|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace AppBundle\Helper; |
|
4
|
|
|
|
|
5
|
|
|
use DateInterval; |
|
6
|
|
|
use Mediawiki\Api\MediawikiApi; |
|
7
|
|
|
use Mediawiki\Api\SimpleRequest; |
|
8
|
|
|
use Mediawiki\Api\FluentRequest; |
|
9
|
|
|
use Psr\Cache\CacheItemPoolInterface; |
|
10
|
|
|
use Symfony\Component\Config\Definition\Exception\Exception; |
|
11
|
|
|
use Symfony\Component\Debug\Exception\FatalErrorException; |
|
12
|
|
|
use Symfony\Component\DependencyInjection\ContainerInterface; |
|
13
|
|
|
use Xtools\ProjectRepository; |
|
14
|
|
|
|
|
15
|
|
|
class ApiHelper extends HelperBase |
|
16
|
|
|
{ |
|
17
|
|
|
/** @var MediawikiApi */ |
|
18
|
|
|
private $api; |
|
19
|
|
|
|
|
20
|
|
|
/** @var LabsHelper */ |
|
21
|
|
|
private $labsHelper; |
|
22
|
|
|
|
|
23
|
|
|
/** @var CacheItemPoolInterface */ |
|
24
|
|
|
protected $cache; |
|
25
|
|
|
|
|
26
|
|
|
/** @var ContainerInterface */ |
|
27
|
|
|
protected $container; |
|
28
|
|
|
|
|
29
|
|
|
public function __construct(ContainerInterface $container, LabsHelper $labsHelper) |
|
30
|
|
|
{ |
|
31
|
|
|
$this->container = $container; |
|
32
|
|
|
$this->labsHelper = $labsHelper; |
|
33
|
|
|
$this->cache = $container->get('cache.app'); |
|
34
|
|
|
} |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* Set up the MediawikiApi object for the given project. |
|
38
|
|
|
* |
|
39
|
|
|
* @param string $project |
|
40
|
|
|
*/ |
|
41
|
|
|
private function setUp($project) |
|
42
|
|
|
{ |
|
43
|
|
|
if (!$this->api instanceof MediawikiApi) { |
|
44
|
|
|
$project = ProjectRepository::getProject($project, $this->container); |
|
|
|
|
|
|
45
|
|
|
$this->api = $project->getApi(); |
|
46
|
|
|
} |
|
47
|
|
|
} |
|
48
|
|
|
|
|
49
|
|
|
/** |
|
50
|
|
|
* Get general siteinfo and namespaces for a project and cache it. |
|
51
|
|
|
* @param string [$project] Base project domain with or without protocal, or database name |
|
52
|
|
|
* such as 'en.wikipedia.org', 'https://en.wikipedia.org' or 'enwiki' |
|
53
|
|
|
* Can be left blank for single wikis. |
|
54
|
|
|
* @return string[] with keys 'general' and 'namespaces'. General info will include 'dbName', |
|
55
|
|
|
* 'wikiName', 'url', 'lang', 'articlePath', 'scriptPath', |
|
56
|
|
|
* 'script', 'timezone', and 'timeOffset' |
|
57
|
|
|
*/ |
|
58
|
|
|
public function getSiteInfo($projectName = '') |
|
59
|
|
|
{ |
|
60
|
|
|
if ($this->container->getParameter('app.single_wiki')) { |
|
61
|
|
|
$projectName = $this->container->getParameter('wiki_url'); |
|
62
|
|
|
} |
|
63
|
|
|
$project = ProjectRepository::getProject($projectName, $this->container); |
|
|
|
|
|
|
64
|
|
|
|
|
65
|
|
|
if (!$project->exists()) { |
|
66
|
|
|
throw new Exception("Unable to find project '$projectName'"); |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
$cacheKey = "siteinfo." . $project->getDatabaseName(); |
|
70
|
|
|
if ($this->cacheHas($cacheKey)) { |
|
71
|
|
|
return $this->cacheGet($cacheKey); |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
$params = [ 'meta'=>'siteinfo', 'siprop'=>'general|namespaces' ]; |
