETURN(MATCH_NOMATCH);
/* Start of subject unless notbol, or after internal newline if multiline */
case OP_CIRC:
if (md->notbol && eptr == md->start_subject)
RRETURN(MATCH_NOMATCH);
if ((ims & PCRE_MULTILINE) != 0) {
if (eptr != md->start_subject && eptr[-1] != NEWLINE)
RRETURN(MATCH_NOMATCH);
ecode++;
break;
}
/* ... else fall through */
/* Start of subject assertion */
case OP_SOD:
if (eptr != md->start_subject)
RRETURN(MATCH_NOMATCH);
ecode++;
break;
/* Start of match assertion */
case OP_SOM:
if (eptr != md->start_subject + md->start_offset)
RRETURN(MATCH_NOMATCH);
ecode++;
break;
/* Assert before internal newline if multiline, or before a terminating
newline unless endonly is set, else end of subject unless noteol is set. */
case OP_DOLL:
if ((ims & PCRE_MULTILINE) != 0) {
if (eptr < md->end_subject) {
if (*eptr != NEWLINE)
RRETURN(MATCH_NOMATCH);
} else {
if (md->noteol)
RRETURN(MATCH_NOMATCH);
}
ecode++;
break;
} else {
if (md->noteol)
RRETURN(MATCH_NOMATCH);
if (!md->endonly) {
if (eptr < md->end_subject - 1 ||
(eptr == md->end_subject - 1 && *eptr != NEWLINE))
RRETURN(MATCH_NOMATCH);
ecode++;
break;
}
}
/* ... else fall through */
/* End of subject assertion (\z) */
case OP_EOD:
if (eptr < md->end_subject)
RRETURN(MATCH_NOMATCH);
ecode++;
break;
/* End of subject or ending \n assertion (\Z) */
case OP_EODN:
if (eptr < md->end_subject - 1 ||
(eptr == md->end_subject - 1 && *eptr != NEWLINE))
RRETURN(MATCH_NOMATCH);
ecode++;
break;
/* Word boundary assertions */
case OP_NOT_WORD_BOUNDARY:
case OP_WORD_BOUNDARY:
{
/* Find out if the previous and current characters are "word" characters.
It takes a bit more work in UTF-8 mode. Characters > 255 are assumed to
be "non-word" characters. */
/* More streamlined when not in UTF-8 mode */
{
prev_is_word = (eptr != md->start_subject) &&
((md->ctypes[eptr[-1]] & ctype_word) != 0);
cur_is_word = (eptr < md->end_subject) &&
((md->ctypes[*eptr] & ctype_word) != 0);
}
/* Now see if the situation is what we want */
if ((*ecode++ == OP_WORD_BOUNDARY) ?
cur_is_word == prev_is_word : cur_is_word != prev_is_word)
RRETURN(MATCH_NOMATCH);
}
break;
/* Match a single character type; inline for speed */
case OP_ANY:
if ((ims & PCRE_DOTALL) == 0 && eptr < md->end_subject
&& *eptr == NEWLINE)
RRETURN(MATCH_NOMATCH);
if (eptr++ >= md->end_subject)
RRETURN(MATCH_NOMATCH);
ecode++;
break;
/* Match a single byte, even in UTF-8 mode. This opcode really does match
any byte, even newline, independent of the setting of PCRE_DOTALL. */
case OP_ANYBYTE:
if (eptr++ >= md->end_subject)
RRETURN(MATCH_NOMATCH);
ecode++;
break;
case OP_NOT_DIGIT:
if (eptr >= md->end_subject)
RRETURN(MATCH_NOMATCH);
GETCHARINCTEST(c, eptr);
if ((md->ctypes[c] & ctype_digit) != 0)
RRETURN(MATCH_NOMATCH);
ecode++;
break;
case OP_DIGIT:
if (eptr >= md->end_subject)
RRETURN(MATCH_NOMATCH);
GETCHARINCTEST(c, eptr);
if ((md->ctypes[c] & ctype_digit) == 0)
RRETURN(MATCH_NOMATCH);
ecode++;
break;
case OP_NOT_WHITESPACE:
if (eptr >= md->end_subject)
RRETURN(MATCH_NOMATCH);
GETCHARINCTEST(c, eptr);
if ((md->ctypes[c] & ctype_space) != 0)
RRETURN(MATCH_NOMATCH);
ecode++;
break;
case OP_WHITESPACE:
if (eptr >= md->end_subject)
RRETURN(MATCH_NOMATCH);
GETCHARINCTEST(c, eptr);
if ((md->ctypes[c] & ctype_space) == 0)
RRETURN(MATCH_NOMATCH);
ecode++;
break;
case OP_NOT_WORDCHAR:
if (eptr >= md->end_subject)
RRETURN(MATCH_NOMATCH);
GETCHARINCTEST(c, eptr);
if ((md->ctypes[c] & ctype_word) != 0)
RRETURN(MATCH_NOMATCH);
ecode++;
break;
case OP_WORDCHAR:
if (eptr >= md->end_subject)
RRETURN(MATCH_NOMATCH);
GETCHARINCTEST(c, eptr);
if ((md->ctypes[c] & ctype_word) == 0)
RRETURN(MATCH_NOMATCH);
ecode++;
break;
/* Match a back reference, possibly repeatedly. Look past the end of the
item to see if there is repeat information following. The code is similar
to that for character classes, but repeated for efficiency. Then obey
similar code to character type repeats - written out again for speed.
