|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* This file contains only the ArticleInfoController class. |
|
4
|
|
|
*/ |
|
5
|
|
|
|
|
6
|
|
|
namespace AppBundle\Controller; |
|
7
|
|
|
|
|
8
|
|
|
use AppBundle\Helper\AutomatedEditsHelper; |
|
9
|
|
|
use AppBundle\Helper\LabsHelper; |
|
10
|
|
|
use AppBundle\Helper\PageviewsHelper; |
|
11
|
|
|
use Doctrine\DBAL\Connection; |
|
12
|
|
|
use Sensio\Bundle\FrameworkExtraBundle\Configuration\Route; |
|
13
|
|
|
use Symfony\Bundle\FrameworkBundle\Controller\Controller; |
|
14
|
|
|
use Symfony\Component\HttpFoundation\Request; |
|
15
|
|
|
use Symfony\Component\DependencyInjection\ContainerInterface; |
|
16
|
|
|
use Symfony\Component\HttpFoundation\Response; |
|
17
|
|
|
use Xtools\ProjectRepository; |
|
18
|
|
|
use Xtools\Page; |
|
19
|
|
|
use Xtools\PagesRepository; |
|
20
|
|
|
use Xtools\Edit; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* This controller serves the search form and results for the ArticleInfo tool |
|
24
|
|
|
*/ |
|
25
|
|
|
class ArticleInfoController extends Controller |
|
26
|
|
|
{ |
|
27
|
|
|
/** @var LabsHelper The Labs helper object. */ |
|
28
|
|
|
private $lh; |
|
29
|
|
|
/** @var mixed[] Information about the page in question. */ |
|
30
|
|
|
private $pageInfo; |
|
31
|
|
|
/** @var Edit[] All edits of the page. */ |
|
32
|
|
|
private $pageHistory; |
|
33
|
|
|
/** @var string The fully-qualified name of the revision table. */ |
|
34
|
|
|
private $revisionTable; |
|
35
|
|
|
/** @var Connection The projects' database connection. */ |
|
36
|
|
|
protected $conn; |
|
37
|
|
|
/** @var AutomatedEditsHelper The semi-automated edits helper. */ |
|
38
|
|
|
protected $aeh; |
|
39
|
|
|
/** @var PageviewsHelper The page-views helper. */ |
|
40
|
|
|
protected $ph; |
|
41
|
|
|
|
|
42
|
|
|
/** |
|
43
|
|
|
* Get the tool's shortname. |
|
44
|
|
|
* @return string |
|
45
|
|
|
*/ |
|
46
|
|
|
public function getToolShortname() |
|
47
|
|
|
{ |
|
48
|
|
|
return 'articleinfo'; |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
|
|
* Override method to call ArticleInfoController::containerInitialized() when container set. |
|
53
|
|
|
* @param ContainerInterface|null $container A ContainerInterface instance or null |
|
54
|
|
|
*/ |
|
55
|
|
|
public function setContainer(ContainerInterface $container = null) |
|
56
|
|
|
{ |
|
57
|
|
|
parent::setContainer($container); |
|
58
|
|
|
$this->containerInitialized(); |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
/** |
|
62
|
|
|
* Perform some operations after controller initialized and container set. |
|
63
|
|
|
*/ |
|
64
|
|
|
private function containerInitialized() |
|
65
|
|
|
{ |
|
66
|
|
|
$this->lh = $this->get('app.labs_helper'); |
|
67
|
|
|
$this->conn = $this->getDoctrine()->getManager('replicas')->getConnection(); |
|
|
|
|
|
|
68
|
|
|
$this->ph = $this->get('app.pageviews_helper'); |
|
69
|
|
|
$this->aeh = $this->get('app.automated_edits_helper'); |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
|
|
/** |
|
73
|
|
|
* The search form. |
|
74
|
|
|
* @Route("/articleinfo", name="articleinfo") |
