1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
/* |
6
|
|
|
* This file is part of the TYPO3 CMS project. |
7
|
|
|
* |
8
|
|
|
* It is free software; you can redistribute it and/or modify it under |
9
|
|
|
* the terms of the GNU General Public License, either version 2 |
10
|
|
|
* of the License, or any later version. |
11
|
|
|
* |
12
|
|
|
* For the full copyright and license information, please read the |
13
|
|
|
* LICENSE.txt file that was distributed with this source code. |
14
|
|
|
* |
15
|
|
|
* The TYPO3 project - inspiring people to share! |
16
|
|
|
*/ |
17
|
|
|
|
18
|
|
|
namespace ApacheSolrForTypo3\Solr\IndexQueue; |
19
|
|
|
|
20
|
|
|
use ApacheSolrForTypo3\Solr\ContentObject\Classification; |
21
|
|
|
use ApacheSolrForTypo3\Solr\ContentObject\Multivalue; |
22
|
|
|
use ApacheSolrForTypo3\Solr\ContentObject\Relation; |
23
|
|
|
use ApacheSolrForTypo3\Solr\System\Solr\Document\Document; |
24
|
|
|
use TYPO3\CMS\Core\Core\Environment; |
25
|
|
|
use TYPO3\CMS\Core\TypoScript\Parser\TypoScriptParser; |
26
|
|
|
use TYPO3\CMS\Core\Utility\GeneralUtility; |
27
|
|
|
use TYPO3\CMS\Frontend\Controller\TypoScriptFrontendController; |
28
|
|
|
use UnexpectedValueException; |
29
|
|
|
|
30
|
|
|
/** |
31
|
|
|
* An abstract indexer class to collect a few common methods shared with other |
32
|
|
|
* indexers. |
33
|
|
|
* |
34
|
|
|
* @author Ingo Renner <[email protected]> |
35
|
|
|
*/ |
36
|
|
|
abstract class AbstractIndexer |
37
|
|
|
{ |
38
|
|
|
/** |
39
|
|
|
* Holds the type of the data to be indexed, usually that is the table name. |
40
|
|
|
* |
41
|
|
|
* @var string |
42
|
|
|
*/ |
43
|
|
|
protected string $type = ''; |
44
|
|
|
|
45
|
|
|
/** |
46
|
|
|
* Holds field names that are denied to overwrite in thy indexing configuration. |
47
|
|
|
* |
48
|
|
|
* @var array |
49
|
|
|
*/ |
50
|
|
|
protected static array $unAllowedOverrideFields = ['type']; |
51
|
|
|
|
52
|
|
|
/** |
53
|
|
|
* @param string $solrFieldName |
54
|
|
|
* @return bool |
55
|
|
|
*/ |
56
|
62 |
|
public static function isAllowedToOverrideField(string $solrFieldName): bool |
57
|
|
|
{ |
58
|
62 |
|
return !in_array($solrFieldName, static::$unAllowedOverrideFields); |
59
|
|
|
} |
60
|
|
|
|
61
|
|
|
/** |
62
|
|
|
* Adds fields to the document as defined in $indexingConfiguration |
63
|
|
|
* |
64
|
|
|
* @param Document $document base document to add fields to |
65
|
|
|
* @param array $indexingConfiguration Indexing configuration / mapping |
66
|
|
|
* @param array $data Record data |
67
|
|
|
* @return Document Modified document with added fields |
68
|
|
|
*/ |
69
|
20 |
|
protected function addDocumentFieldsFromTyposcript(Document $document, array $indexingConfiguration, array $data, TypoScriptFrontendController $tsfe): Document |
70
|
|
|
{ |
71
|
20 |
|
$data = static::addVirtualContentFieldToRecord($document, $data); |
72
|
|
|
|
73
|
|
|
// mapping of record fields => solr document fields, resolving cObj |
74
|
20 |
|
foreach ($indexingConfiguration as $solrFieldName => $recordFieldName) { |
75
|
20 |
|
if (is_array($recordFieldName)) { |
76
|
|
|
// configuration for a content object, skipping |
77
|
16 |
|
continue; |
78
|
|
|
} |
79
|
|
|
|
80
|
20 |
|
if (!static::isAllowedToOverrideField($solrFieldName)) { |
81
|
|
|
throw new InvalidFieldNameException( |
82
|
|
|
'Must not overwrite field .' . $solrFieldName, |
83
|
|
|
1435441863 |
84
|
|
|
); |
85
|
|
|
} |
86
|
|
|
|
87
|
20 |
|
$fieldValue = $this->resolveFieldValue($indexingConfiguration, $solrFieldName, $data, $tsfe); |
