|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* Zend Framework |
|
4
|
|
|
* |
|
5
|
|
|
* LICENSE |
|
6
|
|
|
* |
|
7
|
|
|
* This source file is subject to version 1.0 of the Zend Framework |
|
8
|
|
|
* license, that is bundled with this package in the file LICENSE, and |
|
9
|
|
|
* is available through the world-wide-web at the following URL: |
|
10
|
|
|
* http://www.zend.com/license/framework/1_0.txt. If you did not receive |
|
11
|
|
|
* a copy of the Zend Framework license and are unable to obtain it |
|
12
|
|
|
* through the world-wide-web, please send a note to [email protected] |
|
13
|
|
|
* so we can mail you a copy immediately. |
|
14
|
|
|
* |
|
15
|
|
|
* @package Zend_Search_Lucene |
|
16
|
|
|
* @subpackage Search |
|
17
|
|
|
* @copyright Copyright (c) 2005-2006 Zend Technologies USA Inc. (http://www.zend.com) |
|
18
|
|
|
* @license http://www.zend.com/license/framework/1_0.txt Zend Framework License version 1.0 |
|
19
|
|
|
*/ |
|
20
|
|
|
|
|
21
|
|
|
|
|
22
|
|
|
/** Zend_Search_Lucene_Search_Query */ |
|
23
|
|
|
require_once 'Zend/Search/Lucene/Search/Query.php'; |
|
24
|
|
|
|
|
25
|
|
|
/** Zend_Search_Lucene_Search_Weight_MultiTerm */ |
|
26
|
|
|
require_once 'Zend/Search/Lucene/Search/Weight/MultiTerm.php'; |
|
27
|
|
|
|
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* @package Zend_Search_Lucene |
|
31
|
|
|
* @subpackage Search |
|
32
|
|
|
* @copyright Copyright (c) 2005-2006 Zend Technologies USA Inc. (http://www.zend.com) |
|
33
|
|
|
* @license http://www.zend.com/license/framework/1_0.txt Zend Framework License version 1.0 |
|
34
|
|
|
*/ |
|
35
|
|
|
class Zend_Search_Lucene_Search_Query_MultiTerm extends Zend_Search_Lucene_Search_Query |
|
36
|
|
|
{ |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* Terms to find. |
|
40
|
|
|
* Array of Zend_Search_Lucene_Index_Term |
|
41
|
|
|
* |
|
42
|
|
|
* @var array |
|
43
|
|
|
*/ |
|
44
|
|
|
private $_terms = array(); |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* Term signs. |
|
48
|
|
|
* If true then term is required. |
|
49
|
|
|
* If false then term is prohibited. |
|
50
|
|
|
* If null then term is neither prohibited, nor required |
|
51
|
|
|
* |
|
52
|
|
|
* If array is null then all terms are required |
|
53
|
|
|
* |
|
54
|
|
|
* @var array |
|
55
|
|
|
*/ |
|
56
|
|
|
|
|
57
|
|
|
private $_signs = array(); |
|
58
|
|
|
|
|
59
|
|
|
/** |
|
60
|
|
|
* Result vector. |
|
61
|
|
|
* Bitset or array of document IDs |
|
62
|
|
|
* (depending from Bitset extension availability). |
|
63
|
|
|
* |
|
64
|
|
|
* @var mixed |
|
65
|
|
|
*/ |
|
66
|
|
|
private $_resVector = null; |
|
67
|
|
|
|
|
68
|
|
|
/** |
|
69
|
|
|
* Terms positions vectors. |
|
70
|
|
|
* Array of Arrays: |
|
71
|
|
|
* term1Id => (docId => array( pos1, pos2, ... ), ...) |
|
72
|
|
|
* term2Id => (docId => array( pos1, pos2, ... ), ...) |
|
73
|
|
|
* |
|
74
|
|
|
* @var array |
|
75
|
|
|
*/ |
|
76
|
|
|
private $_termsPositions = array(); |
|
77
|
|
|
|
|
78
|
|
|
|
|
79
|
|
|
/** |
|
80
|
|
|
* A score factor based on the fraction of all query terms |
|
81
|
|
|
* that a document contains. |
