1
|
|
|
<?php |
2
|
|
|
namespace ApacheSolrForTypo3\Solr; |
3
|
|
|
|
4
|
|
|
/*************************************************************** |
5
|
|
|
* Copyright notice |
6
|
|
|
* |
7
|
|
|
* (c) 2009-2015 Ingo Renner <[email protected]> |
8
|
|
|
* All rights reserved |
9
|
|
|
* |
10
|
|
|
* This script is part of the TYPO3 project. The TYPO3 project is |
11
|
|
|
* free software; you can redistribute it and/or modify |
12
|
|
|
* it under the terms of the GNU General Public License as published by |
13
|
|
|
* the Free Software Foundation; either version 2 of the License, or |
14
|
|
|
* (at your option) any later version. |
15
|
|
|
* |
16
|
|
|
* The GNU General Public License can be found at |
17
|
|
|
* http://www.gnu.org/copyleft/gpl.html. |
18
|
|
|
* |
19
|
|
|
* This script is distributed in the hope that it will be useful, |
20
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of |
21
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
22
|
|
|
* GNU General Public License for more details. |
23
|
|
|
* |
24
|
|
|
* This copyright notice MUST APPEAR in all copies of the script! |
25
|
|
|
***************************************************************/ |
26
|
|
|
|
27
|
|
|
use ApacheSolrForTypo3\Solr\System\Configuration\TypoScriptConfiguration; |
28
|
|
|
use ApacheSolrForTypo3\Solr\System\Solr\Parser\SchemaParser; |
29
|
|
|
use ApacheSolrForTypo3\Solr\System\Solr\Parser\StopWordParser; |
30
|
|
|
use ApacheSolrForTypo3\Solr\System\Solr\Parser\SynonymParser; |
31
|
|
|
use ApacheSolrForTypo3\Solr\System\Solr\Schema\Schema; |
32
|
|
|
use TYPO3\CMS\Core\Utility\GeneralUtility; |
33
|
|
|
use Apache_Solr_HttpTransportException; |
34
|
|
|
|
35
|
|
|
/** |
36
|
|
|
* Solr Service Access |
37
|
|
|
* |
38
|
|
|
* @author Ingo Renner <[email protected]> |
39
|
|
|
*/ |
40
|
|
|
class SolrService extends \Apache_Solr_Service |
41
|
|
|
{ |
42
|
|
|
const LUKE_SERVLET = 'admin/luke'; |
43
|
|
|
const SYSTEM_SERVLET = 'admin/system'; |
44
|
|
|
const PLUGINS_SERVLET = 'admin/plugins'; |
45
|
|
|
const CORES_SERVLET = 'admin/cores'; |
46
|
|
|
const SCHEMA_SERVLET = 'schema'; |
47
|
|
|
const SYNONYMS_SERVLET = 'schema/analysis/synonyms/'; |
48
|
|
|
const STOPWORDS_SERVLET = 'schema/analysis/stopwords/'; |
49
|
|
|
|
50
|
|
|
const SCHEME_HTTP = 'http'; |
51
|
|
|
const SCHEME_HTTPS = 'https'; |
52
|
|
|
|
53
|
|
|
/** |
54
|
|
|
* Server connection scheme. http or https. |
55
|
|
|
* |
56
|
|
|
* @var string |
57
|
|
|
*/ |
58
|
|
|
protected $_scheme = self::SCHEME_HTTP; |
59
|
|
|
|
60
|
|
|
/** |
61
|
|
|
* Constructed servlet URL for Luke |
62
|
|
|
* |
63
|
|
|
* @var string |
64
|
|
|
*/ |
65
|
|
|
protected $_lukeUrl; |
66
|
|
|
|
67
|
|
|
/** |
68
|
|
|
* Constructed servlet URL for plugin information |
69
|
|
|
* |
70
|
|
|
* @var string |
71
|
|
|
*/ |
72
|
|
|
protected $_pluginsUrl; |
73
|
|
|
|
74
|
|
|
/** |
75
|
|
|
* @var string |
76
|
|
|
*/ |
77
|
|
|
protected $_coresUrl; |
78
|
|
|
|
79
|
|
|
/** |
80
|
|
|
* @var string |
81
|
|
|
*/ |
82
|
|
|
protected $_extractUrl; |
83
|
|
|
|
84
|
|
|
/** |
85
|
|
|
* @var string |
86
|
|
|
*/ |
87
|
|
|
protected $_synonymsUrl; |
88
|
|
|
|
89
|
|
|
/** |
90
|
|
|
* @var string |
91
|
|
|
*/ |
92
|
|
|
protected $_stopWordsUrl; |
93
|
|
|
|
94
|
|
|
/** |
95
|
|
|
* @var string |
96
|
|
|
*/ |
97
|
|
|
protected $_schemaUrl; |
98
|
|
|
|
99
|
|
|
/** |
100
|
|
|
* @var bool |
101
|
|
|
*/ |
102
|
|
|
protected $debug = false; |
103
|
|
|
|
104
|
|
|
/** |
105
|
|
|
* @var \Apache_Solr_Response |
106
|
|
|
*/ |
107
|
|
|
protected $responseCache = null; |
108
|
|
|
|
109
|
|
|
/** |
110
|
|
|
* @var bool |
111
|
|
|
*/ |
112
|
|
|
protected $hasSearched = false; |
113
|
|
|
|
114
|
|
|
/** |
115
|
|
|
* @var array |
116
|
|
|
*/ |
117
|
|
|
protected $lukeData = []; |
118
|
|
|
|
119
|
|
|
protected $systemData = null; |
120
|
|
|
protected $pluginsData = null; |
121
|
|
|
|
122
|
|
|
protected $solrconfigName = null; |
123
|
|
|
|
124
|
|
|
/** |
125
|
|
|
* @var TypoScriptConfiguration |
126
|
|
|
*/ |
127
|
|
|
protected $configuration; |
128
|
|
|
|
129
|
|
|
/** |
130
|
|
|
* @var array |
131
|
|
|
*/ |
132
|
|
|
protected static $pingCache = []; |
133
|
|
|
|
134
|
|
|
/** |
135
|
|
|
* @var SynonymParser |
136
|
|
|
*/ |
137
|
|
|
protected $synonymParser = null; |
138
|
|
|
|
139
|
|
|
/** |
140
|
|
|
* @var StopWordParser |
141
|
|
|
*/ |
142
|
|
|
protected $stopWordParser = null; |
143
|
|
|
|
144
|
|
|
/** |
145
|
|
|
* @var SchemaParser |
146
|
|
|
*/ |
147
|
|
|
protected $schemaParser = null; |
148
|
|
|
|
149
|
|
|
/** |
150
|
|
|
* @var Schema |
151
|
|
|
*/ |
152
|
|
|