|
75
|
|
|
* @Route("/articleinfo", name="articleInfo") |
|
76
|
|
|
* @Route("/articleinfo/", name="articleInfoSlash") |
|
77
|
|
|
* @Route("/articleinfo/index.php", name="articleInfoIndexPhp") |
|
78
|
|
|
* @Route("/articleinfo/{project}", name="ArticleInfoProject") |
|
79
|
|
|
* @param Request $request The HTTP request. |
|
80
|
|
|
* @return Response |
|
81
|
|
|
*/ |
|
82
|
|
|
public function indexAction(Request $request) |
|
83
|
|
|
{ |
|
84
|
|
|
$projectQuery = $request->query->get('project'); |
|
85
|
|
|
$article = $request->query->get('article'); |
|
86
|
|
|
|
|
87
|
|
|
if ($projectQuery != '' && $article != '') { |
|
88
|
|
|
return $this->redirectToRoute('ArticleInfoResult', [ 'project'=>$projectQuery, 'article' => $article ]); |
|
89
|
|
|
} elseif ($article != '') { |
|
90
|
|
|
return $this->redirectToRoute('ArticleInfoProject', [ 'project'=>$projectQuery ]); |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
|
|
if ($projectQuery == '') { |
|
94
|
|
|
$projectQuery = $this->container->getParameter('default_project'); |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
$project = ProjectRepository::getProject($projectQuery, $this->container); |
|
98
|
|
|
|
|
99
|
|
|
return $this->render('articleInfo/index.html.twig', [ |
|
100
|
|
|
'xtPage' => 'articleinfo', |
|
101
|
|
|
'xtPageTitle' => 'tool-articleinfo', |
|
102
|
|
|
'xtSubtitle' => 'tool-articleinfo-desc', |
|
103
|
|
|
'project' => $project, |
|
104
|
|
|
]); |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
|
|
/** |
|
108
|
|
|
* Display the results. |
|
109
|
|
|
* @Route("/articleinfo/{project}/{article}", name="ArticleInfoResult", requirements={"article"=".+"}) |
|
110
|
|
|
* @param Request $request The HTTP request. |
|
111
|
|
|
* @return Response |
|
112
|
|
|
*/ |
|
113
|
|
|
public function resultAction(Request $request) |
|
114
|
|
|
{ |
|
115
|
|
|
$projectQuery = $request->attributes->get('project'); |
|
116
|
|
|
$project = ProjectRepository::getProject($projectQuery, $this->container); |
|
117
|
|
View Code Duplication |
if (!$project->exists()) { |
|
|
|
|
|
|
118
|
|
|
$this->addFlash('notice', ['invalid-project', $projectQuery]); |
|
|
|
|
|
|
119
|
|
|
return $this->redirectToRoute('articleInfo'); |
|
120
|
|
|
} |
|
121
|
|
|
$projectUrl = $project->getUrl(); |
|
122
|
|
|
$dbName = $project->getDatabaseName(); |
|
123
|
|
|
|
|
124
|
|
|
$pageQuery = $request->attributes->get('article'); |
|
125
|
|
|
$page = new Page($project, $pageQuery); |
|
126
|
|
|
$pageRepo = new PagesRepository(); |
|
127
|
|
|
$pageRepo->setContainer($this->container); |
|
|
|
|
|
|
128
|
|
|
$page->setRepository($pageRepo); |
|
129
|
|
|
|
|
130
|
|
|
if (!$page->exists()) { |
|
131
|
|
|
$this->addFlash('notice', ['no-exist', $pageQuery]); |
|
|
|
|
|
|
132
|
|
|
return $this->redirectToRoute('articleInfo'); |
|
133
|
|
|
} |
|
134
|
|
|
|
|
135
|
|
|
$this->revisionTable = $project->getRepository()->getTableName( |
|
136
|
|
|
$project->getDatabaseName(), |
|
137
|
|
|
'revision' |
|
138
|
|
|
); |
|
139
|
|
|
|
|
140
|
|
|
// TODO: throw error if $basicInfo['missing'] is set |
