1 | /**********************************************************************
|
---|
2 | *
|
---|
3 | * mgsearch.cpp --
|
---|
4 | * Copyright (C) 1999 The New Zealand Digital Library Project
|
---|
5 | *
|
---|
6 | * A component of the Greenstone digital library software
|
---|
7 | * from the New Zealand Digital Library Project at the
|
---|
8 | * University of Waikato, New Zealand.
|
---|
9 | *
|
---|
10 | * This program is free software; you can redistribute it and/or modify
|
---|
11 | * it under the terms of the GNU General Public License as published by
|
---|
12 | * the Free Software Foundation; either version 2 of the License, or
|
---|
13 | * (at your option) any later version.
|
---|
14 | *
|
---|
15 | * This program is distributed in the hope that it will be useful,
|
---|
16 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
17 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
18 | * GNU General Public License for more details.
|
---|
19 | *
|
---|
20 | * You should have received a copy of the GNU General Public License
|
---|
21 | * along with this program; if not, write to the Free Software
|
---|
22 | * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
23 | *
|
---|
24 | *********************************************************************/
|
---|
25 |
|
---|
26 | #include "gsdlconf.h"
|
---|
27 | #include "mgsearch.h"
|
---|
28 | #include "fileutil.h"
|
---|
29 |
|
---|
30 | #include <string.h>
|
---|
31 | #include <stdio.h>
|
---|
32 | #include <stdlib.h>
|
---|
33 | #include <ctype.h>
|
---|
34 |
|
---|
35 | #if defined(GSDL_USE_OBJECTSPACE)
|
---|
36 | # include <ospace\std\iostream>
|
---|
37 | #elif defined(GSDL_USE_IOS_H)
|
---|
38 | # include <iostream.h>
|
---|
39 | #else
|
---|
40 | # include <iostream>
|
---|
41 | #endif
|
---|
42 |
|
---|
43 | #if defined(__WIN32__)
|
---|
44 | // gdbm stuff
|
---|
45 | # include "autoconf.h"
|
---|
46 | # include "systems.h"
|
---|
47 | # include "gdbmconst.h"
|
---|
48 | # include "gdbm.h"
|
---|
49 | #else
|
---|
50 | # include <gdbm.h>
|
---|
51 | #endif
|
---|
52 |
|
---|
53 |
|
---|
54 | #include <assert.h>
|
---|
55 |
|
---|
56 | #include "mgq.h"
|
---|
57 | // #include "locateinfo.h"
|
---|
58 | #include "gsdlunicode.h"
|
---|
59 | #include "unitool.h"
|
---|
60 |
|
---|
61 |
|
---|
62 | /////////////
|
---|
63 | // globals //
|
---|
64 | /////////////
|
---|
65 |
|
---|
66 | static char *tempdoc = NULL;
|
---|
67 | static int templen = 0;
|
---|
68 |
|
---|
69 |
|
---|
70 | //////////////////////
|
---|
71 | // useful functions //
|
---|
72 | //////////////////////
|
---|
73 |
|
---|
74 |
|
---|
75 | // input and output are in utf8
|
---|
76 | text_t mgsearch_stemword (const text_t &word) {
|
---|
77 | // allocate working stem space
|
---|
78 | int maxstemlen = mgq_getmaxstemlen ();
|
---|
79 | unsigned char *word_stem = new unsigned char [maxstemlen + 2];
|
---|
80 | if (word_stem == NULL) return "";
|
---|
81 |
|
---|
82 | // copy word to word_stem
