|
1
|
<?php
|
|
2
|
|
|
3
|
/**
|
|
4
|
* @file
|
|
5
|
* FeedsImporter class and related.
|
|
6
|
*/
|
|
7
|
|
|
8
|
/**
|
|
9
|
* Class for a Feeds importer.
|
|
10
|
*
|
|
11
|
* A FeedsImporter object describes how an external source should be fetched,
|
|
12
|
* parsed and processed. Feeds can manage an arbitrary amount of importers.
|
|
13
|
*
|
|
14
|
* A FeedsImporter holds a pointer to a FeedsFetcher, a FeedsParser and a
|
|
15
|
* FeedsProcessor plugin. It further contains the configuration for itself and
|
|
16
|
* each of the three plugins.
|
|
17
|
*
|
|
18
|
* Its most important responsibilities are configuration management, interfacing
|
|
19
|
* with the job scheduler and expiring of all items produced by this
|
|
20
|
* importer.
|
|
21
|
*
|
|
22
|
* When a FeedsImporter is instantiated, it loads its configuration. Then it
|
|
23
|
* instantiates one fetcher, one parser and one processor plugin depending on
|
|
24
|
* the configuration information. After instantiating them, it sets them to
|
|
25
|
* the configuration information it holds for them.
|
|
26
|
*/
|
|
27
|
class FeedsImporter extends FeedsConfigurable {
|
|
28
|
|
|
29
|
/**
|
|
30
|
* Every feed has a fetcher, a parser and a processor.
|
|
31
|
*
|
|
32
|
* These variable names match the possible return values of
|
|
33
|
* FeedsPlugin::typeOf().
|
|
34
|
*/
|
|
35
|
|
|
36
|
/**
|
|
37
|
* The selected fetcher for this importer.
|
|
38
|
*
|
|
39
|
* @var FeedsFetcher|null
|
|
40
|
*/
|
|
41
|
protected $fetcher;
|
|
42
|
|
|
43
|
/**
|
|
44
|
* The selected parser for this importer.
|
|
45
|
*
|
|
46
|
* @var FeedsParser|null
|
|
47
|
*/
|
|
48
|
protected $parser;
|
|
49
|
|
|
50
|
/**
|
|
51
|
* The selected processor for this importer.
|
|
52
|
*
|
|
53
|
* @var FeedsProcessor|null
|
|
54
|
*/
|
|
55
|
protected $processor;
|
|
56
|
|
|
57
|
/**
|
|
58
|
* This array defines the variable names of the plugins above.
|
|
59
|
*
|
|
60
|
* @var string[]
|
|
61
|
*/
|
|
62
|
protected $plugin_types = array('fetcher', 'parser', 'processor');
|
|
63
|
|
|
64
|
/**
|
|
65
|
* Instantiate class variables, initialize and configure plugins.
|
|
66
|
*
|
|
67
|
* @param string $id
|
|
68
|
* The importer ID.
|
|
69
|
*/
|
|
70
|
protected function __construct($id) {
|
|
71
|
parent::__construct($id);
|
|
72
|
|
|
73
|
// Try to load information from database.
|
|
74
|
$this->load();
|
|
75
|
|
|
76
|
// Instantiate fetcher, parser and processor, set their configuration if
|
|
77
|
// stored info is available.
|
|
78
|
foreach ($this->plugin_types as $type) {
|
|
79
|
$plugin = feeds_plugin($this->config[$type]['plugin_key'], $this->id);
|
|
80
|
|
|
81
|
if (isset($this->config[$type]['config'])) {
|
|
82
|
$plugin->setConfig($this->config[$type]['config']);
|
|
83
|
}
|
|
84
|
$this->$type = $plugin;
|
|
85
|
}
|
|
86
|
}
|
|
87
|
|
|
88
|
/**
|
|
89
|
* Reports how many items *should* be created on one request by this importer.
|
|
90
|
*
|
|
91
|
* Note:
|
|
92
|
*
|
|
93
|
* It depends on whether parser implements batching if this limit is actually
|
|
94
|
* respected. Further, if no limit is reported it doesn't mean that the
|
|
95
|
* number of items that can be created on one page load is actually without
|
|
96
|
* limit.