|
75
|
|
|
$query = new SimpleRequest('query', $params); |
|
76
|
|
|
|
|
77
|
|
|
$result = [ |
|
78
|
|
|
'general' => [], |
|
79
|
|
|
'namespaces' => [] |
|
80
|
|
|
]; |
|
81
|
|
|
|
|
82
|
|
|
try { |
|
83
|
|
|
$res = $project->getApi()->getRequest($query); |
|
84
|
|
|
|
|
85
|
|
|
if (isset($res['query']['general'])) { |
|
86
|
|
|
$info = $res['query']['general']; |
|
87
|
|
|
$result['general'] = [ |
|
88
|
|
|
'wikiName' => $info['sitename'], |
|
89
|
|
|
'dbName' => $info['wikiid'], |
|
90
|
|
|
'url' => $info['server'], |
|
91
|
|
|
'lang' => $info['lang'], |
|
92
|
|
|
'articlePath' => $info['articlepath'], |
|
93
|
|
|
'scriptPath' => $info['scriptpath'], |
|
94
|
|
|
'script' => $info['script'], |
|
95
|
|
|
'timezone' => $info['timezone'], |
|
96
|
|
|
'timeOffset' => $info['timeoffset'], |
|
97
|
|
|
]; |
|
98
|
|
|
|
|
99
|
|
|
if ($this->container->getParameter('app.is_labs') && substr($result['general']['dbName'], -2) != '_p') { |
|
100
|
|
|
$result['general']['dbName'] .= '_p'; |
|
101
|
|
|
} |
|
102
|
|
|
} |
|
103
|
|
|
|
|
104
|
|
View Code Duplication |
if (isset($res['query']['namespaces'])) { |
|
|
|
|
|
|
105
|
|
|
foreach ($res['query']['namespaces'] as $namespace) { |
|
106
|
|
|
if ($namespace['id'] < 0) { |
|
107
|
|
|
continue; |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
|
|
if (isset($namespace['name'])) { |
|
111
|
|
|
$name = $namespace['name']; |
|
112
|
|
|
} elseif (isset($namespace['*'])) { |
|
113
|
|
|
$name = $namespace['*']; |
|
114
|
|
|
} else { |
|
115
|
|
|
continue; |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
// FIXME: Figure out a way to i18n-ize this |
|
119
|
|
|
if ($name === '') { |
|
120
|
|
|
$name = 'Article'; |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
$result['namespaces'][$namespace['id']] = $name; |
|
124
|
|
|
} |
|
125
|
|
|
} |
|
126
|
|
|
|
|
127
|
|
|
$this->cacheSave($cacheKey, $result, 'P7D'); |
|
|
|
|
|
|
128
|
|
|
} catch (Exception $e) { |
|
129
|
|
|
// The api returned an error! Ignore |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
|
|
return $result; |
|
|
|
|
|
|
133
|
|
|
} |
|
134
|
|
|
|
|
135
|
|
View Code Duplication |
public function groups($project, $username) |
|
|
|
|
|
|
136
|
|
|
{ |
|
137
|
|
|
$this->setUp($project); |
|
138
|
|
|
$params = [ "list"=>"users", "ususers"=>$username, "usprop"=>"groups" ]; |
|
139
|
|
|
$query = new SimpleRequest('query', $params); |
|
140
|
|
|
$result = []; |
|
141
|
|
|
|
|
142
|
|
|
try { |
|
143
|
|
|
$res = $this->api->getRequest($query); |
|
144
|
|
|
if (isset($res["batchcomplete"]) && isset($res["query"]["users"][0]["groups"])) { |
|
145
|
|
|
$result = $res["query"]["users"][0]["groups"]; |
|
146
|
|
|
} |
|
147
|
|
|
} catch (Exception $e) { |
|
148
|
|
|
// The api returned an error! Ignore |
|
149
|
|
|
} |
|
150
|
|
|
|
|
151
|
|
|
return $result; |
|
152
|
|
|
} |
|
153
|
|
|
|
|
154
|
|
View Code Duplication |
public function globalGroups($project, $username) |
|
|
|
|
|
|
155
|
|
|
{ |
|
156
|
|
|
$this->setUp($project); |
|
157
|
|
|