|
---|
83 | int len = 0;
|
---|
84 | text_t::const_iterator here = word.begin();
|
---|
85 | text_t::const_iterator end = word.end();
|
---|
86 | while (len < maxstemlen && here != end) {
|
---|
87 | word_stem[len+1] = (unsigned char)(*here);
|
---|
88 | len++; here++;
|
---|
89 | }
|
---|
90 | word_stem[len+1] = '\0';
|
---|
91 | word_stem[0] = len;
|
---|
92 |
|
---|
93 | mgq_stemword (word_stem);
|
---|
94 |
|
---|
95 | // copy word_stem back to tempstr
|
---|
96 | text_t tempstr;
|
---|
97 | tempstr.setcarr((char *)(&word_stem[1]), word_stem[0]);
|
---|
98 |
|
---|
99 | delete [] word_stem;
|
---|
100 |
|
---|
101 | return tempstr;
|
---|
102 | }
|
---|
103 |
|
---|
104 |
|
---|
105 |
|
---|
106 | ////////////////////////
|
---|
107 | // callback functions //
|
---|
108 | ////////////////////////
|
---|
109 |
|
---|
110 | // This routine is called for each document found in a search
|
---|
111 | // it assumes that cache_num is set up correctly to point to
|
---|
112 | // a suitable result cache
|
---|
113 | int ourquerycallback(char * /*UDoc*/, int /*ULen*/, int DocNum,
|
---|
114 | float Weight, void *info) {
|
---|
115 |
|
---|
116 |
|
---|
117 | queryresultsclass *queryresults = (queryresultsclass * )info;
|
---|
118 |
|
---|
119 | // append this entry to the document results
|
---|
120 | docresultclass docresult;
|
---|
121 | docresult.docnum = DocNum;
|
---|
122 | docresult.num_query_terms_matched = (int)(Weight/100.0); // will always be 0 on some versions of mg...
|
---|
123 | docresult.docweight = Weight - docresult.num_query_terms_matched*100;
|
---|
124 |
|
---|
125 | queryresults->docs.docset[DocNum] = docresult;
|
---|
126 | queryresults->docs.docorder.push_back(DocNum);
|
---|
127 |
|
---|
128 | return 0;
|
---|
129 | }
|
---|
130 |
|
---|
131 | int termequivcallback(char *Word, int ULen, int /*Freq*/,
|
---|
132 | float /*Weight*/, void *info) {
|
---|
133 | text_tset *equivterms = (text_tset *)info;
|
---|
134 | if (equivterms == NULL) return 0;
|
---|
135 |
|
---|
136 | text_t thisterm;
|
---|
137 | thisterm.setcarr(Word, ULen);
|
---|
138 |
|
---|
139 | equivterms->insert(thisterm);
|
---|
140 |
|
---|
141 | return 0;
|
---|
142 | }
|
---|
143 |
|
---|
144 |
|
---|
145 | void mgsearch_equivterms (const text_t &word, text_tset &equivterms) {
|
---|
146 | // allocate working stem space
|
---|
147 | int maxstemlen = mgq_getmaxstemlen ();
|
---|
148 | unsigned char *word_stem = new unsigned char [maxstemlen + 2];
|
---|
149 | if (word_stem == NULL) return;
|
---|
150 |
|
---|
151 | // copy word to word_stem
|
---|
152 | int len = 0;
|
---|
153 | text_t::const_iterator here = word.begin();
|
---|
154 | text_t::const_iterator end = word.end();
|
---|
155 | while (len < maxstemlen && here != end) {
|
---|
156 | word_stem[len+1] = (unsigned char)(*here);
|
---|
157 | len++; here++;
|
---|
158 | }
|
---|
159 | word_stem[len+1] = '\0';
|
---|
160 | word_stem[0] = len;
|
---|
161 |
|
---|
162 | // get the equivalent terms