protected $schema; |
153
|
|
|
|
154
|
|
|
/** |
155
|
|
|
* Constructor |
156
|
|
|
* |
157
|
|
|
* @param string $host Solr host |
158
|
|
|
* @param string $port Solr port |
159
|
|
|
* @param string $path Solr path |
160
|
|
|
* @param string $scheme Scheme, defaults to http, can be https |
161
|
|
|
* @param TypoScriptConfiguration $typoScriptConfiguration |
162
|
|
|
* @param SynonymParser $synonymParser |
163
|
|
|
* @param StopWordParser $stopWordParser |
164
|
|
|
* @param SchemaParser $schemaParser |
165
|
|
|
*/ |
166
|
82 |
|
public function __construct( |
167
|
|
|
$host = '', |
168
|
|
|
$port = '8983', |
169
|
|
|
$path = '/solr/', |
170
|
|
|
$scheme = 'http', |
171
|
|
|
TypoScriptConfiguration $typoScriptConfiguration = null, |
172
|
|
|
SynonymParser $synonymParser = null, |
173
|
|
|
StopWordParser $stopWordParser = null, |
174
|
|
|
SchemaParser $schemaParser = null |
175
|
|
|
) { |
176
|
82 |
|
$this->setScheme($scheme); |
177
|
81 |
|
$this->configuration = is_null($typoScriptConfiguration) ? Util::getSolrConfiguration() : $typoScriptConfiguration; |
178
|
81 |
|
$this->synonymParser = is_null($synonymParser) ? GeneralUtility::makeInstance(SynonymParser::class) : $synonymParser; |
179
|
81 |
|
$this->stopWordParser = is_null($stopWordParser) ? GeneralUtility::makeInstance(StopWordParser::class) : $stopWordParser; |
180
|
81 |
|
$this->schemaParser = is_null($schemaParser) ? GeneralUtility::makeInstance(SchemaParser::class) : $schemaParser; |
181
|
|
|
|
182
|
81 |
|
$this->initializeTimeoutFromConfiguration(); |
183
|
|
|
|
184
|
81 |
|
parent::__construct($host, $port, $path); |
185
|
81 |
|
} |
186
|
|
|
|
187
|
|
|
/** |
188
|
|
|
* Initializes the timeout from TypoScript when configuration is present. |
189
|
|
|
* |
190
|
|
|
* @return void |
191
|
|
|
*/ |
192
|
81 |
|
protected function initializeTimeoutFromConfiguration() |
193
|
|
|
{ |
194
|
81 |
|
$timeout = $this->configuration->getSolrTimeout(); |
195
|
81 |
|
if ($timeout > 0) { |
196
|
1 |
|
$this->getHttpTransport()->setDefaultTimeout($timeout); |
197
|
|
|
} |
198
|
81 |
|
} |
199
|
|
|
|
200
|
|
|
/** |
201
|
|
|
* Creates a string representation of the Solr connection. Specifically |
202
|
|
|
* will return the Solr URL. |
203
|
|
|
* |
204
|
|
|
* @return string The Solr URL. |
205
|
|
|
*/ |
206
|
50 |
|
public function __toString() |
207
|
|
|
{ |
208
|
50 |
|
return $this->_scheme . '://' . $this->_host . ':' . $this->_port . $this->_path; |
209
|
|
|
} |
210
|
|
|
|
211
|
|
|
/** |
212
|
|
|
* Returns the current time in milliseconds. |
213
|
|
|
* |
214
|
|
|
* @return double |
215
|
|
|
*/ |
216
|
1 |
|
protected function getMilliseconds() |
217
|
|
|
{ |
218
|
1 |
|
return GeneralUtility::milliseconds(); |
219
|
|
|
} |
220
|
|
|
|
221
|
|
|
/** |
222
|
|
|
* Performs a search. |
223
|
|
|
* |
224
|
|
|
* @param string $query query string / search term |
225
|
|
|
* @param int $offset result offset for pagination |
226
|
|
|
* @param int $limit number of results to retrieve |
227
|
|
|
* @param array $params additional HTTP GET parameters |
228
|
|
|
* @param string $method The HTTP method (Apache_Solr_Service::METHOD_GET or Apache_Solr_Service::METHOD::POST) |
229
|
|
|
* @return \Apache_Solr_Response Solr response |
230
|
|
|
* @throws \RuntimeException if Solr returns a HTTP status code other than 200 |
231
|
|
|
*/ |
232
|
26 |
|
public function search($query, $offset = 0, $limit = 10, $params = [], $method = self::METHOD_GET) |
233
|
|
|
{ |
234
|
26 |
|
$response = parent::search($query, $offset, $limit, $params, $method); |
235
|
26 |
|
$this->hasSearched = true; |
236
|
|
|
|
237
|
26 |
|
$this->responseCache = $response; |
238
|
|
|
|
239
|
26 |
|
if ($response->getHttpStatus() != 200) { |
240
|
1 |
|
throw new \RuntimeException( |
241
|
|
|
'Invalid query. Solr returned an error: ' |
242
|
1 |
|
. $response->getHttpStatus() . ' ' |
243
|
1 |
|
. $response->getHttpStatusMessage(), |
244
|
1 |
|
1293109870 |
245
|
|
|
); |
246
|
|
|
} |
247
|
|
|
|
248
|
25 |
|
return $response; |
249
|
|
|
} |
250
|
|
|
|
251
|
|
|
/** |
252
|
|
|
* Call the /admin/ping servlet, can be used to quickly tell if a connection to the |
253
|
|
|
* server is available. |
254
|
|
|
* |
255
|
|
|
* Simply overrides the SolrPhpClient implementation, changing ping from a |
256
|
|
|
* HEAD to a GET request, see http://forge.typo3.org/issues/44167 |
257
|
|
|
* |
258
|
|
|
* Also does not report the time, see https://forge.typo3.org/issues/64551 |
259
|
|
|
* |
260
|
|
|
* @param float|int $timeout maximum time to wait for ping in seconds, -1 for unlimited (default is 2) |
261
|
|