$params = [ "meta"=>"globaluserinfo", "guiuser"=>$username, "guiprop"=>"groups" ]; |
|
158
|
|
|
$query = new SimpleRequest('query', $params); |
|
159
|
|
|
$result = []; |
|
160
|
|
|
|
|
161
|
|
|
try { |
|
162
|
|
|
$res = $this->api->getRequest($query); |
|
163
|
|
|
if (isset($res["batchcomplete"]) && isset($res["query"]["globaluserinfo"]["groups"])) { |
|
164
|
|
|
$result = $res["query"]["globaluserinfo"]["groups"]; |
|
165
|
|
|
} |
|
166
|
|
|
} catch (Exception $e) { |
|
167
|
|
|
// The api returned an error! Ignore |
|
168
|
|
|
} |
|
169
|
|
|
|
|
170
|
|
|
return $result; |
|
171
|
|
|
} |
|
172
|
|
|
|
|
173
|
|
|
/** |
|
174
|
|
|
* Get a list of namespaces on the given project. |
|
175
|
|
|
* |
|
176
|
|
|
* @param string $project such as en.wikipedia.org |
|
177
|
|
|
* @return string[] Array of namespace IDs (keys) to names (values). |
|
178
|
|
|
*/ |
|
179
|
|
|
public function namespaces($project) |
|
180
|
|
|
{ |
|
181
|
|
|
return $this->getSiteInfo($project)['namespaces']; |
|
182
|
|
|
} |
|
183
|
|
|
|
|
184
|
|
|
public function getAdmins($project) |
|
185
|
|
|
{ |
|
186
|
|
|
$params = [ |
|
187
|
|
|
'list' => 'allusers', |
|
188
|
|
|
'augroup' => 'sysop|bureaucrat|steward|oversight|checkuser', |
|
189
|
|
|
'auprop' => 'groups', |
|
190
|
|
|
'aulimit' => '500', |
|
191
|
|
|
]; |
|
192
|
|
|
|
|
193
|
|
|
$result = []; |
|
194
|
|
|
$admins = $this->massApi($params, $project, 'allusers', 'aufrom')['allusers']; |
|
195
|
|
|
|
|
196
|
|
|
foreach ($admins as $admin) { |
|
197
|
|
|
$groups = []; |
|
198
|
|
|
if (in_array("sysop", $admin["groups"])) { |
|
199
|
|
|
$groups[] = "A"; |
|
200
|
|
|
} |
|
201
|
|
|
if (in_array("bureaucrat", $admin["groups"])) { |
|
202
|
|
|
$groups[] = "B"; |
|
203
|
|
|
} |
|
204
|
|
|
if (in_array("steward", $admin["groups"])) { |
|
205
|
|
|
$groups[] = "S" ; |
|
206
|
|
|
} |
|
207
|
|
|
if (in_array("checkuser", $admin["groups"])) { |
|
208
|
|
|
$groups[] = "CU"; |
|
209
|
|
|
} |
|
210
|
|
|
if (in_array("oversight", $admin["groups"])) { |
|
211
|
|
|
$groups[] = "OS"; |
|
212
|
|
|
} |
|
213
|
|
|
if (in_array("bot", $admin["groups"])) { |
|
214
|
|
|
$groups[] = "Bot"; |
|
215
|
|
|
} |
|
216
|
|
|
$result[ $admin["name"] ] = [ |
|
217
|
|
|
"groups" => implode('/', $groups) |
|
218
|
|
|
]; |
|
219
|
|
|
} |
|
220
|
|
|
|
|
221
|
|
|
return $result; |
|
222
|
|
|
} |
|
223
|
|
|
|
|
224
|
|
|
/** |
|
225
|
|
|
* Get basic info about a page via the API |
|
226
|
|
|
* @param string $project Full domain of project (en.wikipedia.org) |
|
227
|
|
|
* @param string $page Page title |
|
228
|
|
|
* @param boolean $followRedir Whether or not to resolve redirects |
|
229
|
|
|
* @return array Associative array of data |
|
230
|
|
|
*/ |
|
231
|
|
|
public function getBasicPageInfo($project, $page, $followRedir) |
|
232
|
|
|
{ |
|
233
|
|
|
$this->setUp($project); |
|
234
|
|
|
|
|
235
|
|
|
// @TODO: Also include 'extlinks' prop when we start checking for dead external links. |
|
236
|
|
|
$params = [ |
|
237
|
|
|
'prop' => 'info|pageprops', |
|
238
|
|
|