88
|
|
|
if ($fieldValue === null) { |
89
|
20 |
|
continue; |
90
|
|
|
} |
91
|
9 |
|
|
92
|
|
|
if (is_array($fieldValue)) { |
93
|
20 |
|
// multi value |
94
|
19 |
|
$document->setField($solrFieldName, $fieldValue); |
95
|
|
|
} else { |
96
|
|
|
if ($fieldValue !== '' && $fieldValue !== null) { |
97
|
|
|
$document->setField($solrFieldName, $fieldValue); |
98
|
|
|
} |
99
|
20 |
|
} |
100
|
|
|
} |
101
|
|
|
|
102
|
|
|
return $document; |
103
|
|
|
} |
104
|
|
|
|
105
|
|
|
/** |
106
|
|
|
* Adds the content of the field 'content' from the solr document as virtual field __solr_content in the record, |
107
|
|
|
* to have it available in typoscript. |
108
|
|
|
* |
109
|
|
|
* @param Document $document |
110
|
28 |
|
* @param array $data |
111
|
|
|
* @return array |
112
|
28 |
|
*/ |
113
|
8 |
|
public static function addVirtualContentFieldToRecord(Document $document, array $data): array |
114
|
8 |
|
{ |
115
|
|
|
if (isset($document['content'])) { |
116
|
20 |
|
$data['__solr_content'] = $document['content']; |
117
|
|
|
return $data; |
118
|
|
|
} |
119
|
|
|
return $data; |
120
|
|
|
} |
121
|
|
|
|
122
|
|
|
/** |
123
|
|
|
* Resolves a field to its value depending on its configuration. |
124
|
|
|
* |
125
|
|
|
* This enables you to configure the indexer to put the item/record through |
126
|
|
|
* cObj processing if wanted/needed. Otherwise, the plain item/record value |
127
|
|
|
* is taken. |
128
|
|
|
* |
129
|
|
|
* @param array $indexingConfiguration Indexing configuration as defined in plugin.tx_solr_index.queue.[indexingConfigurationName].fields |
130
|
|
|
* @param string $solrFieldName A Solr field name that is configured in the indexing configuration |
131
|
|
|
* @param array $data A record or item's data |
132
|
20 |
|
* @param TypoScriptFrontendController $tsfe |
133
|
|
|
* @return string|null The resolved string value to be indexed; null if value could not be resolved |
134
|
|
|
*/ |
135
|
|
|
protected function resolveFieldValue( |
136
|
|
|
array $indexingConfiguration, |
137
|
|
|
string $solrFieldName, |
138
|
20 |
|
array $data, |
139
|
|
|
TypoScriptFrontendController $tsfe |
140
|
|
|
) { |
141
|
|
|
if (isset($indexingConfiguration[$solrFieldName . '.'])) { |
142
|
|
|
// configuration found => need to resolve a cObj |
143
|
16 |
|
|
144
|
16 |
|
// need to change directory to make IMAGE content objects work in BE context |
145
|
|
|
// see http://blog.netzelf.de/lang/de/tipps-und-tricks/tslib_cobj-image-im-backend |
146
|
16 |
|
$backupWorkingDirectory = getcwd(); |
147
|
16 |
|
chdir(Environment::getPublicPath() . '/'); |
148
|
16 |
|
|
149
|
16 |
|
$tsfe->cObj->start($data, $this->type); |
150
|
|
|
$fieldValue = $tsfe->cObj->cObjGetSingle( |
151
|
|
|
$indexingConfiguration[$solrFieldName], |
152
|
16 |
|
$indexingConfiguration[$solrFieldName . '.'] |
153
|
|
|
); |
154
|
16 |
|
|
155
|
|
|
chdir($backupWorkingDirectory); |
156
|
|
|
|
157
|
|
|
if ($this->isSerializedValue( |
158
|
|
|
$indexingConfiguration, |
159
|
16 |
|
$solrFieldName |
160
|
|
|
) |
161
|
|
|
) { |
162
|
19 |
|
$fieldValue = unserialize($fieldValue); |
163
|
|
|
} |
164
|
|
|
} elseif ( |
165
|
|
|
substr($indexingConfiguration[$solrFieldName], 0, 1) === '<' |
166
|
|
|
) { |
167
|
|
|
$referencedTsPath = trim(substr( |
168
|
|
|
$indexingConfiguration[$solrFieldName], |
169
|
|
|
1 |
170
|
|
|
)); |
171
|
|
|
$typoScriptParser = GeneralUtility::makeInstance(TypoScriptParser::class); |
172
|
|
|
// $name and $conf is loaded with the referenced values. |
173
|
|
|