|
* @param boolean $useCache indicates if the ping result should be cached in the instance or not |
262
|
|
|
* @return bool TRUE if Solr can be reached, FALSE if not |
263
|
|
|
*/ |
264
|
47 |
|
public function ping($timeout = 2, $useCache = true) |
265
|
|
|
{ |
266
|
47 |
|
$httpResponse = $this->performPingRequest($timeout, $useCache); |
267
|
47 |
|
return ($httpResponse->getStatusCode() === 200); |
268
|
|
|
} |
269
|
|
|
|
270
|
|
|
/** |
271
|
|
|
* Call the /admin/ping servlet, can be used to get the runtime of a ping request. |
272
|
|
|
* |
273
|
|
|
* @param float|int $timeout maximum time to wait for ping in seconds, -1 for unlimited (default is 2) |
274
|
|
|
* @param boolean $useCache indicates if the ping result should be cached in the instance or not |
275
|
|
|
* @return double runtime in milliseconds |
276
|
|
|
* @throws \ApacheSolrForTypo3\Solr\PingFailedException |
277
|
|
|
*/ |
278
|
1 |
|
public function getPingRoundTripRuntime($timeout = 2, $useCache = true) |
279
|
|
|
{ |
280
|
1 |
|
$start = $this->getMilliseconds(); |
281
|
1 |
|
$httpResponse = $this->performPingRequest($timeout, $useCache); |
282
|
1 |
|
$end = $this->getMilliseconds(); |
283
|
|
|
|
284
|
1 |
|
if ($httpResponse->getStatusCode() !== 200) { |
285
|
|
|
$message = 'Solr ping failed with unexpected response code: ' . $httpResponse->getStatusCode(); |
286
|
|
|
/** @var $exception \ApacheSolrForTypo3\Solr\PingFailedException */ |
287
|
|
|
$exception = GeneralUtility::makeInstance(PingFailedException::class, $message); |
288
|
|
|
$exception->setHttpResponse($httpResponse); |
289
|
|
|
throw $exception; |
290
|
|
|
} |
291
|
|
|
|
292
|
1 |
|
return $end - $start; |
293
|
|
|
} |
294
|
|
|
|
295
|
|
|
/** |
296
|
|
|
* Performs a ping request and returns the result. |
297
|
|
|
* |
298
|
|
|
* @param int $timeout |
299
|
|
|
* @param boolean $useCache indicates if the ping result should be cached in the instance or not |
300
|
|
|
* @return \Apache_Solr_HttpTransport_Response |
301
|
|
|
*/ |
302
|
48 |
|
protected function performPingRequest($timeout = 2, $useCache = true) |
303
|
|
|
{ |
304
|
48 |
|
$cacheKey = (string) ($this); |
305
|
48 |
|
if ($useCache && isset(static::$pingCache[$cacheKey])) { |
306
|
42 |
|
return static::$pingCache[$cacheKey]; |
307
|
|
|
} |
308
|
|
|
|
309
|
48 |
|
$pingResult = $this->getHttpTransport()->performGetRequest($this->_pingUrl, $timeout); |
310
|
|
|
|
311
|
48 |
|
if ($useCache) { |
312
|
47 |
|
static::$pingCache[$cacheKey] = $pingResult; |
313
|
|
|
} |
314
|
|
|
|
315
|
48 |
|
return $pingResult; |
316
|
|
|
} |
317
|
|
|
|
318
|
|
|
/** |
319
|
|
|
* Performs a content and meta data extraction request. |
320
|
|
|
* |
321
|
|
|
* @param ExtractingQuery $query An extraction query |
322
|
|
|
* @return array An array containing the extracted content [0] and meta data [1] |
323
|
|
|
*/ |
324
|
1 |
|
public function extractByQuery(ExtractingQuery $query) |
325
|
|
|
{ |
326
|
|
|
$headers = [ |
327
|
1 |
|
'Content-Type' => 'multipart/form-data; boundary=' . $query->getMultiPartPostDataBoundary() |
328
|
|
|
]; |
329
|
|
|
|
330
|
|
|
try { |
331
|
1 |
|
$response = $this->requestServlet( |
332
|
1 |
|
self::EXTRACT_SERVLET, |
333
|
1 |
|
$query->getQueryParameters(), |
334
|
1 |
|
'POST', |
335
|
|
|
$headers, |
336
|
1 |
|
$query->getRawPostFileData() |
337
|
|
|
); |
338
|
|
|
} catch (\Exception $e) { |
339
|
|
|
GeneralUtility::devLog('Extracting text and meta data through Solr Cell over HTTP POST', |
340
|
|
|
'solr', 3, [ |
341
|
|
|
'query' => (array)$query, |
342
|
|
|
'parameters' => $query->getQueryParameters(), |
343
|
|
|
'file' => $query->getFile(), |
344
|
|
|
'headers' => $headers, |
345
|
|
|
'query url' => self::EXTRACT_SERVLET, |
346
|
|
|
'exception' => $e->getMessage() |
347
|
|
|
]); |
348
|
|
|
} |
349
|
|
|
|
350
|
|
|
return [ |
351
|
1 |
|
$response->extracted, |
352
|
1 |
|
(array)$response->extracted_metadata |
353
|
|
|
]; |
354
|
|
|
} |
355
|
|
|
|
356
|
|
|
/** |
357
|
|
|
* Make a request to a servlet (a path) that's not a standard path. |
358
|
|
|
* |
359
|
|
|
* @param string $servlet Path to be added to the base Solr path. |
360
|
|
|
* @param array $parameters Optional, additional request parameters when constructing the URL. |
361
|
|
|
* @param string $method HTTP method to use, defaults to GET. |
362
|
|
|
* @param array $requestHeaders Key value pairs of header names and values. Should include 'Content-Type' for POST and PUT. |
363
|
|
|
* @param string $rawPost Must be an empty string unless method is POST or PUT. |
364
|
|
|
* @param float|bool $timeout Read timeout in seconds, defaults to FALSE. |