|
---|
163 | mgq_equivterms (word_stem, termequivcallback, (void *)(&equivterms));
|
---|
164 |
|
---|
165 | delete [] word_stem;
|
---|
166 |
|
---|
167 | return;
|
---|
168 | }
|
---|
169 |
|
---|
170 | text_tset utf8equivterms; // kept as utf8 string for fast matching
|
---|
171 |
|
---|
172 |
|
---|
173 | // This callback is called once for each term in the query
|
---|
174 | int termfreqcallback(char *Word, int ULen, int Freq,
|
---|
175 | float /*Weight*/, void *info) {
|
---|
176 | queryresultsclass *queryresults = (queryresultsclass *)info;
|
---|
177 | if (queryresults == NULL) return 0;
|
---|
178 |
|
---|
179 | text_t term;
|
---|
180 | term.setcarr(Word, ULen);
|
---|
181 | termfreqclass termfreq;
|
---|
182 |
|
---|
183 | termfreq.termstr = to_uni(term);
|
---|
184 | text_t utf8termstem = mgsearch_stemword (term);
|
---|
185 | termfreq.termstemstr = to_uni (utf8termstem);
|
---|
186 |
|
---|
187 | mgsearch_equivterms (utf8termstem, termfreq.utf8equivterms);
|
---|
188 |
|
---|
189 | termfreq.termfreq = Freq;
|
---|
190 | queryresults->orgterms.push_back(termfreq);
|
---|
191 |
|
---|
192 | return 0;
|
---|
193 | }
|
---|
194 |
|
---|
195 | // this callback is called once for each variation of each term
|
---|
196 | int termvariantscallback(char *Word, int ULen, int /*Freq*/,
|
---|
197 | float /*Weight*/, void *info) {
|
---|
198 |
|
---|
199 | text_t term;
|
---|
200 | term.setcarr(Word, ULen);
|
---|
201 | queryresultsclass *queryresults = (queryresultsclass *)info;
|
---|
202 | queryresults->termvariants.insert(to_uni(term));
|
---|
203 |
|
---|
204 | return 0;
|
---|
205 | }
|
---|
206 |
|
---|
207 | // This callback is for getting document text
|
---|
208 | int doctextcallback(char *Doc, int ULen, int /*Freq*/,
|
---|
209 | float /*Weight*/, void * /*info*/) {
|
---|
210 | tempdoc = Doc;
|
---|
211 | templen = ULen;
|
---|
212 |
|
---|
213 | return 0;
|
---|
214 | }
|
---|
215 |
|
---|
216 |
|
---|
217 | static text_t getindexsuffix (const text_t &collection,
|
---|
218 | const text_t &index) {
|
---|
219 |
|
---|
220 | text_t indexsuffix = "index";
|
---|
221 | indexsuffix = filename_cat (indexsuffix, index);
|
---|
222 | indexsuffix = filename_cat (indexsuffix, collection);
|
---|
223 | return indexsuffix;
|
---|
224 | }
|
---|
225 |
|
---|
226 |
|
---|
227 |
|
---|
228 |
|
---|
229 | ////////////////////
|
---|
230 | // mgsearch class //
|
---|
231 | ////////////////////
|
---|
232 |
|
---|
233 | mgsearchclass::mgsearchclass ()
|
---|
234 | : searchclass() {
|
---|
235 |
|
---|
236 | }
|
---|
237 |
|
---|
238 | mgsearchclass::~mgsearchclass ()
|
---|
239 | {
|
---|
240 | if (cache != NULL)
|
---|
241 | {
|
---|
242 | delete cache;
|
---|
243 | cache = NULL;
|
---|
244 | }
|
---|
245 | }
|
---|
246 |
|
---|
247 | // you only need to use this function before doing any stemming
|
---|
248 | // casefolding and stemming will be set if values for them are
|
---|
249 | // provided (0 or 1).
|
---|
250 | // makeindexcurrent returns true if it was able to load the database
|
---|
251 | bool mgsearchclass::makeindexcurrent (const text_t &index,
|
---|
252 | const text_t &subcollection,
|
---|
253 | const text_t &language,
|
---|
254 | const text_t &collection,
|
---|
255 | int casefolding,
|
---|
256 | int stemming) {
|
---|
257 | bool databaseloaded = true;
|
---|
258 |
|
---|
259 | // get the names of the collection, index and text suffixes
|
---|
260 | char *ccollection = collection.getcstr();
|
---|
261 | assert (ccollection != NULL);
|
---|
262 | char *idxsuffix = (getindexsuffix (collection, (index+subcollection+language))).getcstr();
|
---|
263 | assert (idxsuffix != NULL);
|
---|
264 | char *txtsuffix = (getindexsuffix (collection, "text")).getcstr();
|
---|
265 | assert (txtsuffix != NULL);
|
---|
266 |
|
---|
267 | #ifdef __WIN32__
|
---|
268 | char *ccollectdir = (collectdir+"\\").getcstr(); assert (ccollectdir != NULL);
|
---|
269 | #else
|
---|
270 | char *ccollectdir = collectdir.getcstr(); assert (ccollectdir != NULL);
|
---|
271 | #endif
|
---|
272 |
|
---|
273 | if (load_database(ccollection, ccollectdir, idxsuffix, txtsuffix)) {
|
---|
274 | if (casefolding == 0) mgq_ask(".set casefold off");
|
---|
275 | else if (casefolding > 0) mgq_ask(".set casefold on");
|
---|
276 | if (stemming == 0) mgq_ask(".set stem off");
|
---|
277 | else if (stemming > 0) mgq_ask(".set stem on");
|
---|
278 |
|
---|
279 | } else databaseloaded = false;
|
---|
280 |
|
---|
281 | // free up the c strings
|
---|
282 | delete ccollection;
|
---|
283 | delete idxsuffix;
|
---|
284 | delete txtsuffix;
|
---|
285 | delete ccollectdir;
|
---|
286 |
|
---|
287 | return databaseloaded;
|
---|
288 | }
|
---|
289 |
|
---|
290 |
|
---|
291 | // stem word uses the values set in the last call to makeindexcurrent
|
---|
292 | // to stem the word. It is assumed that word is in unicode
|
---|
293 | text_t mgsearchclass::stemword (const text_t &word) {
|
---|
294 | return to_uni (mgsearch_stemword (to_utf8 (word)));
|
---|
295 | }
|
---|
296 |
|
---|
297 | text_t mgsearchclass::stemword (text_t::const_iterator here, text_t::const_iterator end) {
|
---|
298 | return to_uni (mgsearch_stemword (to_utf8 (here, end)));
|
---|
299 | }
|
---|
300 |
|
---|
301 |
|
---|
302 | bool mgsearchclass::search(const queryparamclass &queryparams,
|
---|
303 | queryresultsclass &queryresults) {
|
---|
304 | // assert (cache != NULL);
|
---|
305 |
|
---|
306 | queryresults.clear();
|
---|
307 | cerr << "mgsearch start of search"<<endl;
|
---|
308 | // first check the cache
|
---|
309 | if (cache != NULL) {
|
---|
310 | if (cache->find(queryparams, queryresults)) return true;
|
---|
311 | }
|
---|
312 | // make sure there is a query to be processed
|
---|
313 | if (!has_unicode_letdig(queryparams.querystring)) return true;
|
---|
314 |
|
---|
315 | if (makeindexcurrent (queryparams.index, queryparams.subcollection,
|
---|
316 | queryparams.language, queryparams.collection)) {
|
---|
317 | cerr << "made index current "<<endl;
|
---|
318 | setsearchmode (queryparams);
|
---|
319 | submitquery (queryparams);
|
---|
320 | getresults (queryparams, queryresults);
|
---|
321 | cerr << "got results"<<endl;
|
---|
322 | return true;
|
---|
323 | }
|
---|
324 |
|
---|
325 | return false;
|
---|
326 | }
|
---|
327 |
|
---|
328 |
|
---|
329 | void mgsearchclass::setsearchmode (const queryparamclass &queryparams)
|
---|
330 | {
|
---|
331 | mgq_ask(".set expert true");
|
---|
