source: trunk/mgpp/text/GSDLQueryParser.cpp@ 5638

Last change on this file since 5638 was 5449, checked in by kjdon, 21 years ago

added a new token type to the Lex module - UnknownE.
this will match any unrecognisable char - can now just skip over unknown chars rather than spitting the dummy and stopping the parsing
  • Property svn:executable set to *
  • Property svn:keywords set to Author Date Id Revision
File size: 12.6 KB
Line 
1/**************************************************************************
2 *
3 * QueryParser.cpp -- Query parser for a simple query language
4 * Copyright (C) 2000 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#include "GSDLQueryParser.h"
23#include "GSDLQueryLex.h"
24
25
26static QueryNode *ParseExpression (UCArray::const_iterator &here,
27 UCArray::const_iterator end,
28 int defaultBoolCombine,
29 int defaultStemMethod);
30
31static QueryNode *AndAdd (QueryNode *t1, QueryNode *t2) {
32 if (t1 == NULL) return t2;
33 if (t2 == NULL) return t1;
34
35 AndQueryNode *andNode = new AndQueryNode;
36 andNode->leftNode = t1;
37 andNode->rightNode = t2;
38 return andNode;
39}
40
41static QueryNode *OrAdd (QueryNode *t1, QueryNode *t2) {
42 if (t1 == NULL) return t2;
43 if (t2 == NULL) return t1;
44
45 OrQueryNode *orNode = new OrQueryNode;
46 orNode->leftNode = t1;
47 orNode->rightNode = t2;
48 return orNode;
49}
50
51static QueryNode *NotAdd (QueryNode *t1, QueryNode *t2) {
52 if (t1 == NULL) return t2;
53 if (t2 == NULL) return t1;
54
55 NotQueryNode *notNode = new NotQueryNode;
56 notNode->queryNode = t1;
57 notNode->notNode = t2;
58 return notNode;
59}
60
61// expects the opening bracket to have already been parsed
62// and discarded
63static QueryNode *ParseBracketExpression (UCArray::const_iterator &here,
64 UCArray::const_iterator end,
65 int defaultBoolCombine,
66 int defaultStemMethod) {
67 // get everything in the expression
68 QueryNode *curTree = ParseExpression (here, end, defaultBoolCombine,
69 defaultStemMethod);
70
71 // gobble up tokens until a closing bracket is found
72 // or the end of the string
73 LexEl el;
74 while (ParseLexEl (here, end, el)) {
75 if (el.lexType == CloseBracketE) break;
76 }
77
78 return curTree;
79}
80
81static int ParseInt (UCArray::const_iterator &here,
82 UCArray::const_iterator end) {
83 LexEl el;
84 UCArray::const_iterator oldHere = here;
85 if (ParseLexEl (here, end, el) && el.lexType == IntegerE)
86 return el.num;
87
88 here = oldHere; // not an integer
89 return 0;
90}
91
92// default is within 20 words
93static void SetRangeValues (TermNode &termNode,
94 UCArray &nearby) {
95 UCArray NEARBY; SetCStr(NEARBY, "NEAR");
96 if (nearby == NEARBY) { // no modifier
97 termNode.startRange = (NEAR_DEFAULT+1)*-1;
98 termNode.endRange = NEAR_DEFAULT;
99
100 }
101 else { // extract number
102 UCArray::const_iterator here = nearby.begin()+4;
103 UCArray::const_iterator end = nearby.end();
104 int size=0;
105 while (here != end) {
106 size = size*10 + (*here-'0');
107 here++;
108 }
109 termNode.startRange = -1 * (size+1);
110 termNode.endRange = size;
111 }
112}
113
114static unsigned long GetStemMethod(LexEl &el, int defaultStemMethod) {
115 // here expect el to contain some of c,s,i,u
116 // stem method 0 = c,u 00
117 // stem method 1 = i,u 01 - default for DL
118 // stem method 2 = c, s 10
119 // stem method 3 = i,s 11
120
121 unsigned long stem = (unsigned long)defaultStemMethod;
122
123 UCArray::const_iterator here = el.text.begin();
124 UCArray::const_iterator end = el.text.end();
125
126 unsigned char c1 = *here;
127 if (!(c1 == 'c'| c1 == 'i' | c1 == 'u' | c1 == 's'))
128 return 4; // incorrect format
129
130 here++;
131 unsigned char c2 = 'a';
132 if (here !=end) {
133 c2 = *here;
134 if (!(c2 == 'c'| c2 == 'i' | c2 == 'u' | c2 == 's'))
135 return 4; // incorrect format
136 }
137
138 if (c1 == 'i'|| c2=='i') stem |= 1; // set bit 0 to 1
139 if (c1 == 'c' || c2 == 'c') stem &=0xe; //set bit 0 to 0
140 if (c1 == 's'|| c2 == 's') stem |= 2; // set bit 1 to 1
141 if (c1 == 'u' || c2 =='u') stem &=0xd; // set bit 1 to 0
142 return stem;
143}
144
145
146static void ParseTermModifiers (UCArray::const_iterator &here,
147 UCArray::const_iterator end,
148 TermNode &termNode,
149 int defaultStemMethod) {
