[3745] | 1 | /**************************************************************************
|
---|
| 2 | *
|
---|
| 3 | * invf_get.c -- Inverted file reading routines
|
---|
| 4 | * Copyright (C) 1994 Neil Sharman
|
---|
| 5 | *
|
---|
| 6 | * This program is free software; you can redistribute it and/or modify
|
---|
| 7 | * it under the terms of the GNU General Public License as published by
|
---|
| 8 | * the Free Software Foundation; either version 2 of the License, or
|
---|
| 9 | * (at your option) any later version.
|
---|
| 10 | *
|
---|
| 11 | * This program is distributed in the hope that it will be useful,
|
---|
| 12 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
| 13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
| 14 | * GNU General Public License for more details.
|
---|
| 15 | *
|
---|
| 16 | * You should have received a copy of the GNU General Public License
|
---|
| 17 | * along with this program; if not, write to the Free Software
|
---|
| 18 | * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
| 19 | *
|
---|
| 20 | * $Id: invf_get.c 3745 2003-02-20 21:20:24Z mdewsnip $
|
---|
| 21 | *
|
---|
| 22 | **************************************************************************/
|
---|
| 23 |
|
---|
| 24 | #include "sysfuncs.h"
|
---|
| 25 |
|
---|
| 26 | #include "memlib.h"
|
---|
| 27 | #include "filestats.h"
|
---|
| 28 | #include "timing.h"
|
---|
| 29 | #include "bitio_gen.h"
|
---|
| 30 | #include "bitio_m_mem.h"
|
---|
| 31 | #include "bitio_m.h"
|
---|
| 32 | #include "sptree.h"
|
---|
| 33 | #include "messages.h"
|
---|
| 34 | #include "netorder.h" /* [RPAP - Jan 97: Endian Ordering] */
|
---|
| 35 |
|
---|
| 36 | #include "mg.h"
|
---|
| 37 | #include "invf.h"
|
---|
| 38 | #include "text.h"
|
---|
| 39 | #include "lists.h"
|
---|
| 40 | #include "backend.h"
|
---|
| 41 | #include "invf_get.h"
|
---|
| 42 | #include "mg_errors.h"
|
---|
| 43 |
|
---|
| 44 | #define SKIPPING
|
---|
| 45 | #define SHORT_CUT
|
---|
| 46 |
|
---|
| 47 | struct pool
|
---|
| 48 | {
|
---|
| 49 | int length, pos;
|
---|
| 50 | struct pool *next;
|
---|
| 51 | Invf_Doc_Entry ide[1];
|
---|
| 52 | };
|
---|
| 53 |
|
---|
| 54 | static int skip_header = 0;
|
---|
| 55 |
|
---|
| 56 | static char skip_header_text0[] =
|
---|
| 57 | "#######################################################################\n"
|
---|
| 58 | "#\n"
|
---|
| 59 | "# 1\tWord number.\n"
|
---|
| 60 | "# 2\tNumber of time the word occurs in the collection.\n"
|
---|
| 61 | "# 3\tNumber of documents the word occurs in.\n"
|
---|
| 62 | "# 4\tThe word.\n"
|
---|
| 63 | "# 5\tNumber of docnum/word_count entries decoded.\n"
|
---|
| 64 | "# 6\tNumber of hits.\n"
|
---|
| 65 | "# 7\tNumber of entries in the splay tree or hash table.\n"
|
---|
| 66 | "# 8\tAmount added to the accumulators while while processing this word.\n"
|
---|
| 67 | "#\n"
|
---|
| 68 | "#######################################################################\n";
|
---|
| 69 |
|
---|
| 70 | static char skip_header_text1[] =
|
---|
| 71 | "#######################################################################\n"
|
---|
| 72 | "#\n"
|
---|
| 73 | "# 1\tWord number.\n"
|
---|
| 74 | "# 2\tNumber of time the word occurs in the collection.\n"
|
---|
| 75 | "# 3\tNumber of documents the word occurs in.\n"
|
---|
| 76 | "# 4\tThe word.\n"
|
---|
| 77 | "# 5\tNumber of skips taken in evaluating the inverted file entry.\n"
|
---|
| 78 | "# 6\tNumber of docnum/word_count entries decoded.\n"
|
---|
| 79 | "# 7\tNumber of hits.\n"
|
---|
| 80 | "# 8\tNumber of entries in the splay tree or hash table.\n"
|
---|
| 81 | "# 9\tAmount added to the accumulators while while processing this word.\n"
|
---|
| 82 | "#\n"
|
---|
| 83 | "#######################################################################\n";
|
---|
| 84 |
|
---|
| 85 |
|
---|
| 86 | static char skip_header_text2[] =
|
---|
| 87 | "#######################################################################\n"
|
---|
| 88 | "#\n"
|
---|
| 89 | "# 1\tWord number.\n"
|
---|
| 90 | "# 2\tNumber of time the word occurs in the collection.\n"
|
---|
| 91 | "# 3\tNumber of documents the word occurs in.\n"
|
---|
| 92 | "# 4\tThe word.\n"
|
---|
| 93 | "# 5\tSkip size.\n"
|
---|
| 94 | "# 6\tNumber of skips taken in evaluating the inverted file entry.\n"
|
---|
| 95 | "# 7\tNumber of docnum/word_count entries decoded.\n"
|
---|
| 96 | "# 8\tNumber of hits.\n"
|
---|
| 97 | "# 9\tNumber of entries in the splay tree or hash table.\n"
|
---|
| 98 | "# 10\tAmount added to the accumulators while while processing this word.\n"
|
---|
| 99 | "#\n"
|
---|
| 100 | "#######################################################################\n";
|
---|
| 101 |
|
---|
| 102 | static char *skip_header_text[] =
|
---|
| 103 | {skip_header_text0,
|
---|
| 104 | skip_header_text1,
|
---|
| 105 | skip_header_text2};
|
---|
| 106 |
|
---|
| 107 |
|
---|
| 108 |
|
---|
| 109 |
|
---|
| 110 |
|
---|
| 111 |
|
---|
| 112 |
|
---|
| 113 |
|
---|
| 114 | /**************************************************************************
|
---|
| 115 | *
|
---|
| 116 | * Code for the hash table.
|
---|
| 117 | *
|
---|
| 118 | */
|
---|
| 119 |
|
---|
| 120 | static Hash_Table *
|
---|
| 121 | HT_create (query_data * qd, unsigned max_entries,
|
---|
| 122 | unsigned max_size)
|
---|
| 123 | {
|
---|
| 124 | unsigned es, hts;
|
---|
| 125 | Hash_Table *HT;
|
---|
| 126 | if (max_entries == 0)
|
---|
| 127 | max_entries = 8192;
|
---|
| 128 | es = sizeof (Hash_Table) + sizeof (Invf_Doc_EntryH) * (max_entries - 1);
|
---|
| 129 | hts = sizeof (Invf_Doc_EntryH *) * max_size;
|
---|
| 130 |
|
---|
| 131 | HT = Xmalloc (es);
|
---|
| 132 | if (!HT)
|
---|
| 133 | return NULL;
|
---|
| 134 |
|
---|
| 135 | HT->HashTable = Xmalloc (hts);
|
---|
| 136 | if (!(HT->HashTable))
|
---|
| 137 | {
|
---|
| 138 | Xfree (HT);
|
---|
| 139 | return NULL;
|
---|
| 140 | }
|
---|
| 141 |
|
---|
| 142 | ChangeMemInUse (qd, es + hts);
|
---|
| 143 | bzero ((char *) (HT->HashTable), hts);
|
---|
| 144 | HT->max = max_entries;
|
---|
| 145 | HT->num = 0;
|
---|
| 146 | HT->Suplimentary_Entries = NULL;
|
---|
| 147 | HT->Suplimentary_Size = 0;
|
---|
| 148 | HT->Suplimentary_Num = 0;
|
---|
| 149 | HT->Head = NULL;
|
---|
| 150 | HT->size = max_size;
|
---|
| 151 | return HT;
|
---|
| 152 | }
|
---|
| 153 |
|
---|
| 154 |
|
---|
| 155 | static void
|
---|
| 156 | HT_Sup_create (query_data * qd, Hash_Table * HT, unsigned max_size)
|
---|
| 157 | {
|
---|
| 158 | int es = sizeof (Invf_Doc_EntryH) * max_size;
|
---|
| 159 | HT->Suplimentary_Entries = Xmalloc (es);
|
---|
| 160 | if (!HT->Suplimentary_Entries)
|
---|
| 161 | return;
|
---|
| 162 | HT->Suplimentary_Size = max_size;
|
---|
| 163 | HT->Suplimentary_Num = 0;
|
---|
| 164 | ChangeMemInUse (qd, es);
|
---|
| 165 | }
|
---|
| 166 |
|
---|
| 167 | static void
|
---|
| 168 | HT_Sup_Add (Hash_Table * HT, unsigned long DN, float Sum)
|
---|
| 169 | {
|
---|
| 170 | Invf_Doc_EntryH *IDEH;
|
---|
| 171 | IDEH = &HT->Suplimentary_Entries[HT->Suplimentary_Num++];
|
---|
| 172 | IDEH->next = NULL;
|
---|
| 173 | IDEH->IDE.DocNum = DN;
|
---|
| 174 | IDEH->IDE.Sum = Sum;
|
---|
| 175 | }
|
---|
| 176 |
|
---|
| 177 | Invf_Doc_Entry *
|
---|
| 178 | HT_find (Hash_Table * HT, unsigned long DN)
|
---|
| 179 | {
|
---|
| 180 | Invf_Doc_EntryH *IDEH;
|
---|
| 181 | if (!HT->HashTable)
|
---|
| 182 | FatalError (1, "Accessing the hash_table through HT_find after HT_sort");
|
---|
| 183 | IDEH = HT->HashTable[DN % HT->size];
|
---|
| 184 | while (IDEH)
|
---|
| 185 | {
|
---|
| 186 | if (IDEH->IDE.DocNum == DN)
|
---|
| 187 | return &(IDEH->IDE);
