|
1
|
<?php
|
|
2
|
|
|
3
|
/**
|
|
4
|
* @file
|
|
5
|
* FeedsImporter class and related.
|
|
6
|
*/
|
|
7
|
|
|
8
|
/**
|
|
9
|
* A FeedsImporter object describes how an external source should be fetched,
|
|
10
|
* parsed and processed. Feeds can manage an arbitrary amount of importers.
|
|
11
|
*
|
|
12
|
* A FeedsImporter holds a pointer to a FeedsFetcher, a FeedsParser and a
|
|
13
|
* FeedsProcessor plugin. It further contains the configuration for itself and
|
|
14
|
* each of the three plugins.
|
|
15
|
*
|
|
16
|
* Its most important responsibilities are configuration management, interfacing
|
|
17
|
* with the job scheduler and expiring of all items produced by this
|
|
18
|
* importer.
|
|
19
|
*
|
|
20
|
* When a FeedsImporter is instantiated, it loads its configuration. Then it
|
|
21
|
* instantiates one fetcher, one parser and one processor plugin depending on
|
|
22
|
* the configuration information. After instantiating them, it sets them to
|
|
23
|
* the configuration information it holds for them.
|
|
24
|
*/
|
|
25
|
class FeedsImporter extends FeedsConfigurable {
|
|
26
|
|
|
27
|
// Every feed has a fetcher, a parser and a processor.
|
|
28
|
// These variable names match the possible return values of
|
|
29
|
// FeedsPlugin::typeOf().
|
|
30
|
protected $fetcher, $parser, $processor;
|
|
31
|
|
|
32
|
// This array defines the variable names of the plugins above.
|
|
33
|
protected $plugin_types = array('fetcher', 'parser', 'processor');
|
|
34
|
|
|
35
|
/**
|
|
36
|
* Instantiate class variables, initialize and configure
|
|
37
|
* plugins.
|
|
38
|
*/
|
|
39
|
protected function __construct($id) {
|
|
40
|
parent::__construct($id);
|
|
41
|
|
|
42
|
// Try to load information from database.
|
|
43
|
$this->load();
|
|
44
|
|
|
45
|
// Instantiate fetcher, parser and processor, set their configuration if
|
|
46
|
// stored info is available.
|
|
47
|
foreach ($this->plugin_types as $type) {
|
|
48
|
$plugin = feeds_plugin($this->config[$type]['plugin_key'], $this->id);
|
|
49
|
|
|
50
|
if (isset($this->config[$type]['config'])) {
|
|
51
|
$plugin->setConfig($this->config[$type]['config']);
|
|
52
|
}
|
|
53
|
$this->$type = $plugin;
|
|
54
|
}
|
|
55
|
}
|
|
56
|
|
|
57
|
/**
|
|
58
|
* Remove items older than $time.
|
|
59
|
*
|
|
60
|
* @param $time
|
|
61
|
* All items older than REQUEST_TIME - $time will be deleted. If not
|
|
62
|
* given, internal processor settings will be used.
|
|
63
|
*
|
|
64
|
* @return
|
|
65
|
* FEEDS_BATCH_COMPLETE if the expiry process finished. A decimal between
|
|
66
|
* 0.0 and 0.9 periodic if expiry is still in progress.
|
|
67
|
*
|
|
68
|
* @throws
|
|
69
|
* Throws Exception if an error occurs when expiring items.
|
|
70
|
*/
|
|
71
|
public function expire($time = NULL) {
|
|
72
|
return $this->processor->expire($time);
|
|
73
|
}
|
|
74
|
|
|
75
|
/**
|
|
76
|
* Schedule all periodic tasks for this importer.
|
|
77
|
*/
|
|
78
|
public function schedule() {
|
|
79
|
$this->scheduleExpire();
|
|
80
|
}
|
|
81
|
|
|
82
|
/**
|
|
83
|
* Schedule expiry of items.
