[856] | 1 | /**************************************************************************
|
---|
| 2 | *
|
---|
| 3 | * mg_weights_build.cpp -- Program to build the document weights file
|
---|
| 4 | * Copyright (C) 1999 Rodger McNab
|
---|
| 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: mg_weights_build.cpp 856 2000-01-14 02:26:25Z sjboddie $
|
---|
| 21 | *
|
---|
| 22 | **************************************************************************/
|
---|
| 23 |
|
---|
| 24 | #include "sysfuncs.h"
|
---|
| 25 | #include "memlib.h"
|
---|
| 26 | #include "messages.h"
|
---|
| 27 | #include "local_strings.h"
|
---|
| 28 | #include "bitio_gen.h"
|
---|
| 29 | #include "bitio_m_stdio.h"
|
---|
| 30 |
|
---|
| 31 | #include "mg_files.h"
|
---|
| 32 | #include "locallib.h"
|
---|
| 33 | #include "invf.h"
|
---|
| 34 | // #include "WordData.h"
|
---|
| 35 | #include "UCArray.h"
|
---|
| 36 | #include "FIvfLevelInfo.h"
|
---|
| 37 | #include "FragLevelConvert.h"
|
---|
| 38 |
|
---|
| 39 | #if defined(GSDL_USE_OBJECTSPACE)
|
---|
| 40 | # include <ospace\std\map>
|
---|
| 41 | #elif defined(GSDL_USE_STL_H)
|
---|
| 42 | # include <map.h>
|
---|
| 43 | #else
|
---|
| 44 | # include <map>
|
---|
| 45 | #endif
|
---|
| 46 |
|
---|
| 47 |
|
---|
| 48 | /*
|
---|
| 49 | $Log$
|
---|
| 50 | Revision 1.1 2000/01/14 02:26:21 sjboddie
|
---|
| 51 | Rodgers new C++ mg
|
---|
| 52 |
|
---|
| 53 | Revision 1.1 1999/10/11 02:58:06 cs025
|
---|
| 54 | Base install of MG-PP
|
---|
| 55 |
|
---|
| 56 | Revision 1.1 1999/08/10 21:18:16 sjboddie
|
---|
| 57 | renamed mg-1.3d directory mg
|
---|
| 58 |
|
---|
| 59 | Revision 1.2 1998/11/25 07:55:49 rjmcnab
|
---|
| 60 |
|
---|
| 61 | Modified mg to that you can specify the stemmer you want
|
---|
| 62 | to use via a command line option. You specify it to
|
---|
| 63 | mg_passes during the build process. The number of the
|
---|
| 64 | stemmer that you used is stored within the inverted
|
---|
| 65 | dictionary header and the stemmed dictionary header so
|
---|
| 66 | the correct stemmer is used in later stages of building
|
---|
| 67 | and querying.
|
---|
| 68 |
|
---|
| 69 | Revision 1.1 1998/11/17 09:35:22 rjmcnab
|
---|
| 70 | *** empty log message ***
|
---|
| 71 |
|
---|
| 72 | * Revision 1.4 1994/11/29 00:32:05 tes
|
---|
| 73 | * Committing the new merged files and changes.
|
---|
| 74 | *
|
---|
| 75 | * Revision 1.3 1994/10/20 03:57:00 tes
|
---|
| 76 | * I have rewritten the boolean query optimiser and abstracted out the
|
---|
| 77 | * components of the boolean query.