However, if the referenced string is the empty string, always treat
it as matched, any number of times (otherwise there could be infinite
loops). */
case OP_REF:
{
offset = GET2(ecode, 1) << 1; /* Doubled ref number */
ecode += 3; /* Advance past item */
/* If the reference is unset, set the length to be longer than the amount
of subject left; this ensures that every attempt at a match fails. We
can't just fail here, because of the possibility of quantifiers with zero
minima. */
length = (offset >= offset_top || md->offset_vector[offset] < 0) ?
md->end_subject - eptr + 1 :
md->offset_vector[offset + 1] - md->offset_vector[offset];
/* Set up for repetition, or handle the non-repeated case */
switch (*ecode) {
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRPLUS:
case OP_CRMINPLUS:
case OP_CRQUERY:
case OP_CRMINQUERY:
c = *ecode++ - OP_CRSTAR;
minimize = (c & 1) != 0;
min = rep_min[c]; /* Pick up values from tables; */
max = rep_max[c]; /* zero for max => infinity */
if (max == 0)
max = INT_MAX;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
minimize = (*ecode == OP_CRMINRANGE);
min = GET2(ecode, 1);
max = GET2(ecode, 3);
if (max == 0)
max = INT_MAX;
ecode += 5;
break;
default: /* No repeat follows */
if (!match_ref(offset, eptr, length, md, ims))
RRETURN(MATCH_NOMATCH);
eptr += length;
continue; /* With the main loop */
}
/* If the length of the reference is zero, just continue with the
main loop. */
if (length == 0)
continue;
/* First, ensure the minimum number of matches are present. We get back
the length of the reference string explicitly rather than passing the
address of eptr, so that eptr can be a register variable. */
for (i = 1; i <= min; i++) {
if (!match_ref(offset, eptr, length, md, ims))
RRETURN(MATCH_NOMATCH);
eptr += length;
}
/* If min = max, continue at the same level without recursion.
They are not both allowed to be zero. */
if (min == max)
continue;
/* If minimizing, keep trying and advancing the pointer */
if (minimize) {
for (fi = min;; fi++) {
RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
if (fi >= max || !match_ref(offset, eptr, length, md, ims))
RRETURN(MATCH_NOMATCH);
eptr += length;
}
/* Control never gets here */
}
/* If maximizing, find the longest string and work backwards */
else {
pp = eptr;
for (i = min; i < max; i++) {
if (!match_ref(offset, eptr, length, md, ims))
break;
eptr += length;
}
while (eptr >= pp) {
RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
eptr -= length;
}
RRETURN(MATCH_NOMATCH);
}
}
/* Control never gets here */
/* Match a bit-mapped character class, possibly repeatedly. This op code is
used when all the characters in the class have values in the range 0-255.
The only difference between OP_CLASS and OP_NCLASS occurs when a data
character outside the range is encountered.
First, look past the end of the item to see if there is repeat information
following. Then obey similar code to character type repeats - written out
again for speed. */
case OP_NCLASS:
case OP_CLASS:
{
data = ecode + 1; /* Save for matching */
ecode += 33; /* Advance past the item */
switch (*ecode) {
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRPLUS:
case OP_CRMINPLUS:
case OP_CRQUERY:
case OP_CRMINQUERY:
c = *ecode++ - OP_CRSTAR;
minimize = (c & 1) != 0;
min = rep_min[c]; /* Pick up values from tables; */
max = rep_max[c]; /* zero for max => infinity */
if (max == 0)
max = INT_MAX;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
minimize = (*ecode == OP_CRMINRANGE);
min = GET2(ecode, 1);
max = GET2(ecode, 3);
if (max == 0)
max = INT_MAX;
ecode += 5;
break;
default: /* No repeat follows */
min = max = 1;
break;
}
/* First, ensure the minimum number of matches are present. */
/* Not UTF-8 mode */
{
for (i = 1; i <= min; i++) {
if (eptr >= md->end_subject)
RRETURN(MATCH_NOMATCH);
c = *eptr++;
if ((data[c / 8] & (1 << (c & 7))) == 0)
RRETURN(MATCH_NOMATCH);
}
}
/* If max == min we can continue with the main loop without the
need to recurse. */
if (min == max)
continue;
/* If minimizing, keep testing the rest of the expression and advancing
the pointer while it matches the class. */
if (minimize) {
/* Not UTF-8 mode */
{
for (fi = min;; fi++) {
RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
if (fi >= max || eptr >= md->end_subject)
RRETURN(MATCH_NOMATCH);
c = *eptr++;
if ((data[c / 8] & (1 << (c & 7))) == 0)
RRETURN(MATCH_NOMATCH);
}
}
/* Control never gets here */
}
/* If maximizing, find the longest possible run, then work backwards. */
else {
pp = eptr;
/* Not UTF-8 mode */
{
for (i = min; i < max; i++) {
if (eptr >= md->end_subject)
break;
c = *eptr;
if ((data[c / 8] & (1 << (c & 7))) == 0)
break;
eptr++;
}
while (eptr >= pp) {
RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);
eptr--;
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
}
}
RRETURN(MATCH_NOMATCH);
}
}
/* Control never gets here */
/* Match an extended character class. This opcode is encountered only
in UTF-8 mode, because that's the only time it is compiled. */
/* Match a run of characters */
case OP_CHARS:
{
register int slen = ecode[1];
ecode += 2;
if (slen > md->end_subject - eptr)
RRETURN(MATCH_NOMATCH);
if ((ims & PCRE_CASELESS) != 0) {
while (slen-- > 0)
if (md->lcc[*ecode++] != md->lcc[*eptr++])
RRETURN(MATCH_NOMATCH);
} else {
while (slen-- > 0)
if (*ecode++ != *eptr++)
RRETURN(MATCH_NOMATCH);
}
}
break;
/* Match a single character repeatedly; different opcodes share code. */
case OP_EXACT:
min = max = GET2(ecode, 1);
ecode += 3;
goto REPEATCHAR;
case OP_UPTO:
case OP_MINUPTO:
min = 0;
max = GET2(ecode, 1);
minimize = *ecode == OP_MINUPTO;
ecode += 3;
goto REPEATCHAR;
case OP_STAR:
case OP_MINSTAR:
case OP_PLUS:
case OP_MINPLUS:
case OP_QUERY:
case OP_MINQUERY:
c = *ecode++ - OP_STAR;
minimize = (c & 1) != 0;
min = rep_min[c]; /* Pick up values from tables; */
max = rep_max[c]; /* zero for max => infinity */
if (max == 0)
max = INT_MAX;
/* Common code for all repeated single-character matches. We can give
up quickly if there are fewer than the minimum number of characters left in
the subject. */
REPEATCHAR:
/* When not in UTF-8 mode, load a single-byte character. */
{
if (min > md->end_subject - eptr)
RRETURN(MATCH_NOMATCH);
fc = *ecode++;
}
/* The value of fc at this point is always less than 256, though we may or
may not be in UTF-8 mode. The code is duplicated for the caseless and
caseful cases, for speed, since matching characters is likely to be quite
common. First, ensure the minimum number of matches are present. If min =
max, continue at the same level without recursing. Otherwise, if
minimizing, keep trying the rest of the expression and advancing one
matching character if failing, up to the maximum. Alternatively, if
maximizing, find the maximum number of characters and work backwards. */
DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
