|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* @package midcom.services |
|
4
|
|
|
* @author The Midgard Project, http://www.midgard-project.org |
|
5
|
|
|
* @copyright The Midgard Project, http://www.midgard-project.org |
|
6
|
|
|
* @license http://www.gnu.org/licenses/lgpl.html GNU Lesser General Public License |
|
7
|
|
|
*/ |
|
8
|
|
|
|
|
9
|
|
|
use Symfony\Component\EventDispatcher\EventSubscriberInterface; |
|
10
|
|
|
use midcom\events\dbaevent; |
|
11
|
|
|
|
|
12
|
|
|
/** |
|
13
|
|
|
* This class is the main access point into the MidCOM Indexer subsystem. |
|
14
|
|
|
* |
|
15
|
|
|
* It allows you to maintain and query the document index. |
|
16
|
|
|
* |
|
17
|
|
|
* Do not instantiate this class directly. Instead use the get_service |
|
18
|
|
|
* method on midcom_application using the service name 'indexer' to obtain |
|
19
|
|
|
* a running instance. |
|
20
|
|
|
* |
|
21
|
|
|
* @see midcom_services_indexer_document |
|
22
|
|
|
* @see midcom_services_indexer_backend |
|
23
|
|
|
* @see midcom_services_indexer_filter |
|
24
|
|
|
* |
|
25
|
|
|
* @todo Batch indexing support |
|
26
|
|
|
* @todo Write code examples |
|
27
|
|
|
* @todo More elaborate class introduction. |
|
28
|
|
|
* @package midcom.services |
|
29
|
|
|
*/ |
|
30
|
|
|
class midcom_services_indexer implements EventSubscriberInterface |
|
31
|
|
|
{ |
|
32
|
|
|
/** |
|
33
|
|
|
* The backend indexer implementation |
|
34
|
|
|
* |
|
35
|
|
|
* @var midcom_services_indexer_backend |
|
36
|
|
|
*/ |
|
37
|
|
|
private $_backend; |
|
38
|
|
|
|
|
39
|
|
|
/** |
|
40
|
|
|
* Flag for disabled indexing, set by the constructor. |
|
41
|
|
|
* |
|
42
|
|
|
* @var boolean |
|
43
|
|
|
*/ |
|
44
|
|
|
private $_disabled = false; |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* Initialization |
|
48
|
|
|
* |
|
49
|
|
|
* The constructor will initialize the indexer backend using the MidCOM |
|
50
|
|
|
* configuration by default. If you need a different indexer backend, you |
|
51
|
|
|
* can always explicitly instantiate a backend and pass it to the |
|
52
|
|
|
* constructor. In that case you have to load the corresponding PHP file |
|
53
|
|
|
* manually. |
|
54
|
|
|
* |
|
55
|
|
|
* @param midcom_services_indexer_backend $backend An explicit indexer to initialize with. |
|
56
|
|
|
*/ |
|
57
|
1 |
|
public function __construct($backend = null) |
|
58
|
|
|
{ |
|
59
|
1 |
|
if (!midcom::get()->config->get('indexer_backend')) { |
|
60
|
1 |
|
$this->_disabled = true; |
|
61
|
1 |
|
return; |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
if ($backend === null) { |
|
65
|
|
|
$class = midcom::get()->config->get('indexer_backend'); |
|
66
|
|
|
if (strpos($class, '_') === false) { |
|
67
|
|
|
// Built-in backend called using the shorthand notation |
|
68
|
|
|
$class = "midcom_services_indexer_backend_" . $class; |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
$this->_backend = new $class(); |
|
72
|
|
|
} else { |
|
73
|
|
|
$this->_backend = $backend; |
|
74
|
|
|
} |
|
75
|
|
|
midcom::get()->dispatcher->addSubscriber($this); |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
|
|
public static function getSubscribedEvents() |
|
79
|
|
|
{ |
|
80
|
|
|
return [dbaevent::DELETE => ['handle_delete']]; |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
public function handle_delete(dbaevent $event) |
|
84
|
|
|
{ |
|
85
|
|
|
$this->delete($event->get_object()->guid); |
