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