list($name, $conf) = $typoScriptParser->getVal($referencedTsPath, $GLOBALS['TSFE']->tmpl->setup); |
174
|
|
|
|
175
|
|
|
// need to change directory to make IMAGE content objects work in BE context |
176
|
|
|
// see http://blog.netzelf.de/lang/de/tipps-und-tricks/tslib_cobj-image-im-backend |
177
|
|
|
$backupWorkingDirectory = getcwd(); |
178
|
|
|
chdir(Environment::getPublicPath() . '/'); |
179
|
|
|
|
180
|
|
|
$tsfe->cObj->start($data, $this->type); |
181
|
|
|
$fieldValue = $tsfe->cObj->cObjGetSingle($name, $conf); |
182
|
|
|
|
183
|
|
|
chdir($backupWorkingDirectory); |
184
|
|
|
|
185
|
|
|
if ($this->isSerializedValue( |
186
|
|
|
$indexingConfiguration, |
187
|
|
|
$solrFieldName |
188
|
|
|
) |
189
|
|
|
) { |
190
|
19 |
|
$fieldValue = unserialize($fieldValue); |
191
|
|
|
} |
192
|
|
|
} else { |
193
|
|
|
$indexingFieldName = $indexingConfiguration[$solrFieldName] ?? null; |
194
|
|
|
if (empty($indexingFieldName) || |
195
|
|
|
!is_string($indexingFieldName) || |
196
|
20 |
|
!array_key_exists($indexingFieldName, $data)) { |
197
|
|
|
return null; |
198
|
20 |
|
} |
199
|
|
|
$fieldValue = $data[$indexingFieldName]; |
200
|
|
|
} |
201
|
20 |
|
|
202
|
9 |
|
// detect and correct type for dynamic fields |
203
|
9 |
|
|
204
|
|
|
// find last underscore, substr from there, cut off last character (S/M) |
205
|
|
|
$fieldType = substr( |
206
|
|
|
$solrFieldName, |
207
|
|
|
strrpos($solrFieldName, '_') + 1, |
208
|
|
|
-1 |
209
|
20 |
|
); |
210
|
|
|
if (is_array($fieldValue)) { |
211
|
|
|
foreach ($fieldValue as $key => $value) { |
212
|
20 |
|
$fieldValue[$key] = $this->ensureFieldValueType( |
213
|
|
|
$value, |
214
|
|
|
$fieldType |
215
|
|
|
); |
216
|
|
|
} |
217
|
|
|
} else { |
218
|
|
|
$fieldValue = $this->ensureFieldValueType($fieldValue, $fieldType); |
219
|
|
|
} |
220
|
|
|
|
221
|
|
|
return $fieldValue; |
222
|
|
|
} |
223
|
|
|
|
224
|
|
|
// Utility methods |
225
|
|
|
|
226
|
27 |
|
/** |
227
|
|
|
* Uses a field's configuration to detect whether its value returned by a |
228
|
27 |
|
* content object is expected to be serialized and thus needs to be |
229
|
26 |
|
* unserialized. |
230
|
|
|
* |
231
|
|
|
* @param array $indexingConfiguration Current item's indexing configuration |
232
|
|
|
* @param string $solrFieldName Current field being indexed |
233
|
|
|
* @return bool TRUE if the value is expected to be serialized, FALSE otherwise |
234
|
|
|
*/ |
235
|
|
|
public static function isSerializedValue(array $indexingConfiguration, string $solrFieldName): bool |
236
|
|
|
{ |
237
|
|
|
return static::isSerializedResultFromRegisteredHook($indexingConfiguration, $solrFieldName) |
238
|
|
|
|| static::isSerializedResultFromCustomContentElement($indexingConfiguration, $solrFieldName); |
239
|
25 |
|
} |
240
|
|
|
|
241
|
25 |
|
/** |
242
|
|
|
* Checks if the response comes from a custom content element that returns a serialized value. |
243
|
|
|
* |
244
|
25 |
|
* @param array $indexingConfiguration |
245
|
1 |
|
* @param string $solrFieldName |
246
|
|
|
* @return bool |
247
|
|
|
*/ |
248
|
|
|
protected static function isSerializedResultFromCustomContentElement(array $indexingConfiguration, string $solrFieldName): bool |
249
|
25 |
|
{ |
250
|
2 |
|
$isSerialized = false; |
251
|
|
|
|
252
|
|
|
// SOLR_CLASSIFICATION - always returns serialized array |
253
|
|
|
if (($indexingConfiguration[$solrFieldName] ?? null) == Classification::CONTENT_OBJECT_NAME) { |
254
|
25 |
|
$isSerialized = true; |
255
|
12 |
|
} |
256
|
|
|
|
257
|
|
|