'inprop' => 'protection|talkid|watched|watchers|notificationtimestamp|subjectid|url|readable', |
|
239
|
|
|
'converttitles' => '', |
|
240
|
|
|
// 'ellimit' => 20, |
|
|
|
|
|
|
241
|
|
|
// 'elexpandurl' => '', |
|
|
|
|
|
|
242
|
|
|
'titles' => $page, |
|
243
|
|
|
'formatversion' => 2 |
|
244
|
|
|
// 'pageids' => $pageIds // FIXME: allow page IDs |
|
|
|
|
|
|
245
|
|
|
]; |
|
246
|
|
|
|
|
247
|
|
|
if ($followRedir) { |
|
248
|
|
|
$params['redirects'] = ''; |
|
249
|
|
|
} |
|
250
|
|
|
|
|
251
|
|
|
$query = new SimpleRequest('query', $params); |
|
252
|
|
|
$result = []; |
|
253
|
|
|
|
|
254
|
|
|
try { |
|
255
|
|
|
$res = $this->api->getRequest($query); |
|
256
|
|
|
if (isset($res['query']['pages'])) { |
|
257
|
|
|
$result = $res['query']['pages'][0]; |
|
258
|
|
|
} |
|
259
|
|
|
} catch (Exception $e) { |
|
260
|
|
|
// The api returned an error! Ignore |
|
261
|
|
|
} |
|
262
|
|
|
|
|
263
|
|
|
return $result; |
|
264
|
|
|
} |
|
265
|
|
|
|
|
266
|
|
|
/** |
|
267
|
|
|
* Get HTML display titles of a set of pages (or the normal title if there's no display title). |
|
268
|
|
|
* This will send t/50 API requests where t is the number of titles supplied. |
|
269
|
|
|
* @param string $project The project. |
|
270
|
|
|
* @param string[] $pageTitles The titles to fetch. |
|
271
|
|
|
* @return string[] Keys are the original supplied title, and values are the display titles. |
|
272
|
|
|
*/ |
|
273
|
|
|
public function displayTitles($project, $pageTitles) |
|
274
|
|
|
{ |
|
275
|
|
|
$this->setUp($project); |
|
276
|
|
|
$displayTitles = []; |
|
277
|
|
|
for ($n = 0; $n < count($pageTitles); $n += 50) { |
|
|
|
|
|
|
278
|
|
|
$titleSlice = array_slice($pageTitles, $n, 50); |
|
279
|
|
|
$params = [ |
|
280
|
|
|
'prop' => 'info|pageprops', |
|
281
|
|
|
'inprop' => 'displaytitle', |
|
282
|
|
|
'titles' => join('|', $titleSlice), |
|
283
|
|
|
]; |
|
284
|
|
|
$query = new SimpleRequest('query', $params); |
|
285
|
|
|
$result = $this->api->getRequest($query); |
|
286
|
|
|
|
|
287
|
|
|
// Extract normalization info. |
|
288
|
|
|
$normalized = []; |
|
289
|
|
|
if (isset($result['query']['normalized'])) { |
|
290
|
|
|
array_map( |
|
291
|
|
|
function ($e) use (&$normalized) { |
|
292
|
|
|
$normalized[$e['to']] = $e['from']; |
|
293
|
|
|
}, |
|
294
|
|
|
$result['query']['normalized'] |
|
295
|
|
|
); |
|
296
|
|
|
} |
|
297
|
|
|
|
|
298
|
|
|
// Match up the normalized titles with the display titles and the original titles. |
|
299
|
|
|
foreach ($result['query']['pages'] as $pageInfo) { |
|
300
|
|
|
$displayTitle = isset($pageInfo['pageprops']['displaytitle']) |
|
301
|
|
|
? $pageInfo['pageprops']['displaytitle'] |
|
302
|
|
|
: $pageInfo['title']; |
|
303
|
|
|
$origTitle = isset($normalized[$pageInfo['title']]) |
|
304
|
|
|
? $normalized[$pageInfo['title']] : $pageInfo['title']; |
|
305
|
|
|
$displayTitles[$origTitle] = $displayTitle; |
|
306
|
|
|
} |
|
307
|
|
|
} |
|
308
|
|
|
|
|
309
|
|
|
return $displayTitles; |
|
310
|
|
|
} |
|
311
|
|
|
|
|
312
|
|
|
/** |
|
313
|
|
|
* Get assessments of the given pages, if a supported project |
|
314
|
|