|
|
84
|
*/
|
|
85
|
public function scheduleExpire() {
|
|
86
|
$job = array(
|
|
87
|
'type' => $this->id,
|
|
88
|
'period' => 0,
|
|
89
|
'periodic' => TRUE,
|
|
90
|
);
|
|
91
|
if (FEEDS_EXPIRE_NEVER != $this->processor->expiryTime()) {
|
|
92
|
$job['period'] = 3600;
|
|
93
|
JobScheduler::get('feeds_importer_expire')->set($job);
|
|
94
|
}
|
|
95
|
else {
|
|
96
|
JobScheduler::get('feeds_importer_expire')->remove($job);
|
|
97
|
}
|
|
98
|
}
|
|
99
|
|
|
100
|
/**
|
|
101
|
* Report how many items *should* be created on one page load by this
|
|
102
|
* importer.
|
|
103
|
*
|
|
104
|
* Note:
|
|
105
|
*
|
|
106
|
* It depends on whether parser implements batching if this limit is actually
|
|
107
|
* respected. Further, if no limit is reported it doesn't mean that the
|
|
108
|
* number of items that can be created on one page load is actually without
|
|
109
|
* limit.
|
|
110
|
*
|
|
111
|
* @return
|
|
112
|
* A positive number defining the number of items that can be created on
|
|
113
|
* one page load. 0 if this number is unlimited.
|
|
114
|
*/
|
|
115
|
public function getLimit() {
|
|
116
|
return $this->processor->getLimit();
|
|
117
|
}
|
|
118
|
|
|
119
|
/**
|
|
120
|
* Save configuration.
|
|
121
|
*/
|
|
122
|
public function save() {
|
|
123
|
$save = new stdClass();
|
|
124
|
$save->id = $this->id;
|
|
125
|
$save->config = $this->getConfig();
|
|
126
|
|
|
127
|
if ($config = db_query("SELECT config FROM {feeds_importer} WHERE id = :id", array(':id' => $this->id))->fetchField()) {
|
|
128
|
drupal_write_record('feeds_importer', $save, 'id');
|
|
129
|
// Only rebuild menu if content_type has changed. Don't worry about
|
|
130
|
// rebuilding menus when creating a new importer since it will default
|
|
131
|
// to the standalone page.
|
|
132
|
$config = unserialize($config);
|
|
133
|
if ($config['content_type'] != $save->config['content_type']) {
|
|
134
|
variable_set('menu_rebuild_needed', TRUE);
|
|
135
|
}
|
|
136
|
}
|
|
137
|
else {
|
|
138
|
drupal_write_record('feeds_importer', $save);
|
|
139
|
}
|
|
140
|
}
|
|
141
|
|
|
142
|
/**
|
|
143
|
* Load configuration and unpack.
|
|
144
|
*/
|
|
145
|
public function load() {
|
|
146
|
ctools_include('export');
|
|
147
|
if ($config = ctools_export_load_object('feeds_importer', 'conditions', array('id' => $this->id))) {
|
|
148
|
$config = array_shift($config);
|
|
149
|
$this->export_type = $config->export_type;
|
|
150
|
$this->disabled = isset($config->disabled) ? $config->disabled : FALSE;
|
|
151
|
$this->config = $config->config;
|
|
152
|
return TRUE;
|
|
153
|
}
|
|
154
|
return FALSE;
|
|
155
|
}
|
|
156
|
|
|
157
|
/**
|
|
158
|
* Delete configuration. Removes configuration information
|
|
159
|
* from database, does not delete configuration itself.
|
|
160
|
*/
|
|
161
|
public function delete() {
|
|
162
|
db_delete('feeds_importer')
|
|
163
|
->condition('id', $this->id)
|
|
164
|
->execute();
|
|
165
|
$job = array(
|
|
166
|
'type' => $this->id,
|
|
167
|
'id' => 0,
|
|
168
|
);
|
|
169
|
if ($this->export_type & EXPORT_IN_CODE) {
|
|
170
|
feeds_reschedule($this->id);
|
|
171
|
}
|
|
172
|
else {
|
|
173
|
JobScheduler::get('feeds_importer_expire')->remove($job);
|
|
174
|
}
|
|
175
|
}
|
|
176
|
|
|
177
|
/**
|
|
178
|
* Set plugin.