|
---|
| 188 | IDEH = IDEH->next;
|
---|
| 189 | }
|
---|
| 190 | return NULL;
|
---|
| 191 | }
|
---|
| 192 |
|
---|
| 193 |
|
---|
| 194 | static int
|
---|
| 195 | HT_insert (Hash_Table * HT,
|
---|
| 196 | unsigned long DN, float Sum)
|
---|
| 197 | {
|
---|
| 198 | Invf_Doc_EntryH **IDEH;
|
---|
| 199 | if (HT->num == HT->max)
|
---|
| 200 | return 0;
|
---|
| 201 | IDEH = &(HT->HashTable[DN % HT->size]);
|
---|
| 202 | while (*IDEH)
|
---|
| 203 | IDEH = &((*IDEH)->next);
|
---|
| 204 | *IDEH = &(HT->IDEH[HT->num++]);
|
---|
| 205 | (*IDEH)->next = NULL;
|
---|
| 206 | (*IDEH)->IDE.DocNum = DN;
|
---|
| 207 | (*IDEH)->IDE.Sum = Sum;
|
---|
| 208 | return 1;
|
---|
| 209 | }
|
---|
| 210 |
|
---|
| 211 |
|
---|
| 212 | void
|
---|
| 213 | HT_free (query_data * qd, Hash_Table * HT)
|
---|
| 214 | {
|
---|
| 215 | unsigned es;
|
---|
| 216 | es = sizeof (Hash_Table) + sizeof (Invf_Doc_EntryH) * HT->max +
|
---|
| 217 | sizeof (Invf_Doc_EntryH *) * HT->size;
|
---|
| 218 | if (HT->HashTable)
|
---|
| 219 | Xfree (HT->HashTable);
|
---|
| 220 | if (HT->Suplimentary_Entries)
|
---|
| 221 | {
|
---|
| 222 | Xfree (HT->Suplimentary_Entries);
|
---|
| 223 | es += sizeof (Invf_Doc_EntryH) * HT->Suplimentary_Size;
|
---|
| 224 | }
|
---|
| 225 | Xfree (HT);
|
---|
| 226 | ChangeMemInUse (qd, -es);
|
---|
| 227 | }
|
---|
| 228 |
|
---|
| 229 |
|
---|
| 230 | static int
|
---|
| 231 | IDEH_comp (const void *a, const void *b)
|
---|
| 232 | {
|
---|
| 233 | const Invf_Doc_EntryH *A = a;
|
---|
| 234 | const Invf_Doc_EntryH *B = b;
|
---|
| 235 | return A->IDE.DocNum - B->IDE.DocNum;
|
---|
| 236 | }
|
---|
| 237 |
|
---|
| 238 | static void
|
---|
| 239 | HT_sort (query_data * qd, Hash_Table * HT)
|
---|
| 240 | {
|
---|
| 241 | int i;
|
---|
| 242 | Invf_Doc_EntryH *l, *r, **b;
|
---|
| 243 | unsigned es = sizeof (Invf_Doc_EntryH *) * HT->size;
|
---|
| 244 | if (HT->HashTable)
|
---|
| 245 | Xfree (HT->HashTable);
|
---|
| 246 | HT->size = 0;
|
---|
| 247 | HT->HashTable = NULL;
|
---|
| 248 | ChangeMemInUse (qd, -es);
|
---|
| 249 |
|
---|
| 250 | qsort (HT->IDEH, HT->num, sizeof (Invf_Doc_EntryH), IDEH_comp);
|
---|
| 251 |
|
---|
| 252 | for (i = 0; i < HT->num - 1; i++)
|
---|
| 253 | HT->IDEH[i].next = &HT->IDEH[i + 1];
|
---|
| 254 | HT->IDEH[i].next = NULL;
|
---|
| 255 | l = HT->IDEH;
|
---|
| 256 |
|
---|
| 257 | if (HT->Suplimentary_Entries)
|
---|
| 258 | {
|
---|
| 259 | for (i = 0; i < HT->Suplimentary_Num - 1; i++)
|
---|
| 260 | HT->Suplimentary_Entries[i].next = &HT->Suplimentary_Entries[i + 1];
|
---|
| 261 | HT->Suplimentary_Entries[i].next = NULL;
|
---|
| 262 | }
|
---|
| 263 | r = HT->Suplimentary_Entries;
|
---|
| 264 |
|
---|
| 265 | b = &HT->Head;
|
---|
| 266 |
|
---|
| 267 | while (l || r)
|
---|
| 268 | {
|
---|
| 269 | if (r && (!l || (l->IDE.DocNum > r->IDE.DocNum)))
|
---|
| 270 | {
|
---|
| 271 | *b = r;
|
---|
| 272 | b = &r->next;
|
---|
| 273 | r = *b;
|
---|
| 274 | }
|
---|
| 275 | else
|
---|
| 276 | {
|
---|
| 277 | *b = l;
|
---|
| 278 | b = &l->next;
|
---|
| 279 | l = *b;
|
---|
| 280 | }
|
---|
| 281 | }
|
---|
| 282 | *b = NULL;
|
---|
| 283 | }
|
---|
| 284 |
|
---|
| 285 |
|
---|
| 286 |
|
---|
| 287 |
|
---|
| 288 |
|
---|
| 289 |
|
---|
| 290 | /**************************************************************************
|
---|
| 291 | *
|
---|
| 292 | * Code for the hash table.
|
---|
| 293 | *
|
---|
| 294 | */
|
---|
| 295 |
|
---|
| 296 | static List_Table *
|
---|
| 297 | LT_create (query_data * qd, unsigned max_entries)
|
---|
| 298 | {
|
---|
| 299 | unsigned es;
|
---|
| 300 | List_Table *LT;
|
---|
| 301 | if (max_entries == 0)
|
---|
| 302 | max_entries = 8192;
|
---|
| 303 | es = sizeof (List_Table) + sizeof (Invf_Doc_Entry) * (max_entries - 1);
|
---|
| 304 |
|
---|
| 305 | LT = Xmalloc (es);
|
---|
| 306 | if (!LT)
|
---|
| 307 | return NULL;
|
---|
| 308 |
|
---|
| 309 | ChangeMemInUse (qd, es);
|
---|
| 310 | LT->max = max_entries;
|
---|
| 311 | LT->num = 0;
|
---|
| 312 | return LT;
|
---|
| 313 | }
|
---|
| 314 |
|
---|
| 315 |
|
---|
| 316 |
|
---|
| 317 | Invf_Doc_Entry *
|
---|
| 318 | LT_find (List_Table * LT, unsigned long DN)
|
---|
| 319 | {
|
---|
| 320 | int l = 0;
|
---|
| 321 | int r = LT->num - 1;
|
---|
| 322 | int m, c;
|
---|
| 323 | while (l <= r)
|
---|
| 324 | {
|
---|
| 325 | m = (l + r) / 2;
|
---|
| 326 | c = DN - LT->IDE[m].DocNum;
|
---|
| 327 | if (c < 0)
|
---|
| 328 | r = m - 1;
|
---|
| 329 | else if (c > 0)
|
---|
| 330 | l = m + 1;
|
---|
| 331 | else
|
---|
| 332 | return <->IDE[m];
|
---|
| 333 | }
|
---|
| 334 | return NULL;
|
---|
| 335 | }
|
---|
| 336 |
|
---|
| 337 | static List_Table *
|
---|
| 338 | LT_add (query_data * qd, List_Table * LT,
|
---|
| 339 | unsigned long DN, float Sum)
|
---|
| 340 | {
|
---|
| 341 | Invf_Doc_Entry *ide;
|
---|
| 342 |
|
---|
| 343 | /* increase the size of the list if there is not enough */
|
---|
| 344 | /* room for this entry */
|
---|
| 345 | if (LT->num == LT->max)
|
---|
| 346 | {
|
---|
| 347 | int old, new, max_entries;
|
---|
| 348 | old = sizeof (List_Table) + sizeof (Invf_Doc_Entry) * LT->max;
|
---|
| 349 | max_entries = LT->max + (LT->max >> 1);
|
---|
| 350 | new = sizeof (List_Table) + sizeof (Invf_Doc_Entry) * max_entries;
|
---|
| 351 |
|
---|
| 352 | LT = Xrealloc (LT, new);
|
---|
| 353 | if (!LT)
|
---|
| 354 | return NULL;
|
---|
| 355 |
|
---|
| 356 | ChangeMemInUse (qd, new - old);
|
---|
| 357 | LT->max = max_entries;
|
---|
| 358 | }
|
---|
| 359 |
|
---|
| 360 | /* add the entry to the end of the list */
|
---|
| 361 | ide = <->IDE[LT->num];
|
---|
| 362 | ide->DocNum = DN;
|
---|
| 363 | ide->Sum = Sum;
|
---|
| 364 | LT->num++;
|
---|
| 365 |
|
---|
| 366 | return LT;
|
---|
| 367 | }
|
---|
| 368 |
|
---|
| 369 |
|
---|
| 370 | void
|
---|
| 371 | LT_free (query_data * qd, List_Table * LT)
|
---|
| 372 | {
|
---|
| 373 | unsigned es;
|
---|
| 374 | es = sizeof (List_Table) + sizeof (Invf_Doc_Entry) * LT->max;
|
---|
| 375 | Xfree (LT);
|
---|
| 376 | ChangeMemInUse (qd, -es);
|
---|
| 377 | }
|
---|
| 378 |
|
---|
| 379 |
|
---|
| 380 | static int
|
---|
| 381 | IDE_comp (const void *a, const void *b)
|
---|
| 382 | {
|
---|
| 383 | const Invf_Doc_Entry *A = a;
|
---|
| 384 | const Invf_Doc_Entry *B = b;
|
---|
| 385 | return A->DocNum - B->DocNum;
|
---|
| 386 | }
|
---|
| 387 |
|
---|
| 388 | static void
|
---|
| 389 | LT_sort (List_Table * LT)
|
---|
| 390 | {
|
---|
| 391 | qsort (LT->IDE, LT->num, sizeof (Invf_Doc_Entry), IDE_comp);
|
---|
| 392 | }
|
---|
| 393 |
|
---|
| 394 |
|
---|
| 395 | /* [RJM 07/97: Ranked Required Terms] */
|
---|
| 396 | /* accumulators with a sum less than or equal to 0.0 are to be deleted */
|
---|
| 397 | static void
|
---|
| 398 | LT_pack (List_Table * LT)
|
---|
| 399 | {
|
---|
| 400 | int s, d = 0;
|
---|
| 401 | for (s = 0; s < LT->num; s++) {
|
---|
| 402 | if (LT->IDE[s].Sum <= 0.0) {
|
---|
| 403 | /* ignore this accumulator */
|
---|
| 404 |
|
---|
| 405 | } else if (d > 0 && LT->IDE[s].DocNum == LT->IDE[d-1].DocNum) {
|
---|
| 406 | LT->IDE[d-1].Sum += LT->IDE[s].Sum;
|
---|
| 407 |
|
---|
| 408 | } else {
|
---|
| 409 | if (d != s) LT->IDE[d] = LT->IDE[s];
|
---|
| 410 | d++;
|
---|
| 411 | }
|
---|
| 412 | }
|
---|
| 413 |
|
---|
| 414 | if (LT->num) LT->num = d;
|
---|
| 415 | }
|
---|
| 416 |
|
---|
| 417 |
|
---|
| 418 | /**************************************************************************/
|
---|
| 419 |
|
---|
| 420 |
|
---|
| 421 |
|
---|
| 422 | invf_data *
|
---|
| 423 | InitInvfFile (File * InvfFile, stemmed_dict * sd)
|
---|
| 424 | {
|
---|
| 425 | invf_data *id;
|
---|
