1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
use Elastica\Document; |
4
|
|
|
use Elastica\Query; |
5
|
|
|
use \SilverStripe\Elastica\ResultList; |
6
|
|
|
use Elastica\Query\QueryString; |
7
|
|
|
use Elastica\Aggregation\Filter; |
8
|
|
|
use Elastica\Filter\Term; |
9
|
|
|
use Elastica\Filter\BoolAnd; |
10
|
|
|
use Elastica\Aggregation\Terms; |
11
|
|
|
use Elastica\Query\Filtered; |
12
|
|
|
use Elastica\Query\Range; |
13
|
|
|
use \SilverStripe\Elastica\ElasticSearcher; |
14
|
|
|
use \SilverStripe\Elastica\Searchable; |
15
|
|
|
use \SilverStripe\Elastica\QueryGenerator; |
16
|
|
|
use \SilverStripe\Elastica\ElasticaUtil; |
17
|
|
|
|
18
|
|
|
class ElasticSearchPage_Controller extends Page_Controller { |
19
|
|
|
|
20
|
|
|
private static $allowed_actions = array('SearchForm', 'submit', 'index', 'similar'); |
21
|
|
|
|
22
|
|
|
public function init() { |
23
|
|
|
parent::init(); |
24
|
|
|
|
25
|
|
|
Requirements::javascript(THIRDPARTY_DIR . '/jquery/jquery.js'); |
26
|
|
|
Requirements::javascript("elastica/javascript/jquery.autocomplete.js"); |
27
|
|
|
Requirements::javascript("elastica/javascript/elastica.js"); |
28
|
|
|
Requirements::css("elastica/css/elastica.css"); |
29
|
|
|
|
30
|
|
|
$this->SearchPage = Controller::curr()->dataRecord; |
31
|
|
|
} |
32
|
|
|
|
33
|
|
|
|
34
|
|
|
|
35
|
|
|
/* |
36
|
|
|
Find DataObjects in Elasticsearch similar to the one selected. Note that aggregations are not |
37
|
|
|
taken into account, merely the text of the selected document. |
38
|
|
|
*/ |
39
|
|
|
public function similar() { |
40
|
|
|
//FIXME double check security, ie if escaping needed |
41
|
|
|
$class = $this->request->param('ID'); |
42
|
|
|
$instanceID = $this->request->param('OtherID'); |
43
|
|
|
|
44
|
|
|
$data = $this->initialiseDataArray(); |
45
|
|
|
|
46
|
|
|
$es = $this->primeElasticSearcherFromRequest(); |
47
|
|
|
|
48
|
|
|
$this->setMoreLikeThisParamsFromRequest($es); |
49
|
|
|
$es->setPageLength($this->SearchPage->ResultsPerPage); |
50
|
|
|
|
51
|
|
|
// filter by class or site tree |
52
|
|
|
if($this->SearchPage->SiteTreeOnly) { |
53
|
|
|
T7; //FIXME test missing |
54
|
|
|
$es->addFilter('IsInSiteTree', true); |
55
|
|
|
} else { |
56
|
|
|
$es->setClasses($this->SearchPage->ClassesToSearch); |
57
|
|
|
} |
58
|
|
|
|
59
|
|
|
// get the edited fields to search from the database for this search page |
60
|
|
|
// Convert this into a name => weighting array |
61
|
|
|
$fieldsToSearch = array(); |
62
|
|
|
$editedSearchFields = $this->ElasticaSearchableFields()->filter(array( |
63
|
|
|
'Active' => true, |
64
|
|
|
'SimilarSearchable' => true |
65
|
|
|
)); |
66
|
|
|
|
67
|
|
|
foreach($editedSearchFields->getIterator() as $searchField) { |
68
|
|
|
$fieldsToSearch[$searchField->Name] = $searchField->Weight; |
69
|
|
|
} |
70
|
|
|
|
71
|
|
|
// Use the standard field for more like this, ie not stemmed |
72
|
|
|