|
82
|
|
|
* float for conjunction queries |
|
83
|
|
|
* array of float for non conjunction queries |
|
84
|
|
|
* |
|
85
|
|
|
* @var mixed |
|
86
|
|
|
*/ |
|
87
|
|
|
private $_coord = null; |
|
88
|
|
|
|
|
89
|
|
|
|
|
90
|
|
|
/** |
|
91
|
|
|
* Terms weights |
|
92
|
|
|
* array of Zend_Search_Lucene_Search_Weight |
|
93
|
|
|
* |
|
94
|
|
|
* @var array |
|
95
|
|
|
*/ |
|
96
|
|
|
private $_weights = array(); |
|
97
|
|
|
|
|
98
|
|
|
|
|
99
|
|
|
/** |
|
100
|
|
|
* Class constructor. Create a new multi-term query object. |
|
101
|
|
|
* |
|
102
|
|
|
* @param array $terms Array of Zend_Search_Lucene_Index_Term objects |
|
103
|
|
|
* @param array $signs Array of signs. Sign is boolean|null. |
|
104
|
|
|
* @return void |
|
|
|
|
|
|
105
|
|
|
*/ |
|
106
|
|
|
public function __construct($terms = null, $signs = null) |
|
107
|
|
|
{ |
|
108
|
|
|
/** |
|
109
|
|
|
* @todo Check contents of $terms and $signs before adding them. |
|
110
|
|
|
*/ |
|
111
|
|
|
if (is_array($terms)) { |
|
112
|
|
|
$this->_terms = $terms; |
|
113
|
|
|
|
|
114
|
|
|
$this->_signs = null; |
|
|
|
|
|
|
115
|
|
|
// Check if all terms are required |
|
116
|
|
|
if (is_array($signs)) { |
|
117
|
|
|
foreach ($signs as $sign ) { |
|
118
|
|
|
if ($sign !== true) { |
|
119
|
|
|
$this->_signs = $signs; |
|
120
|
|
|
continue; |
|
121
|
|
|
} |
|
122
|
|
|
} |
|
123
|
|
|
} |
|
124
|
|
|
} |
|
125
|
|
|
} |
|
126
|
|
|
|
|
127
|
|
|
|
|
128
|
|
|
/** |
|
129
|
|
|
* Add a $term (Zend_Search_Lucene_Index_Term) to this query. |
|
130
|
|
|
* |
|
131
|
|
|
* The sign is specified as: |
|
132
|
|
|
* TRUE - term is required |
|
133
|
|
|
* FALSE - term is prohibited |
|
134
|
|
|
* NULL - term is neither prohibited, nor required |
|
135
|
|
|
* |
|
136
|
|
|
* @param Zend_Search_Lucene_Index_Term $term |
|
137
|
|
|
* @param boolean|null $sign |
|
138
|
|
|
* @return void |
|
139
|
|
|
*/ |
|
140
|
|
|
public function addTerm(Zend_Search_Lucene_Index_Term $term, $sign=null) { |
|
141
|
|
|
$this->_terms[] = $term; |
|
142
|
|
|
|
|
143
|
|
|
/** |
|
144
|
|
|
* @todo This is not good. Sometimes $this->_signs is an array, sometimes |
|
145
|
|
|
* it is null, even when there are terms. It will be changed so that |
|
146
|
|
|
* it is always an array. |
|
147
|
|
|
*/ |
|
148
|
|
|
if ($this->_signs === null) { |
|
149
|
|
|
if ($sign !== null) { |
|
150
|
|
|
$this->_signs = array(); |
|
151
|
|
|
foreach ($this->_terms as $term) { |
|
152
|
|
|
$this->_signs[] = null; |
|
153
|
|
|
} |
|
154
|
|
|
$this->_signs[] = $sign; |
|
155
|
|
|
} |
|
156
|
|
|
} else { |
|
157
|
|
|
$this->_signs[] = $sign; |
|
158
|
|
|
} |
|
159
|
|
|
} |
|
160
|
|
|
|
|
161
|
|
|
|
|
162
|
|
|
/** |
|
163
|
|
|
* Returns query term |
|
164
|
|
|
* |
|
165
|
|
|
* @return array |
|
166
|
|
|
*/ |
|
167
|
|
|
public function getTerms() |
|
168
|
|
|
{ |
|
169
|
|
|
return $this->_terms; |
|
170
|
|
|
} |
|
171
|
|
|
|
|
172
|
|
|
|
|
173
|
|
|
/** |
|
174
|
|
|
* Return terms signs |
|
175
|
|
|
* |
|
176
|
|
|
* @return array |
|
177
|
|
|
*/ |
|
178
|
|
|
public function getSigns() |
|
179
|
|
|
{ |
|
180
|
|
|