// SOLR_MULTIVALUE - always returns serialized array |
258
|
25 |
|
if (($indexingConfiguration[$solrFieldName] ?? null) == Multivalue::CONTENT_OBJECT_NAME) { |
259
|
|
|
$isSerialized = true; |
260
|
|
|
} |
261
|
|
|
|
262
|
|
|
// SOLR_RELATION - returns serialized array if multiValue option is set |
263
|
|
|
if (($indexingConfiguration[$solrFieldName] ?? null) == Relation::CONTENT_OBJECT_NAME && !empty($indexingConfiguration[$solrFieldName . '.']['multiValue'])) { |
264
|
|
|
$isSerialized = true; |
265
|
|
|
} |
266
|
|
|
|
267
|
|
|
return $isSerialized; |
268
|
27 |
|
} |
269
|
|
|
|
270
|
27 |
|
/** |
271
|
24 |
|
* Checks registered hooks if a SerializedValueDetector detects a serialized response. |
272
|
|
|
* |
273
|
|
|
* @param array $indexingConfiguration |
274
|
3 |
|
* @param string $solrFieldName |
275
|
2 |
|
* @return bool |
276
|
2 |
|
*/ |
277
|
1 |
|
protected static function isSerializedResultFromRegisteredHook(array $indexingConfiguration, string $solrFieldName): bool |
278
|
1 |
|
{ |
279
|
|
|
if (!is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['detectSerializedValue'] ?? null)) { |
280
|
|
|
return false; |
281
|
1 |
|
} |
282
|
1 |
|
|
283
|
1 |
|
foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['detectSerializedValue'] as $classReference) { |
284
|
|
|
$serializedValueDetector = GeneralUtility::makeInstance($classReference); |
285
|
|
|
if (!$serializedValueDetector instanceof SerializedValueDetector) { |
286
|
1 |
|
$message = get_class($serializedValueDetector) . ' must implement interface ' . SerializedValueDetector::class; |
287
|
|
|
throw new UnexpectedValueException($message, 1404471741); |
288
|
|
|
} |
289
|
|
|
|
290
|
|
|
$isSerialized = (boolean)$serializedValueDetector->isSerializedValue($indexingConfiguration, $solrFieldName); |
291
|
|
|
if ($isSerialized) { |
292
|
|
|
return true; |
293
|
|
|
} |
294
|
|
|
} |
295
|
|
|
return false; |
296
|
20 |
|
} |
297
|
|
|
|
298
|
20 |
|
/** |
299
|
20 |
|
* Makes sure a field's value matches a (dynamic) field's type. |
300
|
20 |
|
* |
301
|
|
|
* @param mixed $value Value to be added to a document |
302
|
|
|
* @param string $fieldType The dynamic field's type |
303
|
|
|
* @return mixed Returns the value in the correct format for the field type |
304
|
20 |
|
*/ |
305
|
20 |
|
protected function ensureFieldValueType($value, string $fieldType) |
306
|
|
|
{ |
307
|
|
|
switch ($fieldType) { |
308
|
20 |
|
case 'int': |
309
|
|
|
case 'tInt': |
310
|
|
|
$value = (int)$value; |
311
|
|
|
break; |
312
|
20 |
|
|
313
|
|
|
case 'float': |
314
|
|
|
case 'tFloat': |
315
|
|
|
$value = (float)$value; |
316
|
|
|
break; |
317
|
|
|
case 'long': |
318
|
|
|
// long and double do not exist in PHP |
319
|
20 |
|
// simply make sure it somehow looks like a number |
320
|
20 |
|
// <insert PHP rant here> |
321
|
20 |
|
case 'tLong': |
322
|
|
|
// remove anything that's not a number or negative/minus sign |
323
|
|
|
$value = preg_replace('/[^0-9\\-]/', '', $value); |
324
|
|
|
if (trim($value) === '') { |
325
|
|
|
$value = 0; |
326
|
|
|
} |
327
|
|
|
break; |
328
|
|
|
case 'double': |
329
|
|
|
case 'tDouble': |
330
|
|
|
case 'tDouble4': |
331
|
|
|
// as long as it's numeric we'll take it, int or float doesn't matter |
332
|
20 |
|
if (!is_numeric($value)) { |
333
|
|
|
$value = 0; |
334
|
|
|
} |
335
|
|
|
break; |
336
|
|
|
|
337
|
|
|
default: |
338
|
|
|
// assume things are correct for non-dynamic fields |
339
|
|
|
} |
340
|
|
|
|
341
|
|
|
return $value; |
342
|
|
|
} |
343
|
|
|
} |
344
|
|
|
|