1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* File containing the DoctrineDatabase Content search Gateway class. |
4
|
|
|
* |
5
|
|
|
* @copyright Copyright (C) eZ Systems AS. All rights reserved. |
6
|
|
|
* @license For full copyright and license information view LICENSE file distributed with this source code. |
7
|
|
|
*/ |
8
|
|
|
namespace eZ\Publish\Core\Search\Legacy\Content\WordIndexer\Gateway; |
9
|
|
|
|
10
|
|
|
use eZ\Publish\Core\Search\Legacy\Content\WordIndexer\Gateway; |
11
|
|
|
use eZ\Publish\Core\Persistence\Database\DatabaseHandler; |
12
|
|
|
use eZ\Publish\Core\Persistence\TransformationProcessor; |
13
|
|
|
use eZ\Publish\Core\Search\Legacy\Content\WordIndexer\Repository\SearchIndex; |
14
|
|
|
use eZ\Publish\Core\Search\Legacy\Content\FullTextData; |
15
|
|
|
use eZ\Publish\SPI\Persistence\Content; |
16
|
|
|
use eZ\Publish\SPI\Persistence\Content\Type\Handler as SPITypeHandler; |
17
|
|
|
use eZ\Publish\SPI\Search\Field; |
18
|
|
|
|
19
|
|
|
/** |
20
|
|
|
* WordIndexer gateway implementation using the Doctrine database. |
21
|
|
|
*/ |
22
|
|
|
class DoctrineDatabase extends Gateway |
23
|
|
|
{ |
24
|
|
|
/** |
25
|
|
|
* Max acceptable by any DBMS INT value. |
26
|
|
|
* |
27
|
|
|
* Note: 2^31-1 seems to be the most reasonable value that should work in any setup. |
28
|
|
|
*/ |
29
|
|
|
const DB_INT_MAX = 2147483647; |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* Database handler. |
33
|
|
|
* |
34
|
|
|
* @var \eZ\Publish\Core\Persistence\Database\DatabaseHandler |
35
|
|
|
*/ |
36
|
|
|
protected $dbHandler; |
37
|
|
|
|
38
|
|
|
/** |
39
|
|
|
* SPI Content Type Handler. |
40
|
|
|
* |
41
|
|
|
* Need this for being able to pick fields that are searchable. |
42
|
|
|
* |
43
|
|
|
* @var \eZ\Publish\SPI\Persistence\Content\Type\Handler |
44
|
|
|
*/ |
45
|
|
|
protected $typeHandler; |
46
|
|
|
|
47
|
|
|
/** |
48
|
|
|
* Transformation processor. |
49
|
|
|
* |
50
|
|
|
* Need this for being able to transform text to searchable value |
51
|
|
|
* |
52
|
|
|
* @var \eZ\Publish\Core\Persistence\TransformationProcessor |
53
|
|
|
*/ |
54
|
|
|
protected $transformationProcessor; |
55
|
|
|
|
56
|
|
|
/** |
57
|
|
|
* LegacySearchService. |
58
|
|
|
* |
59
|
|
|
* Need this for queries on ezsearch* tables |
60
|
|
|
* |
61
|
|
|
* @var \eZ\Publish\Core\Search\Legacy\Content\WordIndexer\Repository\SearchIndex |
62
|
|
|
*/ |
63
|
|
|
protected $searchIndex; |
64
|
|
|
|
65
|
|
|
/** |
66
|
|
|
* Full text search configuration options. |
67
|
|
|
* |
68
|
|
|
* @var array |
69
|
|
|
*/ |
70
|
|
|
protected $fullTextSearchConfiguration; |
71
|
|
|
|
72
|
|
|
/** |
73
|
|
|
* Construct from handler handler. |
74
|
|
|
* |
75
|
|
|
* @param \eZ\Publish\Core\Persistence\Database\DatabaseHandler $dbHandler |
76
|
|
|
* @param \eZ\Publish\SPI\Persistence\Content\Type\Handler $typeHandler |
77
|
|
|
* @param \eZ\Publish\Core\Persistence\TransformationProcessor $transformationProcessor |
78
|
|
|
* @param \eZ\Publish\Core\Search\Legacy\Content\WordIndexer\Repository\SearchIndex $searchIndex |
79
|
|
|
* @param array $fullTextSearchConfiguration |
80
|
|
|
*/ |
81
|
|
|
public function __construct( |
82
|
|
|
DatabaseHandler $dbHandler, |
83
|
|
|
SPITypeHandler $typeHandler, |
84
|
|
|
TransformationProcessor $transformationProcessor, |
