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
|
|
|
* @var midcom_services_indexer_backend |
34
|
|
|
*/ |
35
|
|
|
private $_backend; |
36
|
|
|
|
37
|
|
|
/** |
38
|
|
|
* @var boolean |
39
|
|
|
*/ |
40
|
|
|
private $_disabled; |
41
|
|
|
|
42
|
|
|
/** |
43
|
|
|
* Initialization |
44
|
|
|
*/ |
45
|
1 |
|
public function __construct(midcom_services_indexer_backend $backend = null) |
46
|
|
|
{ |
47
|
1 |
|
$this->_backend = $backend; |
48
|
1 |
|
$this->_disabled = $this->_backend === null; |
49
|
1 |
|
} |
50
|
|
|
|
51
|
|
|
public static function getSubscribedEvents() |
52
|
|
|
{ |
53
|
|
|
return [dbaevent::DELETE => ['handle_delete']]; |
54
|
|
|
} |
55
|
|
|
|
56
|
|
|
public function handle_delete(dbaevent $event) |
57
|
|
|
{ |
58
|
|
|
$this->delete($event->get_object()->guid); |
|
|
|
|
59
|
|
|
} |
60
|
|
|
|
61
|
|
|
/** |
62
|
|
|
* Simple helper, returns true if the indexer service is online, false if it is disabled. |
63
|
|
|
*/ |
64
|
|
|
public function enabled() : bool |
65
|
|
|
{ |
66
|
|
|
return !$this->_disabled; |
67
|
|
|
} |
68
|
|
|
|
69
|
|
|
/** |
70
|
|
|
* Adds a document to the index. |
71
|
|
|
* |
72
|
|
|
* A finished document object must be passed to this object. If the index |
73
|
|
|
* already contains a record with the same Resource Identifier, the record |
74
|
|
|
* is replaced. |
75
|
|
|
* |
76
|
|
|
* Support of batch-indexing using an Array of documents instead of a single |
77
|
|
|
* document is possible (and strongly advised for performance reasons). |
78
|
|
|
* |
79
|
|
|
* @param mixed $documents One or more documents to be indexed, so this is either a |
80
|
|
|
* midcom_services_indexer_document or an Array of these objects. |
81
|
|
|
* @return boolean Indicating success. |
82
|
|
|
*/ |
83
|
6 |
|
public function index($documents) : bool |
84
|
|
|
{ |
85
|
6 |
|
if ($this->_disabled) { |
86
|
6 |
|
return true; |
87
|
|
|
} |
88
|
|
|
|
89
|
|
|
$batch = is_array($documents); |
90
|
|
|
if (!$batch) { |
91
|
|
|
$documents = [$documents]; |
92
|
|
|
} |
93
|
|
|
if (empty($documents)) { |
94
|
|
|
// Nothing to do. |
95
|
|
|
return true; |
96
|
|
|
} |
97
|
|
|
|
98
|
|
|
foreach ($documents as $value) { |
99
|
|
|
$value->members_to_fields(); |
100
|
|
|
} |
101
|
|
|
|
102
|
|
|
try { |
103
|
|
|
$this->_backend->index($documents); |
104
|
|
|
return true; |
105
|
|
|
} catch (Exception $e) { |
106
|
|
|
if ($batch) { |
107
|
|
|
throw $e; |
108
|
|
|
} |
109
|
|
|
debug_add("Indexing error: " . $e->getMessage(), MIDCOM_LOG_ERROR); |
110
|
|
|
return false; |
111
|
|
|
} |
112
|
|
|
} |
113
|
|
|
|
114
|
|
|
/** |
115
|
|
|
* Removes the document(s) with the given resource identifier(s) from the index. |
116
|
|
|
* Using GUIDs instead of RIs will delete all language versions |
117
|
|
|
* |
118
|
|
|
* @param array $RIs The resource identifier(s) of the document(s) that should be deleted. |
119
|
|
|
* @return boolean Indicating success. |
120
|
|
|
*/ |
121
|
|
|
public function delete($RIs) : bool |
122
|
|
|
{ |
123
|
|
|
if ($this->_disabled) { |