150
151 termNode.stemMethod = defaultStemMethod;
152
153 LexEl el;
154 UCArray::const_iterator oldHere = here;
155 while (ParseLexEl (here, end, el)) {
156 if (el.lexType == TermWeightE) {
157 termNode.termWeight = ParseInt (here, end);
158
159 } else if (el.lexType == StemMethodE) {
160 oldHere = here;
161 LexEl stem;
162 if (ParseLexEl (here, end, stem) && stem.lexType == TermE) {
163 termNode.stemMethod = GetStemMethod(stem, defaultStemMethod);
164 if (termNode.stemMethod == 4) { // error so backtrack
165 here = oldHere;
166 termNode.stemMethod = (unsigned long)defaultStemMethod;
167 }
168 }else here = oldHere; //ignore - wrong syntax
169
170 } else if (el.lexType == RangeE) {
171 termNode.startRange = ParseInt (here, end);
172 termNode.endRange = ParseInt (here, end);
173
174 } else if (el.lexType == AtE) {
175 termNode.startRange = termNode.endRange = ParseInt (here, end);
176
177 } else {
178 // no term modifiers
179 here = oldHere;
180 break;
181 }
182
183 oldHere = here;
184 }
185}
186
187static void ParseProxModifiers (UCArray::const_iterator &here,
188 UCArray::const_iterator end,
189 ProxMatchQueryNode *proxNode) {
190 // so far only have one - the tag stuff
191 LexEl el;
192 UCArray::const_iterator oldHere = here;
193 while (ParseLexEl (here, end, el)) {
194 if (el.lexType == TagE) {
195 oldHere = here; // don't backtrack past here
196 if (ParseLexEl (here, end, el) && el.lexType == TermE) {
197 proxNode->tagNodePtr = new TagNode;
198 proxNode->tagNodePtr->tagName = el.text;
199
200 }
201 else { // error in tag
202 here = oldHere;
203 }
204 } // TagE
205 // add in other cases here
206 else {
207 // no modifiers
208 here = oldHere;
209 break;
210 }
211 oldHere = here;
212 }//while
213
214
215}
216
217// expects starting brackets to have been parsed
218// sets error to true if something has gone wrong
219static void ParseSquareBrackets(UCArray::const_iterator &here,
220 UCArray::const_iterator end,
221 ProxMatchQueryNode *proxNode,
222 int defaultStemMethod,
223 bool & error) {
224
225 LexEl el;
226 bool phrase=false;
227 bool first=true;
228 while (ParseLexEl (here, end, el)) {
229 if (el.lexType == TermE || el.lexType == IntegerE) {
230 TermNode termNode;
231 termNode.term = el.text;
232 ParseTermModifiers (here, end, termNode, defaultStemMethod);
233 if (phrase) {
234 if (first) first=false;
235 else {
236 termNode.startRange = -2;
237 termNode.endRange = -1;
238 }
239 }
240 proxNode->terms.push_back(termNode);
241 }
242 else if (el.lexType == CloseSquareBracketE) {
243 break;
244 }
245 else if (el.lexType == QuoteE) {
246 // phrase inside square brackets
247 if (phrase) phrase=false;
248 else phrase=true;
249 } else if (el.lexType == UnknownE) {
250 // just ignore it
251 }
252 else {
253 //error - we set the proxNode to NULL,
254 cerr <<"GSDLQueryParser: bad syntax inside []\n";
255 error = true;
256 return;
257 }
258 } // while
259
260}
261// expects the starting quote to have been parsed
262// and discarded
263// now phrases use the case and stem preference options
264// ie can search for a phrase ignoring case
265static void ParsePhrase (UCArray::const_iterator &here,
266 UCArray::const_iterator end,
267 ProxMatchQueryNode &proxNode,
268 int defaultStemMethod,
269 bool &error) {
270 LexEl el;
271 bool first = true;
272 while (ParseLexEl (here, end, el)) {
273 if (el.lexType == TermE || el.lexType == IntegerE) {
274 TermNode termNode;
275 termNode.term = el.text;
276 //termNode.stemMethod = defaultStemMethod;
277 ParseTermModifiers (here, end, termNode, defaultStemMethod);
278 if (first) {
279 first = false;
280 }
281 else {
282 termNode.startRange = -2;
283 termNode.endRange = -1;
284 }
285 proxNode.terms.push_back (termNode);
286
287 } else if (el.lexType == QuoteE) {
288 break;
289
290 } else if (el.lexType == UnknownE) {
291 // just ignore it
292 } else {
293 // error
294 error = true;
295 return;
296 }
297 }
298}
299
300static QueryNode *ParseTerm (UCArray::const_iterator &here,
301 UCArray::const_iterator end,
302 int defaultBoolCombine,
303 int defaultStemMethod) {
304 LexEl el;
305
306 UCArray::const_iterator oldHere = here;
307 if (!ParseLexEl (here, end, el)) return NULL;
308
309 if (el.lexType == OpenBracketE)
310 return ParseBracketExpression (here, end, defaultBoolCombine,