max, eptr));
if ((ims & PCRE_CASELESS) != 0) {
fc = md->lcc[fc];
for (i = 1; i <= min; i++)
if (fc != md->lcc[*eptr++])
RRETURN(MATCH_NOMATCH);
if (min == max)
continue;
if (minimize) {
for (fi = min;; fi++) {
RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
if (fi >= max || eptr >= md->end_subject || fc != md->lcc[*eptr++])
RRETURN(MATCH_NOMATCH);
}
/* Control never gets here */
} else {
pp = eptr;
for (i = min; i < max; i++) {
if (eptr >= md->end_subject || fc != md->lcc[*eptr])
break;
eptr++;
}
while (eptr >= pp) {
RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);
eptr--;
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
}
RRETURN(MATCH_NOMATCH);
}
/* Control never gets here */
}
/* Caseful comparisons (includes all multi-byte characters) */
else {
for (i = 1; i <= min; i++)
if (fc != *eptr++)
RRETURN(MATCH_NOMATCH);
if (min == max)
continue;
if (minimize) {
for (fi = min;; fi++) {
RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
if (fi >= max || eptr >= md->end_subject || fc != *eptr++)
RRETURN(MATCH_NOMATCH);
}
/* Control never gets here */
} else {
pp = eptr;
for (i = min; i < max; i++) {
if (eptr >= md->end_subject || fc != *eptr)
break;
eptr++;
}
while (eptr >= pp) {
RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);
eptr--;
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
}
RRETURN(MATCH_NOMATCH);
}
}
/* Control never gets here */
/* Match a negated single one-byte character. The character we are
checking can be multibyte. */
case OP_NOT:
if (eptr >= md->end_subject)
RRETURN(MATCH_NOMATCH);
ecode++;
GETCHARINCTEST(c, eptr);
if ((ims & PCRE_CASELESS) != 0) {
c = md->lcc[c];
if (md->lcc[*ecode++] == c)
RRETURN(MATCH_NOMATCH);
} else {
if (*ecode++ == c)
RRETURN(MATCH_NOMATCH);
}
break;
/* Match a negated single one-byte character repeatedly. This is almost a
repeat of the code for a repeated single character, but I haven't found a
nice way of commoning these up that doesn't require a test of the
positive/negative option for each character match. Maybe that wouldn't add
very much to the time taken, but character matching *is* what this is all
about... */
case OP_NOTEXACT:
min = max = GET2(ecode, 1);
ecode += 3;
goto REPEATNOTCHAR;
case OP_NOTUPTO:
case OP_NOTMINUPTO:
min = 0;
max = GET2(ecode, 1);
minimize = *ecode == OP_NOTMINUPTO;
ecode += 3;
goto REPEATNOTCHAR;
case OP_NOTSTAR:
case OP_NOTMINSTAR:
case OP_NOTPLUS:
case OP_NOTMINPLUS:
case OP_NOTQUERY:
case OP_NOTMINQUERY:
c = *ecode++ - OP_NOTSTAR;
minimize = (c & 1) != 0;
min = rep_min[c]; /* Pick up values from tables; */
max = rep_max[c]; /* zero for max => infinity */
if (max == 0)
max = INT_MAX;
/* Common code for all repeated single-character (less than 255) matches.
We can give up quickly if there are fewer than the minimum number of
characters left in the subject. */
REPEATNOTCHAR:
if (min > md->end_subject - eptr)
RRETURN(MATCH_NOMATCH);
fc = *ecode++;
/* The code is duplicated for the caseless and caseful cases, for speed,
since matching characters is likely to be quite common. First, ensure the
minimum number of matches are present. If min = max, continue at the same
level without recursing. Otherwise, if minimizing, keep trying the rest of
the expression and advancing one matching character if failing, up to the
maximum. Alternatively, if maximizing, find the maximum number of
characters and work backwards. */
DPRINTF(("negative matching %c{%d,%d} against subject %.*s\n", fc, min,
max, max, eptr));
if ((ims & PCRE_CASELESS) != 0) {
fc = md->lcc[fc];
/* Not UTF-8 mode */
{
for (i = 1; i <= min; i++)
if (fc == md->lcc[*eptr++])
RRETURN(MATCH_NOMATCH);
}
if (min == max)
continue;
if (minimize) {
/* Not UTF-8 mode */
{
for (fi = min;; fi++) {
RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
if (fi >= max || eptr >= md->end_subject
|| fc == md->lcc[*eptr++])
RRETURN(MATCH_NOMATCH);
}
}
/* Control never gets here */
}
/* Maximize case */
else {
pp = eptr;
/* Not UTF-8 mode */
{
for (i = min; i < max; i++) {
