[3365] | 1 | /**************************************************************************
|
---|
| 2 | *
|
---|
| 3 | * mgpp_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 | **************************************************************************/
|
---|
| 21 |
|
---|
| 22 | #define _XOPEN_SOURCE 1
|
---|
[19822] | 23 | // This was added for Solaris, but it makes things worse on Solaris for me...
|
---|
| 24 | // #define _XOPEN_SOURCE_EXTENDED 1
|
---|
[3365] | 25 |
|
---|
| 26 | /* getopt is in posix.2, so cygwin should have it in unistd, but doesn't */
|
---|
| 27 | #if defined (__WIN32__) || defined (__CYGWIN__)
|
---|
| 28 | # include "getopt_old.h"
|
---|
| 29 | #else
|
---|
| 30 | # include <unistd.h>
|
---|
| 31 | #endif
|
---|
| 32 |
|
---|
| 33 | #include "UCArray.h"
|
---|
| 34 | #include "sysfuncs.h"
|
---|
| 35 | #include "memlib.h"
|
---|
| 36 | #include "messages.h"
|
---|
| 37 | #include "local_strings.h"
|
---|
| 38 | #include "bitio_gen.h"
|
---|
| 39 | #include "bitio_m_stdio.h"
|
---|
| 40 | #include "mg_files.h"
|
---|
| 41 | #include "locallib.h"
|
---|
| 42 | #include "invf.h"
|
---|
| 43 | #include "FIvfLevelInfo.h"
|
---|
| 44 | #include "FragLevelConvert.h"
|
---|
| 45 |
|
---|
| 46 | #if defined(GSDL_USE_OBJECTSPACE)
|
---|
| 47 | # include <ospace\std\map>
|
---|
| 48 | #elif defined(GSDL_USE_STL_H)
|
---|
| 49 | # include <map.h>
|
---|
| 50 | #else
|
---|
| 51 | # include <map>
|
---|
| 52 | #endif
|
---|
| 53 |
|
---|
[25147] | 54 | #define MAXBITS (sizeof(mg_u_long) * 8)
|
---|
[3365] | 55 |
|
---|
| 56 | struct WBTagPtr {
|
---|
[25147] | 57 | mg_u_long tagNum;
|
---|
| 58 | mg_u_long tagPtr;
|
---|
| 59 | mg_u_long fragOccur;
|
---|
[3365] | 60 |
|
---|
| 61 | WBTagPtr () {
|
---|
| 62 | tagNum = 0;
|
---|
| 63 | tagPtr = 0;
|
---|
| 64 | fragOccur = 0;
|
---|
| 65 | }
|
---|
| 66 | };
|
---|
| 67 |
|
---|
| 68 | // maps tags to tag information
|
---|
| 69 | typedef map<UCArray, WBTagPtr, DictLTUCArray> WBTagDict;
|
---|
| 70 |
|
---|
| 71 | typedef vector<float> Weights;
|
---|
| 72 |
|
---|
| 73 |
|
---|
| 74 | static void ReadTagDict (const invf_dict_header &idh,
|
---|
| 75 | FILE *dictFile,
|
---|
| 76 | FILE *invfIdxFile,
|
---|
| 77 | WBTagDict &tagDict) {
|
---|
| 78 | tagDict.erase (tagDict.begin(), tagDict.end());
|
---|
| 79 |
|
---|
| 80 | // seek to the start of the tag information
|
---|
| 81 | fseek (dictFile, idh.tag_dict_start, SEEK_SET);
|
---|
[25147] | 82 | fseek (invfIdxFile, sizeof(mg_u_long) +
|
---|
| 83 | idh.word_dict_size*sizeof(mg_u_long), SEEK_SET);
|
---|
[3365] | 84 |
|
---|
[25147] | 85 | mg_u_long tagNum;
|
---|
| 86 | mg_u_long tagPtr;
|
---|
[3365] | 87 | dict_el thisEl;
|
---|
[9613] | 88 | for (tagNum = 0; tagNum < idh.tag_dict_size; ++tagNum) {
|
---|
[3365] | 89 | thisEl.Read (dictFile);
|
---|
| 90 | ReadUL (invfIdxFile, tagPtr);
|
---|
| 91 | tagDict[thisEl.el].tagNum = tagNum;
|
---|
| 92 | tagDict[thisEl.el].tagPtr = tagPtr;
|
---|
| 93 | tagDict[thisEl.el].fragOccur = thisEl.frag_occur;
|
---|
| 94 | }
|
---|