365
|
|
|
* @return \Apache_Solr_Response Response object |
366
|
|
|
* @throws \Apache_Solr_HttpTransportException if returned HTTP status is other than 200 |
367
|
|
|
*/ |
368
|
1 |
|
public function requestServlet($servlet, $parameters = [], $method = 'GET', $requestHeaders = [], $rawPost = '', $timeout = false) |
369
|
|
|
{ |
370
|
|
|
// Add default parameters |
371
|
1 |
|
$parameters['wt'] = self::SOLR_WRITER; |
372
|
1 |
|
$parameters['json.nl'] = $this->_namedListTreatment; |
373
|
1 |
|
$url = $this->_constructUrl($servlet, $parameters); |
374
|
|
|
|
375
|
1 |
|
$httpResponse = $this->getResponseFromTransport($url, $method, $requestHeaders, $rawPost, $timeout); |
376
|
1 |
|
$solrResponse = new \Apache_Solr_Response($httpResponse, $this->_createDocuments, $this->_collapseSingleValueArrays); |
377
|
1 |
|
if ($solrResponse->getHttpStatus() != 200) { |
378
|
|
|
throw new \Apache_Solr_HttpTransportException($solrResponse); |
379
|
|
|
} |
380
|
|
|
|
381
|
1 |
|
return $solrResponse; |
382
|
|
|
} |
383
|
|
|
|
384
|
|
|
/** |
385
|
|
|
* Decides which transport method to used, depending on the request method and retrieves the response. |
386
|
|
|
* |
387
|
|
|
* @param string $url |
388
|
|
|
* @param string $method |
389
|
|
|
* @param array $requestHeaders |
390
|
|
|
* @param string $rawPost |
391
|
|
|
* @param float|bool $timeout |
392
|
|
|
* @return \Apache_Solr_HttpTransport_Response |
393
|
|
|
*/ |
394
|
1 |
|
protected function getResponseFromTransport($url, $method, $requestHeaders, $rawPost, $timeout) |
395
|
|
|
{ |
396
|
1 |
|
$httpTransport = $this->getHttpTransport(); |
397
|
|
|
|
398
|
1 |
|
if ($method == self::METHOD_GET) { |
399
|
|
|
return $httpTransport->performGetRequest($url, $timeout); |
400
|
|
|
} |
401
|
1 |
|
if ($method == self::METHOD_POST) { |
402
|
|
|
// FIXME should respect all headers, not only Content-Type |
403
|
1 |
|
return $httpTransport->performPostRequest($url, $rawPost, $requestHeaders['Content-Type'], $timeout); |
404
|
|
|
} |
405
|
|
|
|
406
|
|
|
throw new \InvalidArgumentException('$method should be GET or POST'); |
407
|
|
|
} |
408
|
|
|
|
409
|
|
|
/** |
410
|
|
|
* Return a valid http URL given this server's scheme, host, port, and path |
411
|
|
|
* and a provided servlet name. |
412
|
|
|
* |
413
|
|
|
* @param string $servlet Servlet name |
414
|
|
|
* @param array $params Additional URL parameters to attach to the end of the URL |
415
|
|
|
* @return string Servlet URL |
416
|
|
|
*/ |
417
|
81 |
|
protected function _constructUrl($servlet, $params = []) |
418
|
|
|
{ |
419
|
81 |
|
$url = parent::_constructUrl($servlet, $params); |
420
|
|
|
|
421
|
81 |
|
if (!GeneralUtility::isFirstPartOfStr($url, $this->_scheme)) { |
422
|
2 |
|
$parsedUrl = parse_url($url); |
423
|
|
|
|
424
|
|
|
// unfortunately can't use str_replace as it replace all |
425
|
|
|
// occurrences of $needle and can't be limited to replace only once |
426
|
2 |
|
$url = $this->_scheme . substr($url, strlen($parsedUrl['scheme'])); |
427
|
|
|
} |
428
|
|
|
|
429
|
81 |
|
return $url; |
430
|
|
|
} |
431
|
|
|
|
432
|
|
|
/** |
433
|
|
|
* Returns the set scheme |
434
|
|
|
* |
435
|
|
|
* @return string |
436
|
|
|
*/ |
437
|
1 |
|
public function getScheme() |
438
|
|
|
{ |
439
|
1 |
|
return $this->_scheme; |
440
|
|
|
} |
441
|
|
|
|
442
|
|
|
/** |
443
|
|
|
* Set the scheme used. If empty will fallback to constants |
444
|
|
|
* |
445
|
|
|
* @param string $scheme Either http or https |
446
|
|
|
* @throws \UnexpectedValueException |
447
|
|
|
*/ |
448
|
82 |
|
public function setScheme($scheme) |
449
|
|
|
{ |
450
|
|
|
// Use the provided scheme or use the default |
451
|
82 |
|
if (empty($scheme)) { |
452
|
2 |
|
throw new \UnexpectedValueException('Scheme parameter is empty', 1380756390); |
453
|
|
|
} |
454
|
|
|
|
455
|
81 |
|
$isHttpOrHttps = in_array($scheme, [self::SCHEME_HTTP, self::SCHEME_HTTPS]); |
456
|
81 |
|
if (!$isHttpOrHttps) { |
457
|
1 |
|
throw new \UnexpectedValueException('Unsupported scheme parameter, scheme must be http or https', 1380756442); |
458
|
|
|
} |
459
|
|
|
|
460
|
|
|
// we have a valid scheme |
461
|
81 |
|
$this->_scheme = $scheme; |
462
|
|
|
|
463
|
81 |
|
if ($this->_urlsInited) { |
464
|
1 |
|
$this->_initUrls(); |
465
|
|
|
} |
466
|
81 |
|
} |
467
|
|
|
|
468
|
|
|
/** |
469
|
|
|
* get field meta data for the index |
470
|
|
|
* |
471
|
|
|
* @param int $numberOfTerms Number of top terms to fetch for each field |
472
|
|
|
* @return array |
473
|
|
|
*/ |
474
|
|
|
public function getFieldsMetaData($numberOfTerms = 0) |
475
|
|
|
{ |
476
|
|
|