if (eptr >= md->end_subject || fc == md->lcc[*eptr])
break;
eptr++;
}
while (eptr >= pp) {
RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
eptr--;
}
}
RRETURN(MATCH_NOMATCH);
}
/* Control never gets here */
}
/* Caseful comparisons */
else {
/* Not UTF-8 mode */
{
for (i = 1; i <= min; i++)
if (fc == *eptr++)
RRETURN(MATCH_NOMATCH);
}
if (min == max)
continue;
if (minimize) {
/* Not UTF-8 mode */
{
for (fi = min;; fi++) {
RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
if (fi >= max || eptr >= md->end_subject || fc == *eptr++)
RRETURN(MATCH_NOMATCH);
}
}
/* Control never gets here */
}
/* Maximize case */
else {
pp = eptr;
/* Not UTF-8 mode */
{
for (i = min; i < max; i++) {
if (eptr >= md->end_subject || fc == *eptr)
break;
eptr++;
}
while (eptr >= pp) {
RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
eptr--;
}
}
RRETURN(MATCH_NOMATCH);
}
}
/* Control never gets here */
/* Match a single character type repeatedly; several different opcodes
share code. This is very similar to the code for single characters, but we
repeat it in the interests of efficiency. */
case OP_TYPEEXACT:
min = max = GET2(ecode, 1);
minimize = TRUE;
ecode += 3;
goto REPEATTYPE;
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
min = 0;
max = GET2(ecode, 1);
minimize = *ecode == OP_TYPEMINUPTO;
ecode += 3;
goto REPEATTYPE;
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
c = *ecode++ - OP_TYPESTAR;
minimize = (c & 1) != 0;
min = rep_min[c]; /* Pick up values from tables; */
max = rep_max[c]; /* zero for max => infinity */
if (max == 0)
max = INT_MAX;
/* Common code for all repeated single character type matches. Note that
in UTF-8 mode, '.' matches a character of any length, but for the other
character types, the valid characters are all one-byte long. */
REPEATTYPE:
ctype = *ecode++; /* Code for the character type */
/* First, ensure the minimum number of matches are present. Use inline
code for maximizing the speed, and do the type test once at the start
(i.e. keep it out of the loop). Also we can test that there are at least
the minimum number of bytes before we start. This isn't as effective in
UTF-8 mode, but it does no harm. Separate the UTF-8 code completely as that
is tidier. */
if (min > md->end_subject - eptr)
RRETURN(MATCH_NOMATCH);
if (min > 0) {
/* Code for the non-UTF-8 case for minimum matching */
switch (ctype) {
case OP_ANY:
if ((ims & PCRE_DOTALL) == 0) {
for (i = 1; i <= min; i++)
if (*eptr++ == NEWLINE)
RRETURN(MATCH_NOMATCH);
} else
eptr += min;
break;
case OP_ANYBYTE:
eptr += min;
break;
case OP_NOT_DIGIT:
for (i = 1; i <= min; i++)
if ((md->ctypes[*eptr++] & ctype_digit) != 0)
RRETURN(MATCH_NOMATCH);
break;
case OP_DIGIT:
for (i = 1; i <= min; i++)
if ((md->ctypes[*eptr++] & ctype_digit) == 0)
RRETURN(MATCH_NOMATCH);
break;
case OP_NOT_WHITESPACE:
for (i = 1; i <= min; i++)
if ((md->ctypes[*eptr++] & ctype_space) != 0)
RRETURN(MATCH_NOMATCH);
break;
case OP_WHITESPACE:
for (i = 1; i <= min; i++)
if ((md->ctypes[*eptr++] & ctype_space) == 0)
RRETURN(MATCH_NOMATCH);
break;
case OP_NOT_WORDCHAR:
for (i = 1; i <= min; i++)
if ((md->ctypes[*eptr++] & ctype_word) != 0)
RRETURN(MATCH_NOMATCH);
break;
case OP_WORDCHAR:
for (i = 1; i <= min; i++)
if ((md->ctypes[*eptr++] & ctype_word) == 0)
RRETURN(MATCH_NOMATCH);
break;
}
}
/* If min = max, continue at the same level without recursing */
if (min == max)
continue;
/* If minimizing, we have to test the rest of the pattern before each
subsequent match. Again, separate the UTF-8 case for speed. */
if (minimize) {
/* Not UTF-8 mode */
{
for (fi = min;; fi++) {
RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
if (fi >= max || eptr >= md->end_subject)
RRETURN(MATCH_NOMATCH);
c = *eptr++;
switch (ctype) {
case OP_ANY:
if ((ims & PCRE_DOTALL) == 0 && c == NEWLINE)
RRETURN(MATCH_NOMATCH);
break;
case OP_ANYBYTE:
break;
case OP_NOT_DIGIT:
if ((md->ctypes[c] & ctype_digit) != 0)
RRETURN(MATCH_NOMATCH);
break;
case OP_DIGIT:
if ((md->ctypes[c] & ctype_digit) == 0)
RRETURN(MATCH_NOMATCH);
break;
case OP_NOT_WHITESPACE:
if ((md->ctypes[c] & ctype_space) != 0)
RRETURN(MATCH_NOMATCH);
break;
case OP_WHITESPACE:
if ((md->ctypes[c] & ctype_space) == 0)
RRETURN(MATCH_NOMATCH);
break;
case OP_NOT_WORDCHAR:
if ((md->ctypes[c] & ctype_word) != 0)
RRETURN(MATCH_NOMATCH);
break;
case OP_WORDCHAR:
if ((md->ctypes[c] & ctype_word) == 0)
RRETURN(MATCH_NOMATCH);
break;
}
}
}
/* Control never gets here */
}
/* If maximizing it is worth using inline code for speed, doing the type
test once at the start (i.e. keep it out of the loop). Again, keep the
UTF-8 stuff separate. */
else {
pp = eptr;
/* Not UTF-8 mode */
{
switch (ctype) {
case OP_ANY:
if ((ims & PCRE_DOTALL) == 0) {
for (i = min; i < max; i++) {
if (eptr >= md->end_subject || *eptr == NEWLINE)
break;
eptr++;
}
break;
}
/* For DOTALL case, fall through and treat as \C */
case OP_ANYBYTE:
c = max - min;
if (c > md->end_subject - eptr)
c = md->end_subject - eptr;
eptr += c;
break;
case OP_NOT_DIGIT:
for (i = min; i < max; i++) {
if (eptr >= md->end_subject
|| (md->ctypes[*eptr] & ctype_digit) != 0)
break;
eptr++;
}
break;
case OP_DIGIT:
for (i = min; i < max; i++) {
if (eptr >= md->end_subject
|| (md->ctypes[*eptr] & ctype_digit) == 0)
break;
eptr++;
}
break;
case OP_NOT_WHITESPACE:
for (i = min; i < max; i++) {
if (eptr >= md->end_subject
|| (md->ctypes[*eptr] & ctype_space) != 0)
break;
eptr++;
}
break;
case OP_WHITESPACE:
for (i = min; i < max; i++) {
if (eptr >= md->end_subject
|| (md->ctypes[*eptr] & ctype_space) == 0)
break;
eptr++;
}
break;
case OP_NOT_WORDCHAR:
for (i = min; i < max; i++) {
if (eptr >= md->end_subject
|| (md->ctypes[*eptr] & ctype_word) != 0)
break;
eptr++;
}
break;
case OP_WORDCHAR:
for (i = min; i < max; i++) {
if (eptr >= md->end_subject
|| (md->ctypes[*eptr] & ctype_word) == 0)
break;
eptr++;
}
break;
}
/* eptr is now past the end of the maximum run */
while (eptr >= pp) {
RMATCH(rrc, eptr, ecode, offset_top, md, ims, eptrb, 0);
eptr--;
if (rrc != MATCH_NOMATCH)
RRETURN(rrc);
}
}
/* Get here if we can't make it match with any permitted repetitions */
RRETURN(MATCH_NOMATCH);
}
/* Control never gets here */
/* There's been some horrible disaster. Since all codes > OP_BRA are
for capturing brackets, and there shouldn't be any gaps between 0 and
OP_BRA, arrival here can only mean there is something seriously wrong
in the code above or the OP_xxx definitions. */
default:
DPRINTF(("Unknown opcode %d\n", *ecode));
RRETURN(PCRE_ERROR_UNKNOWN_NODE);
}
/* Do not stick any code in here without much thought; it is assumed
that "continue" in the code above comes out to here to repeat the main
loop. */
} /* End of main loop */
/* Control never reaches here */
}
/***************************************************************************
****************************************************************************
RECURSION IN THE match() FUNCTION
Undefine all the macros that were defined above to handle this. */
#ifdef NO_RECURSE
#undef eptr
#undef ecode
#undef offset_top
#undef ims
#undef eptrb
#undef flags
#undef callpat
#undef charptr
#undef data
#undef lastptr
#undef next
#undef pp
#undef prev
#undef saved_eptr
#undef new_recursive
#undef cur_is_word
#undef condition
#undef minimize
#undef prev_is_word
#undef original_ims
#undef ctype
#undef length
#undef max
#undef min
#undef number
#undef offset
#undef op
#undef save_capture_last
#undef save_offset1
#undef save_offset2
#undef save_offset3
#undef stacksave
#undef newptrb
#endif
/* These two are defined as macros in both cases */
#undef fc
#undef fi
/***************************************************************************
***************************************************************************/
/*************************************************
* Execute a Regular Expression *
*************************************************/
/* This function applies a compiled re to a subject string and picks out
portions of the string if it matches. Two elements in the vector are set for
each substring: the offsets to the start and end of the substring.