|
|
97
|
*
|
|
98
|
* @return int
|
|
99
|
* A positive number defining the number of items that can be created on
|
|
100
|
* one page load. 0 if this number is unlimited.
|
|
101
|
*/
|
|
102
|
public function getLimit() {
|
|
103
|
return $this->processor->getLimit();
|
|
104
|
}
|
|
105
|
|
|
106
|
/**
|
|
107
|
* Save configuration.
|
|
108
|
*/
|
|
109
|
public function save() {
|
|
110
|
$save = new stdClass();
|
|
111
|
$save->id = $this->id;
|
|
112
|
$save->config = $this->getConfig();
|
|
113
|
|
|
114
|
if ($config = db_query("SELECT config FROM {feeds_importer} WHERE id = :id", array(':id' => $this->id))->fetchField()) {
|
|
115
|
drupal_write_record('feeds_importer', $save, 'id');
|
|
116
|
// Only rebuild menu if content_type has changed. Don't worry about
|
|
117
|
// rebuilding menus when creating a new importer since it will default
|
|
118
|
// to the standalone page.
|
|
119
|
$config = unserialize($config);
|
|
120
|
if ($config['content_type'] != $save->config['content_type']) {
|
|
121
|
variable_set('menu_rebuild_needed', TRUE);
|
|
122
|
}
|
|
123
|
}
|
|
124
|
else {
|
|
125
|
drupal_write_record('feeds_importer', $save);
|
|
126
|
}
|
|
127
|
}
|
|
128
|
|
|
129
|
/**
|
|
130
|
* Load configuration and unpack.
|
|
131
|
*/
|
|
132
|
public function load() {
|
|
133
|
ctools_include('export');
|
|
134
|
if ($config = ctools_export_load_object('feeds_importer', 'conditions', array('id' => $this->id))) {
|
|
135
|
$config = array_shift($config);
|
|
136
|
$this->export_type = $config->export_type;
|
|
137
|
$this->disabled = isset($config->disabled) ? $config->disabled : FALSE;
|
|
138
|
$this->config = $config->config;
|
|
139
|
return TRUE;
|
|
140
|
}
|
|
141
|
return FALSE;
|
|
142
|
}
|
|
143
|
|
|
144
|
/**
|
|
145
|
* Deletes configuration.
|
|
146
|
*
|
|
147
|
* Removes configuration information from database, does not delete
|
|
148
|
* configuration itself.
|
|
149
|
*/
|
|
150
|
public function delete() {
|
|
151
|
db_delete('feeds_importer')
|
|
152
|
->condition('id', $this->id)
|
|
153
|
->execute();
|
|
154
|
|
|
155
|
feeds_reschedule($this->id);
|
|
156
|
}
|
|
157
|
|
|
158
|
/**
|
|
159
|
* Set plugin.
|
|
160
|
*
|
|
161
|
* @param string $plugin_key
|
|
162
|
* The name of a fetcher, parser or processor plugin.
|
|
163
|
*
|
|
164
|
* @todo Error handling, handle setting to the same plugin.
|
|
165
|
*/
|
|
166
|
public function setPlugin($plugin_key) {
|
|
167
|
// $plugin_type can be either 'fetcher', 'parser' or 'processor'.
|
|
168
|
if ($plugin_type = FeedsPlugin::typeOf($plugin_key)) {
|
|
169
|
if ($plugin = feeds_plugin($plugin_key, $this->id)) {
|
|
170
|
// Unset existing plugin, switch to new plugin.
|
|
171
|
unset($this->$plugin_type);
|
|
172
|
$this->$plugin_type = $plugin;
|
|
173
|
// Set configuration information, blow away any previous information on
|
|
174
|
// this spot.
|
|
175
|
$this->config[$plugin_type] = array('plugin_key' => $plugin_key);
|
|
176
|
}
|
|
177
|
}
|
|
178
|
}
|
|
179
|
|
|
180
|
/**
|
|
181
|
* Copy a FeedsImporter configuration into this importer.
|
|
182
|
*
|
|
183
|
* @param FeedsConfigurable $configurable
|
|
184
|
* The feeds importer object to copy from.