|
* @param string $project Project such as en.wikipedia.org |
|
315
|
|
|
* @param string|array $pageTitles Single page title or array of titles |
|
316
|
|
|
* @return array|null Page assessments info or null if none found |
|
317
|
|
|
*/ |
|
318
|
|
|
public function getPageAssessments($project, $pageTitles) |
|
319
|
|
|
{ |
|
320
|
|
|
// From config/assessments.yml |
|
321
|
|
|
$config = $this->getAssessmentsConfig(); |
|
322
|
|
|
|
|
323
|
|
|
// return null if unsupported project |
|
324
|
|
|
if (!in_array($project, array_keys($config))) { |
|
325
|
|
|
return null; |
|
326
|
|
|
} |
|
327
|
|
|
|
|
328
|
|
|
$config = $config[$project]; |
|
329
|
|
|
|
|
330
|
|
|
$params = [ |
|
331
|
|
|
'prop' => 'pageassessments', |
|
332
|
|
|
'titles' => is_string($pageTitles) ? $pageTitles : implode('|', $pageTitles), |
|
333
|
|
|
'palimit' => 500, |
|
334
|
|
|
]; |
|
335
|
|
|
|
|
336
|
|
|
// get assessments for this page from the API |
|
337
|
|
|
$assessments = $this->massApi($params, $project, function ($data) { |
|
|
|
|
|
|
338
|
|
|
return isset($data['pages'][0]['pageassessments']) ? $data['pages'][0]['pageassessments'] : []; |
|
339
|
|
|
}, 'pacontinue')['pages']; |
|
340
|
|
|
|
|
341
|
|
|
$decoratedAssessments = []; |
|
342
|
|
|
|
|
343
|
|
|
// Set the default decorations for the overall quality assessment |
|
344
|
|
|
// This will be replaced with the first valid class defined for any WikiProject |
|
345
|
|
|
$overallQuality = $config['class']['Unknown']; |
|
346
|
|
|
$overallQuality['value'] = '???'; |
|
347
|
|
|
|
|
348
|
|
|
if (empty($assessments)) { |
|
349
|
|
|
return null; |
|
350
|
|
|
} |
|
351
|
|
|
|
|
352
|
|
|
// loop through each assessment and decorate with colors, category URLs and images, if applicable |
|
353
|
|
|
foreach ($assessments as $wikiproject => $assessment) { |
|
354
|
|
|
$classValue = $assessment['class']; |
|
355
|
|
|
|
|
356
|
|
|
// Use ??? as the presented value when the class is unknown or is not defined in the config |
|
357
|
|
|
if ($classValue === 'Unknown' || $classValue === '' || !isset($config['class'][$classValue])) { |
|
358
|
|
|
$classAttrs = $config['class']['Unknown']; |
|
359
|
|
|
$assessment['class']['value'] = '???'; |
|
360
|
|
|
$assessment['class']['category'] = $classAttrs['category']; |
|
361
|
|
|
$assessment['class']['badge'] = "https://upload.wikimedia.org/wikipedia/commons/". $classAttrs['badge']; |
|
362
|
|
|
} else { |
|
363
|
|
|
$classAttrs = $config['class'][$classValue]; |
|
364
|
|
|
$assessment['class'] = [ |
|
365
|
|
|
'value' => $classValue, |
|
366
|
|
|
'color' => $classAttrs['color'], |
|
367
|
|
|
'category' => $classAttrs['category'], |
|
368
|
|
|
]; |
|
369
|
|
|
|
|
370
|
|
|
// add full URL to badge icon |
|
371
|
|
|
if ($classAttrs['badge'] !== '') { |
|
372
|
|
|
$assessment['class']['badge'] = "https://upload.wikimedia.org/wikipedia/commons/" . |
|
373
|
|
|
$classAttrs['badge']; |
|
374
|
|
|
} |
|
375
|
|
|
|
|
376
|
|
|
if ($overallQuality['value'] === '???') { |
|
377
|
|
|
$overallQuality = $assessment['class']; |
|
378
|
|
|