foreach($fieldsToSearch as $field => $value) { |
73
|
|
|
$fieldsToSearch[$field . '.standard'] = $value; |
74
|
|
|
unset($fieldsToSearch[$field]); |
75
|
|
|
} |
76
|
|
|
|
77
|
|
|
try { |
78
|
|
|
// Simulate server being down for testing purposes |
79
|
|
|
if($this->request->getVar('ServerDown')) { |
80
|
|
|
throw new Elastica\Exception\Connection\HttpException('Unable to reach search server'); |
81
|
|
|
} |
82
|
|
|
if(class_exists($class)) { |
83
|
|
|
$instance = \DataObject::get_by_id($class, $instanceID); |
84
|
|
|
|
85
|
|
|
$paginated = $es->moreLikeThis($instance, $fieldsToSearch); |
86
|
|
|
|
87
|
|
|
$this->Aggregations = $es->getAggregations(); |
88
|
|
|
$this->successfulSearch($data, $paginated); |
89
|
|
|
$data['SimilarTo'] = $instance; |
90
|
|
|
|
91
|
|
|
|
92
|
|
|
$moreLikeThisTerms = $paginated->getList()->MoreLikeThisTerms; |
93
|
|
|
$fieldToTerms = new ArrayList(); |
94
|
|
|
foreach(array_keys($moreLikeThisTerms) as $fieldName) { |
95
|
|
|
$readableFieldName = str_replace('.standard', '', $fieldName); |
96
|
|
|
$fieldTerms = new ArrayList(); |
97
|
|
|
foreach($moreLikeThisTerms[$fieldName] as $value) { |
98
|
|
|
$do = new DataObject(); |
99
|
|
|
$do->Term = $value; |
100
|
|
|
$fieldTerms->push($do); |
101
|
|
|
} |
102
|
|
|
|
103
|
|
|
$do = new DataObject(); |
104
|
|
|
$do->FieldName = $readableFieldName; |
105
|
|
|
$do->Terms = $fieldTerms; |
106
|
|
|
$fieldToTerms->push($do); |
107
|
|
|
} |
108
|
|
|
|
109
|
|
|
$data['SimilarSearchTerms'] = $fieldToTerms; |
110
|
|
|
} else { |
111
|
|
|
// class does not exist |
112
|
|
|
$data['ErrorMessage'] = "Class $class is either not found or not searchable\n"; |
113
|
|
|
} |
114
|
|
|
} catch (\InvalidArgumentException $e) { |
115
|
|
|
$data['ErrorMessage'] = "Class $class is either not found or not searchable\n"; |
116
|
|
|
} catch (Elastica\Exception\Connection\HttpException $e) { |
117
|
|
|
$data['ErrorMessage'] = 'Unable to connect to search server'; |
118
|
|
|
$data['SearchPerformed'] = false; |
119
|
|
|
} |
120
|
|
|
|
121
|
|
|
|
122
|
|
|
$elapsed = $this->calculateTime(); |
123
|
|
|
$data['ElapsedTime'] = $elapsed; |
124
|
|
|
|
125
|
|
|
|
126
|
|
|
return $this->renderResults($data); |
127
|
|
|
} |
128
|
|
|
|
129
|
|
|
|
130
|
|
|
|
131
|
|
|
/* |
132
|
|
|
Display the search form. If the query parameter exists, search against Elastica |
133
|
|
|
and render results accordingly. |
134
|
|
|
*/ |
135
|
|
|
public function index() { |
136
|
|
|
$data = $this->initialiseDataArray(); |
137
|
|
|
$es = $this->primeElasticSearcherFromRequest(); |
138
|
|
|
$es->setPageLength($this->SearchPage->ResultsPerPage); |
139
|
|
|
|
140
|
|
|
// Do not show suggestions if this flag is set |
141
|
|
|
$ignoreSuggestions = null !== $this->request->getVar('is'); |
142
|
|
|
|
143
|
|
|
|
144
|
|
|
// query string |
145
|
|
|
$queryTextParam = $this->request->getVar('q'); |
146
|
|
|