| 426 | if (!(id = Xmalloc (sizeof (invf_data))))
|
---|
| 427 | {
|
---|
| 428 | mg_errno = MG_NOMEM;
|
---|
| 429 | return NULL;
|
---|
| 430 | }
|
---|
| 431 |
|
---|
| 432 | id->InvfFile = InvfFile;
|
---|
| 433 |
|
---|
| 434 | Fread (&id->ifh, sizeof (id->ifh), 1, InvfFile);
|
---|
| 435 | /* [RPAP - Jan 97: Endian Ordering] */
|
---|
| 436 | {
|
---|
| 437 | int i;
|
---|
| 438 |
|
---|
| 439 | NTOHUL(id->ifh.no_of_words);
|
---|
| 440 | NTOHUL(id->ifh.no_of_ptrs);
|
---|
| 441 | NTOHUL(id->ifh.skip_mode);
|
---|
| 442 | for (i = 0; i < 16; i++)
|
---|
| 443 | NTOHUL(id->ifh.params[i]);
|
---|
| 444 | NTOHUL(id->ifh.InvfLevel);
|
---|
| 445 | }
|
---|
| 446 |
|
---|
| 447 | id->N = sd->sdh.num_of_docs;
|
---|
| 448 | id->Nstatic = sd->sdh.static_num_of_docs;
|
---|
| 449 |
|
---|
| 450 | return (id);
|
---|
| 451 | }
|
---|
| 452 |
|
---|
| 453 |
|
---|
| 454 | void
|
---|
| 455 | FreeInvfData (invf_data * id)
|
---|
| 456 | {
|
---|
| 457 | Xfree (id);
|
---|
| 458 | }
|
---|
| 459 |
|
---|
| 460 | /****************************************************************************/
|
---|
| 461 |
|
---|
| 462 |
|
---|
| 463 | static void
|
---|
| 464 | CalcBlks (query_data * qd, WordEntry * we, int *blk,
|
---|
| 465 | int *dn_blk, int *len_blk, int *k, int *p)
|
---|
| 466 | {
|
---|
| 467 | *p = we->doc_count;
|
---|
| 468 |
|
---|
| 469 | *blk = BIO_Bblock_Init (qd->id->Nstatic, *p);
|
---|
| 470 | switch (qd->id->ifh.skip_mode)
|
---|
| 471 | {
|
---|
| 472 | case 1:
|
---|
| 473 | {
|
---|
| 474 | unsigned long len;
|
---|
| 475 | if (*p <= qd->id->ifh.params[0])
|
---|
| 476 | *k = 0;
|
---|
| 477 | else
|
---|
| 478 | {
|
---|
| 479 | *k = qd->id->ifh.params[0];
|
---|
| 480 | len = BIO_Bblock_Bound (qd->id->N, *p);
|
---|
| 481 | if (qd->id->ifh.InvfLevel >= 2)
|
---|
| 482 | len += we->count;
|
---|
| 483 | *dn_blk = BIO_Bblock_Init (qd->sd->sdh.num_of_docs, (*p + *k - 1) / *k);
|
---|
| 484 | *len_blk = BIO_Bblock_Init (len, (*p + *k - 1) / *k);
|
---|
| 485 | }
|
---|
| 486 | break;
|
---|
| 487 | }
|
---|
| 488 | case 2:
|
---|
| 489 | {
|
---|
| 490 | unsigned long len;
|
---|
| 491 | if (*p <= qd->id->ifh.params[1])
|
---|
| 492 | *k = 0;
|
---|
| 493 | else
|
---|
| 494 | {
|
---|
| 495 | *k = (int) (2 * sqrt (((double) (*p)) / qd->id->ifh.params[0]));
|
---|
| 496 | if (*k <= qd->id->ifh.params[1])
|
---|
| 497 | *k = qd->id->ifh.params[1];
|
---|
| 498 | len = BIO_Bblock_Bound (qd->id->N, *p);
|
---|
| 499 | if (qd->id->ifh.InvfLevel >= 2)
|
---|
| 500 | len += we->count;
|
---|
| 501 | *dn_blk = BIO_Bblock_Init (qd->sd->sdh.num_of_docs,
|
---|
| 502 | (*p + *k - 1) / *k);
|
---|
| 503 | *len_blk = BIO_Bblock_Init (len, (*p + *k - 1) / *k);
|
---|
| 504 | }
|
---|
| 505 | break;
|
---|
| 506 | }
|
---|
| 507 | default:
|
---|
| 508 | *k = 0;
|
---|
| 509 | }
|
---|
| 510 | }
|
---|
| 511 |
|
---|
| 512 |
|
---|
| 513 | /* =========================================================================
|
---|
| 514 | * Function: GetDocsOp
|
---|
| 515 | * Description:
|
---|
| 516 | * if op == op_term then
|
---|
| 517 | * look up term in invf and decode all its entries/documents
|
---|
| 518 | * if op == op_and_terms then
|
---|
| 519 | * go thru L, testing if it exists in invf using skips
|
---|
| 520 | * if it does exist then overwrite the entry nearer the start of the list
|
---|
| 521 | * if op == op_and_or_terms then
|
---|
| 522 | * go thru L, testing if it exists in invf using skips
|
---|
| 523 | * BUT only mark ones which are found - keep list intact
|
---|
| 524 | * if op == op_diff_terms then >>>> NOT tested <<<<
|
---|
| 525 | * Input:
|
---|
| 526 | * Output:
|
---|
| 527 | * ========================================================================= */
|
---|
| 528 |
|
---|
| 529 | DocList *
|
---|
| 530 | GetDocsOp (query_data * qd, WordEntry * we, Op_types op, DocList * L)
|
---|
| 531 | {
|
---|
| 532 | u_char *buffer;
|
---|
| 533 | DocEntry *src = NULL, *dest = NULL;
|
---|
| 534 | int i; /* index counter into inverted list */
|
---|
| 535 | unsigned long pos;
|
---|
| 536 | int next_start, next_doc_num, next_i;
|
---|
| 537 | int block_counter; /* within counter in block */
|
---|
| 538 | int dn_blk = 0, len_blk = 0;
|
---|
| 539 | int skip_size, have_skipping;
|
---|
| 540 | int blk; /* bblock control parameter */
|
---|
| 541 | int ft; /* The number of documents the word occurs in */
|
---|
| 542 | unsigned long CurrDoc = 0; /* NOTE: Document numbers start at 1 */
|
---|
| 543 | float Weight;
|
---|
| 544 |
|
---|
| 545 | /* ... AND empty */
|
---|
| 546 | if (op == op_and_terms && (!L || !L->num))
|
---|
| 547 | {
|
---|
| 548 | if (L)
|
---|
| 549 | Xfree (L);
|
---|
| 550 | return MakeDocList (0);
|
---|
| 551 | }
|
---|
| 552 |
|
---|
| 553 | if (op == op_term)
|
---|
| 554 | L = MakeDocList (we->doc_count);
|
---|
| 555 |
|
---|
| 556 | CalcBlks (qd, we, &blk, &dn_blk, &len_blk, &skip_size, &ft);
|
---|
| 557 | have_skipping = (skip_size != 0);
|
---|
| 558 | if (!(buffer = Xmalloc (we->invf_len)))
|
---|
| 559 | return MakeDocList (0);
|
---|
| 560 |
|
---|
| 561 | ChangeMemInUse (qd, we->invf_len);
|
---|
| 562 |
|
---|
| 563 | qd->num_of_terms++;
|
---|
| 564 |
|
---|
| 565 | /* --- Read an inverted file entry for the given word --- */
|
---|
| 566 | Fseek (qd->File_invf, we->invf_ptr, 0);
|
---|
| 567 | Fread (buffer, sizeof (char), we->invf_len, qd->File_invf);
|
---|
| 568 |
|
---|
| 569 | /* Read from the buffer */
|
---|
| 570 | DECODE_START (buffer, we->invf_len);
|
---|
| 571 |
|
---|
| 572 | Weight = qd->sd->sdh.num_of_words / (100 * log2 ((float) we->count + 1));
|
---|
| 573 |
|
---|
| 574 | if (op != op_term)
|
---|
| 575 | dest = src = L->DE;
|
---|
| 576 | i = pos = 0;
|
---|
| 577 | next_i = next_start = next_doc_num = 0;
|
---|
| 578 | block_counter = skip_size - 1;
|
---|
| 579 |
|
---|
| 580 | /* Go thru the elements of an inverted list for a term */
|
---|
| 581 | /* May not go sequentially if skipping is done */
|
---|
| 582 | while (i < ft && ((op == op_term) || src - L->DE < L->num))
|
---|
| 583 | {
|
---|
| 584 | unsigned long diff, Temp;
|
---|
| 585 |
|
---|
| 586 | if (have_skipping)
|
---|
| 587 | {
|
---|
| 588 |
|
---|
| 589 | /* --- Skip to next block --- */
|
---|
| 590 | if (op != op_term && i + skip_size < ft &&
|
---|
| 591 | src->DocNum > next_doc_num && next_doc_num >= 0)
|
---|
| 592 | {
|
---|
| 593 | ShortCut:
|
---|
| 594 | i = next_i;
|
---|
| 595 | DECODE_SEEK (next_start);
|
---|
| 596 | pos = next_start;
|
---|
| 597 | CurrDoc = next_doc_num;
|
---|
| 598 | qd->hops_taken++;
|
---|
| 599 | block_counter = skip_size - 1;
|
---|
| 600 | }
|
---|
| 601 |
|
---|
| 602 | /* --- Decode the skip block header --- */
|
---|
| 603 | if (block_counter == skip_size - 1)
|
---|
| 604 | {
|
---|
| 605 | if (i + skip_size < ft)
|
---|
| 606 | {
|
---|
| 607 | /* --- Decode the skip block header --- */
|
---|
| 608 | int doc_diff, pos_diff;
|
---|
| 609 | BBLOCK_DECODE_L (doc_diff, dn_blk, pos);
|
---|
| 610 | next_doc_num += doc_diff;
|
---|
| 611 | next_i += skip_size;
|
---|
| 612 | BBLOCK_DECODE_L (pos_diff, len_blk, pos);
|
---|
| 613 | next_start = pos + pos_diff;
|
---|
| 614 | }
|
---|
| 615 | else
|
---|
| 616 | next_doc_num = -1;
|
---|
| 617 | }
|
---|
| 618 |
|
---|
| 619 | /* --- Do a skip if c-set doc is higher than block header doc --- */
|
---|
| 620 | if (op != op_term && i + skip_size < ft &&
|
---|
| 621 | src->DocNum > next_doc_num && next_doc_num >= 0)
|
---|
| 622 | goto ShortCut;
|
---|
| 623 |
|
---|
| 624 |
|
---|
| 625 | } /*if skipping */
|
---|
| 626 |
|
---|
| 627 | /* --- increment CurrDoc --- */
|
---|
| 628 | if (have_skipping && block_counter == 0 && i != ft - 1)