| 95 | }
|
---|
| 96 |
|
---|
| 97 | static void AddWeight (Weights &w,
|
---|
[25147] | 98 | mg_u_long levelDocNum,
|
---|
| 99 | mg_u_long termFreq,
|
---|
[3365] | 100 | float idf) {
|
---|
| 101 | double weight = termFreq * idf;
|
---|
| 102 | w[levelDocNum-1] += weight * weight;
|
---|
| 103 | }
|
---|
| 104 |
|
---|
| 105 | static void GenerateLevelWeights (const invf_dict_header &idh,
|
---|
| 106 | const invf_file_header &ifh,
|
---|
[25147] | 107 | mg_u_long numLevelDocs,
|
---|
| 108 | mg_u_long levelNum,
|
---|
[3365] | 109 | FILE *dictFile,
|
---|
| 110 | FILE *invfFile,
|
---|
| 111 | FILE *invfIdxFile,
|
---|
| 112 | const FragLevelConvert &fragLevelConvert,
|
---|
| 113 | Weights &w) {
|
---|
| 114 | // pre-allocate the right number of weights
|
---|
| 115 | w.erase (w.begin(), w.end());
|
---|
| 116 | w.insert (w.end(), (Weights::size_type)numLevelDocs, (float)0.0);
|
---|
| 117 |
|
---|
| 118 | double logN = log ((double) numLevelDocs);
|
---|
| 119 |
|
---|
| 120 | // reset the files
|
---|
| 121 | fseek (dictFile, idh.word_dict_start, SEEK_SET);
|
---|
[25147] | 122 | fseek (invfIdxFile, sizeof (mg_u_long), SEEK_SET);
|
---|
[3365] | 123 |
|
---|
| 124 | // process each word adding its contributions to the document weights
|
---|
[25147] | 125 | mg_u_long wordNum;
|
---|
| 126 | mg_u_long wordStart;
|
---|
[3365] | 127 | word_dict_el wordEl;
|
---|
| 128 | wordEl.SetNumLevels (idh.num_levels);
|
---|
[9613] | 129 | for (wordNum=0; wordNum<idh.word_dict_size; ++wordNum) {
|
---|
[3365] | 130 | #ifndef SILENT
|
---|
| 131 | // give a little feedback every 4096 words
|
---|
| 132 | if ((wordNum & 0xfff) == 0) fprintf (stderr, ".");
|
---|
| 133 | #endif
|
---|
| 134 |
|
---|
| 135 | wordEl.Read (dictFile, idh.num_levels);
|
---|
| 136 | ReadUL (invfIdxFile, wordStart);
|
---|
| 137 |
|
---|
| 138 | float idf = logN - log ((double) wordEl.levelFreqs[levelNum]);
|
---|
| 139 |
|
---|
| 140 | // seek to the appropriate place in the inverted file
|
---|
| 141 | fseek (invfFile, wordStart, SEEK_SET);
|
---|
| 142 | stdio_bitio_buffer buffer (invfFile);
|
---|
| 143 |
|
---|
[25147] | 144 | mg_u_long B = BIO_Bblock_Init (idh.num_frags, wordEl.frag_occur);
|
---|
| 145 | mg_u_long fragNum = 0;
|
---|
| 146 | mg_u_long levelDocNum = 0;
|
---|
| 147 | mg_u_long lastLevelDocNum = 0;
|
---|
| 148 | mg_u_long termFreq = 0;
|
---|
| 149 | mg_u_long checkLevelFreq = 0;
|
---|
[3365] | 150 |
|
---|
[25147] | 151 | mg_u_long count, i;
|
---|
[9613] | 152 | for (i=0; i<wordEl.frag_occur; ++i) {
|
---|
[3365] | 153 | fragNum += buffer.bblock_decode (B, NULL);
|
---|
| 154 | if (!ifh.word_level_index) count = buffer.gamma_decode (NULL);
|
---|
| 155 | else count = 1;
|
---|
| 156 |
|
---|
| 157 | if (fragNum > idh.num_frags)
|
---|
| 158 | FatalError (1, "fragNum = %d, "
|
---|
| 159 | "number of fragments = %d\n"
|
---|
| 160 | "wordNum = %d\n"
|
---|
| 161 | "i = %d, frag_occur = %d\n",
|
---|
| 162 | fragNum, idh.num_frags, wordNum, i, wordEl.frag_occur);
|
---|
| 163 |
|
---|