311 defaultStemMethod);
312
313 ProxMatchQueryNode *proxNode = new ProxMatchQueryNode;
314
315 if (el.lexType == TermE || el.lexType == IntegerE) {
316 TermNode termNode;
317 termNode.term = el.text;
318 ParseTermModifiers (here, end, termNode, defaultStemMethod);
319 oldHere = here; // dont backtrack past here
320 if (ParseLexEl(here, end, el) && el.lexType == NearOpE) {
321 delete proxNode;
322 oldHere = here;
323 // this is calling ParseTerm again, but only a subset of the things accepted by ParseTerm are appropriate here. add in some hacks to avoid segmentation faults - kjdon, 04/2003
324
325 // if the next element is a '(' have a syntax error, return NULL
326 LexEl temp_el;
327 if (ParseLexEl(here, end, temp_el) && temp_el.lexType == OpenBracketE) {
328 cerr << "GSDLQueryParser: NEAR cannot be followed by a '('\n";
329 return NULL;
330 }
331 here = oldHere; // else backtrack
332
333 proxNode = (ProxMatchQueryNode *)ParseTerm(here, end, defaultBoolCombine,
334 defaultStemMethod);
335 SetRangeValues(termNode, el.text);
336 proxNode->terms.push_back (termNode);
337 return proxNode;
338
339 } else {
340 here = oldHere; // backtrack
341 proxNode->terms.push_back (termNode);
342 ParseProxModifiers(here, end, proxNode);
343 return proxNode;
344 }
345 } else if (el.lexType == QuoteE) {
346 bool error = false;
347 ParsePhrase (here, end, *proxNode, defaultStemMethod, error);
348 if (error) {
349 delete proxNode;
350 return NULL;
351 }
352 return proxNode;
353 }
354 else if (el.lexType == OpenSquareBracketE) {
355 bool error = false;
356 ParseSquareBrackets (here, end, proxNode, defaultStemMethod, error);
357 if (error) {
358 delete proxNode;
359 return NULL;
360 }
361 ParseProxModifiers (here, end, proxNode);
362 return proxNode;
363 }
364
365 // not a term
366 here = oldHere;
367 delete proxNode;
368 return NULL;
369}
370
371
372static QueryNode *ParseExpression (UCArray::const_iterator &here,
373 UCArray::const_iterator end,
374 int defaultBoolCombine,
375 int defaultStemMethod) {
376 LexEl el;
377 QueryNode *curTree = NULL;
378 UCArray::const_iterator oldHere = here;
379 while (ParseLexEl (here, end, el)) {
380 if (el.lexType == CloseBracketE) {
381 // parsebracketexpression is waiting for the last bracket, so put it back
382 here = oldHere;
383 break;
384
385 } else if (el.lexType == OpenSquareBracketE ||
386 el.lexType == OpenBracketE ||
387 el.lexType == TermE ||
388 el.lexType == QuoteE ||
389 el.lexType == IntegerE ) {
390
391 // some type of term, back track and parse it
392 here = oldHere;
393
394 // parse the term
395 QueryNode * newTerm = ParseTerm (here, end, defaultBoolCombine,
396 defaultStemMethod);
397 if (newTerm == NULL) {
398 delete curTree;
399 return NULL;
400 }
401
402 // if default==1, AND, else if==0, OR
403 if (defaultBoolCombine) {
404 curTree = AndAdd (curTree, newTerm);
405 }
406 else {
407 curTree = OrAdd (curTree, newTerm);
408 }
409
410 } else if (el.lexType == AndOpE) {
411 QueryNode * newTerm = ParseTerm (here, end, defaultBoolCombine,
412 defaultStemMethod);
413 if (newTerm == NULL) {
414 delete curTree;
415 return NULL;
416 }
417 curTree = AndAdd (curTree, newTerm);
418
419 } else if (el.lexType == OrOpE) {
420 QueryNode * newTerm = ParseTerm (here, end, defaultBoolCombine,
421 defaultStemMethod);
422 if (newTerm == NULL) {
423 delete curTree;
424 return NULL;
425 }
426 curTree = OrAdd (curTree, newTerm);
427
428 } else if (el.lexType == NotOpE) {
429 QueryNode * newTerm = ParseTerm (here, end, defaultBoolCombine,
430 defaultStemMethod);
431 if (newTerm == NULL) {
432 delete curTree;
433 return NULL;
434 }
435 curTree = NotAdd (curTree, newTerm);
436
437 } else if (el.lexType == UnknownE) {
438 // just ignore it
439 } else {
440
441 // syntax error, return NUll
442 delete curTree;
443 return NULL;
444 }
445
446 oldHere = here;
447 }
448
449 return curTree;
450}
451
452QueryNode *ParseQuery (const UCArray &queryStr, int defaultBoolCombine,
453 int defaultStemMethod) {
454 UCArray::const_iterator here = queryStr.begin();
455 UCArray::const_iterator end = queryStr.end();
456 return ParseExpression (here, end, defaultBoolCombine, defaultStemMethod);
457}
Note: See TracBrowser for help on using the repository browser.