|
|
179
|
*
|
|
180
|
* @param $plugin_key
|
|
181
|
* A fetcher, parser or processor plugin.
|
|
182
|
*
|
|
183
|
* @todo Error handling, handle setting to the same plugin.
|
|
184
|
*/
|
|
185
|
public function setPlugin($plugin_key) {
|
|
186
|
// $plugin_type can be either 'fetcher', 'parser' or 'processor'
|
|
187
|
if ($plugin_type = FeedsPlugin::typeOf($plugin_key)) {
|
|
188
|
if ($plugin = feeds_plugin($plugin_key, $this->id)) {
|
|
189
|
// Unset existing plugin, switch to new plugin.
|
|
190
|
unset($this->$plugin_type);
|
|
191
|
$this->$plugin_type = $plugin;
|
|
192
|
// Set configuration information, blow away any previous information on
|
|
193
|
// this spot.
|
|
194
|
$this->config[$plugin_type] = array('plugin_key' => $plugin_key);
|
|
195
|
}
|
|
196
|
}
|
|
197
|
}
|
|
198
|
|
|
199
|
/**
|
|
200
|
* Copy a FeedsImporter configuration into this importer.
|
|
201
|
*
|
|
202
|
* @param FeedsImporter $importer
|
|
203
|
* The feeds importer object to copy from.
|
|
204
|
*/
|
|
205
|
public function copy(FeedsConfigurable $configurable) {
|
|
206
|
parent::copy($configurable);
|
|
207
|
|
|
208
|
if ($configurable instanceof FeedsImporter) {
|
|
209
|
// Instantiate new fetcher, parser and processor and initialize their
|
|
210
|
// configurations.
|
|
211
|
foreach ($this->plugin_types as $plugin_type) {
|
|
212
|
$this->setPlugin($configurable->config[$plugin_type]['plugin_key']);
|
|
213
|
$this->$plugin_type->setConfig($configurable->config[$plugin_type]['config']);
|
|
214
|
}
|
|
215
|
}
|
|
216
|
}
|
|
217
|
|
|
218
|
/**
|
|
219
|
* Get configuration of this feed.
|
|
220
|
*/
|
|
221
|
public function getConfig() {
|
|
222
|
foreach (array('fetcher', 'parser', 'processor') as $type) {
|
|
223
|
$this->config[$type]['config'] = $this->$type->getConfig();
|
|
224
|
}
|
|
225
|
return parent::getConfig();
|
|
226
|
}
|
|
227
|
|
|
228
|
/**
|
|
229
|
* Return defaults for feed configuration.
|
|
230
|
*/
|
|
231
|
public function configDefaults() {
|
|
232
|
return array(
|
|
233
|
'name' => '',
|
|
234
|
'description' => '',
|
|
235
|
'fetcher' => array(
|
|
236
|
'plugin_key' => 'FeedsHTTPFetcher',
|
|
237
|
),
|
|
238
|
'parser' => array(
|
|
239
|
'plugin_key' => 'FeedsSyndicationParser',
|
|
240
|
),
|
|
241
|
'processor' => array(
|
|
242
|
'plugin_key' => 'FeedsNodeProcessor',
|
|
243
|
),
|
|
244
|
'content_type' => '',
|
|
245
|
'update' => 0,
|
|
246
|
'import_period' => 1800, // Refresh every 30 minutes by default.
|
|
247
|
'expire_period' => 3600, // Expire every hour by default, this is a hidden setting.
|
|
248
|
'import_on_create' => TRUE, // Import on submission.
|
|
249
|
'process_in_background' => FALSE,
|
|
250
|
);
|
|
251
|
}
|
|
252
|
|
|
253
|
/**
|
|
254
|
* Override parent::configForm().
|
|
255
|
*/
|
|
256
|
public function configForm(&$form_state) {
|
|
257
|
$config = $this->getConfig();
|
|
258
|
$form = array();
|
|
259
|
$form['name'] = array(
|
|
260
|
'#type' => 'textfield',
|
|
261
|
'#title' => t('Name'),
|
|
262
|
'#description' => t('A human readable name of this importer.'),
|
|
263
|
'#default_value' => $config['name'],
|
|
264
|
'#required' => TRUE,
|
|
265
|
);
|
|
266
|
$form['description'] = array(
|
|
267
|
'#type' => 'textfield',
|
|
268
|
'#title' => t('Description'),
|
|
269
|
'#description' => t('A description of this importer.'),
|
|
270
|
'#default_value' => $config['description'],
|
|
271
|
);
|
|
272
|
$node_types = node_type_get_names();
|
|
273
|
array_walk($node_types, 'check_plain');
|
|
274
|
$form['content_type'] = array(
|
|
275
|
'#type' => 'select',
|
|
276
|
'#title' => t('Attach to content type'),
|
|
277
|
'#description' => t('If "Use standalone form" is selected a source is imported by using a form under !import_form.
|
|
278
|
If a content type is selected a source is imported by creating a node of that content type.',
|
|
279
|
array('!import_form' => l(url('import', array('absolute' => TRUE)), 'import', array('attributes' => array('target' => '_new'))))),
|
|
280
|
'#options' => array('' => t('Use standalone form')) + $node_types,
|
|
281
|
'#default_value' => $config['content_type'],
|
|
282
|
);
|
|
283
|
$cron_required = ' ' . l(t('Requires cron to be configured.'), 'http://drupal.org/cron', array('attributes' => array('target' => '_new')));
|
|
284
|
$period = drupal_map_assoc(array(900, 1800, 3600, 10800, 21600, 43200, 86400, 259200, 604800, 2419200), 'format_interval');
|
|
285
|
foreach ($period as &$p) {
|
|
286
|
$p = t('Every !p', array('!p' => $p));
|
|
287
|
}
|
|
288
|
$period = array(
|
|
289
|
FEEDS_SCHEDULE_NEVER => t('Off'),
|
|
290
|
0 => t('As often as possible'),
|
|
291
|
) + $period;
|
|
292
|
$form['import_period'] = array(
|
|
293
|
'#type' => 'select',
|
|
294
|
'#title' => t('Periodic import'),
|
|
295
|
'#options' => $period,
|
|
296
|
'#description' => t('Choose how often a source should be imported periodically.') . $cron_required,
|
|
297
|
'#default_value' => $config['import_period'],
|
|
298
|
);
|
|
299
|
$form['import_on_create'] = array(
|
|
300
|
'#type' => 'checkbox',
|
|
301
|
'#title' => t('Import on submission'),
|
|
302
|
'#description' => t('Check if import should be started at the moment a standalone form or node form is submitted.'),
|
|
303
|
'#default_value' => $config['import_on_create'],
|
|
304
|
);
|
|
305
|
$form['process_in_background'] = array(
|
|
306
|
'#type' => 'checkbox',
|
|
307
|
'#title' => t('Process in background'),
|
|
308
|
'#description' => t('For very large imports. If checked, import and delete tasks started from the web UI will be handled by a cron task in the background rather than by the browser. This does not affect periodic imports, they are handled by a cron task in any case.') . $cron_required,
|
|
309
|
'#default_value' => $config['process_in_background'],
|
|
310
|
);
|
|
311
|
return $form;
|
|
312
|
}
|
|
313
|
|
|
314
|
/**
|
|
315
|
* Reschedule if import period changes.
|
|
316
|
*/
|
|
317
|
public function configFormSubmit(&$values) {
|
|
318
|
if ($this->config['import_period'] != $values['import_period']) {
|
|
319
|
feeds_reschedule($this->id);
|
|
320
|
}
|
|
321
|
parent::configFormSubmit($values);
|
|
322
|
}
|
|
323
|
}
|
|
324
|
|
|
325
|
/**
|
|
326
|
* Helper, see FeedsDataProcessor class.
|
|
327
|
*/
|
|
328
|
function feeds_format_expire($timestamp) {
|
|
329
|
if ($timestamp == FEEDS_EXPIRE_NEVER) {
|
|
330
|
return t('Never');
|
|
331
|
}
|
|
332
|
return t('after !time', array('!time' => format_interval($timestamp)));
|
|
333
|
}
|