$overallQuality['category'] = $classAttrs['category']; |
|
379
|
|
|
} |
|
380
|
|
|
} |
|
381
|
|
|
|
|
382
|
|
|
$importanceValue = $assessment['importance']; |
|
383
|
|
|
$importanceUnknown = $importanceValue === 'Unknown' || $importanceValue === ''; |
|
384
|
|
|
|
|
385
|
|
|
if ($importanceUnknown || !isset($config['importance'][$importanceValue])) { |
|
386
|
|
|
$importanceAttrs = $config['importance']['Unknown']; |
|
387
|
|
|
$assessment['importance'] = $importanceAttrs; |
|
388
|
|
|
$assessment['importance']['value'] = '???'; |
|
389
|
|
|
$assessment['importance']['category'] = $importanceAttrs['category']; |
|
390
|
|
|
} else { |
|
391
|
|
|
$importanceAttrs = $config['importance'][$importanceValue]; |
|
392
|
|
|
$assessment['importance'] = [ |
|
393
|
|
|
'value' => $importanceValue, |
|
394
|
|
|
'color' => $importanceAttrs['color'], |
|
395
|
|
|
'weight' => $importanceAttrs['weight'], // numerical weight for sorting purposes |
|
396
|
|
|
'category' => $importanceAttrs['category'], |
|
397
|
|
|
]; |
|
398
|
|
|
} |
|
399
|
|
|
|
|
400
|
|
|
$decoratedAssessments[$wikiproject] = $assessment; |
|
401
|
|
|
} |
|
402
|
|
|
|
|
403
|
|
|
return [ |
|
404
|
|
|
'assessment' => $overallQuality, |
|
405
|
|
|
'wikiprojects' => $decoratedAssessments, |
|
406
|
|
|
'wikiproject_prefix' => $config['wikiproject_prefix'] |
|
407
|
|
|
]; |
|
408
|
|
|
} |
|
409
|
|
|
|
|
410
|
|
|
/** |
|
411
|
|
|
* Get the image URL of the badge for the given page assessment |
|
412
|
|
|
* @param string $project Project such as en.wikipedia.org |
|
413
|
|
|
* @param string $class Valid classification for project, such as 'Start', 'GA', etc. |
|
414
|
|
|
* @return string URL to image |
|
415
|
|
|
*/ |
|
416
|
|
|
public function getAssessmentBadgeURL($project, $class) |
|
417
|
|
|
{ |
|
418
|
|
|
$config = $this->getAssessmentsConfig(); |
|
419
|
|
|
|
|
420
|
|
|
if (isset($config[$project]['class'][$class])) { |
|
421
|
|
|
return "https://upload.wikimedia.org/wikipedia/commons/" . $config[$project]['class'][$class]['badge']; |
|
422
|
|
|
} elseif (isset($config[$project]['class']['Unknown'])) { |
|
423
|
|
|
return "https://upload.wikimedia.org/wikipedia/commons/" . $config[$project]['class']['Unknown']['badge']; |
|
424
|
|
|
} else { |
|
425
|
|
|
return ""; |
|
426
|
|
|
} |
|
427
|
|
|
} |
|
428
|
|
|
|
|
429
|
|
|
/** |
|
430
|
|
|
* Fetch assessments data from config/assessments.yml and cache in static variable |
|
431
|
|
|
* @return array Mappings of project/quality/class with badges, colors and category links |
|
432
|
|
|
*/ |
|
433
|
|
|
private function getAssessmentsConfig() |
|
434
|
|
|
{ |
|
435
|
|
|
static $assessmentsConfig = null; |
|
436
|
|
|
if ($assessmentsConfig === null) { |
|
437
|
|
|
$assessmentsConfig = $this->container->getParameter('assessments'); |
|
438
|
|
|
} |
|
439
|
|
|
return $assessmentsConfig; |
|
440
|
|
|
} |
|
441
|
|
|
|
|
442
|
|
|
/** |
|
443
|
|
|
* Does the given project support page assessments? |
|
444
|
|
|
* @param string $project Project to query, e.g. en.wikipedia.org |
|
445
|
|
|