85
|
|
|
SearchIndex $searchIndex, |
86
|
|
|
array $fullTextSearchConfiguration |
87
|
|
|
) { |
88
|
|
|
$this->dbHandler = $dbHandler; |
89
|
|
|
$this->typeHandler = $typeHandler; |
90
|
|
|
$this->transformationProcessor = $transformationProcessor; |
91
|
|
|
$this->searchIndex = $searchIndex; |
92
|
|
|
$this->fullTextSearchConfiguration = $fullTextSearchConfiguration; |
93
|
|
|
} |
94
|
|
|
|
95
|
|
|
/** |
96
|
|
|
* Index search engine full text data corresponding to content object field values. |
97
|
|
|
* |
98
|
|
|
* Ported from the legacy code |
99
|
|
|
* @see https://github.com/ezsystems/ezpublish-legacy/blob/master/kernel/search/plugins/ezsearchengine/ezsearchengine.php#L45 |
100
|
|
|
* |
101
|
|
|
* @param \eZ\Publish\Core\Search\Legacy\Content\FullTextData $fullTextData |
102
|
|
|
*/ |
103
|
|
|
public function index(FullTextData $fullTextData) |
104
|
|
|
{ |
105
|
|
|
$indexArray = []; |
106
|
|
|
$indexArrayOnlyWords = []; |
107
|
|
|
$wordCount = 0; |
108
|
|
|
$placement = 0; |
109
|
|
|
|
110
|
|
|
// Remove previously indexed content if exists to avoid keeping in index removed field values |
111
|
|
|
$this->remove($fullTextData->id); |
112
|
|
|
foreach ($fullTextData->values as $fullTextValue) { |
113
|
|
|
/** @var \eZ\Publish\Core\Search\Legacy\Content\FullTextValue $fullTextValue */ |
114
|
|
|
if (is_numeric(trim($fullTextValue->value))) { |
115
|
|
|
$integerValue = (int)$fullTextValue->value; |
116
|
|
|
if ($integerValue > self::DB_INT_MAX) { |
117
|
|
|
$integerValue = 0; |
118
|
|
|
} |
119
|
|
|
} else { |
120
|
|
|
$integerValue = 0; |
121
|
|
|
} |
122
|
|
|
// Split transformed text on whitespace |
123
|
|
|
$wordArray = explode(' ', $this->transformationProcessor->transform($fullTextValue->value, $this->fullTextSearchConfiguration['commands'])); |
124
|
|
|
foreach ($wordArray as $word) { |
125
|
|
|
if (trim($word) === '') { |
126
|
|
|
continue; |
127
|
|
|
} |
128
|
|
|
// words stored in search index are limited to 150 characters |
129
|
|
|
if (mb_strlen($word) > 150) { |
130
|
|
|
$word = mb_substr($word, 0, 150); |
131
|
|
|
} |
132
|
|
|
$indexArray[] = ['Word' => $word, |
133
|
|
|
'ContentClassAttributeID' => $fullTextValue->fieldDefinitionId, |
134
|
|
|
'identifier' => $fullTextValue->id, |
135
|
|
|
'integer_value' => $integerValue, ]; |
136
|
|
|
$indexArrayOnlyWords[$word] = 1; |
137
|
|
|
++$wordCount; |
138
|
|
|
// if we have "www." before word than |
139
|
|
|
// treat it as url and add additional entry to the index |
140
|
|
|
if (mb_substr(mb_strtolower($word), 0, 4) === 'www.') { |
141
|
|
|
$additionalUrlWord = substr($word, 4); |
142
|
|
|
$indexArray[] = ['Word' => $additionalUrlWord, |
143
|
|
|
'ContentClassAttributeID' => $fullTextValue->fieldDefinitionId, |
144
|
|
|
'identifier' => $fullTextValue->id, |
145
|
|
|
'integer_value' => $integerValue, ]; |
146
|
|
|
$indexArrayOnlyWords[$additionalUrlWord] = 1; |
147
|
|
|
++$wordCount; |
148
|
|
|
} |
149
|
|
|
} |
150
|
|
|
} |
151
|
|
|
|
152
|
|
|
$wordIDArray = $this->buildWordIDArray(array_keys($indexArrayOnlyWords)); |
153
|
|
|
$this->dbHandler->beginTransaction(); |
154
|
|
|
for ($arrayCount = 0; $arrayCount < $wordCount; $arrayCount += 1000) { |
155
|
|
|
$placement = $this->indexWords( |
156
|
|
|