return $this->_signs; |
|
181
|
|
|
} |
|
182
|
|
|
|
|
183
|
|
|
|
|
184
|
|
|
/** |
|
185
|
|
|
* Set weight for specified term |
|
186
|
|
|
* |
|
187
|
|
|
* @param integer $num |
|
188
|
|
|
* @param Zend_Search_Lucene_Search_Weight_Term $weight |
|
189
|
|
|
*/ |
|
190
|
|
|
public function setWeight($num, $weight) |
|
191
|
|
|
{ |
|
192
|
|
|
$this->_weights[$num] = $weight; |
|
193
|
|
|
} |
|
194
|
|
|
|
|
195
|
|
|
|
|
196
|
|
|
/** |
|
197
|
|
|
* Constructs an appropriate Weight implementation for this query. |
|
198
|
|
|
* |
|
199
|
|
|
* @param Zend_Search_Lucene $reader |
|
200
|
|
|
* @return Zend_Search_Lucene_Search_Weight |
|
201
|
|
|
*/ |
|
202
|
|
|
protected function _createWeight($reader) |
|
203
|
|
|
{ |
|
204
|
|
|
return new Zend_Search_Lucene_Search_Weight_MultiTerm($this, $reader); |
|
205
|
|
|
} |
|
206
|
|
|
|
|
207
|
|
|
|
|
208
|
|
|
/** |
|
209
|
|
|
* Calculate result vector for Conjunction query |
|
210
|
|
|
* (like '+something +another') |
|
211
|
|
|
* |
|
212
|
|
|
* @param Zend_Search_Lucene $reader |
|
213
|
|
|
*/ |
|
214
|
|
View Code Duplication |
private function _calculateConjunctionResult($reader) |
|
|
|
|
|
|
215
|
|
|
{ |
|
216
|
|
|
if (extension_loaded('bitset')) { |
|
217
|
|
|
foreach( $this->_terms as $termId=>$term ) { |
|
218
|
|
|
if($this->_resVector === null) { |
|
219
|
|
|
$this->_resVector = bitset_from_array($reader->termDocs($term)); |
|
220
|
|
|
} else { |
|
221
|
|
|
$this->_resVector = bitset_intersection( |
|
222
|
|
|
$this->_resVector, |
|
223
|
|
|
bitset_from_array($reader->termDocs($term)) ); |
|
224
|
|
|
} |
|
225
|
|
|
|
|
226
|
|
|
$this->_termsPositions[$termId] = $reader->termPositions($term); |
|
227
|
|
|
} |
|
228
|
|
|
} else { |
|
229
|
|
|
foreach( $this->_terms as $termId=>$term ) { |
|
230
|
|
|
if($this->_resVector === null) { |
|
231
|
|
|
$this->_resVector = array_flip($reader->termDocs($term)); |
|
232
|
|
|
} else { |
|
233
|
|
|
$termDocs = array_flip($reader->termDocs($term)); |
|
234
|
|
|
foreach($this->_resVector as $key=>$value) { |
|
235
|
|
|
if (!isset( $termDocs[$key] )) { |
|
236
|
|
|
unset( $this->_resVector[$key] ); |
|
237
|
|
|
} |
|
238
|
|
|
} |
|
239
|
|
|
} |
|
240
|
|
|
|
|
241
|
|
|
$this->_termsPositions[$termId] = $reader->termPositions($term); |
|
242
|
|
|
} |
|
243
|
|
|
} |
|
244
|
|
|
} |
|
245
|
|
|
|
|
246
|
|
|
|
|
247
|
|
|
/** |
|
248
|
|
|
* Calculate result vector for non Conjunction query |
|
249
|
|
|
* (like '+something -another') |
|
250
|
|
|
* |
|
251
|
|
|
* @param Zend_Search_Lucene $reader |
|
252
|
|
|
*/ |
|
253
|
|
|
private function _calculateNonConjunctionResult($reader) |
|
254
|
|
|
{ |
|
255
|
|
|
if (extension_loaded('bitset')) { |
|
256
|
|
|
$required = null; |
|
257
|
|
|
$neither = bitset_empty(); |
|
258
|
|
|
$prohibited = bitset_empty(); |
|
259
|
|
|
|
|
260
|
|
|
foreach ($this->_terms as $termId => $term) { |
|
261
|
|
|
$termDocs = bitset_from_array($reader->termDocs($term)); |
|
262
|
|
|
|
|
263
|
|
|
if ($this->_signs[$termId] === true) { |
|
264
|
|
|
// required |
|
265
|
|
|
if ($required !== null) { |
|
266
|
|
|