332 | mgq_ask(".set sorted_terms true");
|
---|
333 | mgq_ask(".set accumulator_method list");
|
---|
334 | mgq_ask(".set max_accumulators 500000");
|
---|
335 | mgq_ask(".set maxparas 500000");
|
---|
336 | mgq_ask(".set verbatim true");
|
---|
337 | mgq_ask(".unset skip_dump");
|
---|
338 | mgq_ask(".set mode docnums");
|
---|
339 |
|
---|
340 | switch (queryparams.search_type)
|
---|
341 | {
|
---|
342 | case 0: mgq_ask(".set query boolean"); break;
|
---|
343 | case 1: mgq_ask(".set query ranked"); break;
|
---|
344 | }
|
---|
345 | switch (queryparams.casefolding)
|
---|
346 | {
|
---|
347 | case 1: mgq_ask(".set casefold on"); break;
|
---|
348 | case 0: mgq_ask(".set casefold off"); break;
|
---|
349 | }
|
---|
350 | switch (queryparams.stemming)
|
---|
351 | {
|
---|
352 | case 1: mgq_ask(".set stem on"); break;
|
---|
353 | case 0: mgq_ask(".set stem off"); break;
|
---|
354 | }
|
---|
355 | mgq_ask(".set heads_length 150");
|
---|
356 |
|
---|
357 | if (queryparams.maxdocs == -1) {
|
---|
358 | mgq_ask(".set maxdocs all");
|
---|
359 | } else {
|
---|
360 | char maxdocstr[32];
|
---|
361 | sprintf(maxdocstr, ".set maxdocs %i", queryparams.maxdocs);
|
---|
362 | mgq_ask(maxdocstr);
|
---|
363 | }
|
---|
364 | }
|
---|
365 |
|
---|
366 |
|
---|
367 | void mgsearchclass::submitquery (const queryparamclass &queryparams)
|
---|
368 | {
|
---|
369 | // sort out the query string
|
---|
370 | text_t ttquerystring = queryparams.querystring;
|
---|
371 | filterquery (ttquerystring);
|
---|
372 | char *querystring = to_utf8(ttquerystring).getcstr();
|
---|
373 |
|
---|
374 | // submit the query
|
---|
375 | mgq_ask(querystring);
|
---|
376 |
|
---|
377 | delete querystring;
|
---|
378 | }
|
---|
379 |
|
---|
380 |
|
---|
381 | void mgsearchclass::getresults (const queryparamclass &queryparams,
|
---|
382 | queryresultsclass &queryresults) {
|
---|
383 |
|
---|
384 | int howmany = queryparams.maxdocs;
|
---|
385 | if (howmany == -1) howmany = MAXNUMDOCS;
|
---|
386 | mgq_results(result_docnums, 0, howmany,
|
---|
387 | ourquerycallback, (void *)(&queryresults));
|
---|
388 |
|
---|
389 | // get the term frequencies
|
---|
390 | mgq_results(result_termfreqs, 0, MAXNUMTERMS,
|
---|
391 | termfreqcallback, (void *)(&queryresults));
|
---|
392 | queryresults.sortuniqqueryterms();
|
---|
393 |
|
---|
394 | // get term variants
|
---|
395 | mgq_results(result_terms, 0, MAXNUMTERMS,
|
---|
396 | termvariantscallback, (void *)(&queryresults));
|
---|
397 |
|
---|
398 | // get the number of documents retrieved
|
---|
399 | int total_retrieved = 0, is_approx = 0;
|
---|
400 | mgq_docsretrieved (&total_retrieved, &is_approx);
|
---|
401 |
|
---|
402 | if (total_retrieved == 0) {
|
---|
403 | // not available (or really was zero)
|
---|
404 | queryresults.docs_matched = queryresults.docs.docset.size();
|
---|
405 | if ((queryparams.maxdocs == -1) ||
|
---|
406 | (queryresults.docs_matched < queryparams.maxdocs))
|
---|
407 | queryresults.is_approx = Exact;
|
---|
408 | else
|
---|
409 | queryresults.is_approx = MoreThan;
|
---|
410 | } else {
|
---|
411 | queryresults.docs_matched = total_retrieved;
|
---|
412 | if (is_approx) queryresults.is_approx = Approximate;
|
---|
413 | else queryresults.is_approx = Exact;
|
---|
414 | }
|
---|
415 | }
|
---|
416 |
|
---|
417 | void mgsearchclass::filterquery (text_t &ttquerystring) {
|
---|
418 | text_t::iterator ithere = ttquerystring.begin ();
|
---|
419 | text_t::iterator itend = ttquerystring.end ();
|
---|