|
|
|
|
|
|
141
|
|
|
|
|
142
|
|
|
$this->pageInfo = [ |
|
|
|
|
|
|
143
|
|
|
'project' => $project, |
|
144
|
|
|
'projectUrl' => $projectUrl, |
|
145
|
|
|
'page' => $page, |
|
146
|
|
|
'dbName' => $dbName, |
|
147
|
|
|
'lang' => $project->getLang(), |
|
148
|
|
|
]; |
|
149
|
|
|
|
|
150
|
|
|
if ($page->getWikidataId()) { |
|
151
|
|
|
$this->pageInfo['numWikidataItems'] = $this->getNumWikidataItems(); |
|
152
|
|
|
} |
|
153
|
|
|
|
|
154
|
|
|
// TODO: Adapted from legacy code; may be used to indicate how many dead ext links there are |
|
155
|
|
|
// if ( isset( $basicInfo->extlinks ) ){ |
|
|
|
|
|
|
156
|
|
|
// foreach ( $basicInfo->extlinks as $i => $link ){ |
|
|
|
|
|
|
157
|
|
|
// $this->extLinks[] = array("link" => $link->{'*'}, "status" => "unchecked" ); |
|
|
|
|
|
|
158
|
|
|
// } |
|
159
|
|
|
// } |
|
160
|
|
|
|
|
161
|
|
|
$this->pageHistory = $page->getRevisions(); |
|
162
|
|
|
$this->pageInfo['firstEdit'] = new Edit($this->pageInfo['page'], $this->pageHistory[0]); |
|
163
|
|
|
$this->pageInfo['lastEdit'] = new Edit( |
|
164
|
|
|
$this->pageInfo['page'], |
|
165
|
|
|
$this->pageHistory[$page->getNumRevisions() - 1] |
|
166
|
|
|
); |
|
167
|
|
|
|
|
168
|
|
|
// NOTE: bots are fetched first in case we want to restrict some stats to humans editors only |
|
169
|
|
|
$this->pageInfo['bots'] = $this->getBotData(); |
|
170
|
|
|
$this->pageInfo['general']['bot_count'] = count($this->pageInfo['bots']); |
|
171
|
|
|
|
|
172
|
|
|
$this->pageInfo = array_merge($this->pageInfo, $this->parseHistory()); |
|
|
|
|
|
|
173
|
|
|
$this->pageInfo['general']['top_ten_count'] = $this->getTopTenCount(); |
|
174
|
|
|
$this->pageInfo['general']['top_ten_percentage'] = round( |
|
175
|
|
|
($this->pageInfo['general']['top_ten_count'] / $page->getNumRevisions()) * 100, |
|
176
|
|
|
1 |
|
177
|
|
|
); |
|
178
|
|
|
$this->pageInfo = array_merge($this->pageInfo, $this->getLinksAndRedirects()); |
|
|
|
|
|
|
179
|
|
|
$this->pageInfo['general']['pageviews_offset'] = 60; |
|
180
|
|
|
$this->pageInfo['general']['pageviews'] = $this->ph->sumLastDays( |
|
181
|
|
|
$this->pageInfo['project']->getDomain(), |
|
182
|
|
|
$this->pageInfo['page']->getTitle(), |
|
183
|
|
|
$this->pageInfo['general']['pageviews_offset'] |
|
184
|
|
|
); |
|
185
|
|
|
$api = $this->get('app.api_helper'); |
|
186
|
|
|
$assessments = $api->getPageAssessments($projectQuery, $pageQuery); |
|
187
|
|
|
if ($assessments) { |
|
188
|
|
|
$this->pageInfo['assessments'] = $assessments; |
|
189
|
|
|
} |
|
190
|
|
|
$this->setLogsEvents(); |
|
191
|
|
|
|
|
192
|
|
|
$bugs = array_merge($this->getCheckWikiErrors(), $this->getWikidataErrors()); |
|
193
|
|
|
if (!empty($bugs)) { |
|
194
|
|
|
$this->pageInfo['bugs'] = $bugs; |
|
195
|
|
|
} |
|
196
|
|
|
|
|
197
|
|
|
$this->pageInfo['xtPage'] = 'articleinfo'; |
|
198
|
|
|
$this->pageInfo['xtTitle'] = $this->pageInfo['page']->getTitle(); |
|
199
|
|
|
|
|
200
|
|
|
return $this->render("articleInfo/result.html.twig", $this->pageInfo); |
|
201
|
|
|
} |
|
202
|
|
|
|
|
203
|
|
|
/** |
|
204
|
|
|