* @return boolean True or false |
|
446
|
|
|
*/ |
|
447
|
|
|
public function projectHasPageAssessments($project) |
|
448
|
|
|
{ |
|
449
|
|
|
return in_array($project, array_keys($this->getAssessmentsConfig())); |
|
450
|
|
|
} |
|
451
|
|
|
|
|
452
|
|
|
/** |
|
453
|
|
|
* Make mass API requests to MediaWiki API |
|
454
|
|
|
* The API normally limits to 500 pages, but gives you a 'continue' value |
|
455
|
|
|
* to finish iterating through the resource. |
|
456
|
|
|
* Adapted from https://github.com/MusikAnimal/pageviews |
|
457
|
|
|
* @param array $params Associative array of params to pass to API |
|
458
|
|
|
* @param string $project Project to query, e.g. en.wikipedia.org |
|
459
|
|
|
* @param string|func $dataKey The key for the main chunk of data, in the query hash |
|
460
|
|
|
* (e.g. 'categorymembers' for API:Categorymembers). |
|
461
|
|
|
* If this is a function it is given the response data, |
|
462
|
|
|
* and expected to return the data we want to concatentate. |
|
463
|
|
|
* @param string [$continueKey] the key to look in the continue hash, if present |
|
464
|
|
|
* (e.g. 'cmcontinue' for API:Categorymembers) |
|
465
|
|
|
* @param integer [$limit] Max number of pages to fetch |
|
466
|
|
|
* @return array Associative array with data |
|
467
|
|
|
*/ |
|
468
|
|
|
public function massApi($params, $project, $dataKey, $continueKey = 'continue', $limit = 5000) |
|
469
|
|
|
{ |
|
470
|
|
|
$this->setUp($project); |
|
471
|
|
|
|
|
472
|
|
|
// Passed by reference to massApiInternal so we can keep track of |
|
473
|
|
|
// everything we need during the recursive calls |
|
474
|
|
|
// The magically essential part here is $data['promise'] which we'll |
|
475
|
|
|
// wait to be resolved |
|
476
|
|
|
$data = [ |
|
477
|
|
|
'params' => $params, |
|
478
|
|
|
'project' => $project, |
|
479
|
|
|
'continueKey' => $continueKey, |
|
480
|
|
|
'dataKey' => $dataKey, |
|
481
|
|
|
'limit' => $limit, |
|
482
|
|
|
'resolveData' => [ |
|
483
|
|
|
'pages' => [] |
|
484
|
|
|
], |
|
485
|
|
|
'continueValue' => null, |
|
486
|
|
|
'promise' => new \GuzzleHttp\Promise\Promise(), |
|
487
|
|
|
]; |
|
488
|
|
|
|
|
489
|
|
|
// wait for all promises to complete, even if some of them fail |
|
490
|
|
|
\GuzzleHttp\Promise\settle($this->massApiInternal($data))->wait(); |
|
491
|
|
|
|
|
492
|
|
|
return $data['resolveData']; |
|
493
|
|
|
} |
|
494
|
|
|
|
|
495
|
|
|
/** |
|
496
|
|
|
* Internal function used by massApi() to make recursive calls |
|
497
|
|
|
* @param array &$data Everything we need to keep track of, as defined in massApi() |
|
498
|
|
|
* @return null Nothing. $data['promise']->then is used to continue flow of |
|
499
|
|
|
* execution after all recursive calls are complete |
|
500
|
|
|
*/ |
|
501
|
|
|
private function massApiInternal(&$data) |
|
502
|
|
|
{ |
|
503
|
|
|
$requestData = array_merge([ |
|
504
|
|
|
'action' => 'query', |
|
505
|
|
|
'format' => 'json', |
|
506
|
|
|
'formatversion' => '2', |
|
507
|
|
|
], $data['params']); |
|
508
|
|
|
|
|
509
|
|
|
if ($data['continueValue']) { |
|
510
|
|
|