Arguments:
external_re points to the compiled expression
extra_data points to extra data or is NULL
subject points to the subject string
length length of subject string (may contain binary zeros)
start_offset where to start in the subject string
options option bits
offsets points to a vector of ints to be filled in with offsets
offsetcount the number of elements in the vector
Returns: > 0 => success; value is the number of elements filled in
= 0 => success, but offsets is not big enough
-1 => failed to match
< -1 => some kind of unexpected problem
*/
EXPORT int
pcre_exec(const pcre * external_re, const pcre_extra * extra_data,
const char *subject, int length, int start_offset, int options,
int *offsets, int offsetcount)
{
int rc, resetcount, ocount;
int first_byte = -1;
int req_byte = -1;
int req_byte2 = -1;
unsigned long int ims = 0;
BOOL using_temporary_offsets = FALSE;
BOOL anchored;
BOOL startline;
BOOL first_byte_caseless = FALSE;
BOOL req_byte_caseless = FALSE;
match_data match_block;
const uschar *start_bits = NULL;
const uschar *start_match = (const uschar *) subject + start_offset;
const uschar *end_subject;
const uschar *req_byte_ptr = start_match - 1;
const pcre_study_data *study;
const real_pcre *re = (const real_pcre *) external_re;
/* Plausibility checks */
if ((options & ~PUBLIC_EXEC_OPTIONS) != 0)
return PCRE_ERROR_BADOPTION;
if (re == NULL || subject == NULL || (offsets == NULL && offsetcount > 0))
return PCRE_ERROR_NULL;
/* Fish out the optional data from the extra_data structure, first setting
the default values. */
study = NULL;
match_block.match_limit = MATCH_LIMIT;
match_block.callout_data = NULL;
if (extra_data != NULL) {
register unsigned int flags = extra_data->flags;
if ((flags & PCRE_EXTRA_STUDY_DATA) != 0)
study = (const pcre_study_data *) extra_data->study_data;
if ((flags & PCRE_EXTRA_MATCH_LIMIT) != 0)
match_block.match_limit = extra_data->match_limit;
if ((flags & PCRE_EXTRA_CALLOUT_DATA) != 0)
match_block.callout_data = extra_data->callout_data;
}
/* Now we have re supposedly pointing to the regex */
if (re->magic_number != MAGIC_NUMBER)
return PCRE_ERROR_BADMAGIC;
anchored = ((re->options | options) & PCRE_ANCHORED) != 0;
startline = (re->options & PCRE_STARTLINE) != 0;
match_block.start_code =
(const uschar *) re + sizeof(real_pcre) +
re->name_count * re->name_entry_size;
match_block.start_subject = (const uschar *) subject;
match_block.start_offset = start_offset;
match_block.end_subject = match_block.start_subject + length;
end_subject = match_block.end_subject;
match_block.endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;
match_block.utf8 = (re->options & PCRE_UTF8) != 0;
match_block.notbol = (options & PCRE_NOTBOL) != 0;
match_block.noteol = (options & PCRE_NOTEOL) != 0;
match_block.notempty = (options & PCRE_NOTEMPTY) != 0;
match_block.recursive = NULL; /* No recursion at top level */
match_block.lcc = re->tables + lcc_offset;
match_block.ctypes = re->tables + ctypes_offset;
/* The ims options can vary during the matching as a result of the presence
of (?ims) items in the pattern. They are kept in a local variable so that
restoring at the exit of a group is easy. */
ims = re->options & (PCRE_CASELESS | PCRE_MULTILINE | PCRE_DOTALL);
/* If the expression has got more back references than the offsets supplied can
hold, we get a temporary bit of working store to use during the matching.
Otherwise, we can use the vector supplied, rounding down its size to a multiple
of 3. */
ocount = offsetcount - (offsetcount % 3);
if (re->top_backref > 0 && re->top_backref >= ocount / 3) {
ocount = re->top_backref * 3 + 3;
match_block.offset_vector = (int *) malloc(ocount * sizeof(int));
if (match_block.offset_vector == NULL)
return PCRE_ERROR_NOMEMORY;
using_temporary_offsets = TRUE;
DPRINTF(("Got memory to hold back references\n"));
} else
match_block.offset_vector = offsets;
match_block.offset_end = ocount;
match_block.offset_max = (2 * ocount) / 3;
match_block.offset_overflow = FALSE;
match_block.capture_last = -1;
/* Compute the minimum number of offsets that we need to reset each time. Doing
this makes a huge difference to execution time when there aren't many brackets
in the pattern. */
resetcount = 2 + re->top_bracket * 2;
if (resetcount > offsetcount)
resetcount = ocount;
/* Reset the working variable associated with each extraction. These should
never be used unless previously set, but they get saved and restored, and so we
initialize them to avoid reading uninitialized locations. */
if (match_block.offset_vector != NULL) {
register int *iptr = match_block.offset_vector + ocount;
register int *iend = iptr - resetcount / 2 + 1;
while (--iptr >= iend)
*iptr = -1;
}
/* Set up the first character to match, if available. The first_byte value is
never set for an anchored regular expression, but the anchoring may be forced
at run time, so we have to test for anchoring. The first char may be unset for
an unanchored pattern, of course. If there's no first char and the pattern was
studied, there may be a bitmap of possible first characters. */
if (!anchored) {
if ((re->options & PCRE_FIRSTSET) != 0) {
first_byte = re->first_byte & 255;
if ((first_byte_caseless =
((re->first_byte & REQ_CASELESS) != 0)) == TRUE)
first_byte = match_block.lcc[first_byte];
} else
if (!startline && study != NULL &&
(study->options & PCRE_STUDY_MAPPED) != 0)
start_bits = study->start_bits;
}
/* For anchored or unanchored matches, there may be a "last known required
character" set. */
if ((re->options & PCRE_REQCHSET) != 0) {
req_byte = re->req_byte & 255;
req_byte_caseless = (re->req_byte & REQ_CASELESS) != 0;
req_byte2 = (re->tables + fcc_offset)[req_byte]; /* case flipped */
}
/* Loop for handling unanchored repeated matching attempts; for anchored regexs
the loop runs just once. */
do {
register int *iptr = match_block.offset_vector;
register int *iend = iptr + resetcount;
/* Reset the maximum number of extractions we might see. */
while (iptr < iend)
*iptr++ = -1;
/* Advance to a unique first char if possible */
if (first_byte >= 0) {
if (first_byte_caseless)
while (start_match < end_subject &&
match_block.lcc[*start_match] != first_byte)
start_match++;
else
while (start_match < end_subject && *start_match != first_byte)
start_match++;
}
/* Or to just after \n for a multiline match if possible */
else if (startline) {
if (start_match > match_block.start_subject + start_offset) {
while (start_match < end_subject && start_match[-1] != NEWLINE)
start_match++;
}
}
/* Or to a non-unique first char after study */
else if (start_bits != NULL) {
while (start_match < end_subject) {
register int c = *start_match;
if ((start_bits[c / 8] & (1 << (c & 7))) == 0)
start_match++;
else
break;
}
}
/* If req_byte is set, we know that that character must appear in the subject
for the match to succeed. If the first character is set, req_byte must be
later in the subject; otherwise the test starts at the match point. This
optimization can save a huge amount of backtracking in patterns with nested
unlimited repeats that aren't going to match. Writing separate code for
cased/caseless versions makes it go faster, as does using an autoincrement
and backing off on a match.
HOWEVER: when the subject string is very, very long, searching to its end can
take a long time, and give bad performance on quite ordinary patterns. This
showed up when somebody was matching /^C/ on a 32-megabyte string... so we
don't do this when the string is sufficiently long. */
if (req_byte >= 0 && end_subject - start_match < REQ_BYTE_MAX) {
register const uschar *p = start_match + ((first_byte >= 0) ? 1 : 0);
/* We don't need to repeat the search if we haven't yet reached the
place we found it at last time. */
if (p > req_byte_ptr) {
if (req_byte_caseless) {
while (p < end_subject) {
register int pp = *p++;
if (pp == req_byte || pp == req_byte2) {
p--;
break;
}
}
} else {
while (p < end_subject) {
if (*p++ == req_byte) {
p--;
break;
}
}
}
/* If we can't find the required character, break the matching loop */
if (p >= end_subject)
break;
/* If we have found the required character, save the point where we
found it, so that we don't search again next time round the loop if
the start hasn't passed this character yet. */
req_byte_ptr = p;
}
}
/* When a match occurs, substrings will be set for all internal extractions;
we just need to set up the whole thing as substring 0 before returning. If
there were too many extractions, set the return code to zero. In the case
where we had to get some local store to hold offsets for backreferences, copy
those back references that we can. In this case there need not be overflow
if certain parts of the pattern were not used. */
match_block.start_match = start_match;
match_block.match_call_count = 0;
rc = match(start_match, match_block.start_code, 2, &match_block, ims, NULL,
match_isgroup);
if (rc == MATCH_NOMATCH) {
start_match++;
continue;
}
if (rc != MATCH_MATCH) {
DPRINTF((">>>> error: returning %d\n", rc));
return rc;
}
/* We have a match! Copy the offset information from temporary store if
necessary */
if (using_temporary_offsets) {
if (offsetcount >= 4) {
memcpy(offsets + 2, match_block.offset_vector + 2,
(offsetcount - 2) * sizeof(int));
DPRINTF(("Copied offsets from temporary memory\n"));
}
if (match_block.end_offset_top > offsetcount)
match_block.offset_overflow = TRUE;
DPRINTF(("Freeing temporary memory\n"));
free(match_block.offset_vector);
}
rc = match_block.offset_overflow ? 0 : match_block.end_offset_top / 2;
if (offsetcount < 2)
rc = 0;
else {
offsets[0] = start_match - match_block.start_subject;
offsets[1] = match_block.end_match_ptr - match_block.start_subject;
}
DPRINTF((">>>> returning %d\n", rc));
return rc;
}
/* This "while" is the end of the "do" above */
while (!anchored && start_match <= end_subject);
if (using_temporary_offsets) {
DPRINTF(("Freeing temporary memory\n"));
free(match_block.offset_vector);
}
DPRINTF((">>>> returning PCRE_ERROR_NOMATCH\n"));
return PCRE_ERROR_NOMATCH;
}
/* End of pcre.c */
|