* Get number of wikidata items (not just languages of sister projects) |
|
205
|
|
|
* @return integer Number of items |
|
206
|
|
|
*/ |
|
207
|
|
|
private function getNumWikidataItems() |
|
208
|
|
|
{ |
|
209
|
|
|
$query = "SELECT COUNT(*) AS count |
|
210
|
|
|
FROM wikidatawiki_p.wb_items_per_site |
|
211
|
|
|
WHERE ips_item_id = ". ltrim($this->pageInfo['page']->getWikidataId(), 'Q'); |
|
212
|
|
|
$res = $this->conn->query($query)->fetchAll(); |
|
213
|
|
|
return $res[0]['count']; |
|
214
|
|
|
} |
|
215
|
|
|
|
|
216
|
|
|
/** |
|
217
|
|
|
* Get info about bots that edited the page |
|
218
|
|
|
* This also sets $this->pageInfo['bot_revision_count'] and $this->pageInfo['bot_percentage'] |
|
219
|
|
|
* @return array Associative array containing the bot's username, edit count to the page |
|
220
|
|
|
* and whether or not they are currently a bot |
|
221
|
|
|
*/ |
|
222
|
|
|
private function getBotData() |
|
223
|
|
|
{ |
|
224
|
|
|
$userGroupsTable = $this->lh->getTable('user_groups', $this->pageInfo['dbName']); |
|
225
|
|
|
$userFromerGroupsTable = $this->lh->getTable('user_former_groups', $this->pageInfo['dbName']); |
|
226
|
|
|
$query = "SELECT COUNT(rev_user_text) AS count, rev_user_text AS username, ug_group AS current |
|
227
|
|
|
FROM $this->revisionTable |
|
228
|
|
|
LEFT JOIN $userGroupsTable ON rev_user = ug_user |
|
229
|
|
|
LEFT JOIN $userFromerGroupsTable ON rev_user = ufg_user |
|
230
|
|
|
WHERE rev_page = " . $this->pageInfo['page']->getId() . " AND (ug_group = 'bot' OR ufg_group = 'bot') |
|
231
|
|
|
GROUP BY rev_user_text"; |
|
232
|
|
|
$res = $this->conn->query($query)->fetchAll(); |
|
233
|
|
|
|
|
234
|
|
|
// Parse the botedits |
|
235
|
|
|
$bots = []; |
|
236
|
|
|
$sum = 0; |
|
237
|
|
|
foreach ($res as $bot) { |
|
238
|
|
|
$bots[$bot['username']] = [ |
|
239
|
|
|
'count' => (int) $bot['count'], |
|
240
|
|
|
'current' => $bot['current'] === 'bot' |
|
241
|
|
|
]; |
|
242
|
|
|
$sum += $bot['count']; |
|
243
|
|
|
} |
|
244
|
|
|
|
|
245
|
|
|
uasort($bots, function ($a, $b) { |
|
246
|
|
|
return $b['count'] - $a['count']; |
|
247
|
|
|
}); |
|
248
|
|
|
|
|
249
|
|
|
$this->pageInfo['general']['bot_revision_count'] = $sum; |
|
250
|
|
|
$this->pageInfo['general']['bot_percentage'] = round( |
|
251
|
|
|
($sum / $this->pageInfo['page']->getNumRevisions()) * 100, |
|
252
|
|
|
1 |
|
253
|
|
|
); |
|
254
|
|
|
|
|
255
|
|
|
return $bots; |
|
256
|
|
|
} |
|
257
|
|
|
|
|
258
|
|
|
/** |
|
259
|
|
|
* Get the number of edits made to the page by the top 10% of editors |
|
260
|
|
|
* This is ran *after* parseHistory() since we need the grand totals first. |
|
261
|
|
|
* Various stats are also set for each editor in $this->pageInfo['editors'] |
|
262
|
|
|
* and top ten editors are stored in $this->pageInfo['general']['top_ten'] |
|
263
|
|
|
* to be used in the charts |
|
264
|
|
|
* @return integer Number of edits |
|
265
|
|
|
*/ |
|
266
|
|
|
private function getTopTenCount() |
|
267
|
|
|
{ |
|
268
|
|
|
$topTenCount = $counter = 0; |
|
269
|
|
|
$topTenEditors = []; |
|
270
|
|
|
|
|
271
|
|
|
foreach ($this->pageInfo['editors'] as $editor => $info) { |