124
|
|
|
return true; |
125
|
|
|
} |
126
|
|
|
$RIs = (array) $RIs; |
127
|
|
|
if (empty($RIs)) { |
128
|
|
|
// Nothing to do. |
129
|
|
|
return true; |
130
|
|
|
} |
131
|
|
|
try { |
132
|
|
|
$this->_backend->delete($RIs); |
133
|
|
|
return true; |
134
|
|
|
} catch (Exception $e) { |
135
|
|
|
debug_add("Deleting error: " . $e->getMessage(), MIDCOM_LOG_ERROR); |
136
|
|
|
return false; |
137
|
|
|
} |
138
|
|
|
} |
139
|
|
|
|
140
|
|
|
/** |
141
|
|
|
* Clear the index completely. |
142
|
|
|
* |
143
|
|
|
* This will drop the current index. |
144
|
|
|
*/ |
145
|
|
|
public function delete_all(string $constraint = '') : bool |
146
|
|
|
{ |
147
|
|
|
if ($this->_disabled) { |
148
|
|
|
return true; |
149
|
|
|
} |
150
|
|
|
|
151
|
|
|
try { |
152
|
|
|
$this->_backend->delete_all($constraint); |
153
|
|
|
return true; |
154
|
|
|
} catch (Exception $e) { |
155
|
|
|
debug_add("Deleting error: " . $e->getMessage(), MIDCOM_LOG_ERROR); |
156
|
|
|
return false; |
157
|
|
|
} |
158
|
|
|
} |
159
|
|
|
|
160
|
|
|
/** |
161
|
|
|
* Query the index and, if set, restrict the query by a given filter. |
162
|
|
|
* |
163
|
|
|
* The filter argument is optional and may be a subclass of indexer_filter. |
164
|
|
|
* The backend determines what filters are supported and how they are |
165
|
|
|
* treated. |
166
|
|
|
* |
167
|
|
|
* The query syntax is also dependent on the backend. Refer to its documentation |
168
|
|
|
* how queries should be built. |
169
|
|
|
* |
170
|
|
|
* @param string $query The query, which must suit the backends query syntax. It is assumed to be in the site charset. |
171
|
|
|
* @param array $options Options that are passed straight to the backend |
172
|
|
|
* @return midcom_services_indexer_document[] An array of documents matching the query |
173
|
|
|
* @todo Refactor into multiple methods |
174
|
|
|
*/ |
175
|
3 |
|
public function query(string $query, midcom_services_indexer_filter $filter = null, array $options = []) : array |
176
|
|
|
{ |
177
|
3 |
|
$result = []; |
178
|
3 |
|
if ($this->_disabled) { |
179
|
3 |
|
return $result; |
180
|
|
|
} |
181
|
|
|
|
182
|
|
|
// Do charset translations |
183
|
|
|
$query = midcom::get()->i18n->convert_to_utf8($query); |
184
|
|
|
|
185
|
|
|
try { |
186
|
|
|
$result_raw = $this->_backend->query($query, $filter, $options); |
187
|
|
|
} catch (Exception $e) { |
188
|
|
|
debug_add("Query error: " . $e->getMessage(), MIDCOM_LOG_ERROR); |
189
|
|
|
return $result; |
190
|
|
|
} |
191
|
|
|
|
192
|
|
|
foreach ($result_raw as $document) { |
193
|
|
|
$document->fields_to_members(); |
194
|
|
|
/** |
195
|
|
|
* FIXME: Rethink program flow and especially take into account that not all documents are |
196
|
|
|
* created by midcom or even served by midgard |
197
|
|
|
*/ |
198
|
|
|
|
199
|
|
|
// midgard:read verification, we simply try to create an object instance |
200
|
|
|
// In the case, we distinguish between MidCOM documents, where we can check |
201
|
|
|
// the RI identified object directly, and pure documents, where we use the |