|
---|
| 629 | CurrDoc = next_doc_num;
|
---|
| 630 | else
|
---|
| 631 | {
|
---|
| 632 | BBLOCK_DECODE_L (diff, blk, pos);
|
---|
| 633 | CurrDoc += diff;
|
---|
| 634 | }
|
---|
| 635 |
|
---|
| 636 | /* --- If have f_dt's then eat them up --- */
|
---|
| 637 | if (qd->id->ifh.InvfLevel >= 2)
|
---|
| 638 | GAMMA_DECODE_L (Temp, pos); /* discard word count */
|
---|
| 639 |
|
---|
| 640 | /* --- Increment counters --- */
|
---|
| 641 | qd->num_of_ptrs++;
|
---|
| 642 | i++;
|
---|
| 643 | block_counter--;
|
---|
| 644 | if (block_counter < 0) /* ??? Why would this happen ? */
|
---|
| 645 | block_counter = skip_size - 1;
|
---|
| 646 |
|
---|
| 647 | switch (op)
|
---|
| 648 | {
|
---|
| 649 | case op_term:
|
---|
| 650 | /* --- Add the doc from invf onto the c-set list --- */
|
---|
| 651 | L->DE[L->num].DocNum = CurrDoc;
|
---|
| 652 | L->DE[L->num].Weight = Weight;
|
---|
| 653 | L->DE[L->num].SeekPos = 0;
|
---|
| 654 | L->DE[L->num].Len = 0;
|
---|
| 655 | L->DE[L->num].CompTextBuffer = NULL;
|
---|
| 656 | L->DE[L->num].Next = NULL;
|
---|
| 657 | L->DE[L->num].or_included = 0;
|
---|
| 658 | L->num++;
|
---|
| 659 | break;
|
---|
| 660 |
|
---|
| 661 | case op_and_terms:
|
---|
| 662 | case op_and_or_terms:
|
---|
| 663 | /* --- see if have a match and save it --- */
|
---|
| 664 |
|
---|
| 665 | while (src < (L->DE + L->num) && CurrDoc > src->DocNum)
|
---|
| 666 | {
|
---|
| 667 | src++;
|
---|
| 668 | }
|
---|
| 669 | /* Unless at end of list, assert(CurrDoc <= src->DocNum) */
|
---|
| 670 |
|
---|
| 671 | /* --- Accept the doc as a candidate --- */
|
---|
| 672 | if (src < (L->DE + L->num) && CurrDoc == src->DocNum)
|
---|
| 673 | {
|
---|
| 674 | if (op == op_and_terms)
|
---|
| 675 | {
|
---|
| 676 | *dest++ = *src;
|
---|
| 677 | }
|
---|
| 678 | else
|
---|
| 679 | {
|
---|
| 680 | src->or_included = 1;
|
---|
| 681 | }
|
---|
| 682 | src++; /* Move onto next one */
|
---|
| 683 | }
|
---|
| 684 |
|
---|
| 685 | /* If the next one to try has already been tried then move on again */
|
---|
| 686 | if (op == op_and_or_terms)
|
---|
| 687 | {
|
---|
| 688 | while (src < (L->DE + L->num) && src->or_included)
|
---|
| 689 | src++;
|
---|
| 690 | }
|
---|
| 691 |
|
---|
| 692 | break;
|
---|
| 693 |
|
---|
| 694 | case op_diff_terms:
|
---|
| 695 | while (src < (L->DE + L->num) && src->DocNum < CurrDoc)
|
---|
| 696 | *dest++ = *src++;
|
---|
| 697 | if (src < (L->DE + L->num) && src->DocNum == CurrDoc)
|
---|
| 698 | src++;
|
---|
| 699 | break;
|
---|
| 700 |
|
---|
| 701 | default:
|
---|
| 702 | break;
|
---|
| 703 | } /*switch */
|
---|
| 704 |
|
---|
| 705 | } /*while traversing invf */
|
---|
| 706 |
|
---|
| 707 | if (op == op_and_terms || op == op_diff_terms)
|
---|
| 708 | {
|
---|
| 709 | if (op == op_diff_terms)
|
---|
| 710 | while (src - L->DE < L->num)
|
---|
| 711 | *dest++ = *src++;
|
---|
| 712 |
|
---|
| 713 | /* update size of c-set */
|
---|
| 714 | L->num = dest - L->DE;
|
---|
| 715 | }
|
---|
| 716 | DECODE_DONE
|
---|
| 717 |
|
---|
| 718 | Xfree (buffer);
|
---|
| 719 | ChangeMemInUse (qd, -we->invf_len);
|
---|
| 720 |
|
---|
| 721 | return (L);
|
---|
| 722 | }
|
---|
| 723 |
|
---|
| 724 |
|
---|
| 725 |
|
---|
| 726 |
|
---|
| 727 |
|
---|
| 728 |
|
---|
| 729 |
|
---|
| 730 | float *
|
---|
| 731 | CosineDecode (query_data * qd, TermList * Terms, RankedQueryInfo * rqi)
|
---|
| 732 | {
|
---|
| 733 | float *AccumulatedWeights;
|
---|
| 734 | int j, num_docs, num_terms;
|
---|
| 735 | int seen_must_match = 0; /* [RJM 07/97: Ranked Required Terms] says whether to 'and' queries or not */
|
---|
| 736 |
|
---|
| 737 | num_docs = qd->sd->sdh.num_of_docs;
|
---|
| 738 | /* Create the array for the documents */
|
---|
| 739 | if (!(AccumulatedWeights = Xmalloc (num_docs * sizeof (float))))
|
---|
| 740 | return (NULL);
|
---|
| 741 |
|
---|
| 742 | bzero ((char *) AccumulatedWeights, sizeof (float) * num_docs);
|
---|
| 743 | ChangeMemInUse (qd, sizeof (float) * num_docs);
|
---|
| 744 |
|
---|
| 745 | if (rqi->MaxTerms == -1)
|
---|
| 746 | num_terms = Terms->num;
|
---|
| 747 | else
|
---|
| 748 | num_terms = rqi->MaxTerms < Terms->num ? rqi->MaxTerms : Terms->num;
|
---|
| 749 |
|
---|
| 750 |
|
---|
| 751 | /* For each word in the list . . . */
|
---|
| 752 | for (j = 0; j < num_terms; j++)
|
---|
| 753 | {
|
---|
| 754 | u_char *buffer;
|
---|
| 755 | int i, kd;
|
---|
| 756 | unsigned long next_mjr_dn = 0;
|
---|
| 757 | unsigned long CurrDocNum = 0; /* NOTE: Document numbers start at 1 */
|
---|
| 758 | unsigned long LastDocNum = 0; /* [RJM 07/97: Ranked Required Terms] */
|
---|
| 759 | unsigned long TempDocNum = 0; /* [RJM 07/97: Ranked Required Terms] */
|
---|
| 760 | double Wqt, WordLog;
|
---|
| 761 | int dn_blk = 0, len_blk = 0, k;
|
---|
| 762 | int blk; /* bblock control parameter */
|
---|
| 763 | int p; /* The number of documents the word occurs in */
|
---|
| 764 | int require_match = Terms->TE[j].require_match; /* [RJM 07/97: Ranked Required Terms] */
|
---|
| 765 | WordEntry *we = &Terms->TE[j].WE;
|
---|
| 766 | CalcBlks (qd, we, &blk, &dn_blk, &len_blk, &k, &p);
|
---|
| 767 | if (!(buffer = Xmalloc (we->invf_len)))
|
---|
| 768 | break;
|
---|
| 769 |
|
---|
| 770 | ChangeMemInUse (qd, we->invf_len);
|
---|
| 771 |
|
---|
| 772 | Fseek (qd->File_invf, we->invf_ptr, 0);
|
---|
| 773 | Fread (buffer, sizeof (char), we->invf_len, qd->File_invf);
|
---|
| 774 |
|
---|
| 775 | /* Read from the buffer */
|
---|
| 776 | DECODE_START (buffer, we->invf_len);
|
---|
| 777 |
|
---|
| 778 | WordLog = log ((double) num_docs / (double) we->max_doc_count); /* [RPAP - Jan 97: Stem Index Change] */
|
---|
| 779 |
|
---|
| 780 | Wqt = (rqi->QueryFreqs ? Terms->TE[j].Count : 1) * WordLog;
|
---|
| 781 |
|
---|
| 782 | qd->num_of_terms++;
|
---|
| 783 |
|
---|
| 784 | kd = 0;
|
---|
| 785 | for (i = 0; i < p; i++, kd++)
|
---|
| 786 | {
|
---|
| 787 | int Count; /* The number of times the word occurs in a document */
|
---|
| 788 | double Wdt;
|
---|
| 789 | unsigned long diff;
|
---|
| 790 | if (kd == k)
|
---|
| 791 | {
|
---|
| 792 | kd = 0;
|
---|
| 793 | }
|
---|
| 794 | if (k && kd == 0)
|
---|
| 795 | if (i + k < p)
|
---|
| 796 | {
|
---|
| 797 | int temp;
|
---|
| 798 | BBLOCK_DECODE (next_mjr_dn, dn_blk);
|
---|
| 799 | next_mjr_dn += CurrDocNum;
|
---|
| 800 | BBLOCK_DECODE (temp, len_blk);
|
---|
| 801 | }
|
---|
| 802 | if (k && kd == k - 1 && i != p - 1)
|
---|
| 803 | CurrDocNum = next_mjr_dn;
|
---|
| 804 | else
|
---|
| 805 | {
|
---|
| 806 | BBLOCK_DECODE (diff, blk);
|
---|
| 807 | CurrDocNum += diff;
|
---|
| 808 | }
|
---|
| 809 | if (qd->id->ifh.InvfLevel >= 2)
|
---|
| 810 | GAMMA_DECODE (Count);
|
---|
| 811 | else
|
---|
| 812 | Count = (double) we->count / p;
|
---|
| 813 | Wdt = Count * WordLog;
|
---|
| 814 |
|
---|
| 815 | /* [RJM 07/97: Ranked Required Terms] */
|
---|
| 816 | /* zero the weights of intermediate documents if this term is required to match */
|
---|
| 817 | if (require_match == 1) {
|
---|
| 818 | for (TempDocNum = LastDocNum+1; TempDocNum < CurrDocNum; TempDocNum++)
|
---|
| 819 | AccumulatedWeights[TempDocNum - 1] = 0.0;
|
---|
| 820 | }
|
---|
| 821 |
|
---|
| 822 | /* [RJM 07/97: Ranked Required Terms] */
|
---|
| 823 | /* if a term has been required to match before, the weight we are */
|
---|
| 824 | /* adding to must be non-zero */
|
---|
| 825 | if (!seen_must_match || (AccumulatedWeights[CurrDocNum - 1] > 0.0)) {
|
---|
| 826 | AccumulatedWeights[CurrDocNum - 1] += Wqt * Wdt;
|
---|
| 827 | }
|
---|
| 828 |
|
---|
| 829 | qd->num_of_ptrs++;
|
---|
| 830 | LastDocNum = CurrDocNum; /* [RJM 07/97: Ranked Required Terms] */