|
272
|
|
|
// Count how many users are in the top 10% by number of edits |
|
273
|
|
|
if ($counter < 10) { |
|
274
|
|
|
$topTenCount += $info['all']; |
|
275
|
|
|
$counter++; |
|
276
|
|
|
|
|
277
|
|
|
// To be used in the Top Ten charts |
|
278
|
|
|
$topTenEditors[] = [ |
|
279
|
|
|
'label' => $editor, |
|
280
|
|
|
'value' => $info['all'], |
|
281
|
|
|
'percentage' => ( |
|
282
|
|
|
100 * ($info['all'] / $this->pageInfo['page']->getNumRevisions()) |
|
283
|
|
|
) |
|
284
|
|
|
]; |
|
285
|
|
|
} |
|
286
|
|
|
|
|
287
|
|
|
// Compute the percentage of minor edits the user made |
|
288
|
|
|
$this->pageInfo['editors'][$editor]['minor_percentage'] = $info['all'] |
|
289
|
|
|
? ($info['minor'] / $info['all']) * 100 |
|
290
|
|
|
: 0; |
|
291
|
|
|
|
|
292
|
|
|
if ($info['all'] > 1) { |
|
293
|
|
|
// Number of seconds between first and last edit |
|
294
|
|
|
$secs = intval(strtotime($info['last']) - strtotime($info['first']) / $info['all']); |
|
295
|
|
|
|
|
296
|
|
|
// Average time between edits (in days) |
|
297
|
|
|
$this->pageInfo['editors'][$editor]['atbe'] = $secs / ( 60 * 60 * 24 ); |
|
298
|
|
|
} |
|
299
|
|
|
|
|
300
|
|
|
if (count($info['sizes'])) { |
|
301
|
|
|
// Average Total KB divided by number of stored sizes (user's edit count to this page) |
|
302
|
|
|
$this->pageInfo['editors'][$editor]['size'] = array_sum($info['sizes']) / count($info['sizes']); |
|
303
|
|
|
} else { |
|
304
|
|
|
$this->pageInfo['editors'][$editor]['size'] = 0; |
|
305
|
|
|
} |
|
306
|
|
|
} |
|
307
|
|
|
|
|
308
|
|
|
$this->pageInfo['topTenEditors'] = $topTenEditors; |
|
309
|
|
|
|
|
310
|
|
|
// First sort editors array by the amount of text they added |
|
311
|
|
|
$topTenEditorsByAdded = $this->pageInfo['editors']; |
|
312
|
|
|
uasort($topTenEditorsByAdded, function ($a, $b) { |
|
313
|
|
|
if ($a['added'] === $b['added']) { |
|
314
|
|
|
return 0; |
|
315
|
|
|
} |
|
316
|
|
|
return $a['added'] > $b['added'] ? -1 : 1; |
|
317
|
|
|
}); |
|
318
|
|
|
|
|
319
|
|
|
// Then build a new array of top 10 editors by added text, |
|
320
|
|
|
// in the data structure needed for the chart |
|
321
|
|
|
$this->pageInfo['topTenEditorsByAdded'] = array_map(function ($editor) { |
|
322
|
|
|
$added = $this->pageInfo['editors'][$editor]['added']; |
|
323
|
|
|
return [ |
|
324
|
|
|
'label' => $editor, |
|
325
|
|
|
'value' => $added, |
|
326
|
|
|
'percentage' => ( |
|
327
|
|
|
100 * ($added / $this->pageInfo['general']['added']) |
|
328
|
|
|
) |
|
329
|
|
|
]; |
|
330
|
|
|
}, array_keys(array_slice($topTenEditorsByAdded, 0, 10))); |
|
331
|
|
|
|
|
332
|
|
|
return $topTenCount; |
|
333
|
|
|
} |
|
334
|
|
|
|
|
335
|
|
|
/** |
|
336
|
|
|
* Get number of in and outgoing links and redirects to the page |
|
337
|
|
|
* @return array Associative array containing counts |
|
338
|
|
|
*/ |
|
339
|
|
|
private function getLinksAndRedirects() |
|
340
|
|
|
{ |
|
341
|
|
|
$pageId = $this->pageInfo['page']->getId(); |
|
342
|
|
|
$namespace = $this->pageInfo['page']->getNamespace(); |
|
343
|
|
|
$title = str_replace(' ', '_', $this->pageInfo['page']->getTitle()); |