|
|
185
|
*/
|
|
186
|
public function copy(FeedsConfigurable $configurable) {
|
|
187
|
parent::copy($configurable);
|
|
188
|
|
|
189
|
if ($configurable instanceof FeedsImporter) {
|
|
190
|
// Instantiate new fetcher, parser and processor and initialize their
|
|
191
|
// configurations.
|
|
192
|
foreach ($this->plugin_types as $plugin_type) {
|
|
193
|
$this->setPlugin($configurable->config[$plugin_type]['plugin_key']);
|
|
194
|
$this->$plugin_type->setConfig($configurable->config[$plugin_type]['config']);
|
|
195
|
}
|
|
196
|
}
|
|
197
|
}
|
|
198
|
|
|
199
|
/**
|
|
200
|
* Get configuration of this feed.
|
|
201
|
*/
|
|
202
|
public function getConfig() {
|
|
203
|
foreach (array('fetcher', 'parser', 'processor') as $type) {
|
|
204
|
$this->config[$type]['config'] = $this->$type->getConfig();
|
|
205
|
}
|
|
206
|
return parent::getConfig();
|
|
207
|
}
|
|
208
|
|
|
209
|
/**
|
|
210
|
* Validates the configuration.
|
|
211
|
*/
|
|
212
|
public function validateConfig() {
|
|
213
|
$errors = parent::validateConfig();
|
|
214
|
$config = $this->getConfig();
|
|
215
|
|
|
216
|
// Check if "Attach to content type" setting is the same as content type on
|
|
217
|
// node processor.
|
|
218
|
if ($this->processor instanceof FeedsNodeProcessor) {
|
|
219
|
$processor_config = $this->processor->getConfig();
|
|
220
|
if (!empty($config['content_type']) && $processor_config['bundle'] == $config['content_type']) {
|
|
221
|
$message = t('The importer is attached to the same content type as the content type selected on the node processor. Unless you have a very advanced use case, these two should never be the same.');
|
|
222
|
if (!module_exists('feeds_news')) {
|
|
223
|
$message .= ' ' . t('Enable the Feeds News module for an example of how the "@content_type" setting can be used.', array(
|
|
224
|
'@content_type' => t('Attach to content type'),
|
|
225
|
));
|
|
226
|
}
|
|
227
|
elseif ($this->id != 'feed') {
|
|
228
|
$message .= ' ' . t('See the importer !importer_name (provided by the Feeds News module) for an example of how the "@content_type" setting can be used.', array(
|
|
229
|
'!importer_name' => l(check_plain(feeds_importer('feed')->config['name']), 'admin/structure/feeds/feed'),
|
|
230
|
'@content_type' => t('Attach to content type'),
|
|
231
|
));
|
|
232
|
}
|
|
233
|
|
|
234
|
$errors[] = $message;
|
|
235
|
}
|
|
236
|
}
|
|
237
|
|
|
238
|
// Check for a combination of incompatible settings.
|
|
239
|
if (!$config['import_on_create'] && empty($config['content_type'])) {
|
|
240
|
if ($config['import_period'] == FEEDS_SCHEDULE_NEVER) {
|
|
241
|
$errors[] = t('"@import_period" and "@import_on_create" are both turned off and the importer is not attached to a content type. Unless you have alternative methods of running imports for this importer, Feeds will not import anything for this importer.', array(
|
|
242
|
'@import_period' => t('Periodic import'),
|
|
243
|
'@import_on_create' => t('Import on submission'),
|
|
244
|
));
|
|
245
|
}
|
|
246
|
elseif ($config['process_in_background']) {
|
|
247
|
$errors[] = t('Since "@import_on_create" is turned off and the importer is not attached to a content type, the "@process_in_background" setting may have no effect. When submitting the standalone form with the "@import_on_create" setting turned off, the feed is only scheduled for periodic import.', array(
|
|
248
|
'@import_on_create' => t('Import on submission'),
|
|
249
|
'@process_in_background' => t('Process in background'),
|
|
250
|
));
|
|
251
|
}
|
|
252
|
}
|
|
253
|
|
|
254
|
// Validate errors of each plugin as well.
|
|
255
|
$plugin_types = array(
|
|
256
|
'fetcher' => t('Fetcher'),
|
|
257
|
'parser' => t('Parser'),
|
|
258
|
'processor' => t('Processor'),
|
|
259
|
);
|
|
260
|
foreach ($plugin_types as $type => $plugin_label) {
|
|
261
|
// Check if plugin exists.
|
|
262
|
$plugin = feeds_plugin($this->config[$type]['plugin_key'], $this->id);
|
|
263
|
if ($plugin instanceof FeedsMissingPlugin) {
|
|
264
|
$errors[] = t('The plugin %plugin is unavailable.', array('%plugin' => $this->config[$type]['plugin_key']));
|
|
265
|
continue;
|
|
266
|
}
|
|
267
|
|
|
268
|
$plugin_errors = $this->$type->validateConfig();
|
|
269
|
foreach ($plugin_errors as $key => $error) {
|
|
270
|
$errors[] = t('@plugin: !error', array(
|
|
271
|
'@plugin' => $plugin_label,
|
|
272
|
'!error' => $error,
|
|
273
|
));
|
|
274
|
}
|
|
275
|
}
|
|
276
|
|
|
277
|
return $errors;
|
|
278
|
}
|
|
279
|
|
|
280
|
/**
|
|
281
|
* Return defaults for feed configuration.
|
|
282
|
*/
|
|
283
|
public function configDefaults() {
|
|
284
|
return array(
|
|
285
|
'name' => '',
|
|
286
|
'description' => '',
|
|
287
|
'fetcher' => array(
|
|
288
|
'plugin_key' => 'FeedsHTTPFetcher',
|
|
289
|
),
|
|
290
|
'parser' => array(
|
|
291
|
'plugin_key' => 'FeedsSyndicationParser',
|
|
292
|
),
|
|
293
|
'processor' => array(
|
|
294
|
'plugin_key' => 'FeedsNodeProcessor',
|
|
295
|
),
|
|
296
|
'content_type' => '',
|
|
297
|
'update' => 0,
|
|
298
|
// Refresh every 30 minutes by default.
|
|
299
|
'import_period' => 1800,
|
|
300
|
// Expire every hour by default, this is a hidden setting.
|
|
301
|
'expire_period' => 3600,
|
|
302
|
// Import on submission.
|
|
303
|
'import_on_create' => TRUE,
|
|
304
|
'process_in_background' => FALSE,
|
|
305
|
) + parent::configDefaults();
|
|
306
|
}
|
|
307
|
|
|
308
|
/**
|
|
309
|
* Overrides parent::configForm().
|
|
310
|
*
|
|
311
|
* @param array $form_state
|
|
312
|
* The current state of the form.
|
|
313
|
*
|
|
314
|
* @return array
|
|
315
|
* The form definition.
|
|
316
|
*/
|
|
317
|
public function configForm(&$form_state) {
|
|
318
|
$config = $this->getConfig();
|
|
319
|
$form = array();
|
|
320
|
$form['name'] = array(
|
|
321
|
'#type' => 'textfield',
|
|
322
|
'#title' => t('Name'),
|
|
323
|
'#description' => t('A human readable name of this importer.'),
|
|
324
|
'#default_value' => $config['name'],
|
|
325
|
'#required' => TRUE,
|
|
326
|
);
|
|
327
|
$form['description'] = array(
|
|
328
|
'#type' => 'textfield',
|
|
329
|
'#title' => t('Description'),
|
|
330
|
'#description' => t('A description of this importer.'),
|
|
331
|
'#default_value' => $config['description'],
|
|
332
|
);
|
|
333
|
$node_types = node_type_get_names();
|
|
334
|
array_walk($node_types, 'check_plain');
|
|
335
|
$form['content_type'] = array(
|
|
336
|
'#type' => 'select',
|
|
337
|
'#title' => t('Attach to content type'),
|
|
338
|
'#description' => '<p>' . t('If "Use standalone form" is selected, a source is imported by using a form under !import_form. If a content type is selected, a source is imported by creating a node of that content type.', array(
|
|
339
|
'!import_form' => l(url('import', array('absolute' => TRUE)), 'import', array('attributes' => array('target' => '_new'))),
|
|
340
|
)) . '</p>',
|
|
341
|
'#options' => array('' => t('Use standalone form')) + $node_types,
|
|
342
|
'#default_value' => $config['content_type'],
|
|
343
|
);
|
|
344
|
$cron_required = ' ' . l(t('Requires cron to be configured.'), 'http://drupal.org/cron', array('attributes' => array('target' => '_new')));
|
|
345
|
$period = drupal_map_assoc(array(
|
|
346
|
900,
|
|
347
|
1800,
|
|
348
|
3600,
|
|
349
|
10800,
|
|
350
|
21600,
|
|
351
|
43200,
|
|
352
|
86400,
|
|
353
|
259200,
|
|
354
|
604800,
|
|
355
|
2419200,
|
|
356
|
), 'format_interval');
|
|
357
|
foreach ($period as &$p) {
|
|
358
|
$p = t('Every !p', array('!p' => $p));
|
|
359
|
}
|
|
360
|
$period = array(
|
|
361
|
FEEDS_SCHEDULE_NEVER => t('Off'),
|
|
362
|
0 => t('As often as possible'),
|
|
363
|
) + $period;
|
|
364
|
$form['import_period'] = array(
|
|
365
|
'#type' => 'select',
|
|
366
|
'#title' => t('Periodic import'),
|
|
367
|
'#options' => $period,
|
|
368
|
'#description' => t('Choose how often a source should be imported periodically.') . $cron_required,
|
|
369
|
'#default_value' => $config['import_period'],
|
|
370
|
);
|
|
371
|
$form['import_on_create'] = array(
|
|
372
|
'#type' => 'checkbox',
|
|
373
|
'#title' => t('Import on submission'),
|
|
374
|
'#description' => t('Check if import should be started at the moment a standalone form or node form is submitted. If not checked and when using the standalone form, be sure to configure periodic import.'),
|
|
375
|
'#default_value' => $config['import_on_create'],
|
|
376
|
);
|
|
377
|
$form['process_in_background'] = array(
|
|
378
|
'#type' => 'checkbox',
|
|
379
|
'#title' => t('Process in background'),
|
|
380
|
'#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,
|
|
381
|
'#default_value' => $config['process_in_background'],
|
|
382
|
);
|
|
383
|
return $form;
|
|
384
|
}
|
|
385
|
|
|
386
|
/**
|
|
387
|
* Overrides parent::configFormSubmit().
|
|
388
|
*
|
|
389
|
* Reschedule if import period changes.
|
|
390
|
*
|
|
391
|
* @param array $values
|
|
392
|
* An array that contains the values entered by the user through configForm.
|
|
393
|
*/
|
|
394
|
public function configFormSubmit(&$values) {
|
|
395
|
if ($this->config['import_period'] != $values['import_period']) {
|
|
396
|
feeds_reschedule($this->id);
|
|
397
|
}
|
|
398
|
parent::configFormSubmit($values);
|
|
399
|
}
|
|
400
|
|
|
401
|
/**
|
|
402
|
* Implements FeedsConfigurable::dependencies().
|
|
403
|
*/
|
|
404
|
public function dependencies() {
|
|
405
|
$dependencies = parent::dependencies();
|
|
406
|
foreach ($this->plugin_types as $plugin_type) {
|
|
407
|
$dependencies = array_merge($dependencies, $this->$plugin_type->dependencies());
|
|
408
|
}
|
|
409
|
return $dependencies;
|
|
410
|
}
|
|
411
|
|
|
412
|
}
|
|
413
|
|
|
414
|
/**
|
|
415
|
* Formats a time interval.
|
|
416
|
*
|
|
417
|
* @param int $timestamp
|
|
418
|
* The length of the interval in seconds.
|
|
419
|
*
|
|
420
|
* @return string
|
|
421
|
* A translated string representation of the interval.
|
|
422
|
*/
|
|
423
|
function feeds_format_expire($timestamp) {
|
|
424
|
if ($timestamp == FEEDS_EXPIRE_NEVER) {
|
|
425
|
return t('Never');
|
|
426
|
}
|
|
427
|
return t('after !time', array('!time' => format_interval($timestamp)));
|
|
428
|
}
|