return $this->getLukeMetaData($numberOfTerms)->fields; |
477
|
|
|
} |
478
|
|
|
|
479
|
|
|
/** |
480
|
|
|
* Retrieves meta data about the index from the luke request handler |
481
|
|
|
* |
482
|
|
|
* @param int $numberOfTerms Number of top terms to fetch for each field |
483
|
|
|
* @return \Apache_Solr_Response Index meta data |
484
|
|
|
*/ |
485
|
|
|
public function getLukeMetaData($numberOfTerms = 0) |
486
|
|
|
{ |
487
|
|
|
if (!isset($this->lukeData[$numberOfTerms])) { |
488
|
|
|
$lukeUrl = $this->_constructUrl( |
489
|
|
|
self::LUKE_SERVLET, |
490
|
|
|
[ |
491
|
|
|
'numTerms' => $numberOfTerms, |
492
|
|
|
'wt' => self::SOLR_WRITER, |
493
|
|
|
'fl' => '*' |
494
|
|
|
] |
495
|
|
|
); |
496
|
|
|
|
497
|
|
|
$this->lukeData[$numberOfTerms] = $this->_sendRawGet($lukeUrl); |
498
|
|
|
} |
499
|
|
|
|
500
|
|
|
return $this->lukeData[$numberOfTerms]; |
501
|
|
|
} |
502
|
|
|
|
503
|
|
|
/** |
504
|
|
|
* Central method for making a get operation against this Solr Server |
505
|
|
|
* |
506
|
|
|
* @param string $url |
507
|
|
|
* @param float|bool $timeout Read timeout in seconds |
508
|
|
|
* @return \Apache_Solr_Response |
509
|
|
|
*/ |
510
|
35 |
|
protected function _sendRawGet($url, $timeout = false) |
511
|
|
|
{ |
512
|
35 |
|
$logSeverity = 0; // info |
513
|
|
|
|
514
|
|
|
try { |
515
|
35 |
|
$response = parent::_sendRawGet($url, $timeout); |
516
|
2 |
|
} catch (Apache_Solr_HttpTransportException $e) { |
|
|
|
|
517
|
2 |
|
$logSeverity = 3; // fatal error |
518
|
2 |
|
$response = $e->getResponse(); |
519
|
|
|
} |
520
|
|
|
|
521
|
35 |
|
if ($this->configuration->getLoggingQueryRawGet() || $response->getHttpStatus() != 200) { |
522
|
|
|
$logData = [ |
523
|
2 |
|
'query url' => $url, |
524
|
2 |
|
'response' => (array)$response |
525
|
|
|
]; |
526
|
|
|
|
527
|
2 |
|
if (!empty($e)) { |
528
|
2 |
|
$logData['exception'] = $e->__toString(); |
529
|
|
|
} else { |
530
|
|
|
// trigger data parsing |
531
|
|
|
$response->response; |
532
|
|
|
$logData['response data'] = print_r($response, true); |
533
|
|
|
} |
534
|
|
|
|
535
|
2 |
|
GeneralUtility::devLog('Querying Solr using GET', 'solr', |
536
|
|
|
$logSeverity, $logData); |
537
|
|
|
} |
538
|
|
|
|
539
|
35 |
|
return $response; |
540
|
|
|
} |
541
|
|
|
|
542
|
|
|
/** |
543
|
|
|
* Returns whether a search has been executed or not. |
544
|
|
|
* |
545
|
|
|
* @return bool TRUE if a search has been executed, FALSE otherwise |
546
|
|
|
*/ |
547
|
|
|
public function hasSearched() |
548
|
|
|
{ |
549
|
|
|
return $this->hasSearched; |
550
|
|
|
} |
551
|
|
|
|
552
|
|
|
/** |
553
|
|
|
* Gets the most recent response (if any) |
554
|
|
|
* |
555
|
|
|
* @return \Apache_Solr_Response Most recent response, or NULL if a search has not been executed yet. |
556
|
|
|
*/ |
557
|
1 |
|
public function getResponse() |
558
|
|
|
{ |
559
|
1 |
|
return $this->responseCache; |
560
|
|
|
} |
561
|
|
|
|
562
|
|
|
/** |
563
|
|
|
* Enable/Disable debug mode |
564
|
|
|
* |
565
|
|
|
* @param bool $debug TRUE to enable debug mode, FALSE to turn off, off by default |
566
|
|
|
*/ |
567
|
|
|
public function setDebug($debug) |
568
|
|
|
{ |
569
|
|
|
$this->debug = (boolean)$debug; |
570
|
|
|
} |
571
|
|
|
|
572
|
|
|
/** |
573
|
|
|
* Gets information about the plugins installed in Solr |
574
|
|
|
* |
575
|
|
|
* @return array A nested array of plugin data. |
576
|
|
|
*/ |
577
|
|
|
public function getPluginsInformation() |
578
|
|
|
{ |
579
|
|
|
if (empty($this->pluginsData)) { |
580
|
|
|
$pluginsInformation = $this->_sendRawGet($this->_pluginsUrl); |
581
|
|
|
|
582
|
|
|
// access a random property to trigger response parsing |
583
|
|
|
$pluginsInformation->responseHeader; |
584
|
|
|
$this->pluginsData = $pluginsInformation; |
585
|
|
|
} |
586
|
|
|
|
587
|
|
|
return $this->pluginsData; |
588
|
|
|
} |
589
|
|
|
|
590
|
|
|
/** |
591
|
|
|
* Gets the name of the schema.xml file installed and in use on the Solr |
592
|
|
|
* server. |
593
|
|
|
* |
594
|
|
|
* @deprecated use getSchema()->getName() instead will be removed in 7.0 |
595
|
|
|
* @return string Name of the active schema.xml |
596
|
|
|
*/ |
597
|
|
|
public function getSchemaName() |
598
|
|
|
{ |
599
|
|
|
GeneralUtility::logDeprecatedFunction(); |
600
|
|
|
return $this->getSchema()->getName(); |
601
|
|
|
} |
602
|
|
|
|
603
|
|
|
/** |
604
|
|
|
* Gets information about the Solr server |
605
|
|
|
* |
606
|
|
|
* @return array A nested array of system data. |
607
|
|
|
*/ |
608
|
3 |
|
public function getSystemInformation() |
609
|
|
|
{ |
610
|
3 |
|
if (empty($this->systemData)) { |
611
|
3 |
|
$systemInformation = $this->system(); |