|
---|
| 78 | *
|
---|
| 79 | * Revision 1.2 1994/09/20 04:41:55 tes
|
---|
| 80 | * For version 1.1
|
---|
| 81 | *
|
---|
| 82 | */
|
---|
| 83 |
|
---|
| 84 | #define MAXBITS (sizeof(unsigned long) * 8)
|
---|
| 85 |
|
---|
| 86 | struct WBTagPtr {
|
---|
| 87 | unsigned long tagNum;
|
---|
| 88 | unsigned long tagPtr;
|
---|
| 89 | unsigned long fragOccur;
|
---|
| 90 |
|
---|
| 91 | WBTagPtr () {
|
---|
| 92 | tagNum = 0;
|
---|
| 93 | tagPtr = 0;
|
---|
| 94 | fragOccur = 0;
|
---|
| 95 | }
|
---|
| 96 | };
|
---|
| 97 |
|
---|
| 98 | // maps tags to tag information
|
---|
| 99 | typedef map<UCArray, WBTagPtr, DictLTUCArray> WBTagDict;
|
---|
| 100 |
|
---|
| 101 | typedef vector<float> Weights;
|
---|
| 102 |
|
---|
| 103 |
|
---|
| 104 | static void ReadTagDict (const invf_dict_header &idh,
|
---|
| 105 | FILE *dictFile,
|
---|
| 106 | FILE *invfIdxFile,
|
---|
| 107 | WBTagDict &tagDict) {
|
---|
| 108 | tagDict.erase (tagDict.begin(), tagDict.end());
|
---|
| 109 |
|
---|
| 110 | // seek to the start of the tag information
|
---|
| 111 | fseek (dictFile, idh.tag_dict_start, SEEK_SET);
|
---|
| 112 | fseek (invfIdxFile, sizeof(unsigned long) +
|
---|
| 113 | idh.word_dict_size*sizeof(unsigned long), SEEK_SET);
|
---|
| 114 |
|
---|
| 115 | unsigned long tagNum;
|
---|
| 116 | unsigned long tagPtr;
|
---|
| 117 | dict_el thisEl;
|
---|
| 118 | for (tagNum = 0; tagNum < idh.tag_dict_size; tagNum++) {
|
---|
| 119 | thisEl.Read (dictFile);
|
---|
| 120 | ReadUL (invfIdxFile, tagPtr);
|
---|
| 121 | tagDict[thisEl.el].tagNum = tagNum;
|
---|
| 122 | tagDict[thisEl.el].tagPtr = tagPtr;
|
---|
| 123 | tagDict[thisEl.el].fragOccur = thisEl.frag_occur;
|
---|
| 124 | }
|
---|
| 125 | }
|
---|
| 126 |
|
---|
| 127 | static void AddWeight (Weights &w,
|
---|
| 128 | unsigned long levelDocNum,
|
---|
| 129 | unsigned long termFreq,
|
---|
| 130 | float idf) {
|
---|
| 131 | double weight = termFreq * idf;
|
---|
| 132 | w[levelDocNum-1] += weight * weight;
|
---|
| 133 | }
|
---|
| 134 |
|
---|
| 135 | static void GenerateLevelWeights (const invf_dict_header &idh,
|
---|
| 136 | const invf_file_header &ifh,
|
---|
| 137 | unsigned long numLevelDocs,
|
---|
| 138 | unsigned long levelNum,
|
---|
| 139 | FILE *dictFile,
|
---|
| 140 | FILE *invfFile,
|
---|
| 141 | FILE *invfIdxFile,
|
---|
| 142 | const FragLevelConvert &fragLevelConvert,
|
---|
| 143 | Weights &w) {
|
---|
| 144 | // pre-allocate the right number of weights
|
---|
| 145 | w.erase (w.begin(), w.end());
|
---|
| 146 | w.insert (w.end(), (Weights::size_type)numLevelDocs, (float)0.0);
|
---|
| 147 |
|
---|
| 148 | double logN = log ((double) numLevelDocs);
|
---|
| 149 |
|
---|