420 |
|
---|
421 | // remove all non alphanumeric characters (except
|
---|
422 | // boolean operators
|
---|
423 | while (ithere != itend) {
|
---|
424 | if ((!is_unicode_letdig(*ithere)) && (*ithere != '!') &&
|
---|
425 | (*ithere != '&') && (*ithere != '|') && (*ithere != '(') &&
|
---|
426 | (*ithere != ')')) (*ithere) = ' ';
|
---|
427 | ithere++;
|
---|
428 | }
|
---|
429 | }
|
---|
430 |
|
---|
431 |
|
---|
432 | // the document text for 'docnum' is placed in 'output'
|
---|
433 | // docTargetDocument returns 'true' if it was able to
|
---|
434 | // try to get a document
|
---|
435 | // collection is needed to see if an index from the
|
---|
436 | // collection is loaded. If no index has been loaded
|
---|
437 | // defaultindex is needed to load one
|
---|
438 | bool mgsearchclass::docTargetDocument(const text_t &defaultindex,
|
---|
439 | const text_t &defaultsubcollection,
|
---|
440 | const text_t &defaultlanguage,
|
---|
441 | const text_t &collection,
|
---|
442 | int docnum,
|
---|
443 | text_t &output) {
|
---|
444 | output.clear();
|
---|
445 |
|
---|
446 | // get the mg version of the document
|
---|
447 | char *mgdoc = NULL;
|
---|
448 | int doclen = 0;
|
---|
449 | if (!mgdocument (defaultindex, defaultsubcollection, defaultlanguage,
|
---|
450 | collection, docnum, mgdoc, doclen)) return false;
|
---|
451 | if (mgdoc == NULL) return false;
|
---|
452 |
|
---|
453 | // replace all control-Cs with spaces
|
---|
454 | char *mgdoc_here = mgdoc;
|
---|
455 | char *mgdoc_end = mgdoc + doclen;
|
---|
456 | while (mgdoc_here < mgdoc_end) {
|
---|
457 | if (*mgdoc_here == '\x3') *mgdoc_here = ' ';
|
---|
458 | mgdoc_here++;
|
---|
459 | }
|
---|
460 |
|
---|
461 | // convert this document to unicode
|
---|
462 | utf8inconvertclass inconvert;
|
---|
463 | convertclass::status_t status;
|
---|
464 | inconvert.reset ();
|
---|
465 | inconvert.setinput (mgdoc, doclen);
|
---|
466 | inconvert.convert (output, status);
|
---|
467 |
|
---|
468 | return true;
|
---|
469 | }
|
---|
470 |
|
---|
471 |
|
---|
472 | bool mgsearchclass::mgdocument (const text_t &defaultindex,
|
---|
473 | const text_t &defaultsubcollection,
|
---|
474 | const text_t &defaultlanguage,
|
---|
475 | const text_t &collection,
|
---|
476 | int docnum,
|
---|
477 | char *&UDoc, int &ULen) {
|
---|
478 | int databaseloaded = 0;
|
---|
479 |
|
---|
480 | UDoc = NULL; ULen = 0;
|
---|
481 |
|
---|
482 | // see if we can make an appropriate database current
|
---|
483 | // char *ccollection = collection.getcstr();
|
---|
484 | // assert (ccollection != NULL);
|
---|
485 | // databaseloaded = load_text_database (ccollection);
|
---|
486 | // delete ccollection;
|
---|
487 |
|
---|
488 | // try and load the database
|
---|
489 | // if (!databaseloaded)
|
---|
490 | databaseloaded = makeindexcurrent (defaultindex, defaultsubcollection,
|
---|
491 | defaultlanguage, collection);
|
---|
492 |
|
---|
493 | if (databaseloaded) {
|
---|
494 | // retrieve the document from mg
|
---|
495 | char docstr[32];
|
---|
496 | sprintf(docstr, "%i", docnum);
|
---|
497 |
|
---|
498 | mgq_ask(".set mode text");
|
---|
499 | mgq_ask(".set query docnums");
|
---|
500 | mgq_ask(docstr);
|
---|
501 |
|
---|
502 | tempdoc = NULL;
|
---|
503 | templen = 0;
|
---|
504 | mgq_results (result_docs, 0, 1, doctextcallback, (void *)NULL);
|
---|
505 | UDoc = tempdoc;
|
---|
506 | ULen = templen;
|
---|
507 | }
|
---|
508 |
|
---|
509 | return (bool)databaseloaded;
|
---|
510 | }
|
---|
511 |
|
---|