612
|
|
|
|
613
|
|
|
// access a random property to trigger response parsing |
614
|
3 |
|
$systemInformation->responseHeader; |
615
|
3 |
|
$this->systemData = $systemInformation; |
616
|
|
|
} |
617
|
|
|
|
618
|
3 |
|
return $this->systemData; |
619
|
|
|
} |
620
|
|
|
|
621
|
|
|
/** |
622
|
|
|
* Gets the name of the solrconfig.xml file installed and in use on the Solr |
623
|
|
|
* server. |
624
|
|
|
* |
625
|
|
|
* @return string Name of the active solrconfig.xml |
626
|
|
|
*/ |
627
|
1 |
|
public function getSolrconfigName() |
628
|
|
|
{ |
629
|
1 |
|
if (is_null($this->solrconfigName)) { |
630
|
1 |
|
$solrconfigXmlUrl = $this->_scheme . '://' |
631
|
1 |
|
. $this->_host . ':' . $this->_port |
632
|
1 |
|
. $this->_path . 'admin/file/?file=solrconfig.xml'; |
633
|
1 |
|
$response= $this->_sendRawGet($solrconfigXmlUrl); |
634
|
|
|
|
635
|
1 |
|
$solrconfigXml = simplexml_load_string($response->getRawResponse()); |
636
|
1 |
|
if ($solrconfigXml === false) { |
637
|
|
|
throw new \InvalidArgumentException('No valid xml response from schema file: ' . $solrconfigXmlUrl); |
638
|
|
|
} |
639
|
1 |
|
$this->solrconfigName = (string)$solrconfigXml->attributes()->name; |
640
|
|
|
} |
641
|
|
|
|
642
|
1 |
|
return $this->solrconfigName; |
643
|
|
|
} |
644
|
|
|
|
645
|
|
|
/** |
646
|
|
|
* Gets the Solr server's version number. |
647
|
|
|
* |
648
|
|
|
* @return string Solr version number |
649
|
|
|
*/ |
650
|
2 |
|
public function getSolrServerVersion() |
651
|
|
|
{ |
652
|
2 |
|
$systemInformation = $this->getSystemInformation(); |
653
|
|
|
|
654
|
|
|
// don't know why $systemInformation->lucene->solr-spec-version won't work |
655
|
2 |
|
$luceneInformation = (array)$systemInformation->lucene; |
656
|
2 |
|
return $luceneInformation['solr-spec-version']; |
657
|
|
|
} |
658
|
|
|
|
659
|
|
|
/** |
660
|
|
|
* Deletes all index documents of a certain type and does a commit |
661
|
|
|
* afterwards. |
662
|
|
|
* |
663
|
|
|
* @param string $type The type of documents to delete, usually a table name. |
664
|
|
|
* @param bool $commit Will commit immediately after deleting the documents if set, defaults to TRUE |
665
|
|
|
*/ |
666
|
|
|
public function deleteByType($type, $commit = true) |
667
|
|
|
{ |
668
|
|
|
$this->deleteByQuery('type:' . trim($type)); |
669
|
|
|
|
670
|
|
|
if ($commit) { |
671
|
|
|
$this->commit(false, false, false); |
672
|
|
|
} |
673
|
|
|
} |
674
|
|
|
|
675
|
|
|
/** |
676
|
|
|
* Raw Delete Method. Takes a raw post body and sends it to the update service. Body should be |
677
|
|
|
* a complete and well formed "delete" xml document |
678
|
|
|
* |
679
|
|
|
* @param string $rawPost Expected to be utf-8 encoded xml document |
680
|
|
|
* @param float|int $timeout Maximum expected duration of the delete operation on the server (otherwise, will throw a communication exception) |
681
|
|
|
* @return \Apache_Solr_Response |
682
|
|
|
*/ |
683
|
11 |
|
public function delete($rawPost, $timeout = 3600) |
684
|
|
|
{ |
685
|
11 |
|
$response = $this->_sendRawPost($this->_updateUrl, $rawPost, $timeout); |
686
|
|
|
|
687
|
11 |
|
GeneralUtility::devLog( |
688
|
11 |
|
'Delete Query sent.', |
689
|
11 |
|
'solr', |
690
|
11 |
|
1, |
691
|
|
|
[ |
692
|
11 |
|
'query' => $rawPost, |
693
|
11 |
|
'query url' => $this->_updateUrl, |
694
|
11 |
|
'response' => (array)$response |
695
|
|
|
] |
696
|
|
|
); |
697
|
|
|
|
698
|
11 |
|
return $response; |
699
|
|
|
} |
700
|
|
|
|
701
|
|
|
/** |
702
|
|
|
* Central method for making a post operation against this Solr Server |
703
|
|
|
* |
704
|
|
|
* @param string $url |
705
|
|
|
* @param string $rawPost |
706
|
|
|
* @param float|bool $timeout Read timeout in seconds |
707
|
|
|
* @param string $contentType |
708
|
|
|
* @return \Apache_Solr_Response |
709
|
|
|
*/ |
710
|
59 |
|
protected function _sendRawPost( |
711
|
|
|
$url, |
712
|
|
|
$rawPost, |
713
|
|
|
$timeout = false, |
714
|
|
|
$contentType = 'text/xml; charset=UTF-8' |
715
|
|
|
) { |
716
|
59 |
|
$logSeverity = 0; // info |
717
|
|
|
|
718
|
|
|
try { |
719
|
59 |
|
$response = parent::_sendRawPost($url, $rawPost, $timeout, |
720
|
|
|
$contentType); |
721
|
|
|
} catch (Apache_Solr_HttpTransportException $e) { |
|
|
|
|
722
|
|
|
$logSeverity = 3; // fatal error |
723
|
|
|
$response = $e->getResponse(); |
724
|
|
|
} |
725
|
|
|
|
726
|
59 |
|
if ($this->configuration->getLoggingQueryRawPost() || $response->getHttpStatus() != 200) { |
727
|
|
|
$logData = [ |
728
|
|
|
'query url' => $url, |
729
|
|
|
'content' => $rawPost, |
730
|
|
|
'response' => (array)$response |
731
|
|
|
]; |
732
|
|
|
|
733