$requestData[$data['continueKey']] = $data['continueValue']; |
|
511
|
|
|
} |
|
512
|
|
|
|
|
513
|
|
|
$query = FluentRequest::factory()->setAction('query')->setParams($requestData); |
|
514
|
|
|
$innerPromise = $this->api->getRequestAsync($query); |
|
515
|
|
|
|
|
516
|
|
|
$innerPromise->then(function ($result) use (&$data) { |
|
517
|
|
|
// some failures come back as 200s, so we still resolve and let the outer function handle it |
|
518
|
|
|
if (isset($result['error']) || !isset($result['query'])) { |
|
519
|
|
|
return $data['promise']->resolve($data); |
|
520
|
|
|
} |
|
521
|
|
|
|
|
522
|
|
|
$dataKey = $data['dataKey']; |
|
523
|
|
|
$isFinished = false; |
|
|
|
|
|
|
524
|
|
|
|
|
525
|
|
|
// allow custom function to parse the data we want, if provided |
|
526
|
|
|
if (is_callable($dataKey)) { |
|
527
|
|
|
$data['resolveData']['pages'] = array_merge( |
|
528
|
|
|
$data['resolveData']['pages'], |
|
529
|
|
|
$data['dataKey']($result['query']) |
|
530
|
|
|
); |
|
531
|
|
|
$isFinished = count($data['resolveData']['pages']) >= $data['limit']; |
|
532
|
|
|
} else { |
|
533
|
|
|
// append new data to data from last request. We might want both 'pages' and dataKey |
|
534
|
|
|
if (isset($result['query']['pages'])) { |
|
535
|
|
|
$data['resolveData']['pages'] = array_merge( |
|
536
|
|
|
$data['resolveData']['pages'], |
|
537
|
|
|
$result['query']['pages'] |
|
538
|
|
|
); |
|
539
|
|
|
} |
|
540
|
|
|
if ($result['query'][$dataKey]) { |
|
541
|
|
|
$newValues = isset($data['resolveData'][$dataKey]) ? $data['resolveData'][$dataKey] : []; |
|
542
|
|
|
$data['resolveData'][$dataKey] = array_merge($newValues, $result['query'][$dataKey]); |
|
543
|
|
|
} |
|
544
|
|
|
|
|
545
|
|
|
// If pages is not the collection we want, it will be either an empty array or one entry with |
|
546
|
|
|
// basic page info depending on what API we're hitting. So resolveData[dataKey] will hit the limit |
|
547
|
|
|
$isFinished = count($data['resolveData']['pages']) >= $data['limit'] || |
|
548
|
|
|
count($data['resolveData'][$dataKey]) >= $data['limit']; |
|
549
|
|
|
} |
|
550
|
|
|
|
|
551
|
|
|
// make recursive call if needed, waiting 100ms |
|
552
|
|
|
if (!$isFinished && isset($result['continue']) && isset($result['continue'][$data['continueKey']])) { |
|
553
|
|
|
usleep(100000); |
|
554
|
|
|
$data['continueValue'] = $result['continue'][$data['continueKey']]; |
|
555
|
|
|
return $this->massApiInternal($data); |
|
556
|
|
|
} else { |
|
557
|
|
|
// indicate there were more entries than the limit |
|
558
|
|
|
if ($result['continue']) { |
|
559
|
|
|
$data['resolveData']['continue'] = true; |
|
560
|
|
|
} |
|
561
|
|
|
$data['promise']->resolve($data); |
|
562
|
|
|
} |
|
563
|
|
|
}); |
|
564
|
|
|
} |
|
565
|
|
|
} |
|
566
|
|
|
|
This check looks for parameters that are defined as one type in their type hint or doc comment but seem to be used as a narrower type, i.e an implementation of an interface or a subclass.
Consider changing the type of the parameter or doing an instanceof check before assuming your parameter is of the expected type.