|
|
|
|
|
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
/** |
|
89
|
|
|
* Simple helper, returns true if the indexer service is online, false if it is disabled. |
|
90
|
|
|
* |
|
91
|
|
|
* @return boolean Service state. |
|
92
|
|
|
*/ |
|
93
|
|
|
public function enabled() : bool |
|
94
|
|
|
{ |
|
95
|
|
|
return !$this->_disabled; |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
/** |
|
99
|
|
|
* Adds a document to the index. |
|
100
|
|
|
* |
|
101
|
|
|
* A finished document object must be passed to this object. If the index |
|
102
|
|
|
* already contains a record with the same Resource Identifier, the record |
|
103
|
|
|
* is replaced. |
|
104
|
|
|
* |
|
105
|
|
|
* Support of batch-indexing using an Array of documents instead of a single |
|
106
|
|
|
* document is possible (and strongly advised for performance reasons). |
|
107
|
|
|
* |
|
108
|
|
|
* @param mixed $documents One or more documents to be indexed, so this is either a |
|
109
|
|
|
* midcom_services_indexer_document or an Array of these objects. |
|
110
|
|
|
* @return boolean Indicating success. |
|
111
|
|
|
*/ |
|
112
|
6 |
|
public function index($documents) : bool |
|
113
|
|
|
{ |
|
114
|
6 |
|
if ($this->_disabled) { |
|
115
|
6 |
|
return true; |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
$batch = is_array($documents); |
|
119
|
|
|
if (!$batch) { |
|
120
|
|
|
$documents = [$documents]; |
|
121
|
|
|
} |
|
122
|
|
|
if (empty($documents)) { |
|
123
|
|
|
// Nothing to do. |
|
124
|
|
|
return true; |
|
125
|
|
|
} |
|
126
|
|
|
|
|
127
|
|
|
foreach ($documents as $value) { |
|
128
|
|
|
$value->members_to_fields(); |
|
129
|
|
|
} |
|
130
|
|
|
|
|
131
|
|
|
try { |
|
132
|
|
|
$this->_backend->index($documents); |
|
133
|
|
|
return true; |
|
134
|
|
|
} catch (Exception $e) { |
|
135
|
|
|
if ($batch) { |
|
136
|
|
|
throw $e; |
|
137
|
|
|
} |
|
138
|
|
|
debug_add("Indexing error: " . $e->getMessage(), MIDCOM_LOG_ERROR); |
|
139
|
|
|
return false; |
|
140
|
|
|
} |
|
141
|
|
|
} |
|
142
|
|
|
|
|
143
|
|
|
/** |
|
144
|
|
|
* Removes the document(s) with the given resource identifier(s) from the index. |
|
145
|
|
|
* Using GUIDs instead of RIs will delete all language versions |
|
146
|
|
|
* |
|
147
|
|
|
* @param array $RIs The resource identifier(s) of the document(s) that should be deleted. |
|
148
|
|
|
* @return boolean Indicating success. |
|
149
|
|
|
*/ |
|
150
|
|
|
public function delete($RIs) : bool |
|
151
|
|
|
{ |
|
152
|
|
|
if ($this->_disabled) { |
|
153
|
|
|
return true; |
|
154
|
|
|
} |
|
155
|
|
|
$RIs = (array) $RIs; |
|
156
|
|
|
if (empty($RIs)) { |
|
157
|
|
|
// Nothing to do. |
|
158
|
|
|
return true; |
|
159
|
|
|
} |
|
160
|
|
|
try { |
|
161
|
|
|
$this->_backend->delete($RIs); |
|
162
|
|
|
return true; |
|
163
|
|
|
} catch (Exception $e) { |
|
164
|
|
|
debug_add("Deleting error: " . $e->getMessage(), MIDCOM_LOG_ERROR); |
|
165
|
|
|
return false; |
|
166
|
|
|
} |
|
167
|
|
|
} |
|
168
|
|
|
|
|
169
|
|
|
/** |
|
170
|
|
|
* Clear the index completely. |
|
171
|
|
|
* |
|
172
|
|
|
* This will drop the current index. |
|
173
|
|
|
* |
|
174
|
|
|
* @return boolean Indicating success. |
|
175
|
|
|
*/ |
|
176
|
|
|
public function delete_all($constraint = '') : bool |
|
177
|
|
|
{ |
|
178
|
|
|
if ($this->_disabled) { |
|
179
|
|
|
return true; |
|
180
|
|
|
} |
|
181
|
|
|
|
|
182
|
|
|
try { |
|
183
|
|
|