$fullTextData, |
157
|
|
|
array_slice($indexArray, $arrayCount, 1000), |
158
|
|
|
$wordIDArray, |
159
|
|
|
$placement |
160
|
|
|
); |
161
|
|
|
} |
162
|
|
|
$this->dbHandler->commit(); |
163
|
|
|
} |
164
|
|
|
|
165
|
|
|
/** |
166
|
|
|
* Indexes an array of FullTextData objects. |
167
|
|
|
* |
168
|
|
|
* Note: on large amounts of data make sure to iterate with several calls to this function with |
169
|
|
|
* a limited set of FullTextData objects. Amount you have memory for depends on server, size |
170
|
|
|
* of FullTextData objects & PHP version. |
171
|
|
|
* |
172
|
|
|
* @param \eZ\Publish\Core\Search\Legacy\Content\FullTextData[] $fullTextBulkData |
173
|
|
|
*/ |
174
|
|
|
public function bulkIndex(array $fullTextBulkData) |
175
|
|
|
{ |
176
|
|
|
foreach ($fullTextBulkData as $fullTextData) { |
177
|
|
|
$this->index($fullTextData); |
178
|
|
|
} |
179
|
|
|
} |
180
|
|
|
|
181
|
|
|
/** |
182
|
|
|
* Remove whole content or a specific version from index. |
183
|
|
|
* |
184
|
|
|
* Ported from the legacy code |
185
|
|
|
* @see https://github.com/ezsystems/ezpublish-legacy/blob/master/kernel/search/plugins/ezsearchengine/ezsearchengine.php#L386 |
186
|
|
|
* |
187
|
|
|
* @param mixed $contentId |
188
|
|
|
* @param mixed|null $versionId |
189
|
|
|
* |
190
|
|
|
* @return bool |
191
|
|
|
*/ |
192
|
|
|
public function remove($contentId, $versionId = null) |
193
|
|
|
{ |
194
|
|
|
$doDelete = false; |
195
|
|
|
$this->dbHandler->beginTransaction(); |
196
|
|
|
// fetch all the words and decrease the object count on all the words |
197
|
|
|
$wordIDList = $this->searchIndex->getContentObjectWords($contentId); |
198
|
|
|
if (count($wordIDList) > 0) { |
199
|
|
|
$this->searchIndex->decrementWordObjectCount($wordIDList); |
200
|
|
|
$doDelete = true; |
201
|
|
|
} |
202
|
|
|
if ($doDelete) { |
203
|
|
|
$this->searchIndex->deleteWordsWithoutObjects(); |
204
|
|
|
$this->searchIndex->deleteObjectWordsLink($contentId); |
205
|
|
|
} |
206
|
|
|
$this->dbHandler->commit(); |
207
|
|
|
|
208
|
|
|
return true; |
209
|
|
|
} |
210
|
|
|
|
211
|
|
|
/** |
212
|
|
|
* Remove entire search index. |
213
|
|
|
*/ |
214
|
|
|
public function purgeIndex() |
215
|
|
|
{ |
216
|
|
|
$this->searchIndex->purge(); |
217
|
|
|
} |
218
|
|
|
|
219
|
|
|
/** |
220
|
|
|
* Index wordIndex. |
221
|
|
|
* |
222
|
|
|
* Ported from the legacy code |
223
|
|
|
* |
224
|
|
|
* @see https://github.com/ezsystems/ezpublish-legacy/blob/master/kernel/search/plugins/ezsearchengine/ezsearchengine.php#L255 |
225
|
|
|
* |
226
|
|
|
* @param \eZ\Publish\Core\Search\Legacy\Content\FullTextData $fullTextData |
227
|
|
|
* @param array $indexArray |
228
|
|
|
* @param array $wordIDArray |
229
|
|
|
* @param int $placement |
230
|
|
|
* |
231
|
|
|
* @return int last placement |
232
|
|
|
*/ |
233
|
|
|
private function indexWords(FullTextData $fullTextData, array $indexArray, array $wordIDArray, $placement = 0) |
234
|
|
|
{ |
235
|
|
|
$contentId = $fullTextData->id; |
236
|
|
|
|
237
|
|
|
$prevWordId = 0; |
238
|
|
|
|
239
|
|
|
for ($i = 0; $i < count($indexArray); ++$i) { |
|
|
|
|
240
|
|
|
$indexWord = $indexArray[$i]['Word']; |
241
|
|
|
$contentFieldId = $indexArray[$i]['ContentClassAttributeID']; |
242
|
|
|
$identifier = $indexArray[$i]['identifier']; |
243
|
|
|