|
344
|
|
|
$externalLinksTable = $this->lh->getTable('externallinks', $this->pageInfo['dbName']); |
|
345
|
|
|
$pageLinksTable = $this->lh->getTable('pagelinks', $this->pageInfo['dbName']); |
|
346
|
|
|
$redirectTable = $this->lh->getTable('redirect', $this->pageInfo['dbName']); |
|
347
|
|
|
|
|
348
|
|
|
// FIXME: Probably need to make the $title mysql-safe or whatever |
|
349
|
|
|
$query = "SELECT COUNT(*) AS value, 'links_ext' AS type |
|
350
|
|
|
FROM $externalLinksTable WHERE el_from = $pageId |
|
351
|
|
|
UNION |
|
352
|
|
|
SELECT COUNT(*) AS value, 'links_out' AS type |
|
353
|
|
|
FROM $pageLinksTable WHERE pl_from = $pageId |
|
354
|
|
|
UNION |
|
355
|
|
|
SELECT COUNT(*) AS value, 'links_in' AS type |
|
356
|
|
|
FROM $pageLinksTable WHERE pl_namespace = $namespace AND pl_title = \"$title\" |
|
357
|
|
|
UNION |
|
358
|
|
|
SELECT COUNT(*) AS value, 'redirects' AS type |
|
359
|
|
|
FROM $redirectTable WHERE rd_namespace = $namespace AND rd_title = \"$title\""; |
|
360
|
|
|
|
|
361
|
|
|
$res = $this->conn->query($query)->fetchAll(); |
|
362
|
|
|
|
|
363
|
|
|
$data = []; |
|
364
|
|
|
|
|
365
|
|
|
// Transform to associative array by 'type' |
|
366
|
|
|
foreach ($res as $row) { |
|
367
|
|
|
$data[$row['type'] . '_count'] = $row['value']; |
|
368
|
|
|
} |
|
369
|
|
|
|
|
370
|
|
|
return $data; |
|
371
|
|
|
} |
|
372
|
|
|
|
|
373
|
|
|
/** |
|
374
|
|
|
* Query for log events during each year of the article's history, |
|
375
|
|
|
* and set the results in $this->pageInfo['year_count'] |
|
376
|
|
|
*/ |
|
377
|
|
|
private function setLogsEvents() |
|
378
|
|
|
{ |
|
379
|
|
|
$loggingTable = $this->lh->getTable('logging', $this->pageInfo['dbName'], 'logindex'); |
|
380
|
|
|
$title = str_replace(' ', '_', $this->pageInfo['page']->getTitle()); |
|
381
|
|
|
$query = "SELECT log_action, log_type, log_timestamp AS timestamp |
|
382
|
|
|
FROM $loggingTable |
|
383
|
|
|
WHERE log_namespace = '" . $this->pageInfo['page']->getNamespace() . "' |
|
384
|
|
|
AND log_title = '$title' AND log_timestamp > 1 |
|
385
|
|
|
AND log_type IN ('delete', 'move', 'protect', 'stable')"; |
|
386
|
|
|
$events = $this->conn->query($query)->fetchAll(); |
|
387
|
|
|
|
|
388
|
|
|
foreach ($events as $event) { |
|
389
|
|
|
$time = strtotime($event['timestamp']); |
|
390
|
|
|
$year = date('Y', $time); |
|
391
|
|
|
if (isset($this->pageInfo['year_count'][$year])) { |
|
392
|
|
|
$yearEvents = $this->pageInfo['year_count'][$year]['events']; |
|
393
|
|
|
|
|
394
|
|
|
// Convert log type value to i18n key |
|
395
|
|
|
switch ($event['log_type']) { |
|
396
|
|
|
case 'protect': |
|
397
|
|
|
$action = 'protections'; |
|
398
|
|
|
break; |
|
399
|
|
|
case 'delete': |
|
400
|
|
|
$action = 'deletions'; |
|
401
|
|
|
break; |
|
402
|
|
|
case 'move': |
|
403
|
|
|
$action = 'moves'; |
|
404
|
|
|
break; |
|
405
|
|
|
// count pending-changes protections along with normal protections |
|
406
|
|
|
case 'stable': |
|
407
|
|
|
$action = 'protections'; |
|
408
|
|
|
break; |
|
409
|
|
|
} |
|
410
|
|
|
|
|
411
|
|
|
if (empty($yearEvents[$action])) { |
|