$queryText = !empty($queryTextParam) ? $queryTextParam : ''; |
147
|
|
|
|
148
|
|
|
$testMode = !empty($this->request->getVar('TestMode')); |
149
|
|
|
|
150
|
|
|
// filters for aggregations |
151
|
|
|
$ignore = \Config::inst()->get('Elastica', 'BlackList'); |
152
|
|
|
foreach($this->request->getVars() as $key => $value) { |
153
|
|
|
if(!in_array($key, $ignore)) { |
154
|
|
|
$es->addFilter($key, $value); |
155
|
|
|
} |
156
|
|
|
} |
157
|
|
|
|
158
|
|
|
// filter by class or site tree |
159
|
|
|
if($this->SearchPage->SiteTreeOnly) { |
160
|
|
|
$es->addFilter('IsInSiteTree', true); |
161
|
|
|
} else { |
162
|
|
|
$es->setClasses($this->SearchPage->ClassesToSearch); |
163
|
|
|
} |
164
|
|
|
|
165
|
|
|
// set the optional aggregation manipulator |
166
|
|
|
// In the event of a manipulator being present, show all the results for search |
167
|
|
|
// Otherwise aggregations are all zero |
168
|
|
|
if($this->SearchHelper) { |
169
|
|
|
$es->setQueryResultManipulator($this->SearchHelper); |
170
|
|
|
$es->showResultsForEmptySearch(); |
171
|
|
|
} else { |
172
|
|
|
$es->hideResultsForEmptySearch(); |
173
|
|
|
} |
174
|
|
|
|
175
|
|
|
// get the edited fields to search from the database for this search page |
176
|
|
|
// Convert this into a name => weighting array |
177
|
|
|
$fieldsToSearch = array(); |
178
|
|
|
$editedSearchFields = $this->ElasticaSearchableFields()->filter(array( |
179
|
|
|
'Active' => true, |
180
|
|
|
'Searchable' => true |
181
|
|
|
)); |
182
|
|
|
|
183
|
|
|
foreach($editedSearchFields->getIterator() as $searchField) { |
184
|
|
|
$fieldsToSearch[$searchField->Name] = $searchField->Weight; |
185
|
|
|
} |
186
|
|
|
|
187
|
|
|
$paginated = null; |
188
|
|
|
try { |
189
|
|
|
// Simulate server being down for testing purposes |
190
|
|
|
if(!empty($this->request->getVar('ServerDown'))) { |
191
|
|
|
throw new Elastica\Exception\Connection\HttpException('Unable to reach search server'); |
192
|
|
|
} |
193
|
|
|
|
194
|
|
|
// now actually perform the search using the original query |
195
|
|
|
$paginated = $es->search($queryText, $fieldsToSearch, $testMode); |
196
|
|
|
|
197
|
|
|
// This is the case of the original query having a better one suggested. Do a |
198
|
|
|
// second search for the suggested query, throwing away the original |
199
|
|
|
if($es->hasSuggestedQuery() && !$ignoreSuggestions) { |
200
|
|
|
$data['SuggestedQuery'] = $es->getSuggestedQuery(); |
201
|
|
|
$data['SuggestedQueryHighlighted'] = $es->getSuggestedQueryHighlighted(); |
202
|
|
|
//Link for if the user really wants to try their original query |
203
|
|
|
$sifLink = rtrim($this->Link(), '/') . '?q=' . $queryText . '&is=1'; |
204
|
|
|
$data['SearchInsteadForLink'] = $sifLink; |
205
|
|
|
$paginated = $es->search($es->getSuggestedQuery(), $fieldsToSearch); |
206
|
|
|
|
207
|
|
|
} |
208
|
|
|
|
209
|
|
|
$elapsed = $this->calculateTime(); |
210
|
|
|