$required = bitset_intersection($required, $termDocs); |
|
267
|
|
|
} else { |
|
268
|
|
|
$required = $termDocs; |
|
269
|
|
|
} |
|
270
|
|
|
} elseif ($this->_signs[$termId] === false) { |
|
271
|
|
|
// prohibited |
|
272
|
|
|
$prohibited = bitset_union($prohibited, $termDocs); |
|
273
|
|
|
} else { |
|
274
|
|
|
// neither required, nor prohibited |
|
275
|
|
|
$neither = bitset_union($neither, $termDocs); |
|
276
|
|
|
} |
|
277
|
|
|
|
|
278
|
|
|
$this->_termsPositions[$termId] = $reader->termPositions($term); |
|
279
|
|
|
} |
|
280
|
|
|
|
|
281
|
|
|
if ($required === null) { |
|
282
|
|
|
$required = $neither; |
|
283
|
|
|
} |
|
284
|
|
|
$this->_resVector = bitset_intersection( $required, |
|
285
|
|
|
bitset_invert($prohibited, $reader->count()) ); |
|
286
|
|
|
} else { |
|
287
|
|
|
$required = null; |
|
288
|
|
|
$neither = array(); |
|
289
|
|
|
$prohibited = array(); |
|
290
|
|
|
|
|
291
|
|
|
foreach ($this->_terms as $termId => $term) { |
|
292
|
|
|
$termDocs = array_flip($reader->termDocs($term)); |
|
293
|
|
|
|
|
294
|
|
|
if ($this->_signs[$termId] === true) { |
|
295
|
|
|
// required |
|
296
|
|
|
if ($required !== null) { |
|
297
|
|
|
// substitute for bitset_intersection |
|
298
|
|
|
foreach ($required as $key => $value) { |
|
|
|
|
|
|
299
|
|
|
if (!isset( $termDocs[$key] )) { |
|
300
|
|
|
unset($required[$key]); |
|
301
|
|
|
} |
|
302
|
|
|
} |
|
303
|
|
|
} else { |
|
304
|
|
|
$required = $termDocs; |
|
305
|
|
|
} |
|
306
|
|
|
} elseif ($this->_signs[$termId] === false) { |
|
307
|
|
|
// prohibited |
|
308
|
|
|
// substitute for bitset_union |
|
309
|
|
|
foreach ($termDocs as $key => $value) { |
|
310
|
|
|
$prohibited[$key] = $value; |
|
311
|
|
|
} |
|
312
|
|
|
} else { |
|
313
|
|
|
// neither required, nor prohibited |
|
314
|
|
|
// substitute for bitset_union |
|
315
|
|
|
foreach ($termDocs as $key => $value) { |
|
316
|
|
|
$neither[$key] = $value; |
|
317
|
|
|
} |
|
318
|
|
|
} |
|
319
|
|
|
|
|
320
|
|
|
$this->_termsPositions[$termId] = $reader->termPositions($term); |
|
321
|
|
|
} |
|
322
|
|
|
|
|
323
|
|
|
if ($required === null) { |
|
324
|
|
|
$required = $neither; |
|
325
|
|
|
} |
|
326
|
|
|
|
|
327
|
|
|
foreach ($required as $key=>$value) { |
|
328
|
|
|
if (isset( $prohibited[$key] )) { |
|
329
|
|
|
unset($required[$key]); |
|
330
|
|
|
} |
|
331
|
|
|
} |
|
332
|
|
|
$this->_resVector = $required; |
|
333
|
|
|
} |
|
334
|
|
|
} |
|
335
|
|
|
|
|
336
|
|
|
|
|
337
|
|
|
/** |
|
338
|
|
|
* Score calculator for conjunction queries (all terms are required) |
|
339
|
|
|
* |
|
340
|
|
|
* @param integer $docId |
|
341
|
|
|
* @param Zend_Search_Lucene $reader |
|
342
|
|
|
* @return float |
|
343
|
|
|
*/ |
|
344
|
|
|
public function _conjunctionScore($docId, $reader) |
|
345
|
|
|
{ |
|
346
|
|
|
if ($this->_coord === null) { |
|
347
|
|
|
$this->_coord = $reader->getSimilarity()->coord(count($this->_terms), |
|
348
|
|
|
count($this->_terms) ); |
|
349
|
|
|
} |
|
350
|
|
|
|
|
351
|
|
|
$score = 0.0; |
|
352
|
|
|
|
|
353
|
|
|
foreach ($this->_terms as $termId=>$term) { |
|
354
|
|
|