| 150 | // reset the files
|
---|
| 151 | fseek (dictFile, idh.word_dict_start, SEEK_SET);
|
---|
| 152 | fseek (invfIdxFile, sizeof (unsigned long), SEEK_SET);
|
---|
| 153 |
|
---|
| 154 | // process each word adding its contributions to the document weights
|
---|
| 155 | unsigned long wordNum;
|
---|
| 156 | unsigned long wordStart;
|
---|
| 157 | word_dict_el wordEl;
|
---|
| 158 | wordEl.SetNumLevels (idh.num_levels);
|
---|
| 159 | for (wordNum=0; wordNum<idh.word_dict_size; wordNum++) {
|
---|
| 160 | // give a little feedback every 4096 words
|
---|
| 161 | if ((wordNum & 0xfff) == 0) fprintf (stderr, ".");
|
---|
| 162 |
|
---|
| 163 | wordEl.Read (dictFile, idh.num_levels);
|
---|
| 164 | ReadUL (invfIdxFile, wordStart);
|
---|
| 165 |
|
---|
| 166 | float idf = logN - log ((double) wordEl.levelFreqs[levelNum]);
|
---|
| 167 |
|
---|
| 168 | // seek to the appropriate place in the inverted file
|
---|
| 169 | fseek (invfFile, wordStart, SEEK_SET);
|
---|
| 170 | stdio_bitio_buffer buffer (invfFile);
|
---|
| 171 |
|
---|
| 172 | unsigned long B = BIO_Bblock_Init (idh.num_frags, wordEl.frag_occur);
|
---|
| 173 | unsigned long fragNum = 0;
|
---|
| 174 | unsigned long levelDocNum = 0;
|
---|
| 175 | unsigned long lastLevelDocNum = 0;
|
---|
| 176 | unsigned long termFreq = 0;
|
---|
| 177 | unsigned long checkLevelFreq = 0;
|
---|
| 178 |
|
---|
| 179 | unsigned long count, i;
|
---|
| 180 | for (i=0; i<wordEl.frag_occur; i++) {
|
---|
| 181 | fragNum += buffer.bblock_decode (B, NULL);
|
---|
| 182 | if (!ifh.word_level_index) count = buffer.gamma_decode (NULL);
|
---|
| 183 | else count = 1;
|
---|
| 184 |
|
---|
| 185 | if (fragNum > idh.num_frags)
|
---|
| 186 | FatalError (1, "fragNum = %d, "
|
---|
| 187 | "number of fragments = %d\n"
|
---|
| 188 | "wordNum = %d\n"
|
---|
| 189 | "i = %d, frag_occur = %d\n",
|
---|
| 190 | fragNum, idh.num_frags, wordNum, i, wordEl.frag_occur);
|
---|
| 191 |
|
---|
| 192 | if (!fragLevelConvert.FragToLevel (fragNum, levelDocNum))
|
---|
| 193 | FatalError (1, "could not convert fragment number %d", fragNum);
|
---|
| 194 |
|
---|
| 195 | if (levelDocNum == 0 || levelDocNum > numLevelDocs)
|
---|
| 196 | FatalError (1, "bad level document number %d", levelDocNum);
|
---|
| 197 |
|
---|
| 198 | if (levelDocNum != lastLevelDocNum) {
|
---|
| 199 | // new level document
|
---|
| 200 | if (lastLevelDocNum > 0) {
|
---|
| 201 | AddWeight (w, lastLevelDocNum, termFreq, idf);
|
---|
| 202 | checkLevelFreq++;
|
---|
| 203 | }
|
---|
| 204 | lastLevelDocNum = levelDocNum;
|
---|
| 205 | termFreq = 0;
|
---|
| 206 | }
|
---|
| 207 | termFreq += count;
|
---|
| 208 | }
|
---|
| 209 |
|
---|
| 210 | if (lastLevelDocNum > 0) {
|