|
|
|
if (!empty($e)) { |
734
|
|
|
$logData['exception'] = $e->__toString(); |
735
|
|
|
} |
736
|
|
|
|
737
|
|
|
GeneralUtility::devLog('Querying Solr using POST', 'solr', |
738
|
|
|
$logSeverity, $logData); |
739
|
|
|
} |
740
|
|
|
|
741
|
59 |
|
return $response; |
742
|
|
|
} |
743
|
|
|
|
744
|
|
|
/** |
745
|
|
|
* Get currently configured synonyms |
746
|
|
|
* |
747
|
|
|
* @param string $baseWord If given a base word, retrieves the synonyms for that word only |
748
|
|
|
* @return array |
749
|
|
|
*/ |
750
|
1 |
|
public function getSynonyms($baseWord = '') |
751
|
|
|
{ |
752
|
1 |
|
$this->initializeSynonymsUrl(); |
753
|
1 |
|
$synonymsUrl = $this->_synonymsUrl; |
754
|
1 |
|
if (!empty($baseWord)) { |
755
|
1 |
|
$synonymsUrl .= '/' . $baseWord; |
756
|
|
|
} |
757
|
|
|
|
758
|
1 |
|
$response = $this->_sendRawGet($synonymsUrl); |
759
|
1 |
|
return $this->synonymParser->parseJson($baseWord, $response->getRawResponse()); |
760
|
|
|
} |
761
|
|
|
|
762
|
|
|
/** |
763
|
|
|
* Add list of synonyms for base word to managed synonyms map |
764
|
|
|
* |
765
|
|
|
* @param string $baseWord |
766
|
|
|
* @param array $synonyms |
767
|
|
|
* |
768
|
|
|
* @return \Apache_Solr_Response |
769
|
|
|
* |
770
|
|
|
* @throws \Apache_Solr_InvalidArgumentException If $baseWord or $synonyms are empty |
771
|
|
|
*/ |
772
|
1 |
|
public function addSynonym($baseWord, array $synonyms) |
773
|
|
|
{ |
774
|
1 |
|
$this->initializeSynonymsUrl(); |
775
|
1 |
|
$json = $this->synonymParser->toJson($baseWord, $synonyms); |
776
|
1 |
|
return $this->_sendRawPost($this->_synonymsUrl, $json, |
777
|
1 |
|
$this->getHttpTransport()->getDefaultTimeout(), 'application/json'); |
778
|
|
|
} |
779
|
|
|
|
780
|
|
|
/** |
781
|
|
|
* Remove a synonym from the synonyms map |
782
|
|
|
* |
783
|
|
|
* @param string $baseWord |
784
|
|
|
* @return \Apache_Solr_Response |
785
|
|
|
* @throws \Apache_Solr_InvalidArgumentException |
786
|
|
|
*/ |
787
|
1 |
|
public function deleteSynonym($baseWord) |
788
|
|
|
{ |
789
|
1 |
|
$this->initializeSynonymsUrl(); |
790
|
1 |
|
if (empty($baseWord)) { |
791
|
|
|
throw new \Apache_Solr_InvalidArgumentException('Must provide base word.'); |
792
|
|
|
} |
793
|
|
|
|
794
|
1 |
|
return $this->_sendRawDelete($this->_synonymsUrl . '/' . $baseWord); |
795
|
|
|
} |
796
|
|
|
|
797
|
|
|
/** |
798
|
|
|
* Central method for making a HTTP DELETE operation against the Solr server |
799
|
|
|
* |
800
|
|
|
* @param string $url |
801
|
|
|
* @param bool|float $timeout Read timeout in seconds |
802
|
|
|
* @return \Apache_Solr_Response |
803
|
|
|
*/ |
804
|
2 |
|
protected function _sendRawDelete($url, $timeout = false) |
805
|
|
|
{ |
806
|
2 |
|
$logSeverity = 0; // info |
807
|
|
|
|
808
|
|
|
try { |
809
|
2 |
|
$httpTransport = $this->getHttpTransport(); |
810
|
|
|
|
811
|
2 |
|
$httpResponse = $httpTransport->performDeleteRequest($url, |
812
|
|
|
$timeout); |
813
|
2 |
|
$solrResponse = new \Apache_Solr_Response($httpResponse, |
814
|
2 |
|
$this->_createDocuments, $this->_collapseSingleValueArrays); |
815
|
|
|
|
816
|
2 |
|
if ($solrResponse->getHttpStatus() != 200) { |
817
|
2 |
|
throw new \Apache_Solr_HttpTransportException($solrResponse); |
818
|
|
|
} |
819
|
2 |
|
} catch (Apache_Solr_HttpTransportException $e) { |
|
|
|
|
820
|
2 |
|
$logSeverity = 3; // fatal error |
821
|
2 |
|
$solrResponse = $e->getResponse(); |
822
|
|
|
} |
823
|
|
|
|
824
|
2 |
|
if ($this->configuration->getLoggingQueryRawDelete() || $solrResponse->getHttpStatus() != 200) { |
825
|
|
|
$logData = [ |
826
|
2 |
|
'query url' => $url, |
827
|
2 |
|
'response' => (array)$solrResponse |
828
|
|
|
]; |
829
|
|
|
|
830
|
2 |
|
if (!empty($e)) { |
831
|
2 |
|
$logData['exception'] = $e->__toString(); |
832
|
|
|
} else { |
833
|
|
|
// trigger data parsing |
834
|
|
|
$solrResponse->response; |
835
|
|
|
$logData['response data'] = print_r($solrResponse, true); |
836
|
|
|
} |
837
|
|
|
|
838
|
2 |
|
GeneralUtility::devLog('Querying Solr using DELETE', 'solr', |
839
|
|
|
$logSeverity, $logData); |
840
|
|
|
} |
841
|
|
|
|
842
|
2 |
|
return $solrResponse; |
843
|
|
|
} |
844
|
|
|
|
845
|
|
|
/** |
846
|
|
|
* Get currently configured stop words |
847
|
|
|
* |
848
|
|
|
* @return array |
849
|
|
|
*/ |
850
|
2 |
|
public function getStopWords() |
851
|
|
|
{ |
852
|
2 |
|
$this->initializeStopWordsUrl(); |
853
|
2 |
|
$response = $this->_sendRawGet($this->_stopWordsUrl); |
854
|
2 |
|
return $this->stopWordParser->parseJson($response->getRawResponse()); |
855
|
|
|
} |
856
|
|
|
|
857
|
|
|
/** |
858
|
|
|
* Adds stop words to the managed stop word list |
859