|
---|
| 831 | }
|
---|
| 832 |
|
---|
| 833 | DECODE_DONE
|
---|
| 834 |
|
---|
| 835 | Xfree (buffer);
|
---|
| 836 | ChangeMemInUse (qd, -we->invf_len);
|
---|
| 837 |
|
---|
| 838 | /* [RJM 07/97: Ranked Required Terms] */
|
---|
| 839 | /* zero the remaining weights if this term is required to match */
|
---|
| 840 | if (require_match == 1) {
|
---|
| 841 | for (TempDocNum = LastDocNum+1; TempDocNum <= num_docs; TempDocNum++)
|
---|
| 842 | AccumulatedWeights[TempDocNum - 1] = 0.0;
|
---|
| 843 | seen_must_match = 1;
|
---|
| 844 | }
|
---|
| 845 | }
|
---|
| 846 | return AccumulatedWeights;
|
---|
| 847 | }
|
---|
| 848 |
|
---|
| 849 |
|
---|
| 850 |
|
---|
| 851 | static int
|
---|
| 852 | doc_comp (void *a, void *b)
|
---|
| 853 | {
|
---|
| 854 | return ((Invf_Doc_Entry *) a)->DocNum - ((Invf_Doc_Entry *) b)->DocNum;
|
---|
| 855 | }
|
---|
| 856 |
|
---|
| 857 |
|
---|
| 858 | static Invf_Doc_Entry *
|
---|
| 859 | get_ide (query_data * qd, Invf_Doc_Entry_Pool * pool)
|
---|
| 860 | {
|
---|
| 861 | struct pool *p = pool->pool;
|
---|
| 862 | if (!p || p->pos == p->length)
|
---|
| 863 | {
|
---|
| 864 | register unsigned size = sizeof (struct pool) +
|
---|
| 865 | (pool->pool_chunks - 1) * sizeof (Invf_Doc_Entry);
|
---|
| 866 | p = malloc (size);
|
---|
| 867 | if (!p)
|
---|
| 868 | return NULL;
|
---|
| 869 | ChangeMemInUse (qd, size);
|
---|
| 870 | p->length = pool->pool_chunks;
|
---|
| 871 | p->pos = 0;
|
---|
| 872 | p->next = pool->pool;
|
---|
| 873 | pool->pool = p;
|
---|
| 874 | }
|
---|
| 875 | return &(p->ide[p->pos++]);
|
---|
| 876 | }
|
---|
| 877 |
|
---|
| 878 |
|
---|
| 879 | void
|
---|
| 880 | free_ide_pool (query_data * qd, Invf_Doc_Entry_Pool * pool)
|
---|
| 881 | {
|
---|
| 882 | while (pool->pool)
|
---|
| 883 | {
|
---|
| 884 | struct pool *p = pool->pool;
|
---|
| 885 | pool->pool = p->next;
|
---|
| 886 | ChangeMemInUse (qd, -(sizeof (struct pool) + (p->length - 1) *
|
---|
| 887 | sizeof (Invf_Doc_Entry)));
|
---|
| 888 | free (p);
|
---|
| 889 | }
|
---|
| 890 | pool->pool = NULL;
|
---|
| 891 | }
|
---|
| 892 |
|
---|
| 893 |
|
---|
| 894 | FILE *
|
---|
| 895 | open_skip (char *sd)
|
---|
| 896 | {
|
---|
| 897 | char buf[256];
|
---|
| 898 | if (sd == NULL || *sd == '\0')
|
---|
| 899 | return NULL;
|
---|
| 900 | sprintf (buf, sd, getpid ());
|
---|
| 901 | return fopen (buf, "a");
|
---|
| 902 | }
|
---|
| 903 |
|
---|
| 904 |
|
---|
| 905 |
|
---|
| 906 | Splay_Tree *
|
---|
| 907 | CosineDecodeSplay (query_data * qd, TermList * Terms,
|
---|
| 908 | RankedQueryInfo * rqi, Invf_Doc_Entry_Pool * pool)
|
---|
| 909 | {
|
---|
| 910 | Splay_Tree *Doc_Set;
|
---|
| 911 | int j, num_docs, num_terms, MoreAccum;
|
---|
| 912 | int Anding = 0;
|
---|
| 913 | FILE *sk = open_skip (rqi->skip_dump);
|
---|
| 914 | u_char indent = 0;
|
---|
| 915 | if (sk)
|
---|
| 916 | {
|
---|
| 917 | if (!skip_header)
|
---|
| 918 | {
|
---|
| 919 | skip_header = 1;
|
---|
| 920 | fprintf (sk, "%s", skip_header_text[qd->id->ifh.skip_mode]);
|
---|
| 921 | fprintf (sk, "\nSkipping method %ld\n", qd->id->ifh.skip_mode);
|
---|
| 922 | switch (qd->id->ifh.skip_mode)
|
---|
| 923 | {
|
---|
| 924 | case 1:
|
---|
| 925 | fprintf (sk, "Skipping is every %ld docnums\n",
|
---|
| 926 | qd->id->ifh.params[0]);
|
---|
| 927 | break;
|
---|
| 928 | case 2:
|
---|
| 929 | fprintf (sk, "Max nodes = %ld\nNo skips smaller or equal to %ld\n",
|
---|
| 930 | qd->id->ifh.params[0], qd->id->ifh.params[1]);
|
---|
| 931 | break;
|
---|
| 932 | }
|
---|
| 933 | }
|
---|
| 934 | fprintf (sk, "\n\t\t\t-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-\n\n");
|
---|
| 935 | for (j = 0; j < Terms->num; j++)
|
---|
| 936 | if (Terms->TE[j].Word[0] > indent)
|
---|
| 937 | indent = Terms->TE[j].Word[0];
|
---|
| 938 | }
|
---|
| 939 |
|
---|
| 940 | num_docs = qd->sd->sdh.num_of_docs;
|
---|
| 941 | /* Create the array for the documents */
|
---|
| 942 | Doc_Set = SP_createset (doc_comp);
|
---|
| 943 |
|
---|
| 944 | pool->pool = NULL;
|
---|
| 945 | pool->pool_chunks = 1024;
|
---|
| 946 |
|
---|
| 947 | if (rqi->MaxTerms == -1)
|
---|
| 948 | num_terms = Terms->num;
|
---|
| 949 | else
|
---|
| 950 | num_terms = rqi->MaxTerms < Terms->num ? rqi->MaxTerms : Terms->num;
|
---|
| 951 |
|
---|
| 952 | MoreAccum = 1;
|
---|
| 953 |
|
---|
| 954 | /* For each word in the list . . . */
|
---|
| 955 | for (j = 0; j < num_terms; j++)
|
---|
| 956 | {
|
---|
| 957 | int Abort = 0;
|
---|
| 958 | u_char *buffer;
|
---|
| 959 | int pos, i;
|
---|
| 960 | unsigned long CurrDocNum = 0; /* NOTE: Document numbers start at 1 */
|
---|
| 961 | double Wqt, WordLog;
|
---|
| 962 | int dn_blk = 0, len_blk = 0, k, kd;
|
---|
| 963 | int blk; /* bblock control parameter */
|
---|
| 964 | int p; /* The number of documents the word occurs in */
|
---|
| 965 | WordEntry *we = &Terms->TE[j].WE;
|
---|
| 966 | Invf_Doc_Entry *next = NULL;
|
---|
| 967 | int next_i, next_doc_num, next_start;
|
---|
| 968 | int skips_taken = 0, ptrs_dec = 0;
|
---|
| 969 | double total_added = 0.0;
|
---|
| 970 | int hits = 0;
|
---|
| 971 | if (sk)
|
---|
| 972 | {
|
---|
| 973 | fprintf (sk, "%3d : %8ld %6ld \"%.*s\"%*s", j, we->count,
|
---|
| 974 | we->doc_count, Terms->TE[j].Word[0], Terms->TE[j].Word + 1,
|
---|
| 975 | indent - Terms->TE[j].Word[0], "");
|
---|
| 976 | }
|
---|
| 977 |
|
---|
| 978 | CalcBlks (qd, we, &blk, &dn_blk, &len_blk, &k, &p);
|
---|
| 979 | if (!(buffer = Xmalloc (we->invf_len)))
|
---|
| 980 | break;
|
---|
| 981 |
|
---|
| 982 | ChangeMemInUse (qd, we->invf_len);
|
---|
| 983 |
|
---|
| 984 | if (sk && qd->id->ifh.skip_mode == 2)
|
---|
| 985 | fprintf (sk, "%4d", k);
|
---|
| 986 |
|
---|
| 987 | Fseek (qd->File_invf, we->invf_ptr, 0);
|
---|
| 988 | Fread (buffer, sizeof (char), we->invf_len, qd->File_invf);
|
---|
| 989 |
|
---|
| 990 | /* Read from the buffer */
|
---|
| 991 | DECODE_START (buffer, we->invf_len);
|
---|
| 992 |
|
---|
| 993 | WordLog = log ((double) num_docs / (double) we->max_doc_count); /* [RPAP - Jan 97: Stem Index Change] */
|
---|
| 994 |
|
---|
| 995 | Wqt = (rqi->QueryFreqs ? Terms->TE[j].Count : 1) * WordLog;
|
---|
| 996 |
|
---|
| 997 | qd->num_of_terms++;
|
---|
| 998 |
|
---|
| 999 | if (rqi->MaxAccums != -1 && Doc_Set->no_of_items >= rqi->MaxAccums)
|
---|
| 1000 | {
|
---|
| 1001 | MoreAccum = 0;
|
---|
| 1002 | Anding = 1;
|
---|
| 1003 | next = SP_get_first (Doc_Set);
|
---|
| 1004 | }
|
---|
| 1005 | i = pos = 0;
|
---|
| 1006 | next_i = next_start = next_doc_num = 0;
|
---|
| 1007 | kd = k - 1;
|
---|
| 1008 | while (i < p && (!Anding || next))
|
---|
| 1009 | {
|
---|
| 1010 | int Count; /* The number of times the occurs in a document */
|
---|
| 1011 | double Wdt;
|
---|
| 1012 | unsigned long diff;
|
---|
| 1013 | Invf_Doc_Entry ent, *mem;
|
---|
| 1014 | if (k)
|
---|
| 1015 | {
|
---|
| 1016 | if (Anding && i + k < p && next->DocNum > next_doc_num &&
|
---|
| 1017 | next_doc_num >= 0)
|
---|
| 1018 | {
|
---|
| 1019 | i = next_i;
|
---|
| 1020 | DECODE_SEEK (next_start);
|
---|
| 1021 | pos = next_start;
|
---|
| 1022 | CurrDocNum = next_doc_num;
|
---|
| 1023 | qd->hops_taken++;
|
---|
| 1024 | skips_taken++;
|
---|
| 1025 | kd = k - 1;
|
---|
| 1026 | }
|
---|
| 1027 | if (kd == k - 1)
|
---|
| 1028 | if (i + k < p)
|
---|
| 1029 | {
|
---|
| 1030 | int doc_diff, pos_diff;
|
---|
| 1031 | BBLOCK_DECODE_L (doc_diff, dn_blk, pos);
|
---|
| 1032 | next_doc_num += doc_diff;
|
---|
| 1033 | next_i += k;
|
---|
| 1034 | BBLOCK_DECODE_L (pos_diff, len_blk, pos);
|
---|
| 1035 | next_start = pos + pos_diff;
|
---|
| 1036 | }
|
---|
| 1037 | else
|