---|
| 211 | AddWeight (w, lastLevelDocNum, termFreq, idf);
|
---|
| 212 | checkLevelFreq++;
|
---|
| 213 | }
|
---|
| 214 |
|
---|
| 215 | if (checkLevelFreq != wordEl.levelFreqs[levelNum]) {
|
---|
| 216 | cerr << "bad level freq " << checkLevelFreq << " != "
|
---|
| 217 | << wordEl.levelFreqs[levelNum] << ", word \""
|
---|
| 218 | << wordEl.el << "\" (" << wordNum << ")\n";
|
---|
| 219 | exit (1);
|
---|
| 220 | }
|
---|
| 221 |
|
---|
| 222 | buffer.done();
|
---|
| 223 | }
|
---|
| 224 |
|
---|
| 225 | if (w.size() != numLevelDocs)
|
---|
| 226 | FatalError (1, "wrong number of weights created %d != %d",
|
---|
| 227 | w.size(), numLevelDocs);
|
---|
| 228 | }
|
---|
| 229 |
|
---|
| 230 | static void WriteExactWeights (FILE *weightsFile,
|
---|
| 231 | unsigned long &diskPtr,
|
---|
| 232 | const Weights &w) {
|
---|
| 233 | diskPtr = ftell(weightsFile);
|
---|
| 234 |
|
---|
| 235 | Weights::const_iterator here = w.begin();
|
---|
| 236 | Weights::const_iterator end = w.end();
|
---|
| 237 | while (here != end) {
|
---|
| 238 | // cout << *here << "\n";
|
---|
| 239 | WriteF (weightsFile, *here);
|
---|
| 240 | here++;
|
---|
| 241 | }
|
---|
| 242 | }
|
---|
| 243 |
|
---|
| 244 | static void WriteApproxWeights (FILE *approxWeightsFile,
|
---|
| 245 | unsigned long &diskPtr,
|
---|
| 246 | const Weights &w,
|
---|
| 247 | unsigned char bits) {
|
---|
| 248 | diskPtr = ftell(approxWeightsFile);
|
---|
| 249 |
|
---|
| 250 | // calculate L, U and B
|
---|
| 251 | double L = 1e300;
|
---|
| 252 | double U = 0;
|
---|
| 253 | float wgt;
|
---|
| 254 | Weights::const_iterator here = w.begin();
|
---|
| 255 | Weights::const_iterator end = w.end();
|
---|
| 256 | while (here != end) {
|
---|
| 257 | wgt = sqrt (*here);
|
---|
| 258 | if (wgt > U) U = wgt;
|
---|
| 259 | if (wgt > 0 && wgt < L) L = wgt;
|
---|
| 260 | here++;
|
---|
| 261 | }
|
---|
| 262 |
|
---|
| 263 | double B = pow (U / L, pow (2.0, -(double) bits));
|
---|
| 264 |
|
---|
| 265 | fprintf (stderr, "L = %f\n", L);
|
---|
| 266 | fprintf (stderr, "U = %f\n", U);
|
---|
| 267 | fprintf (stderr, "B = %f\n", B);
|
---|
| 268 |
|
---|
| 269 | WriteUC (approxWeightsFile, bits);
|
---|
| 270 | WriteD (approxWeightsFile, L);
|
---|
| 271 | WriteD (approxWeightsFile, B);
|
---|
| 272 |
|
---|
| 273 |
|
---|
| 274 | unsigned long max = (bits == 32) ? 0xffffffff : (1 << bits) - 1;;
|
---|
| 275 | unsigned long i=0, buf=0, pos=0;
|
---|
| 276 | here = w.begin();
|
---|
| 277 | end = w.end();
|
---|
| 278 | while (here != end) {
|
---|
| 279 | unsigned long fx;
|
---|
| 280 | wgt = sqrt (*here);
|
---|
| 281 | if (wgt == 0) {
|
---|
| 282 | wgt = L;
|
---|
| 283 | #ifndef QUIET
|
---|