$data['ElapsedTime'] = $elapsed; |
211
|
|
|
|
212
|
|
|
$this->Aggregations = $es->getAggregations(); |
213
|
|
|
$this->successfulSearch($data, $paginated); |
214
|
|
|
|
215
|
|
|
} catch (Elastica\Exception\Connection\HttpException $e) { |
216
|
|
|
$data['ErrorMessage'] = 'Unable to connect to search server'; |
217
|
|
|
$data['SearchPerformed'] = false; |
218
|
|
|
} |
219
|
|
|
|
220
|
|
|
$data['OriginalQuery'] = $queryText; |
221
|
|
|
$data['IgnoreSuggestions'] = $ignoreSuggestions; |
222
|
|
|
|
223
|
|
|
return $this->renderResults($data); |
224
|
|
|
|
225
|
|
|
} |
226
|
|
|
|
227
|
|
|
|
228
|
|
|
private function successfulSearch(&$data, $paginated) { |
229
|
|
|
$data['SearchResults'] = $paginated; |
230
|
|
|
$data['SearchPerformed'] = true; |
231
|
|
|
$data['NumberOfResults'] = $paginated->getTotalItems(); |
232
|
|
|
$data['SearchPageLink'] = $this->SearchPage->Link(); |
233
|
|
|
} |
234
|
|
|
|
235
|
|
|
|
236
|
|
|
/* |
237
|
|
|
Return true if the query is not empty |
238
|
|
|
*/ |
239
|
|
|
public function QueryIsEmpty() { |
240
|
|
|
return empty($this->request->getVar('q')); |
241
|
|
|
} |
242
|
|
|
|
243
|
|
|
|
244
|
|
|
/** |
245
|
|
|
* Process submission of the search form, redirecting to a URL that will render search results |
246
|
|
|
* @param array $data form data |
247
|
|
|
* @param Form $form form |
248
|
|
|
*/ |
249
|
|
|
public function submit($data, $form) { |
250
|
|
|
$queryText = $data['q']; |
251
|
|
|
$url = $this->Link(); |
252
|
|
|
$url = rtrim($url, '/'); |
253
|
|
|
$link = rtrim($url, '/') . '?q=' . $queryText . '&sfid=' . $data['identifier']; |
254
|
|
|
$this->redirect($link); |
255
|
|
|
} |
256
|
|
|
|
257
|
|
|
|
258
|
|
|
/* |
259
|
|
|
Obtain an instance of the form |
260
|
|
|
*/ |
261
|
|
|
public function SearchForm() { |
262
|
|
|
$form = new ElasticSearchForm($this, 'SearchForm'); |
263
|
|
|
$fields = $form->Fields(); |
264
|
|
|
$elasticaSearchPage = Controller::curr()->dataRecord; |
265
|
|
|
$identifierField = new HiddenField('identifier'); |
266
|
|
|
$identifierField->setValue($elasticaSearchPage->Identifier); |
267
|
|
|
|
268
|
|
|
$fields->push($identifierField); |
269
|
|
|
$queryField = $fields->fieldByName('q'); |
270
|
|
|
|
271
|
|
|
if($this->isParamSet('q') && $this->isParamSet('sfid')) { |
272
|
|
|
$sfid = $this->request->getVar('sfid'); |
273
|
|
|
if($sfid == $elasticaSearchPage->Identifier) { |
274
|
|
|
|
275
|
|
|
$queryText = $this->request->getVar('q'); |
276
|
|
|
$queryField->setValue($queryText); |
277
|
|
|
} |
278
|
|
|
|
279
|
|
|
} |
280
|
|
|
|
281
|
|
|
if($this->action == 'similar') { |
282
|
|
|
$queryField->setDisabled(true); |
283
|
|
|
$actions = $form->Actions(); |
284
|
|
|
|
285
|
|
|
if(!empty($actions)) { |
286
|
|
|
foreach($actions as $field) { |
287
|
|
|
$field->setDisabled(true); |
288
|
|
|
} |
289
|
|
|
} |
290
|
|
|
|
291
|
|
|
} |
292
|
|
|
|
293
|
|
|