$integerValue = $indexArray[$i]['integer_value']; |
244
|
|
|
$wordId = $wordIDArray[$indexWord]; |
245
|
|
|
|
246
|
|
|
if (isset($indexArray[$i + 1])) { |
247
|
|
|
$nextIndexWord = $indexArray[$i + 1]['Word']; |
248
|
|
|
$nextWordId = $wordIDArray[$nextIndexWord]; |
249
|
|
|
} else { |
250
|
|
|
$nextWordId = 0; |
251
|
|
|
} |
252
|
|
|
$frequency = 0; |
253
|
|
|
$this->searchIndex->addObjectWordLink($wordId, $contentId, $frequency, $placement, $nextWordId, $prevWordId, $fullTextData->contentTypeId, $contentFieldId, $fullTextData->published, $fullTextData->sectionId, $identifier, $integerValue); |
254
|
|
|
$prevWordId = $wordId; |
255
|
|
|
++$placement; |
256
|
|
|
} |
257
|
|
|
|
258
|
|
|
return $placement; |
259
|
|
|
} |
260
|
|
|
|
261
|
|
|
/** |
262
|
|
|
* Build WordIDArray and update ezsearch_word table. |
263
|
|
|
* |
264
|
|
|
* Ported from the legacy code |
265
|
|
|
* |
266
|
|
|
* @see https://github.com/ezsystems/ezpublish-legacy/blob/master/kernel/search/plugins/ezsearchengine/ezsearchengine.php#L155 |
267
|
|
|
* |
268
|
|
|
* @param array $indexArrayOnlyWords words for object to add |
269
|
|
|
* |
270
|
|
|
* @return array wordIDArray |
271
|
|
|
*/ |
272
|
|
|
private function buildWordIDArray(array $indexArrayOnlyWords) |
273
|
|
|
{ |
274
|
|
|
$wordCount = count($indexArrayOnlyWords); |
275
|
|
|
$wordIDArray = []; |
276
|
|
|
$wordArray = []; |
277
|
|
|
|
278
|
|
|
// store the words in the index and remember the ID |
279
|
|
|
$this->dbHandler->beginTransaction(); |
280
|
|
|
for ($arrayCount = 0; $arrayCount < $wordCount; $arrayCount += 500) { |
281
|
|
|
// Fetch already indexed words from database |
282
|
|
|
$wordArrayChuck = array_slice($indexArrayOnlyWords, $arrayCount, 500); |
283
|
|
|
$wordRes = $this->searchIndex->getWords($wordArrayChuck); |
284
|
|
|
|
285
|
|
|
// Build a has of the existing words |
286
|
|
|
$wordResCount = count($wordRes); |
287
|
|
|
$existingWordArray = []; |
288
|
|
|
for ($i = 0; $i < $wordResCount; ++$i) { |
289
|
|
|
$wordIDArray[] = $wordRes[$i]['id']; |
290
|
|
|
$existingWordArray[] = $wordRes[$i]['word']; |
291
|
|
|
$wordArray[$wordRes[$i]['word']] = $wordRes[$i]['id']; |
292
|
|
|
} |
293
|
|
|
|
294
|
|
|
// Update the object count of existing words by one |
295
|
|
|
if (count($wordIDArray) > 0) { |
296
|
|
|
$this->searchIndex->incrementWordObjectCount($wordIDArray); |
297
|
|
|
} |
298
|
|
|
|
299
|
|
|
// Insert if there is any news words |
300
|
|
|
$newWordArray = array_diff($wordArrayChuck, $existingWordArray); |
301
|
|
|
if (count($newWordArray) > 0) { |
302
|
|
|
$this->searchIndex->addWords($newWordArray); |
303
|
|
|
$newWordRes = $this->searchIndex->getWords($newWordArray); |
304
|
|
|
$newWordCount = count($newWordRes); |
305
|
|
|
for ($i = 0; $i < $newWordCount; ++$i) { |
306
|
|
|
$wordLowercase = $this->transformationProcessor->transformByGroup($newWordRes[$i]['word'], 'lowercase'); |
307
|
|
|
$wordArray[$wordLowercase] = $newWordRes[$i]['id']; |
308
|
|
|
} |
309
|
|
|
} |
310
|
|
|
} |
311
|
|
|
$this->dbHandler->commit(); |
312
|
|
|
|
313
|
|
|
return $wordArray; |
314
|
|
|
} |
315
|
|
|
} |
316
|
|
|
|
If the size of the collection does not change during the iteration, it is generally a good practice to compute it beforehand, and not on each iteration: