File: archer_query_execute.c

package info (click to toggle)
bow 19991122-4
  • links: PTS
  • area: main
  • in suites: woody
  • size: 2,544 kB
  • ctags: 2,987
  • sloc: ansic: 38,660; lisp: 1,072; makefile: 594; perl: 492; yacc: 149; sh: 91
file content (587 lines) | stat: -rw-r--r-- 13,610 bytes parent folder | download
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
#include <bow/libbow.h>
#include <bow/archer.h>
#include <bow/archer_query.h>
#include <bow/archer_query_array.h>
#include <bow/archer_query_execute.h>
#include <bow/archer_query_index.h>
#include <bow/archer_query_table.h>

extern bow_sarray *archer_docs;
archer_query_info *archer_query_last_query = NULL;
int archer_query_doc_restriction = -1;

static int
list_length (archer_query_term * term)
{
  int ret;

  ret = 0;
  while (term && term->proximity)
    {
      ret++;
      term = term->proximity->term;
    }

  if (term)
    ret++;

  return ret;
}

/* moves the index file pointers from where they are to the next di that is
   >= target_di for term `term', and places the final di in result_di. if this
   is not possible (i.e. no more dis that meet that critereon exist) then the
   return value is 1; success returns 0. */
static int
scan_to_di (bow_index * index, archer_query_term * term, int target_di)
{
  int ret, pi;

  archer_query_index_current_di (index, term, &ret, &pi);
  while ((ret != -1) && 
	 ((archer_query_doc_restriction > -1 && 
	   ret != archer_query_doc_restriction) || 
	  (ret < target_di)))
    archer_query_index_next_di (index, term, &ret);

  return ret;
}

/* Insert into pi_array, maintaining sortedness */
static void
insert_pi_into_pi_array(int pi, bow_array *pi_array)
{
  void *ptr;
  int i, opi, len = pi_array->length;

  for (i = 0; i < len; ++i)
  {
    opi = *(int *)bow_array_entry_at_index(pi_array, i);
    if (pi == opi) 
      return;
    if (opi > pi)
      break;
  }

  /* i is now the appropriate index for pi */

  if (i == len)
    bow_array_append(pi_array, &pi);
  else
  {
    /* Make space for a new entry */
    opi = -1;
    bow_array_append(pi_array, &opi);

    /* Shift contents up one space */
    ptr = bow_array_entry_at_index(pi_array, i);
    memmove(ptr + sizeof(int), ptr, sizeof(int) * (len - i));

    /* Insert new pi */
    *((int *)ptr) = pi;
  }
}


static int
insert_new_pi_existing_wo(int pi, archer_query_word_occurence *wop, 
			  bow_array *wo_array)
{
  int i;
  archer_query_word_occurence *wop2 = NULL;

  for (i = 0; i < wo_array->length; ++i)
  {
    wop2 =(archer_query_word_occurence *)bow_array_entry_at_index(wo_array, i);
    if (wop->wi == wop2->wi)
    {
      if (wop->is_li == wop2->is_li)
	break;
      else
	return 0;
    }
    if (wop->wi < wop2->wi)
      return 0;
  }

  if (i == wo_array->length)
    return 0;

  insert_pi_into_pi_array(pi, wop2->pi);
  return 1;
}


static void
insert_new_wo(archer_query_word_occurence *wop, bow_array *wo_array)
{
  void *ptr;
  int i, len = wo_array->length;
  int wosz = sizeof(archer_query_word_occurence);
  archer_query_word_occurence *wop2;

  for (i = 0; i < len; ++i)
  {
    wop2 = (archer_query_word_occurence*)bow_array_entry_at_index(wo_array, i);
    if (wop2->wi > wop->wi || (wop2->wi == wop->wi && !wop2->is_li))
      break;
  }

  /* i is now the appropriate index for wop */

  if (i == len)
    bow_array_append(wo_array, wop);
  else
  {
    /* Make space for a new entry */
    bow_array_append(wo_array, wop);

    /* Shift contents up one space */
    ptr = bow_array_entry_at_index(wo_array, i);
    memmove(ptr + wosz, ptr, wosz * (len - i));

    /* Insert new pi */
    memcpy(ptr, wop, wosz);
  }
}


static void
single_term_result(archer_query_term *term, int di, int pi, bow_array **table)
{
  archer_query_word_occurence wo;

  if (!table[di])
    table[di] = bow_array_new(8, sizeof(archer_query_word_occurence), 
			      archer_query_array_free_wo);

  if (term->word)
  {
    wo.is_li = 0;
    wo.wi = bow_word2int_no_add (term->word);
  }
  else 
  {
    archer_label *lp = 
      bow_sarray_entry_at_keystr(archer_labels, term->labels->string);
    wo.is_li = 1;
    wo.wi = lp->li;
  }

  wo.weight = term->weight;
  wo.term = term;

  if (insert_new_pi_existing_wo(pi, &wo, table[di]))
    return;

  wo.pi = bow_array_new(32, sizeof(int), NULL);
  bow_array_append(wo.pi, &pi);

  insert_new_wo(&wo, table[di]);
}


/* recursively assembles the bow_array of results that satisfies all the
   proximity constraints of the linked list `term' passed. pi_arrays should
   contain the positions of the terms (e.g. pi_arrays[0] contains the
   positions of term->proximity->term, pi_arrays[0] contains those of
   term->proximity->term->proximity->term, etc.)

   possibly some kind of DP approach would be better here; this is going to
   make a lot of redundant calls i think */
static int
search_recursive (archer_query_term * term, int di, int pi,
		  bow_array ** pi_arrays, bow_array **table)
{
  int i, opi, prox, ret, good = 0;

  if (term == NULL)
    return 1;

  /* no recursion needed; just return a single-element bow_array for term */
  else if (term->proximity == NULL)
  {
    if (table)
      single_term_result(term, di, pi, table);
    return 1;
  }

  else
  {
    ret = 0;

    for (i = 0; i < pi_arrays[0]->length; i++)
    {
      opi = *((int *)bow_array_entry_at_index(pi_arrays[0], i));
      prox = term->proximity->proximity;

      switch (term->proximity->position)
      {
      case ARCHER_QUERY_PTERM_BEFORE :
	good = opi > pi && opi - pi <= prox;
	break;
      case ARCHER_QUERY_PTERM_AFTER :
	good = pi > opi && pi - opi <= prox;
	break;
      case ARCHER_QUERY_PTERM_WITHIN :
	good = ABS(pi - opi) <= prox;
	break;
      }

      if (good && 
	  search_recursive(term->proximity->term, di, opi, &pi_arrays[1], 
			   table))
      {
	if (table)
	{
	  ret = 1;
	  single_term_result(term, di, pi, table);
	}
	else
	  return 1;
      }
    }
  }

  return ret;
}


#define good_di(table, di, exclude) \
                  ((table) == NULL || \
		   ((table[di]) && !(exclude)) || \
		   (!(table[di]) && (exclude)))


/* next_term_di advances to next di in which all term components co-occur.
   Does not check that proximity constraints are satisfied. The third 
   argument, if provided, is a shortlist; exclude tells whether to use it
   as an exclusion or inclusion list */

static int 
next_term_di(int current_di, bow_index *index, bow_array **table, int exclude,
	     archer_query_term *term)
{
  archer_query_term *cterm = term;

  ++current_di;

  while (1)
    {
      while (cterm)
	{
	  int di = scan_to_di(index, cterm, current_di);

	  if (di == -1)
	    return -1;

	  if (di == current_di)
	    {
	      /* move on to the next term, recording this term's file pos */
	      cterm = cterm->proximity ? cterm->proximity->term : NULL;
	    }
	  else  /* Try the next di */
	    {
	      current_di = di == current_di ? di + 1 : di;
	      cterm = term;
	    }
	}

      /* 
	 This is a little inaccurate, since the terms may co-occur
	 without satisfying proximity constraints.  Done this way to
	 avoid the cost of verifying prox constraints unnecessarily.
      */
      term->head->df++;

      if (good_di(table, current_di, exclude))
	break;

      ++current_di;
      cterm = term;
    }

  return current_di;
}


static int
satisfies_proximity_constraints(int di, bow_index *index, archer_query_term *term)
{
  int i, j, pi, satisfies;
  int num_terms = list_length(term);
  bow_array *pi_arrays[num_terms];
  archer_query_term *cterm;

  for (cterm = term, i = 0; 
       cterm; 
       cterm = cterm->proximity ? cterm->proximity->term : NULL, ++i)
    pi_arrays[i] = archer_query_index_current_pis(index, cterm);
  
  satisfies = 0;
  for (j = 0; j < pi_arrays[0]->length; ++j)
  {
    pi = *((int *)bow_array_entry_at_index(pi_arrays[0], j));
    if (search_recursive(term, di, pi, &pi_arrays[1], NULL))
    {
      satisfies = 1;
      break;
    }
  }

  while (--i >= 0) bow_array_free(pi_arrays[i]); 

  return satisfies;
}


static int
add_if_satisfies_proximity_constraints(int di, bow_index *index, 
				       archer_query_term *term, 
				       bow_array **table)
{
  int i, j, pi, ret = 0;
  int num_terms = list_length(term);
  bow_array *pi_arrays[num_terms];
  archer_query_term *cterm;

  for (cterm = term, i = 0; 
       cterm; 
       cterm = cterm->proximity ? cterm->proximity->term : NULL, ++i)
    pi_arrays[i] = archer_query_index_current_pis(index, cterm);
  
  for (j = 0; j < pi_arrays[0]->length; ++j)
  {
    pi = *((int *)bow_array_entry_at_index(pi_arrays[0], j));
    if (search_recursive(term, di, pi, &pi_arrays[1], table))
      ret = 1;
  }

  while (--i >= 0) bow_array_free(pi_arrays[i]); 

  return ret;
}


static inline void
delete_intervening_entries(bow_array **table, int lastdi, int di)
{
  int i;
  for (i = lastdi + 1; i < di; ++i)
    if (table[i])
    {
      bow_array_free(table[i]);
      table[i] = NULL;
    }
}


static void
search_restrict(bow_index *index, bow_array **table, archer_query_term *term, 
		int exclude)
{
  int di, lastdi;
  int len = archer_docs->array->length;

  archer_query_index_reset (index);

  di = next_term_di(-1, index, (exclude ? table : NULL), 0, term);
  lastdi = -1;
  while (di != -1)
  {
    if (satisfies_proximity_constraints(di, index, term))
    {
      if (exclude)
      {
	assert(table[di]);
	bow_array_free(table[di]);
	table[di] = NULL;
      }
      else   /* Delete all intervening docs that didn't match */
      {
	delete_intervening_entries(table, lastdi, di);
	if (table[di])
	  add_if_satisfies_proximity_constraints(di, index, term, table);
	lastdi = di;
      }
    }

    di = next_term_di(di, index, (exclude ? table : NULL), 0, term);
  }

  delete_intervening_entries(table, lastdi, len);
}


static void
search (bow_index *index, bow_array **table, archer_query_term *term,
	bow_array **shortlist, int exclude)
{
  int len, di;

  len = archer_docs->array->length;
  archer_query_index_reset(index);
  di = next_term_di(-1, index, shortlist, exclude, term);
  while (di != -1)
  {
    add_if_satisfies_proximity_constraints(di, index, term, table);
    di = next_term_di(di, index, shortlist, exclude, term);
  }
}


/* 
   fill in the `score' elements of a bow_array of results.
   Score used: tfidf(w) = tf(w) * log(|D| / df(w))
*/
static void
calculate_tfidf (bow_index * index, bow_array * array)
{
  extern bow_sarray *archer_docs;
  int doccount = archer_docs->array->length;
  int i;

  for (i = 0; i < array->length; i++)
    {
      archer_query_result *current;
      int j;

      current = (archer_query_result *) bow_array_entry_at_index (array, i);
      current->score = 0.0;

      for (j = 0; j < current->wo->length; j++)
	{
	  archer_query_word_occurence *current_wo;

	  current_wo = (archer_query_word_occurence *)
	    bow_array_entry_at_index (current->wo, j);

	  if (current_wo->term->head->idf < 0.0)
	    current_wo->term->head->idf = 
	      log(((double)doccount) / ((double) current_wo->term->head->df));

	  current->score += 
	    current_wo->pi->length * 
	    current_wo->weight * 
	    current_wo->term->head->idf;
	}
    }
}


/*
  Set all member terms in a proximity list to point to the first
  (for DF calculation).
*/
static void
archer_query_thread(archer_query_term *term)
{
  archer_query_term *pterm;

  while (term)
    {
      for (pterm = term; 
	   pterm; 
	   pterm = pterm->proximity ? pterm->proximity->term : NULL)
	pterm->head = term;
      term = term->next;
    }
}


bow_array *
archer_query_execute (bow_index * index, archer_query_info * query)
{
  int exclude = 0;
  archer_query_term *term;
  bow_array **table = NULL, **shortlist = NULL, *ranking_results = NULL;

  if (query)
  {
    archer_query_doc_restriction = -1;
    archer_query_last_query = query;
  }
  else
    query = archer_query_last_query;

  archer_query_thread(query->inclusion);
  archer_query_thread(query->exclusion);
  archer_query_thread(query->ranking);

  for (term = query->inclusion; term; term = term->next)
  {
    if (table)
      /* Delete any items in table that _don't_ match term */
      search_restrict(index, table, term, 0); 
    else
    {
      /* Do an unconstrained search for matching docs */
      table = archer_query_table_new();
      search(index, table, term, NULL, 0);
    }
  }

  /* If table is non-NULL, we have a short list */

  if (query->exclusion)
  {
    if (query->inclusion)
    {
      /* We have a short list of documents; delete any with exclusion terms */
      for (term = query->exclusion; term; term = term->next)
	search_restrict(index, table, term, 1);
    }
    else
    {
      /* No short list: Create an exclusion table */
      exclude = 1;
      table = archer_query_table_new();
      for (term = query->exclusion; term; term = term->next)
	search(index, table, term, NULL, 0);
    }
  }

  /* If table is non-NULL, it is meant to restrict search in some way.
     If exclude = 1, it contains a list of docs to exclude */

  if (query->ranking)
  {
    shortlist = table;
    table = (exclude || !shortlist) ? 
	      archer_query_table_new() : 
	      archer_query_table_copy(shortlist);

    for (term = query->ranking; term; term = term->next)
      search(index, table, term, shortlist, exclude);

    if (shortlist)
      archer_query_table_free(shortlist);
  }

  /* This happens when only exclusion terms are given.  The right thing
     to do in such a case is just to return an empty list */
  else if (exclude)
  {
    /* archer_query_table_invert(table); */
    archer_query_table_empty(table);
  }

  ranking_results = archer_query_table_to_bow_array_with_freeing(table);
  calculate_tfidf (index, ranking_results);

  return ranking_results;
}



bow_array *
archer_query_repeat_for_document(bow_index *index, int di)
{
  if (!archer_query_last_query)
    return NULL;
  archer_query_doc_restriction = di;
  return archer_query_execute(index, NULL);
}