---|
| 1038 | next_doc_num = -1;
|
---|
| 1039 | if (Anding && i + k < p && next->DocNum > next_doc_num &&
|
---|
| 1040 | next_doc_num >= 0)
|
---|
| 1041 | continue;
|
---|
| 1042 | }
|
---|
| 1043 | if (k && kd == 0 && i != p - 1)
|
---|
| 1044 | CurrDocNum = next_doc_num;
|
---|
| 1045 | else
|
---|
| 1046 | {
|
---|
| 1047 | BBLOCK_DECODE_L (diff, blk, pos);
|
---|
| 1048 | CurrDocNum += diff;
|
---|
| 1049 | }
|
---|
| 1050 | qd->num_of_ptrs++;
|
---|
| 1051 | ptrs_dec++;
|
---|
| 1052 | if (qd->id->ifh.InvfLevel >= 2)
|
---|
| 1053 | GAMMA_DECODE_L (Count, pos);
|
---|
| 1054 | else
|
---|
| 1055 | Count = (double) we->count / p;
|
---|
| 1056 | i++;
|
---|
| 1057 | kd--;
|
---|
| 1058 | if (kd < 0)
|
---|
| 1059 | kd = k - 1;
|
---|
| 1060 | Wdt = Count * WordLog;
|
---|
| 1061 |
|
---|
| 1062 |
|
---|
| 1063 | if (Anding)
|
---|
| 1064 | {
|
---|
| 1065 | while (next && next->DocNum < CurrDocNum - 1)
|
---|
| 1066 | next = SP_get_next (Doc_Set);
|
---|
| 1067 | if (next && next->DocNum == CurrDocNum - 1)
|
---|
| 1068 | {
|
---|
| 1069 | next->Sum += Wqt * Wdt;
|
---|
| 1070 | total_added += Wqt * Wdt;
|
---|
| 1071 | hits++;
|
---|
| 1072 | }
|
---|
| 1073 | if (next && next->DocNum <= CurrDocNum - 1)
|
---|
| 1074 | next = SP_get_next (Doc_Set);
|
---|
| 1075 | goto NextDocNum;
|
---|
| 1076 | }
|
---|
| 1077 | ent.DocNum = CurrDocNum - 1;
|
---|
| 1078 | if ((mem = SP_member (&ent, Doc_Set)) == NULL)
|
---|
| 1079 | {
|
---|
| 1080 | hits++;
|
---|
| 1081 | if (rqi->MaxAccums != -1 &&
|
---|
| 1082 | Doc_Set->no_of_items >= rqi->MaxAccums &&
|
---|
| 1083 | !MoreAccum)
|
---|
| 1084 | goto NextDocNum;
|
---|
| 1085 | if ((mem = get_ide (qd, pool)) == NULL)
|
---|
| 1086 | return NULL;
|
---|
| 1087 | ChangeMemInUse (qd, sizeof (*mem));
|
---|
| 1088 | mem->DocNum = CurrDocNum - 1;
|
---|
| 1089 | mem->Sum = 0;
|
---|
| 1090 | ChangeMemInUse (qd, -Doc_Set->mem_in_use);
|
---|
| 1091 | if (SP_insert (mem, Doc_Set) == NULL)
|
---|
| 1092 | {
|
---|
| 1093 | Message ("Unable to allocate memory for a splay node");
|
---|
| 1094 | Abort = 1;
|
---|
| 1095 | goto FastAbort;
|
---|
| 1096 | }
|
---|
| 1097 | ChangeMemInUse (qd, Doc_Set->mem_in_use);
|
---|
| 1098 | if (rqi->MaxAccums != -1 &&
|
---|
| 1099 | Doc_Set->no_of_items >= rqi->MaxAccums &&
|
---|
| 1100 | rqi->StopAtMaxAccum)
|
---|
| 1101 | Abort = 1;
|
---|
| 1102 | }
|
---|
| 1103 |
|
---|
| 1104 | mem->Sum += Wqt * Wdt;
|
---|
| 1105 | total_added += Wqt * Wdt;
|
---|
| 1106 |
|
---|
| 1107 | NextDocNum:
|
---|
| 1108 | ;
|
---|
| 1109 | }
|
---|
| 1110 |
|
---|
| 1111 | FastAbort:
|
---|
| 1112 |
|
---|
| 1113 | DECODE_DONE
|
---|
| 1114 |
|
---|
| 1115 | Xfree (buffer);
|
---|
| 1116 | ChangeMemInUse (qd, -we->invf_len);
|
---|
| 1117 | if (sk)
|
---|
| 1118 | {
|
---|
| 1119 | if (qd->id->ifh.skip_mode != 0)
|
---|
| 1120 | fprintf (sk, " %6d", skips_taken);
|
---|
| 1121 | fprintf (sk, " %6d %6d %6d%12.2f\n",
|
---|
| 1122 | ptrs_dec, hits, Doc_Set->no_of_items, total_added);
|
---|
| 1123 | }
|
---|
| 1124 | if (Abort)
|
---|
| 1125 | break;
|
---|
| 1126 | }
|
---|
| 1127 | if (sk)
|
---|
| 1128 | fclose (sk);
|
---|
| 1129 | return Doc_Set;
|
---|
| 1130 | }
|
---|
| 1131 |
|
---|
| 1132 |
|
---|
| 1133 |
|
---|
| 1134 |
|
---|
| 1135 | Hash_Table *
|
---|
| 1136 | CosineDecodeHash (query_data * qd, TermList * Terms,
|
---|
| 1137 | RankedQueryInfo * rqi)
|
---|
| 1138 | {
|
---|
| 1139 | int j, num_docs, num_terms, MoreAccum;
|
---|
| 1140 | Hash_Table *HT;
|
---|
| 1141 | int Anding = 0;
|
---|
| 1142 | int Sorted = 0;
|
---|
| 1143 | FILE *sk = open_skip (rqi->skip_dump);
|
---|
| 1144 | u_char indent = 0;
|
---|
| 1145 | if (sk)
|
---|
| 1146 | {
|
---|
| 1147 | if (!skip_header)
|
---|
| 1148 | {
|
---|
| 1149 | skip_header = 1;
|
---|
| 1150 | fprintf (sk, "%s", skip_header_text[qd->id->ifh.skip_mode]);
|
---|
| 1151 | fprintf (sk, "\nSkipping method %ld\n", qd->id->ifh.skip_mode);
|
---|
| 1152 | switch (qd->id->ifh.skip_mode)
|
---|
| 1153 | {
|
---|
| 1154 | case 1:
|
---|
| 1155 | fprintf (sk, "Skipping is every %ld docnums\n",
|
---|
| 1156 | qd->id->ifh.params[0]);
|
---|
| 1157 | break;
|
---|
| 1158 | case 2:
|
---|
| 1159 | fprintf (sk, "Max nodes = %ld\nNo skips smaller or equal to %ld\n",
|
---|
| 1160 | qd->id->ifh.params[0], qd->id->ifh.params[1]);
|
---|
| 1161 | break;
|
---|
| 1162 | }
|
---|
| 1163 | }
|
---|
| 1164 | fprintf (sk, "\n\t\t\t-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-\n\n");
|
---|
| 1165 | for (j = 0; j < Terms->num; j++)
|
---|
| 1166 | if (Terms->TE[j].Word[0] > indent)
|
---|
| 1167 | indent = Terms->TE[j].Word[0];
|
---|
| 1168 | }
|
---|
| 1169 |
|
---|
| 1170 | num_docs = qd->sd->sdh.num_of_docs;
|
---|
| 1171 | /* Create the array for the documents */
|
---|
| 1172 | HT = HT_create (qd, rqi->MaxAccums, rqi->HashTblSize);
|
---|
| 1173 | if (!HT)
|
---|
| 1174 | return NULL;
|
---|
| 1175 |
|
---|
| 1176 | if (rqi->MaxTerms == -1)
|
---|
| 1177 | num_terms = Terms->num;
|
---|
| 1178 | else
|
---|
| 1179 | num_terms = rqi->MaxTerms < Terms->num ? rqi->MaxTerms : Terms->num;
|
---|
| 1180 |
|
---|
| 1181 | MoreAccum = 1;
|
---|
| 1182 |
|
---|
| 1183 | /* For each word in the list . . . */
|
---|
| 1184 | for (j = 0; j < num_terms; j++)
|
---|
| 1185 | {
|
---|
| 1186 | int Abort = 0;
|
---|
| 1187 | u_char *buffer;
|
---|
| 1188 | int pos;
|
---|
| 1189 | int i;
|
---|
| 1190 | unsigned long CurrDocNum = 0; /* NOTE: Document numbers start at 1 */
|
---|
| 1191 | double Wqt, WordLog;
|
---|
| 1192 | int dn_blk = 0, len_blk = 0, k, kd;
|
---|
| 1193 | int blk; /* bblock control parameter */
|
---|
| 1194 | int p; /* The number of documents the word occurs in */
|
---|
| 1195 | WordEntry *we = &Terms->TE[j].WE;
|
---|
| 1196 | Invf_Doc_EntryH *next = NULL;
|
---|
| 1197 | int next_i, next_doc_num, next_start;
|
---|
| 1198 | int skips_taken = 0, ptrs_dec = 0;
|
---|
| 1199 | double total_added = 0.0;
|
---|
| 1200 | int hits = 0;
|
---|
| 1201 | if (sk)
|
---|
| 1202 | {
|
---|
| 1203 | fprintf (sk, "%3d : %8ld %6ld \"%.*s\"%*s", j, we->count,
|
---|
| 1204 | we->doc_count, Terms->TE[j].Word[0], Terms->TE[j].Word + 1,
|
---|
| 1205 | indent - Terms->TE[j].Word[0], "");
|
---|
| 1206 | }
|
---|
| 1207 |
|
---|
| 1208 | CalcBlks (qd, we, &blk, &dn_blk, &len_blk, &k, &p);
|
---|
| 1209 | if (!(buffer = Xmalloc (we->invf_len)))
|
---|
| 1210 | break;
|
---|
| 1211 |
|
---|
| 1212 | ChangeMemInUse (qd, we->invf_len);
|
---|
| 1213 |
|
---|
| 1214 | if (sk && qd->id->ifh.skip_mode == 2)
|
---|
| 1215 | fprintf (sk, "%4d", k);
|
---|
| 1216 |
|
---|
| 1217 | Fseek (qd->File_invf, we->invf_ptr, 0);
|
---|
| 1218 | Fread (buffer, sizeof (char), we->invf_len, qd->File_invf);
|
---|
| 1219 |
|
---|
| 1220 | /* Read from the buffer */
|
---|
| 1221 | DECODE_START (buffer, we->invf_len);
|
---|
| 1222 |
|
---|
| 1223 | WordLog = log ((double) num_docs / (double) we->max_doc_count); /* [RPAP - Jan 97: Stem Index Change] */
|
---|
| 1224 |
|
---|
| 1225 | Wqt = (rqi->QueryFreqs ? Terms->TE[j].Count : 1) * WordLog;
|
---|
| 1226 |
|
---|
| 1227 | qd->num_of_terms++;
|
---|
| 1228 |
|
---|
| 1229 | if (rqi->MaxAccums != -1 && HT->num >= rqi->MaxAccums)
|
---|
| 1230 | {
|
---|
| 1231 | MoreAccum = 0;
|
---|
| 1232 | Anding = 1;
|
---|
| 1233 | if (!Sorted)
|
---|
| 1234 | {
|
---|
| 1235 | HT_sort (qd, HT);
|
---|
| 1236 | Sorted = 1;
|
---|
| 1237 | }
|
---|
| 1238 | next = HT->Head;
|
---|
| 1239 | }
|
---|
| 1240 | i = pos = 0;
|
---|
| 1241 | next_i = next_start = next_doc_num = 0;
|
---|
| 1242 | kd = k - 1;
|
---|
| 1243 | while (i < p && (!Anding || next))
|
---|
| 1244 | {
|
---|