| 284 | Message ("Warning: Document %d had a weight of 0.", i);
|
---|
| 285 | #endif
|
---|
| 286 | }
|
---|
| 287 | fx = (unsigned long) floor (log (wgt / L) / log (B));
|
---|
| 288 |
|
---|
| 289 | if (fx > max) fx = max;
|
---|
| 290 |
|
---|
| 291 | buf |= (fx << pos);
|
---|
| 292 | pos += bits;
|
---|
| 293 |
|
---|
| 294 | if (pos >= MAXBITS) {
|
---|
| 295 | WriteUL (approxWeightsFile, buf);
|
---|
| 296 | buf = fx >> (bits - (pos - MAXBITS));
|
---|
| 297 | pos = pos - MAXBITS;
|
---|
| 298 | }
|
---|
| 299 |
|
---|
| 300 | here++; i++;
|
---|
| 301 | }
|
---|
| 302 |
|
---|
| 303 | // write out the last bits
|
---|
| 304 | if (pos > 0) WriteUL (approxWeightsFile, buf);
|
---|
| 305 | }
|
---|
| 306 |
|
---|
| 307 | int main (int argc, char **argv) {
|
---|
| 308 | unsigned char bits = 8;
|
---|
| 309 | char *filename = "";
|
---|
| 310 | int ch;
|
---|
| 311 | opterr = 0;
|
---|
| 312 | msg_prefix = argv[0];
|
---|
| 313 |
|
---|
| 314 | while ((ch = getopt (argc, argv, "f:d:b:h")) != -1) {
|
---|
| 315 | switch (ch) {
|
---|
| 316 | case 'f': // input file
|
---|
| 317 | filename = optarg;
|
---|
| 318 | break;
|
---|
| 319 | case 'd':
|
---|
| 320 | set_basepath (optarg);
|
---|
| 321 | break;
|
---|
| 322 | case 'b':
|
---|
| 323 | bits = atoi (optarg);
|
---|
| 324 | if (bits > 32) {
|
---|
| 325 | fprintf (stderr, "b may only take values 0-32\n");
|
---|
| 326 | exit (1);
|
---|
| 327 | }
|
---|
| 328 | break;
|
---|
| 329 | case 'h':
|
---|
| 330 | case '?':
|
---|
| 331 | fprintf (stderr, "usage: %s [-f input_file]"
|
---|
| 332 | "[-d data directory] [-b bits] [-h]\n", argv[0]);
|
---|
| 333 | exit (1);
|
---|
| 334 | }
|
---|
| 335 | }
|
---|
| 336 |
|
---|
| 337 |
|
---|
| 338 | // open the dictionary
|
---|
| 339 | FILE *dictFile = open_file (filename, INVF_DICT_SUFFIX, "rb",
|
---|
| 340 | MAGIC_STEM_BUILD, MG_ABORT);
|
---|
| 341 | invf_dict_header idh;
|
---|
| 342 | idh.Read (dictFile);
|
---|
| 343 |
|
---|
| 344 | // open the inverted file
|
---|
| 345 | FILE *invfFile = open_file (filename, INVF_SUFFIX, "rb",
|
---|
| 346 | MAGIC_INVF, MG_ABORT);
|
---|
| 347 | invf_file_header ifh;
|
---|
| 348 | ifh.Read (invfFile);
|
---|
| 349 | if (ifh.skip_mode != SKIP_MODE_NO_SKIPS)
|
---|
| 350 | FatalError (1, "The invf file contains skips. Unable to create weights.");
|
---|
| 351 |
|
---|
| 352 | // open the inverted index file
|
---|
| 353 | FILE *invfIdxFile = open_file (filename, INVF_IDX_SUFFIX, "rb",
|
---|
| 354 | MAGIC_INVI, MG_ABORT);
|
---|
| 355 |
|
---|
| 356 | // read the level information
|
---|
| 357 | FILE *levelFile = open_file (filename, INVF_LEVEL_SUFFIX, "rb",
|
---|
| 358 | MAGIC_INVF_LEVELS, MG_ABORT);
|
---|