| 164 | if (!fragLevelConvert.FragToLevel (fragNum, levelDocNum))
|
---|
| 165 | FatalError (1, "could not convert fragment number %d in level %d", fragNum, levelNum);
|
---|
| 166 |
|
---|
| 167 | if (levelDocNum == 0 || levelDocNum > numLevelDocs)
|
---|
| 168 | FatalError (1, "bad level document number %d in level %d", levelDocNum, levelNum);
|
---|
| 169 |
|
---|
| 170 | if (levelDocNum != lastLevelDocNum) {
|
---|
| 171 | // new level document
|
---|
| 172 | if (lastLevelDocNum > 0) {
|
---|
| 173 | AddWeight (w, lastLevelDocNum, termFreq, idf);
|
---|
[9613] | 174 | ++checkLevelFreq;
|
---|
[3365] | 175 | }
|
---|
| 176 | lastLevelDocNum = levelDocNum;
|
---|
| 177 | termFreq = 0;
|
---|
| 178 | }
|
---|
| 179 | termFreq += count;
|
---|
| 180 | }
|
---|
| 181 |
|
---|
| 182 | if (lastLevelDocNum > 0) {
|
---|
| 183 | AddWeight (w, lastLevelDocNum, termFreq, idf);
|
---|
[9613] | 184 | ++checkLevelFreq;
|
---|
[3365] | 185 | }
|
---|
| 186 |
|
---|
| 187 | if (checkLevelFreq != wordEl.levelFreqs[levelNum]) {
|
---|
| 188 | cerr << "bad level freq at level " <<levelNum<<" "<< checkLevelFreq << " != "
|
---|
| 189 | << wordEl.levelFreqs[levelNum] << ", word \""
|
---|
| 190 | << wordEl.el << "\" (" << wordNum << ")\n";
|
---|
| 191 | exit (1);
|
---|
| 192 | }
|
---|
| 193 |
|
---|
| 194 | buffer.done();
|
---|
| 195 | }
|
---|
| 196 |
|
---|
| 197 | if (w.size() != numLevelDocs)
|
---|
| 198 | FatalError (1, "wrong number of weights created %d != %d",
|
---|
| 199 | w.size(), numLevelDocs);
|
---|
| 200 | }
|
---|
| 201 |
|
---|
| 202 | static void WriteExactWeights (FILE *weightsFile,
|
---|
[25147] | 203 | mg_u_long &diskPtr,
|
---|
[3365] | 204 | const Weights &w) {
|
---|
| 205 | diskPtr = ftell(weightsFile);
|
---|
| 206 |
|
---|
| 207 | Weights::const_iterator here = w.begin();
|
---|
| 208 | Weights::const_iterator end = w.end();
|
---|
| 209 | while (here != end) {
|
---|
| 210 | // cout << *here << "\n";
|
---|
| 211 | WriteF (weightsFile, sqrt (*here));
|
---|
[9613] | 212 | ++here;
|
---|
[3365] | 213 | }
|
---|
| 214 | }
|
---|
| 215 |
|
---|
| 216 | static void WriteApproxWeights (FILE *approxWeightsFile,
|
---|
[25147] | 217 | mg_u_long &diskPtr,
|
---|
[3365] | 218 | const Weights &w,
|
---|
| 219 | unsigned char bits) {
|
---|
| 220 | diskPtr = ftell(approxWeightsFile);
|
---|
| 221 |
|
---|
| 222 | // calculate L, U and B
|
---|
| 223 | double L = 1e300;
|
---|
| 224 | double U = 0;
|
---|
| 225 | float wgt;
|
---|
| 226 | Weights::const_iterator here = w.begin();
|
---|
| 227 | Weights::const_iterator end = w.end();
|
---|
| 228 | while (here != end) {
|
---|
| 229 | wgt = sqrt (*here);
|
---|
| 230 | if (wgt > U) U = wgt;
|
---|
| 231 | if (wgt > 0 && wgt < L) L = wgt;
|
---|
[9613] | 232 | ++here;
|
---|
[3365] | 233 | }
|
---|
| 234 |
|
---|
| 235 | double B = pow (U / L, pow (2.0, -(double) bits));
|
---|
| 236 |
|
---|
| 237 | #ifndef SILENT
|
---|
| 238 | fprintf (stderr, "L = %f\n", L);
|
---|
| 239 | fprintf (stderr, "U = %f\n", U);
|
---|
| 240 | fprintf (stderr, "B = %f\n", B);
|
---|
| 241 | #endif
|
---|
| 242 |
|
---|
| 243 | WriteUC (approxWeightsFile, bits);
|
---|
| 244 | WriteD (approxWeightsFile, L);
|
---|
| 245 | WriteD (approxWeightsFile, B);
|
---|
| 246 |
|
---|
| 247 |
|
---|
[25147] | 248 | mg_u_long max = (bits == 32) ? 0xffffffff : (1 << bits) - 1;;
|
---|
| 249 | mg_u_long i=0, buf=0, pos=0;
|
---|
[3365] | 250 | here = w.begin();
|
---|
| 251 | end = w.end();
|
---|
| 252 | while (here != end) {
|
---|
[25147] | 253 | mg_u_long fx;
|
---|
[3365] | 254 | wgt = sqrt (*here);
|
---|
| 255 | if (wgt == 0) {
|
---|
| 256 | wgt = L;
|
---|
| 257 | #ifndef SILENT
|
---|
| 258 | Message ("Warning: Document %d had a weight of 0.", i);
|
---|
| 259 | #endif
|
---|
| 260 | }
|
---|
[25147] | 261 | fx = (mg_u_long) floor (log (wgt / L) / log (B));
|
---|
[3365] | 262 |
|
---|
| 263 | if (fx > max) fx = max;
|
---|
| 264 |
|
---|
| 265 | buf |= (fx << pos);
|
---|
| 266 | pos += bits;
|
---|
| 267 |
|
---|
| 268 | if (pos >= MAXBITS) {
|
---|
| 269 | WriteUL (approxWeightsFile, buf);
|
---|
| 270 | buf = fx >> (bits - (pos - MAXBITS));
|
---|
| 271 | pos = pos - MAXBITS;
|
---|
| 272 | }
|
---|
| 273 |
|
---|
[9613] | 274 | ++here; ++i;
|
---|
[3365] | 275 | }
|
---|
| 276 |
|
---|
| 277 | // write out the last bits
|
---|
| 278 | if (pos > 0) WriteUL (approxWeightsFile, buf);
|
---|
| 279 | }
|
---|
| 280 |
|
---|
| 281 | int main (int argc, char **argv) {
|
---|
| 282 | unsigned char bits = 8;
|
---|
[25147] | 283 | char *filename = (char*)"";
|
---|
[3365] | 284 | int ch;
|
---|
| 285 | opterr = 0;
|
---|
| 286 | msg_prefix = argv[0];
|
---|
| 287 |
|
---|
| 288 | while ((ch = getopt (argc, argv, "f:d:b:h")) != -1) {
|
---|
| 289 | switch (ch) {
|
---|
| 290 | case 'f': // input file
|
---|
| 291 | filename = optarg;
|
---|
| 292 | break;
|
---|
| 293 | case 'd':
|
---|
| 294 | set_basepath (optarg);
|
---|
| 295 | break;
|
---|
| 296 | case 'b':
|
---|
| 297 | bits = atoi (optarg);
|
---|
| 298 | if (bits > 32) {
|
---|
| 299 | fprintf (stderr, "b may only take values 0-32\n");
|
---|
| 300 | exit (1);
|
---|
| 301 | }
|
---|
| 302 | break;
|
---|
| 303 | case 'h':
|
---|
| 304 | case '?':
|
---|
| 305 | fprintf (stderr, "usage: %s [-f input_file]"
|
---|
| 306 | "[-d data directory] [-b bits] [-h]\n", argv[0]);
|
---|
| 307 | exit (1);
|
---|
| 308 | }
|
---|
| 309 | }
|
---|
| 310 |
|
---|
| 311 |
|
---|
| 312 | // open the dictionary
|
---|
| 313 | FILE *dictFile = open_file (filename, INVF_DICT_SUFFIX, "rb",
|
---|
| 314 | MAGIC_STEM_BUILD, MG_ABORT);
|
---|
| 315 | invf_dict_header idh;
|
---|
| 316 | idh.Read (dictFile);
|
---|
| 317 |
|
---|
| 318 | // open the inverted file
|
---|
| 319 | FILE *invfFile = open_file (filename, INVF_SUFFIX, "rb",
|
---|
| 320 | MAGIC_INVF, MG_ABORT);
|
---|
| 321 | invf_file_header ifh;
|
---|
| 322 | ifh.Read (invfFile);
|
---|
| 323 | if (ifh.skip_mode != SKIP_MODE_NO_SKIPS)
|
---|
| 324 | FatalError (1, "The invf file contains skips. Unable to create weights.");
|
---|
| 325 |
|
---|
| 326 | // open the inverted index file