$score += $reader->getSimilarity()->tf(count($this->_termsPositions[$termId][$docId]) ) * |
|
355
|
|
|
$this->_weights[$termId]->getValue() * |
|
356
|
|
|
$reader->norm($docId, $term->field); |
|
357
|
|
|
} |
|
358
|
|
|
|
|
359
|
|
|
return $score * $this->_coord; |
|
360
|
|
|
} |
|
361
|
|
|
|
|
362
|
|
|
|
|
363
|
|
|
/** |
|
364
|
|
|
* Score calculator for non conjunction queries (not all terms are required) |
|
365
|
|
|
* |
|
366
|
|
|
* @param integer $docId |
|
367
|
|
|
* @param Zend_Search_Lucene $reader |
|
368
|
|
|
* @return float |
|
369
|
|
|
*/ |
|
370
|
|
|
public function _nonConjunctionScore($docId, $reader) |
|
371
|
|
|
{ |
|
372
|
|
|
if ($this->_coord === null) { |
|
373
|
|
|
$this->_coord = array(); |
|
374
|
|
|
|
|
375
|
|
|
$maxCoord = 0; |
|
376
|
|
|
foreach ($this->_signs as $sign) { |
|
377
|
|
|
if ($sign !== false /* not prohibited */) { |
|
378
|
|
|
$maxCoord++; |
|
379
|
|
|
} |
|
380
|
|
|
} |
|
381
|
|
|
|
|
382
|
|
|
for ($count = 0; $count <= $maxCoord; $count++) { |
|
383
|
|
|
$this->_coord[$count] = $reader->getSimilarity()->coord($count, $maxCoord); |
|
384
|
|
|
} |
|
385
|
|
|
} |
|
386
|
|
|
|
|
387
|
|
|
$score = 0.0; |
|
388
|
|
|
$matchedTerms = 0; |
|
389
|
|
|
foreach ($this->_terms as $termId=>$term) { |
|
390
|
|
|
// Check if term is |
|
391
|
|
|
if ($this->_signs[$termId] !== false && // not prohibited |
|
392
|
|
|
isset($this->_termsPositions[$termId][$docId]) // matched |
|
393
|
|
|
) { |
|
394
|
|
|
$matchedTerms++; |
|
395
|
|
|
$score += |
|
396
|
|
|
$reader->getSimilarity()->tf(count($this->_termsPositions[$termId][$docId]) ) * |
|
397
|
|
|
$this->_weights[$termId]->getValue() * |
|
398
|
|
|
$reader->norm($docId, $term->field); |
|
399
|
|
|
} |
|
400
|
|
|
} |
|
401
|
|
|
|
|
402
|
|
|
return $score * $this->_coord[$matchedTerms]; |
|
403
|
|
|
} |
|
404
|
|
|
|
|
405
|
|
|
/** |
|
406
|
|
|
* Score specified document |
|
407
|
|
|
* |
|
408
|
|
|
* @param integer $docId |
|
409
|
|
|
* @param Zend_Search_Lucene $reader |
|
410
|
|
|
* @return float |
|
411
|
|
|
*/ |
|
412
|
|
|
public function score($docId, $reader) |
|
413
|
|
|
{ |
|
414
|
|
|
if($this->_resVector === null) { |
|
415
|
|
|
if ($this->_signs === null) { |
|
416
|
|
|
$this->_calculateConjunctionResult($reader); |
|
417
|
|
|
} else { |
|
418
|
|
|
$this->_calculateNonConjunctionResult($reader); |
|
419
|
|
|
} |
|
420
|
|
|
|
|
421
|
|
|
$this->_initWeight($reader); |
|
422
|
|
|
} |
|
423
|
|
|
|
|
424
|
|
|
if ( (extension_loaded('bitset')) ? |
|
425
|
|
|
bitset_in($this->_resVector, $docId) : |
|
426
|
|
|
isset($this->_resVector[$docId]) ) { |
|
427
|
|
|
if ($this->_signs === null) { |
|
428
|
|
|
return $this->_conjunctionScore($docId, $reader); |
|
429
|
|
|
} else { |
|
430
|
|
|
return $this->_nonConjunctionScore($docId, $reader); |
|
431
|
|
|
} |
|
432
|
|
|
} else { |
|
433
|
|
|
return 0; |
|
434
|
|
|
} |
|
435
|
|
|
} |
|
436
|
|
|
} |
|
437
|
|
|
|
|
438
|
|
|
|
Adding a
@returnannotation to a constructor is not recommended, since a constructor does not have a meaningful return value.Please refer to the PHP core documentation on constructors.