$this->_backend->delete_all($constraint); |
|
184
|
|
|
return true; |
|
185
|
|
|
} catch (Exception $e) { |
|
186
|
|
|
debug_add("Deleting error: " . $e->getMessage(), MIDCOM_LOG_ERROR); |
|
187
|
|
|
return false; |
|
188
|
|
|
} |
|
189
|
|
|
} |
|
190
|
|
|
|
|
191
|
|
|
/** |
|
192
|
|
|
* Query the index and, if set, restrict the query by a given filter. |
|
193
|
|
|
* |
|
194
|
|
|
* The filter argument is optional and may be a subclass of indexer_filter. |
|
195
|
|
|
* The backend determines what filters are supported and how they are |
|
196
|
|
|
* treated. |
|
197
|
|
|
* |
|
198
|
|
|
* The query syntax is also dependent on the backend. Refer to its documentation |
|
199
|
|
|
* how queries should be built. |
|
200
|
|
|
* |
|
201
|
|
|
* @param string $query The query, which must suit the backends query syntax. It is assumed to be in the site charset. |
|
202
|
|
|
* @param midcom_services_indexer_filter $filter An optional filter used to restrict the query. |
|
203
|
|
|
* @param array $options Options that are passed straight to the backend |
|
204
|
|
|
* @return midcom_services_indexer_document[] An array of documents matching the query |
|
205
|
|
|
* @todo Refactor into multiple methods |
|
206
|
|
|
*/ |
|
207
|
3 |
|
public function query($query, midcom_services_indexer_filter $filter = null, array $options = []) : array |
|
208
|
|
|
{ |
|
209
|
3 |
|
$result = []; |
|
210
|
3 |
|
if ($this->_disabled) { |
|
211
|
3 |
|
return $result; |
|
212
|
|
|
} |
|
213
|
|
|
|
|
214
|
|
|
// Do charset translations |
|
215
|
|
|
$query = midcom::get()->i18n->convert_to_utf8($query); |
|
216
|
|
|
|
|
217
|
|
|
try { |
|
218
|
|
|
$result_raw = $this->_backend->query($query, $filter, $options); |
|
219
|
|
|
} catch (Exception $e) { |
|
220
|
|
|
debug_add("Query error: " . $e->getMessage(), MIDCOM_LOG_ERROR); |
|
221
|
|
|
return $result; |
|
222
|
|
|
} |
|
223
|
|
|
|
|
224
|
|
|
foreach ($result_raw as $document) { |
|
225
|
|
|
$document->fields_to_members(); |
|
226
|
|
|
/** |
|
227
|
|
|
* FIXME: Rethink program flow and especially take into account that not all documents are |
|
228
|
|
|
* created by midcom or even served by midgard |
|
229
|
|
|
*/ |
|
230
|
|
|
|
|
231
|
|
|
// midgard:read verification, we simply try to create an object instance |
|
232
|
|
|
// In the case, we distinguish between MidCOM documents, where we can check |
|
233
|
|
|
// the RI identified object directly, and pure documents, where we use the |
|
234
|
|
|
// topic instead. |
|
235
|
|
|
|
|
236
|
|
|
// Try to check topic only if the guid is actually set |
|
237
|
|
|
if (!empty($document->topic_guid)) { |
|
238
|
|
|
try { |
|
239
|
|
|
midcom_db_topic::get_cached($document->topic_guid); |
|
240
|
|
|
} catch (midcom_error $e) { |
|
241
|
|
|
// Skip document, the object is hidden. |
|
242
|
|
|
debug_add("Skipping the generic document {$document->title}, its topic seems to be invisible, we cannot proceed."); |
|
243
|
|
|
continue; |
|
244
|
|
|
} |
|
245
|
|
|
} |
|
246
|
|
|
|
|
247
|
|
|
// this checks acls! |
|
248
|
|
|
if ($document->is_a('midcom')) { |
|
249
|
|
|
// Try to retrieve object: |
|
250
|
|
|
// Strip language code from end of RI if it looks like "<GUID>_<LANG>" |
|
251
|
|
|
try { |
|
252
|
|
|
midcom::get()->dbfactory->get_object_by_guid(preg_replace('/^([0-9a-f]{32,80})_[a-z]{2}$/', '\\1', $document->RI)); |
|
253
|
|
|
} catch (midcom_error $e) { |
|