if($this->AutoCompleteFieldID > 0) { |
294
|
|
|
ElasticaUtil::addAutocompleteToQueryField( |
295
|
|
|
$queryField, |
296
|
|
|
$this->ClassesToSearch, |
297
|
|
|
$this->SiteTreeOnly, |
298
|
|
|
$this->Link(), |
299
|
|
|
$this->AutocompleteFunction()->Slug |
300
|
|
|
); |
301
|
|
|
} |
302
|
|
|
return $form; |
303
|
|
|
} |
304
|
|
|
|
305
|
|
|
|
306
|
|
|
/** |
307
|
|
|
* @param string $paramName |
308
|
|
|
*/ |
309
|
|
|
private function isParamSet($paramName) { |
310
|
|
|
return !empty($this->request->getVar($paramName)); |
311
|
|
|
} |
312
|
|
|
|
313
|
|
|
|
314
|
|
|
/** |
315
|
|
|
* Set the start page from the request and results per page for a given searcher object |
316
|
|
|
*/ |
317
|
|
|
private function primeElasticSearcherFromRequest() { |
318
|
|
|
$elasticSearcher = new ElasticSearcher(); |
319
|
|
|
// start, and page length, i.e. pagination |
320
|
|
|
$startParam = $this->request->getVar('start'); |
321
|
|
|
$start = isset($startParam) ? $startParam : 0; |
322
|
|
|
$elasticSearcher->setStart($start); |
323
|
|
|
$this->StartTime = microtime(true); |
324
|
|
|
return $elasticSearcher; |
325
|
|
|
} |
326
|
|
|
|
327
|
|
|
|
328
|
|
|
/** |
329
|
|
|
* Set the admin configured similarity parameters |
330
|
|
|
* @param \SilverStripe\Elastica\ElasticSearcher &$elasticSearcher ElasticaSearcher object |
331
|
|
|
*/ |
332
|
|
|
private function setMoreLikeThisParamsFromRequest(&$elasticSearcher) { |
333
|
|
|
$elasticSearcher->setMinTermFreq($this->MinTermFreq); |
334
|
|
|
$elasticSearcher->setMaxTermFreq($this->MaxTermFreq); |
335
|
|
|
$elasticSearcher->setMinDocFreq($this->MinDocFreq); |
336
|
|
|
$elasticSearcher->setMaxDocFreq($this->MaxDocFreq); |
337
|
|
|
$elasticSearcher->setMinWordLength($this->MinWordLength); |
338
|
|
|
$elasticSearcher->setMaxWordLength($this->MaxWordLength); |
339
|
|
|
$elasticSearcher->setMinShouldMatch($this->MinShouldMatch); |
340
|
|
|
$elasticSearcher->setSimilarityStopWords($this->SimilarityStopWords); |
341
|
|
|
} |
342
|
|
|
|
343
|
|
|
|
344
|
|
|
private function initialiseDataArray() { |
345
|
|
|
return array( |
346
|
|
|
'Content' => $this->Content, |
347
|
|
|
'Title' => $this->Title, |
348
|
|
|
'SearchPerformed' => false |
349
|
|
|
); |
350
|
|
|
} |
351
|
|
|
|
352
|
|
|
|
353
|
|
|
|
354
|
|
|
private function renderResults($data) { |
355
|
|
|
// allow the optional use of overriding the search result page, e.g. for photos, maps or facets |
356
|
|
|
if($this->hasExtension('PageControllerTemplateOverrideExtension')) { |
357
|
|
|
return $this->useTemplateOverride($data); |
358
|
|
|
} else { |
359
|
|
|
return $data; |
360
|
|
|
} |
361
|
|
|
} |
362
|
|
|
|
363
|
|
|
|
364
|
|
|
private function calculateTime() { |
365
|
|
|
$endTime = microtime(true); |
366
|
|
|
$elapsed = round(100 * ($endTime - $this->StartTime)) / 100; |
367
|
|
|
return $elapsed; |
368
|
|
|
} |
369
|
|
|
} |
370
|
|
|
|