202
|
|
|
// topic instead. |
203
|
|
|
|
204
|
|
|
// Try to check topic only if the guid is actually set |
205
|
|
|
if (!empty($document->topic_guid)) { |
206
|
|
|
try { |
207
|
|
|
midcom_db_topic::get_cached($document->topic_guid); |
208
|
|
|
} catch (midcom_error $e) { |
209
|
|
|
// Skip document, the object is hidden. |
210
|
|
|
debug_add("Skipping the generic document {$document->title}, its topic seems to be invisible, we cannot proceed."); |
211
|
|
|
continue; |
212
|
|
|
} |
213
|
|
|
} |
214
|
|
|
|
215
|
|
|
// this checks acls! |
216
|
|
|
if ($document->is_a('midcom')) { |
217
|
|
|
// Try to retrieve object: |
218
|
|
|
// Strip language code from end of RI if it looks like "<GUID>_<LANG>" |
219
|
|
|
try { |
220
|
|
|
midcom::get()->dbfactory->get_object_by_guid(preg_replace('/^([0-9a-f]{32,80})_[a-z]{2}$/', '\\1', $document->RI)); |
221
|
|
|
} catch (midcom_error $e) { |
222
|
|
|
// Skip document, the object is hidden, deleted or otherwise unavailable. |
223
|
|
|
//@todo Maybe nonexistent objects should be removed from index? |
224
|
|
|
continue; |
225
|
|
|
} |
226
|
|
|
} |
227
|
|
|
$result[] = $document; |
228
|
|
|
} |
229
|
|
|
return $result; |
230
|
|
|
} |
231
|
|
|
|
232
|
|
|
/** |
233
|
|
|
* Try to instantiate the most specific document class for the object given in the parameter. |
234
|
|
|
* |
235
|
|
|
* This class will not return empty document base class instances if nothing |
236
|
|
|
* specific can be found. If you are in this situation, you need to instantiate |
237
|
|
|
* an appropriate document manually and populate it. |
238
|
|
|
* |
239
|
|
|
* This factory method will work even if the indexer is disabled. You can check this |
240
|
|
|
* with the enabled() method of this class. |
241
|
|
|
* |
242
|
|
|
* @todo Move to a full factory pattern here to save document php file parsings where possible. |
243
|
|
|
* This means that all document creations will in the future be handled by this method. |
244
|
|
|
* |
245
|
|
|
* @param object $object The object for which a document instance is required |
246
|
|
|
* @return midcom_services_indexer_document A valid document class as specific as possible. Returns |
247
|
|
|
* false on error or if no specific class match could be found. |
248
|
|
|
*/ |
249
|
2 |
|
function new_document($object) |
|
|
|
|
250
|
|
|
{ |
251
|
|
|
// Scan for datamanager instances. |
252
|
2 |
|
if (is_a($object, 'midcom\datamanager\datamanager')) { |
253
|
2 |
|
return new midcom\datamanager\indexer\document($object); |
254
|
|
|
} |
255
|
|
|
if (is_a($object, 'midcom_helper_datamanager2_datamanager')) { |
256
|
|
|
return new midcom_helper_datamanager2_indexer_document($object); |
|
|
|
|
257
|
|
|
} |
258
|
|
|
|
259
|
|
|
if ($object instanceof midcom_core_dbaobject) { |
260
|
|
|
return new midcom_services_indexer_document_midcom($object); |
261
|
|
|
} |
262
|
|
|
|
263
|
|
|
// No specific match found. |
264
|
|
|
debug_print_r('No match found for this type:', $object); |
265
|
|
|
return false; |
266
|
|
|
} |
267
|
|
|
} |
268
|
|
|
|