254
|
|
|
// Skip document, the object is hidden, deleted or otherwise unavailable. |
|
255
|
|
|
//@todo Maybe nonexistent objects should be removed from index? |
|
256
|
|
|
continue; |
|
257
|
|
|
} |
|
258
|
|
|
} |
|
259
|
|
|
$result[] = $document; |
|
260
|
|
|
} |
|
261
|
|
|
return $result; |
|
262
|
|
|
} |
|
263
|
|
|
|
|
264
|
|
|
/** |
|
265
|
|
|
* Try to instantiate the most specific document class for the object given in the parameter. |
|
266
|
|
|
* |
|
267
|
|
|
* This class will not return empty document base class instances if nothing |
|
268
|
|
|
* specific can be found. If you are in this situation, you need to instantiate |
|
269
|
|
|
* an appropriate document manually and populate it. |
|
270
|
|
|
* |
|
271
|
|
|
* The checking sequence is like this right now: |
|
272
|
|
|
* |
|
273
|
|
|
* 1. If a datamanager instance is passed, it is transformed into a datamanager document. |
|
274
|
|
|
* 2. If a Metadata object is passed, it is transformed into a midcom_services_indexer_document_midcom. |
|
275
|
|
|
* 3. Next, the method tries to retrieve a MidCOM Metadata object using the parameter directly. If successful, |
|
276
|
|
|
* again, a midcom_services_indexer_document_midcom is returned. |
|
277
|
|
|
* |
|
278
|
|
|
* This factory method will work even if the indexer is disabled. You can check this |
|
279
|
|
|
* with the enabled() method of this class. |
|
280
|
|
|
* |
|
281
|
|
|
* @todo Move to a full factory pattern here to save document php file parsings where possible. |
|
282
|
|
|
* This means that all document creations will in the future be handled by this method. |
|
283
|
|
|
* |
|
284
|
|
|
* @param object $object The object for which a document instance is required |
|
285
|
|
|
* @return midcom_services_indexer_document A valid document class as specific as possible. Returns |
|
286
|
|
|
* false on error or if no specific class match could be found. |
|
287
|
|
|
*/ |
|
288
|
2 |
|
function new_document($object) |
|
|
|
|
|
|
289
|
|
|
{ |
|
290
|
|
|
// Scan for datamanager instances. |
|
291
|
2 |
|
if (is_a($object, 'midcom\datamanager\datamanager')) { |
|
292
|
2 |
|
debug_add('This is a datamanager document'); |
|
293
|
2 |
|
return new midcom\datamanager\indexer\document($object); |
|
294
|
|
|
} |
|
295
|
|
|
if (is_a($object, 'midcom_helper_datamanager2_datamanager')) { |
|
296
|
|
|
debug_add('This is a datamanager2 document'); |
|
297
|
|
|
return new midcom_helper_datamanager2_indexer_document($object); |
|
|
|
|
|
|
298
|
|
|
} |
|
299
|
|
|
|
|
300
|
|
|
// Maybe we have a metadata object... |
|
301
|
|
|
if (is_a($object, midcom_helper_metadata::class)) { |
|
302
|
|
|
debug_add('This is a metadata document, built from a metadata object.'); |
|
303
|
|
|
return new midcom_services_indexer_document_midcom($object); |
|
304
|
|
|
} |
|
305
|
|
|
|
|
306
|
|
|
// Try to get a metadata object for the argument passed |
|
307
|
|
|
// This should catch all DBA objects as well. |
|
308
|
|
|
if ($metadata = midcom_helper_metadata::retrieve($object)) { |
|
309
|
|
|
debug_add('Successfully fetched a Metadata object for the argument.'); |
|
310
|
|
|
return new midcom_services_indexer_document_midcom($metadata); |
|
311
|
|
|
} |
|
312
|
|
|
|
|
313
|
|
|
// No specific match found. |
|
314
|
|
|
debug_print_r('No match found for this type:', $object); |
|
315
|
|
|
return false; |
|
316
|
|
|
} |
|
317
|
|
|
} |
|
318
|
|
|
|