| 359 | FIvfLevel ivfLevel;
|
---|
| 360 | ivfLevel.Read (levelFile);
|
---|
| 361 | fclose (levelFile);
|
---|
| 362 |
|
---|
| 363 | // read in the tag dictionary and inverted file pointers
|
---|
| 364 | WBTagDict tagDict;
|
---|
| 365 | ReadTagDict (idh, dictFile, invfIdxFile, tagDict);
|
---|
| 366 |
|
---|
| 367 |
|
---|
| 368 | // create the weights file
|
---|
| 369 | FILE *weightsFile = create_file (filename, WEIGHTS_SUFFIX, "wb",
|
---|
| 370 | MAGIC_WGHT, MG_ABORT);
|
---|
| 371 |
|
---|
| 372 | // create the approx weights file
|
---|
| 373 | FILE *approxWeightsFile = create_file (filename, APPROX_WEIGHTS_SUFFIX, "wb",
|
---|
| 374 | MAGIC_WGHT_APPROX, MG_ABORT);
|
---|
| 375 |
|
---|
| 376 |
|
---|
| 377 | // create weights for each document level
|
---|
| 378 | FragLevelConvert fragLevelConvert;
|
---|
| 379 | Weights w;
|
---|
| 380 | IvfLevelInfoMap::iterator levelHere = ivfLevel.levelInfo.begin();
|
---|
| 381 | IvfLevelInfoMap::iterator levelEnd = ivfLevel.levelInfo.end();
|
---|
| 382 | unsigned long levelNum = 0;
|
---|
| 383 | while (levelHere != levelEnd) {
|
---|
| 384 | const UCArray &levelName = (*levelHere).first;
|
---|
| 385 |
|
---|
| 386 | // read the tag information about this level
|
---|
| 387 | fragLevelConvert.Read(invfFile, tagDict[levelName].tagPtr,
|
---|
| 388 | idh.num_frags, tagDict[levelName].fragOccur);
|
---|
| 389 |
|
---|
| 390 | // create the weights for this level
|
---|
| 391 | GenerateLevelWeights (idh, ifh, tagDict[levelName].fragOccur,
|
---|
| 392 | levelNum, dictFile, invfFile, invfIdxFile,
|
---|
| 393 | fragLevelConvert, w);
|
---|
| 394 |
|
---|
| 395 | // write out the exact weights
|
---|
| 396 | WriteExactWeights (weightsFile,
|
---|
| 397 | (*levelHere).second.exactWeightsDiskPtr,
|
---|
| 398 | w);
|
---|
| 399 |
|
---|
| 400 | // write out the approximate weights
|
---|
| 401 | WriteApproxWeights (approxWeightsFile,
|
---|
| 402 | (*levelHere).second.approxWeightsDiskPtr,
|
---|
| 403 | w, bits);
|
---|
| 404 |
|
---|
| 405 | levelHere++; levelNum++;
|
---|
| 406 | }
|
---|
| 407 |
|
---|
| 408 |
|
---|
| 409 | // close input files
|
---|
| 410 | fclose (dictFile);
|
---|
| 411 | fclose (invfFile);
|
---|
| 412 | fclose (invfIdxFile);
|
---|
| 413 |
|
---|
| 414 | // update the level information
|
---|
| 415 | levelFile = create_file (filename, INVF_LEVEL_SUFFIX, "wb",
|
---|
| 416 | MAGIC_INVF_LEVELS, MG_ABORT);
|
---|
| 417 | ivfLevel.Write (levelFile);
|
---|
| 418 | fclose (levelFile);
|
---|
| 419 |
|
---|
| 420 | // close output files
|
---|
| 421 | fclose (weightsFile);
|
---|
| 422 | fclose (approxWeightsFile);
|
---|
| 423 |
|
---|
| 424 |
|
---|
| 425 | return 0;
|
---|
| 426 | }
|
---|