|
|
|
* |
860
|
|
|
* @param array|string $stopWords string for a single word, array for multiple words |
861
|
|
|
* @return \Apache_Solr_Response |
862
|
|
|
* @throws \Apache_Solr_InvalidArgumentException If $stopWords is empty |
863
|
|
|
*/ |
864
|
1 |
|
public function addStopWords($stopWords) |
865
|
|
|
{ |
866
|
1 |
|
$this->initializeStopWordsUrl(); |
867
|
1 |
|
$json = $this->stopWordParser->toJson($stopWords); |
868
|
1 |
|
return $this->_sendRawPost($this->_stopWordsUrl, $json, |
869
|
1 |
|
$this->getHttpTransport()->getDefaultTimeout(), 'application/json'); |
870
|
|
|
} |
871
|
|
|
|
872
|
|
|
/** |
873
|
|
|
* Deletes a words from the managed stop word list |
874
|
|
|
* |
875
|
|
|
* @param string $stopWord stop word to delete |
876
|
|
|
* @return \Apache_Solr_Response |
877
|
|
|
* @throws \Apache_Solr_InvalidArgumentException If $stopWords is empty |
878
|
|
|
*/ |
879
|
1 |
|
public function deleteStopWord($stopWord) |
880
|
|
|
{ |
881
|
1 |
|
$this->initializeStopWordsUrl(); |
882
|
1 |
|
if (empty($stopWord)) { |
883
|
|
|
throw new \Apache_Solr_InvalidArgumentException('Must provide stop word.'); |
884
|
|
|
} |
885
|
|
|
|
886
|
1 |
|
return $this->_sendRawDelete($this->_stopWordsUrl . '/' . $stopWord); |
887
|
|
|
} |
888
|
|
|
|
889
|
|
|
/** |
890
|
|
|
* Returns the core name from the configured path. |
891
|
|
|
* |
892
|
|
|
* @return string |
893
|
|
|
*/ |
894
|
1 |
|
public function getCoreName() |
895
|
|
|
{ |
896
|
1 |
|
$paths = explode('/', trim($this->_path, '/')); |
897
|
|
|
|
898
|
1 |
|
return (string) array_pop($paths); |
899
|
|
|
} |
900
|
|
|
|
901
|
|
|
/** |
902
|
|
|
* Reloads the current core |
903
|
|
|
* |
904
|
|
|
* @return \Apache_Solr_Response |
905
|
|
|
*/ |
906
|
1 |
|
public function reloadCore() |
907
|
|
|
{ |
908
|
1 |
|
return $this->reloadCoreByName($this->getCoreName()); |
909
|
|
|
} |
910
|
|
|
|
911
|
|
|
/** |
912
|
|
|
* Reloads a core of the connection by a given corename. |
913
|
|
|
* |
914
|
|
|
* @param string $coreName |
915
|
|
|
* @return \Apache_Solr_Response |
916
|
|
|
*/ |
917
|
1 |
|
public function reloadCoreByName($coreName) |
918
|
|
|
{ |
919
|
1 |
|
$coreAdminReloadUrl = $this->_coresUrl . '?action=reload&core=' . $coreName; |
920
|
1 |
|
return $this->_sendRawGet($coreAdminReloadUrl); |
921
|
|
|
} |
922
|
|
|
|
923
|
|
|
/** |
924
|
|
|
* initializes various URLs, including the Luke URL |
925
|
|
|
* |
926
|
|
|
* @return void |
927
|
|
|
*/ |
928
|
81 |
|
protected function _initUrls() |
929
|
|
|
{ |
930
|
81 |
|
parent::_initUrls(); |
931
|
|
|
|
932
|
81 |
|
$this->_lukeUrl = $this->_constructUrl( |
933
|
81 |
|
self::LUKE_SERVLET, |
934
|
|
|
[ |
935
|
81 |
|
'numTerms' => '0', |
936
|
81 |
|
'wt' => self::SOLR_WRITER |
937
|
|
|
] |
938
|
|
|
); |
939
|
|
|
|
940
|
81 |
|
$this->_pluginsUrl = $this->_constructUrl( |
941
|
81 |
|
self::PLUGINS_SERVLET, |
942
|
81 |
|
['wt' => self::SOLR_WRITER] |
943
|
|
|
); |
944
|
|
|
|
945
|
81 |
|
$pathElements = explode('/', trim($this->_path, '/')); |
946
|
81 |
|
$this->_coresUrl = |
947
|
81 |
|
$this->_scheme . '://' . |
948
|
81 |
|
$this->_host . ':' . |
949
|
81 |
|
$this->_port . '/' . |
950
|
81 |
|
$pathElements[0] . '/' . |
951
|
81 |
|
self::CORES_SERVLET; |
952
|
|
|
|
953
|
81 |
|
$this->_schemaUrl = $this->_constructUrl(self::SCHEMA_SERVLET); |
954
|
81 |
|
} |
955
|
|
|
|
956
|
|
|
/** |
957
|
|
|
* @return void |
958
|
|
|
*/ |
959
|
1 |
|
protected function initializeSynonymsUrl() |
960
|
|
|
{ |
961
|
1 |
|
if (trim($this->_synonymsUrl) !== '') { |
962
|
1 |
|
return; |
963
|
|
|
} |
964
|
1 |
|
$this->_synonymsUrl = $this->_constructUrl(self::SYNONYMS_SERVLET) . $this->getSchema()->getLanguage(); |
965
|
1 |
|
} |
966
|
|
|
|
967
|
|
|
/** |
968
|
|
|
* @return void |
969
|
|
|
*/ |
970
|
2 |
|
protected function initializeStopWordsUrl() |
971
|
|
|
{ |
972
|
2 |
|
if (trim($this->_stopWordsUrl) !== '') { |
973
|
1 |
|
return; |
974
|
|
|
} |
975
|
|
|
|
976
|
2 |
|
$this->_stopWordsUrl = $this->_constructUrl(self::STOPWORDS_SERVLET) . $this->getSchema()->getLanguage(); |
977
|
2 |
|
} |
978
|
|
|
|
979
|
|
|
/** |
980
|
|
|
* Get the configured schema for the current core. |
981
|
|
|
* |
982
|
|
|
* @return Schema |
983
|
|
|
*/ |
984
|
4 |
|
public function getSchema() |
985
|
|
|
{ |
986
|
4 |
|
if ($this->schema !== null) { |
987
|
|
|
return $this->schema; |
988
|
|
|
} |
989
|
4 |
|
$response = $this->_sendRawGet($this->_schemaUrl); |
990
|
4 |
|
$this->schema = $this->schemaParser->parseJson($response->getRawResponse()); |
991
|
4 |
|
return $this->schema; |
992
|
|
|
} |
993
|
|
|
} |
994
|
|
|
|