| 1245 | int Count; /* The number of times the occurs in a document */
|
---|
| 1246 | double Wdt;
|
---|
| 1247 | unsigned long diff;
|
---|
| 1248 | Invf_Doc_Entry *mem;
|
---|
| 1249 | if (k)
|
---|
| 1250 | {
|
---|
| 1251 | if (Anding && i + k < p && next->IDE.DocNum > next_doc_num &&
|
---|
| 1252 | next_doc_num >= 0)
|
---|
| 1253 | {
|
---|
| 1254 | #ifdef SHORT_CUT
|
---|
| 1255 | ShortCut:
|
---|
| 1256 | #endif
|
---|
| 1257 | i = next_i;
|
---|
| 1258 | DECODE_SEEK (next_start);
|
---|
| 1259 | pos = next_start;
|
---|
| 1260 | CurrDocNum = next_doc_num;
|
---|
| 1261 | qd->hops_taken++;
|
---|
| 1262 | skips_taken++;
|
---|
| 1263 | kd = k - 1;
|
---|
| 1264 | }
|
---|
| 1265 | if (kd == k - 1)
|
---|
| 1266 | {
|
---|
| 1267 | if (i + k < p)
|
---|
| 1268 | {
|
---|
| 1269 | int doc_diff, pos_diff;
|
---|
| 1270 | BBLOCK_DECODE_L (doc_diff, dn_blk, pos);
|
---|
| 1271 | next_doc_num += doc_diff;
|
---|
| 1272 | next_i += k;
|
---|
| 1273 | BBLOCK_DECODE_L (pos_diff, len_blk, pos);
|
---|
| 1274 | next_start = pos + pos_diff;
|
---|
| 1275 | }
|
---|
| 1276 | else
|
---|
| 1277 | next_doc_num = -1;
|
---|
| 1278 | }
|
---|
| 1279 | #ifdef SHORT_CUT
|
---|
| 1280 | if (Anding && i + k < p && next->IDE.DocNum > next_doc_num &&
|
---|
| 1281 | next_doc_num >= 0)
|
---|
| 1282 | goto ShortCut;
|
---|
| 1283 | #else
|
---|
| 1284 | if (Anding && i + k < p && next->IDE.DocNum > next_doc_num &&
|
---|
| 1285 | next_doc_num >= 0)
|
---|
| 1286 | continue;
|
---|
| 1287 | #endif
|
---|
| 1288 | }
|
---|
| 1289 | if (k && kd == 0 && i != p - 1)
|
---|
| 1290 | {
|
---|
| 1291 | CurrDocNum = next_doc_num;
|
---|
| 1292 | }
|
---|
| 1293 | else
|
---|
| 1294 | {
|
---|
| 1295 | BBLOCK_DECODE_L (diff, blk, pos);
|
---|
| 1296 | CurrDocNum += diff;
|
---|
| 1297 | }
|
---|
| 1298 | qd->num_of_ptrs++;
|
---|
| 1299 | ptrs_dec++;
|
---|
| 1300 | if (qd->id->ifh.InvfLevel >= 2)
|
---|
| 1301 | GAMMA_DECODE_L (Count, pos);
|
---|
| 1302 | else
|
---|
| 1303 | Count = (double) we->count / p;
|
---|
| 1304 | i++;
|
---|
| 1305 | kd--;
|
---|
| 1306 | if (kd < 0)
|
---|
| 1307 | kd = k - 1;
|
---|
| 1308 |
|
---|
| 1309 |
|
---|
| 1310 | if (Anding)
|
---|
| 1311 | {
|
---|
| 1312 | while (next && next->IDE.DocNum < CurrDocNum - 1)
|
---|
| 1313 | next = next->next;
|
---|
| 1314 | if (next && next->IDE.DocNum == CurrDocNum - 1)
|
---|
| 1315 | {
|
---|
| 1316 | Wdt = Count * WordLog;
|
---|
| 1317 | next->IDE.Sum += Wqt * Wdt;
|
---|
| 1318 | total_added += Wqt * Wdt;
|
---|
| 1319 | hits++;
|
---|
| 1320 | }
|
---|
| 1321 | if (next && next->IDE.DocNum <= CurrDocNum - 1)
|
---|
| 1322 | next = next->next;
|
---|
| 1323 | goto NextDocNum;
|
---|
| 1324 | }
|
---|
| 1325 | if ((mem = HT_find (HT, CurrDocNum - 1)) == NULL)
|
---|
| 1326 | {
|
---|
| 1327 | if (rqi->MaxAccums != -1 && HT->num >= rqi->MaxAccums &&
|
---|
| 1328 | !MoreAccum)
|
---|
| 1329 | goto NextDocNum;
|
---|
| 1330 | Wdt = Count * WordLog;
|
---|
| 1331 | if (!HT_insert (HT, CurrDocNum - 1, Wqt * Wdt))
|
---|
| 1332 | {
|
---|
| 1333 | if (!HT->Suplimentary_Entries)
|
---|
| 1334 | HT_Sup_create (qd, HT, we->doc_count - i + 1);
|
---|
| 1335 | HT_Sup_Add (HT, CurrDocNum - 1, Wqt * Wdt);
|
---|
| 1336 | }
|
---|
| 1337 | if (rqi->MaxAccums != -1 && HT->num >= rqi->MaxAccums &&
|
---|
| 1338 | rqi->StopAtMaxAccum)
|
---|
| 1339 | Abort = 1;
|
---|
| 1340 | }
|
---|
| 1341 | else
|
---|
| 1342 | {
|
---|
| 1343 | Wdt = Count * WordLog;
|
---|
| 1344 | mem->Sum += Wqt * Wdt;
|
---|
| 1345 | total_added += Wqt * Wdt;
|
---|
| 1346 | hits++;
|
---|
| 1347 | }
|
---|
| 1348 |
|
---|
| 1349 | NextDocNum:
|
---|
| 1350 | ;
|
---|
| 1351 | }
|
---|
| 1352 |
|
---|
| 1353 | DECODE_DONE
|
---|
| 1354 |
|
---|
| 1355 | Xfree (buffer);
|
---|
| 1356 | ChangeMemInUse (qd, -we->invf_len);
|
---|
| 1357 | if (sk)
|
---|
| 1358 | {
|
---|
| 1359 | if (qd->id->ifh.skip_mode != 0)
|
---|
| 1360 | fprintf (sk, " %6d", skips_taken);
|
---|
| 1361 | fprintf (sk, " %6d %6d %6d%12.2f\n",
|
---|
| 1362 | ptrs_dec, hits, HT->num + HT->Suplimentary_Num, total_added);
|
---|
| 1363 | }
|
---|
| 1364 | if (Abort)
|
---|
| 1365 | break;
|
---|
| 1366 | }
|
---|
| 1367 | if (sk)
|
---|
| 1368 | fclose (sk);
|
---|
| 1369 | if (!HT->Head)
|
---|
| 1370 | HT_sort (qd, HT);
|
---|
| 1371 | return HT;
|
---|
| 1372 | }
|
---|
| 1373 |
|
---|
| 1374 |
|
---|
| 1375 |
|
---|
| 1376 | List_Table *
|
---|
| 1377 | CosineDecodeList (query_data * qd, TermList * Terms,
|
---|
| 1378 | RankedQueryInfo * rqi)
|
---|
| 1379 | {
|
---|
| 1380 | int j, num_docs, num_terms, MoreAccum;
|
---|
| 1381 | List_Table *LT;
|
---|
| 1382 | int Anding = 0;
|
---|
| 1383 | int Sorted = 0;
|
---|
| 1384 | int has_grown = 0; /* [RJM 07/97: Ranked Required Match] whether elements have been added lately */
|
---|
| 1385 | FILE *sk = open_skip (rqi->skip_dump);
|
---|
| 1386 | u_char indent = 0;
|
---|
| 1387 | if (sk)
|
---|
| 1388 | {
|
---|
| 1389 | if (!skip_header)
|
---|
| 1390 | {
|
---|
| 1391 | skip_header = 1;
|
---|
| 1392 | fprintf (sk, "%s", skip_header_text[qd->id->ifh.skip_mode]);
|
---|
| 1393 | fprintf (sk, "\nSkipping method %ld\n", qd->id->ifh.skip_mode);
|
---|
| 1394 | switch (qd->id->ifh.skip_mode)
|
---|
| 1395 | {
|
---|
| 1396 | case 1:
|
---|
| 1397 | fprintf (sk, "Skipping is every %ld docnums\n",
|
---|
| 1398 | qd->id->ifh.params[0]);
|
---|
| 1399 | break;
|
---|
| 1400 | case 2:
|
---|
| 1401 | fprintf (sk, "Max nodes = %ld\nNo skips smaller or equal to %ld\n",
|
---|
| 1402 | qd->id->ifh.params[0], qd->id->ifh.params[1]);
|
---|
| 1403 | break;
|
---|
| 1404 | }
|
---|
| 1405 | }
|
---|
| 1406 | fprintf (sk, "\n\t\t\t-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-\n\n");
|
---|
| 1407 | for (j = 0; j < Terms->num; j++)
|
---|
| 1408 | if (Terms->TE[j].Word[0] > indent)
|
---|
| 1409 | indent = Terms->TE[j].Word[0];
|
---|
| 1410 | }
|
---|
| 1411 |
|
---|
| 1412 | num_docs = qd->sd->sdh.num_of_docs;
|
---|
| 1413 |
|
---|
| 1414 | /* Create the list for the documents */
|
---|
| 1415 | LT = LT_create (qd, rqi->MaxAccums);
|
---|
| 1416 | if (!LT)
|
---|
| 1417 | return NULL;
|
---|
| 1418 |
|
---|
| 1419 | if (rqi->MaxTerms == -1)
|
---|
| 1420 | num_terms = Terms->num;
|
---|
| 1421 | else
|
---|
| 1422 | num_terms = rqi->MaxTerms < Terms->num ? rqi->MaxTerms : Terms->num;
|
---|
| 1423 |
|
---|
| 1424 | MoreAccum = 1;
|
---|
| 1425 |
|
---|
| 1426 | /* For each word in the list . . . */
|
---|
| 1427 | for (j = 0; j < num_terms; j++)
|
---|
| 1428 | {
|
---|
| 1429 | int Abort = 0;
|
---|
| 1430 | u_char *buffer;
|
---|
| 1431 | int pos;
|
---|
| 1432 | int i, n = 0;
|
---|
| 1433 | unsigned long CurrDocNum = 0; /* NOTE: Document numbers start at 1 */
|
---|
| 1434 | unsigned long LastDocNum = 0; /* [RJM 07/97: Ranked Required Match] */
|
---|
| 1435 | int this_item = 0; /* [RJM 07/97: Ranked Required Match] points to the next possible LT to delete */
|
---|
| 1436 | double Wqt, WordLog;
|
---|
| 1437 | int dn_blk = 0, len_blk = 0, k, kd;
|
---|
| 1438 | int blk; /* bblock control parameter */
|
---|
| 1439 | int p; /* The number of documents the word occurs in */
|
---|
| 1440 | int require_match = Terms->TE[j].require_match; /* [RJM 07/97: Ranked Required Match] */
|
---|
| 1441 | WordEntry *we = &Terms->TE[j].WE;
|
---|
| 1442 | Invf_Doc_Entry *next = NULL;
|
---|
| 1443 | int next_i, next_doc_num, next_start;
|
---|
| 1444 | int skips_taken = 0, ptrs_dec = 0;
|
---|
| 1445 | double total_added = 0.0;
|
---|
| 1446 | int hits = 0;
|
---|
| 1447 |
|
---|
| 1448 | if (sk)
|
---|
| 1449 | {
|
---|
| 1450 | fprintf (sk, "%3d : %8ld %6ld \"%.*s\"%*s", j, we->count,
|
---|