|
---|
| 327 | FILE *invfIdxFile = open_file (filename, INVF_IDX_SUFFIX, "rb",
|
---|
| 328 | MAGIC_INVI, MG_ABORT);
|
---|
| 329 |
|
---|
| 330 | // read the level information
|
---|
| 331 | FILE *levelFile = open_file (filename, INVF_LEVEL_SUFFIX, "rb",
|
---|
| 332 | MAGIC_INVF_LEVELS, MG_ABORT);
|
---|
| 333 | FIvfLevel ivfLevel;
|
---|
| 334 | ivfLevel.Read (levelFile);
|
---|
| 335 | fclose (levelFile);
|
---|
| 336 |
|
---|
| 337 | // read in the tag dictionary and inverted file pointers
|
---|
| 338 | WBTagDict tagDict;
|
---|
| 339 | ReadTagDict (idh, dictFile, invfIdxFile, tagDict);
|
---|
| 340 |
|
---|
| 341 |
|
---|
| 342 | // create the weights file
|
---|
| 343 | FILE *weightsFile = create_file (filename, WEIGHTS_SUFFIX, "wb",
|
---|
| 344 | MAGIC_WGHT, MG_ABORT);
|
---|
| 345 |
|
---|
| 346 | // create the approx weights file
|
---|
| 347 | FILE *approxWeightsFile = create_file (filename, APPROX_WEIGHTS_SUFFIX, "wb",
|
---|
| 348 | MAGIC_WGHT_APPROX, MG_ABORT);
|
---|
| 349 |
|
---|
| 350 |
|
---|
| 351 | // create weights for each document level
|
---|
| 352 | FragLevelConvert fragLevelConvert;
|
---|
| 353 | Weights w;
|
---|
| 354 | IvfLevelInfoMap::iterator levelHere = ivfLevel.levelInfo.begin();
|
---|
| 355 | IvfLevelInfoMap::iterator levelEnd = ivfLevel.levelInfo.end();
|
---|
[25147] | 356 | mg_u_long levelNum = 0;
|
---|
[3365] | 357 | while (levelHere != levelEnd) {
|
---|
| 358 | const UCArray &levelName = (*levelHere).first;
|
---|
| 359 |
|
---|
| 360 | // read the tag information about this level
|
---|
| 361 | fragLevelConvert.Read(invfFile, tagDict[levelName].tagPtr,
|
---|
| 362 | idh.num_frags, tagDict[levelName].fragOccur);
|
---|
| 363 |
|
---|
| 364 | // create the weights for this level
|
---|
| 365 | GenerateLevelWeights (idh, ifh, tagDict[levelName].fragOccur,
|
---|
| 366 | levelNum, dictFile, invfFile, invfIdxFile,
|
---|
| 367 | fragLevelConvert, w);
|
---|
| 368 |
|
---|
| 369 | // write out the exact weights
|
---|
| 370 | WriteExactWeights (weightsFile,
|
---|
| 371 | (*levelHere).second.exactWeightsDiskPtr,
|
---|
| 372 | w);
|
---|
| 373 |
|
---|
| 374 | // write out the approximate weights
|
---|
| 375 | WriteApproxWeights (approxWeightsFile,
|
---|
| 376 | (*levelHere).second.approxWeightsDiskPtr,
|
---|
| 377 | w, bits);
|
---|
| 378 |
|
---|
[9613] | 379 | ++levelHere; ++levelNum;
|
---|
[3365] | 380 | }
|
---|
| 381 |
|
---|
| 382 |
|
---|
| 383 | // close input files
|
---|
| 384 | fclose (dictFile);
|
---|
| 385 | fclose (invfFile);
|
---|
| 386 | fclose (invfIdxFile);
|
---|
| 387 |
|
---|
| 388 | // update the level information
|
---|
| 389 | levelFile = create_file (filename, INVF_LEVEL_SUFFIX, "wb",
|
---|
| 390 | MAGIC_INVF_LEVELS, MG_ABORT);
|
---|
| 391 | ivfLevel.Write (levelFile);
|
---|
| 392 | fclose (levelFile);
|
---|
| 393 |
|
---|
| 394 | // close output files
|
---|
| 395 | fclose (weightsFile);
|
---|
| 396 | fclose (approxWeightsFile);
|
---|
| 397 |
|
---|
| 398 |
|
---|
| 399 | return 0;
|
---|
| 400 | }
|
---|