| 1451 | we->doc_count, Terms->TE[j].Word[0], Terms->TE[j].Word + 1,
|
---|
| 1452 | indent - Terms->TE[j].Word[0], "");
|
---|
| 1453 | }
|
---|
| 1454 |
|
---|
| 1455 | CalcBlks (qd, we, &blk, &dn_blk, &len_blk, &k, &p);
|
---|
| 1456 | if (!(buffer = Xmalloc (we->invf_len)))
|
---|
| 1457 | break;
|
---|
| 1458 |
|
---|
| 1459 | ChangeMemInUse (qd, we->invf_len);
|
---|
| 1460 |
|
---|
| 1461 | if (sk && qd->id->ifh.skip_mode == 2)
|
---|
| 1462 | fprintf (sk, "%4d", k);
|
---|
| 1463 |
|
---|
| 1464 | Fseek (qd->File_invf, we->invf_ptr, 0);
|
---|
| 1465 | Fread (buffer, sizeof (char), we->invf_len, qd->File_invf);
|
---|
| 1466 |
|
---|
| 1467 | /* Read from the buffer */
|
---|
| 1468 | DECODE_START (buffer, we->invf_len);
|
---|
| 1469 |
|
---|
| 1470 | WordLog = log ((double) num_docs / (double) we->max_doc_count); /* [RPAP - Jan 97: Stem Index Change] */
|
---|
| 1471 |
|
---|
| 1472 | Wqt = (rqi->QueryFreqs ? Terms->TE[j].Count : 1) * WordLog;
|
---|
| 1473 |
|
---|
| 1474 | qd->num_of_terms++;
|
---|
| 1475 |
|
---|
| 1476 | if (rqi->MaxAccums != -1 && LT->num >= rqi->MaxAccums)
|
---|
| 1477 | MoreAccum = 0;
|
---|
| 1478 |
|
---|
| 1479 | if ((rqi->MaxAccums != -1 && LT->num >= rqi->MaxAccums) || Anding)
|
---|
| 1480 | {
|
---|
| 1481 | Anding = 1;
|
---|
| 1482 | if (!Sorted || has_grown)
|
---|
| 1483 | {
|
---|
| 1484 | LT_sort (LT);
|
---|
| 1485 | LT_pack (LT);
|
---|
| 1486 | Sorted = 1;
|
---|
| 1487 | has_grown = 0;
|
---|
| 1488 | }
|
---|
| 1489 | next = LT->IDE;
|
---|
| 1490 | n = 0;
|
---|
| 1491 | }
|
---|
| 1492 |
|
---|
| 1493 | /* [RJM 07/97: Ranked Required Terms] */
|
---|
| 1494 | /* make sure that the document list is set up correctly for */
|
---|
| 1495 | /* requiring a term to match */
|
---|
| 1496 | if (require_match == 1) {
|
---|
| 1497 | if (!Sorted || has_grown) {
|
---|
| 1498 | LT_sort (LT);
|
---|
| 1499 | LT_pack (LT);
|
---|
| 1500 | Sorted = 1;
|
---|
| 1501 | has_grown = 0;
|
---|
| 1502 | }
|
---|
| 1503 | /* this_item is already set to zero */
|
---|
| 1504 | }
|
---|
| 1505 |
|
---|
| 1506 | i = pos = 0;
|
---|
| 1507 | next_i = next_start = next_doc_num = 0;
|
---|
| 1508 | kd = k - 1;
|
---|
| 1509 | while (i < p && (!Anding || n < LT->num))
|
---|
| 1510 | {
|
---|
| 1511 | int Count; /* The number of times the occurs in a document */
|
---|
| 1512 | double Wdt;
|
---|
| 1513 | unsigned long diff;
|
---|
| 1514 | Invf_Doc_Entry *mem;
|
---|
| 1515 |
|
---|
| 1516 | /* deal with skipping */
|
---|
| 1517 | if (k)
|
---|
| 1518 | {
|
---|
| 1519 | /* see if we should take a skip */
|
---|
| 1520 | if (Anding && i + k < p && next->DocNum > next_doc_num &&
|
---|
| 1521 | next_doc_num >= 0)
|
---|
| 1522 | {
|
---|
| 1523 | i = next_i;
|
---|
| 1524 | DECODE_SEEK (next_start);
|
---|
| 1525 | pos = next_start;
|
---|
| 1526 | CurrDocNum = next_doc_num;
|
---|
| 1527 | qd->hops_taken++;
|
---|
| 1528 | skips_taken++;
|
---|
| 1529 | kd = k - 1;
|
---|
| 1530 | }
|
---|
| 1531 | if (kd == k - 1)
|
---|
| 1532 | if (i + k < p)
|
---|
| 1533 | {
|
---|
| 1534 | int doc_diff, pos_diff;
|
---|
| 1535 | BBLOCK_DECODE_L (doc_diff, dn_blk, pos);
|
---|
| 1536 | next_doc_num += doc_diff;
|
---|
| 1537 | next_i += k;
|
---|
| 1538 | BBLOCK_DECODE_L (pos_diff, len_blk, pos);
|
---|
| 1539 | next_start = pos + pos_diff;
|
---|
| 1540 | }
|
---|
| 1541 | else
|
---|
| 1542 | next_doc_num = -1;
|
---|
| 1543 |
|
---|
| 1544 | /* should we take another skip? */
|
---|
| 1545 | if (Anding && i + k < p && next->DocNum > next_doc_num &&
|
---|
| 1546 | next_doc_num >= 0)
|
---|
| 1547 | continue;
|
---|
| 1548 | }
|
---|
| 1549 | if (k && kd == 0 && i != p - 1)
|
---|
| 1550 | CurrDocNum = next_doc_num;
|
---|
| 1551 | else
|
---|
| 1552 | {
|
---|
| 1553 | BBLOCK_DECODE_L (diff, blk, pos);
|
---|
| 1554 | CurrDocNum += diff;
|
---|
| 1555 | }
|
---|
| 1556 | qd->num_of_ptrs++;
|
---|
| 1557 | ptrs_dec++;
|
---|
| 1558 | if (qd->id->ifh.InvfLevel >= 2)
|
---|
| 1559 | GAMMA_DECODE_L (Count, pos);
|
---|
| 1560 | else
|
---|
| 1561 | Count = (double) we->count / p;
|
---|
| 1562 | i++;
|
---|
| 1563 | kd--;
|
---|
| 1564 | if (kd < 0)
|
---|
| 1565 | kd = k - 1;
|
---|
| 1566 | Wdt = Count * WordLog;
|
---|
| 1567 |
|
---|
| 1568 | /* [RJM 07/97: Ranked Required Terms] */
|
---|
| 1569 | /* if this is a required match we must remove all accumulators */
|
---|
| 1570 | /* of intermediate documents */
|
---|
| 1571 | if (require_match == 1) {
|
---|
| 1572 | while (this_item < LT->num && LT->IDE[this_item].DocNum < CurrDocNum - 1) {
|
---|
| 1573 | if (LastDocNum==0 ||(LT->IDE[this_item].DocNum > LastDocNum - 1))
|
---|
| 1574 | LT->IDE[this_item].Sum = -1000.0; /* mark for deletion */
|
---|
| 1575 | this_item++;
|
---|
| 1576 | }
|
---|
| 1577 | }
|
---|
| 1578 |
|
---|
| 1579 | if (Anding)
|
---|
| 1580 | {
|
---|
| 1581 | while (n < LT->num && next->DocNum < CurrDocNum - 1)
|
---|
| 1582 | {
|
---|
| 1583 | next++;
|
---|
| 1584 | n++;
|
---|
| 1585 | }
|
---|
| 1586 | if (n < LT->num && next->DocNum == CurrDocNum - 1)
|
---|
| 1587 | {
|
---|
| 1588 | next->Sum += Wqt * Wdt;
|
---|
| 1589 | total_added += Wqt * Wdt;
|
---|
| 1590 | hits++;
|
---|
| 1591 | }
|
---|
| 1592 | if (n < LT->num && next->DocNum <= CurrDocNum - 1)
|
---|
| 1593 | {
|
---|
| 1594 | next++;
|
---|
| 1595 | n++;
|
---|
| 1596 | }
|
---|
| 1597 | goto NextDocNum;
|
---|
| 1598 | }
|
---|
| 1599 | if (!Anding)
|
---|
| 1600 | {
|
---|
| 1601 | if (rqi->MaxAccums != -1 && LT->num >= rqi->MaxAccums &&
|
---|
| 1602 | !MoreAccum)
|
---|
| 1603 | goto NextDocNum;
|
---|
| 1604 | LT = LT_add (qd, LT, CurrDocNum - 1, Wqt * Wdt);
|
---|
| 1605 | if (!LT)
|
---|
| 1606 | {
|
---|
| 1607 | Abort = 1;
|
---|
| 1608 | goto FastAbort;
|
---|
| 1609 | }
|
---|
| 1610 |
|
---|
| 1611 | /* the list of accumulators just got bigger */
|
---|
| 1612 | has_grown = 1;
|
---|
| 1613 |
|
---|
| 1614 | if (rqi->MaxAccums != -1 && LT->num >= rqi->MaxAccums &&
|
---|
| 1615 | rqi->StopAtMaxAccum)
|
---|
| 1616 | Abort = 1;
|
---|
| 1617 |
|
---|
| 1618 |
|
---|
| 1619 | }
|
---|
| 1620 | else if ((mem = LT_find (LT, CurrDocNum - 1)))
|
---|
| 1621 | {
|
---|
| 1622 | mem->Sum += Wqt * Wdt;
|
---|
| 1623 | total_added += Wqt * Wdt;
|
---|
| 1624 | hits++;
|
---|
| 1625 | }
|
---|
| 1626 |
|
---|
| 1627 | NextDocNum:
|
---|
| 1628 |
|
---|
| 1629 | LastDocNum = CurrDocNum;
|
---|
| 1630 | }
|
---|
| 1631 |
|
---|
| 1632 | FastAbort:
|
---|
| 1633 |
|
---|
| 1634 | DECODE_DONE
|
---|
| 1635 |
|
---|
| 1636 | Xfree (buffer);
|
---|
| 1637 | ChangeMemInUse (qd, -we->invf_len);
|
---|
| 1638 | if (sk)
|
---|
| 1639 | {
|
---|
| 1640 | if (qd->id->ifh.skip_mode != 0)
|
---|
| 1641 | fprintf (sk, " %6d", skips_taken);
|
---|
| 1642 | fprintf (sk, " %6d %6d %6d%12.2f\n",
|
---|
| 1643 | ptrs_dec, hits, LT->num, total_added);
|
---|
| 1644 | }
|
---|
| 1645 | if (Abort)
|
---|
| 1646 | break;
|
---|
| 1647 |
|
---|
| 1648 | /* [RJM 07/97: Ranked Required Terms] */
|
---|
| 1649 | /* remove the remaining accumulators if this term is required to match */
|
---|
| 1650 | if (require_match == 1) {
|
---|
| 1651 | while (this_item < LT->num) {
|
---|
| 1652 | if (LT->IDE[this_item].DocNum > LastDocNum - 1)
|
---|
| 1653 | LT->IDE[this_item].Sum = -1000.0; /* mark for deletion */
|
---|
| 1654 | this_item++;
|
---|
| 1655 | }
|
---|
| 1656 | Anding = 1; /* no more documents should be added */
|
---|
| 1657 |
|
---|
| 1658 | /* Get rid of any documents which have been deleted */
|
---|
| 1659 | LT_pack (LT);
|
---|
| 1660 | }
|
---|
| 1661 | }
|
---|
| 1662 |
|
---|
| 1663 | if (sk)
|
---|
| 1664 | fclose (sk);
|
---|
| 1665 | return LT;
|
---|
| 1666 | }
|
---|