412
|
|
|
$yearEvents[$action] = 1; |
|
|
|
|
|
|
413
|
|
|
} else { |
|
414
|
|
|
$yearEvents[$action]++; |
|
415
|
|
|
} |
|
416
|
|
|
|
|
417
|
|
|
$this->pageInfo['year_count'][$year]['events'] = $yearEvents; |
|
418
|
|
|
} |
|
419
|
|
|
} |
|
420
|
|
|
} |
|
421
|
|
|
|
|
422
|
|
|
/** |
|
423
|
|
|
* Get any CheckWiki errors |
|
424
|
|
|
* @return array Results from query |
|
425
|
|
|
*/ |
|
426
|
|
|
private function getCheckWikiErrors() |
|
427
|
|
|
{ |
|
428
|
|
|
if ($this->pageInfo['page']->getNamespace() !== 0 || !$this->container->getParameter('app.is_labs')) { |
|
429
|
|
|
return []; |
|
430
|
|
|
} |
|
431
|
|
|
$title = $this->pageInfo['page']->getTitle(); // no underscores |
|
432
|
|
|
$dbName = preg_replace('/_p$/', '', $this->pageInfo['dbName']); // remove _p if present |
|
433
|
|
|
|
|
434
|
|
|
$query = "SELECT error, notice, found, name_trans AS name, prio, text_trans AS explanation |
|
435
|
|
|
FROM s51080__checkwiki_p.cw_error a |
|
436
|
|
|
JOIN s51080__checkwiki_p.cw_overview_errors b |
|
437
|
|
|
WHERE a.project = b.project AND a.project = '$dbName' |
|
438
|
|
|
AND a.title = '$title' AND a.error = b.id |
|
439
|
|
|
AND b.done IS NULL"; |
|
440
|
|
|
|
|
441
|
|
|
$conn = $this->container->get('doctrine')->getManager('toolsdb')->getConnection(); |
|
442
|
|
|
$res = $conn->query($query)->fetchAll(); |
|
443
|
|
|
return $res; |
|
444
|
|
|
} |
|
445
|
|
|
|
|
446
|
|
|
/** |
|
447
|
|
|
* Get basic wikidata on the page: label and description. |
|
448
|
|
|
* Reported as "bugs" if they are missing. |
|
449
|
|
|
* @return array Label and description, if present |
|
450
|
|
|
*/ |
|
451
|
|
|
private function getWikidataErrors() |
|
452
|
|
|
{ |
|
453
|
|
|
if (empty($this->pageInfo['wikidataId'])) { |
|
454
|
|
|
return []; |
|
455
|
|
|
} |
|
456
|
|
|
|
|
457
|
|
|
$wikidataId = ltrim($this->pageInfo['wikidataId'], 'Q'); |
|
458
|
|
|
$lang = $this->pageInfo['lang']; |
|
459
|
|
|
|
|
460
|
|
|
$query = "SELECT IF(term_type = 'label', 'label', 'description') AS term, term_text |
|
461
|
|
|
FROM wikidatawiki_p.wb_entity_per_page |
|
462
|
|
|
JOIN wikidatawiki_p.page ON epp_page_id = page_id |
|
463
|
|
|
JOIN wikidatawiki_p.wb_terms ON term_entity_id = epp_entity_id |
|
464
|
|
|
AND term_language = '$lang' AND term_type IN ('label', 'description') |
|
465
|
|
|
WHERE epp_entity_id = $wikidataId |
|
466
|
|
|
UNION |
|
467
|
|
|
SELECT pl_title AS term, wb_terms.term_text |
|
468
|
|
|
FROM wikidatawiki_p.pagelinks |
|
469
|
|
|
JOIN wikidatawiki_p.wb_terms ON term_entity_id = SUBSTRING(pl_title, 2) |
|
470
|
|
|
AND term_entity_type = (IF(SUBSTRING(pl_title, 1, 1) = 'Q', 'item', 'property')) |
|
471
|
|
|
AND term_language = '$lang' |
|
472
|
|
|
AND term_type = 'label' |
|
473
|
|
|
WHERE pl_namespace IN (0,120 ) |
|
474
|
|
|
AND pl_from = ( |
|
475
|
|
|
SELECT page_id FROM page |
|
476
|
|
|
WHERE page_namespace = 0 AND page_title = 'Q$wikidataId' |
|
477
|
|
|
)"; |
|
478
|
|
|
|
|
479
|
|
|
$conn = $this->container->get('doctrine')->getManager('replicas')->getConnection(); |
|
480
|
|
|
$res = $conn->query($query)->fetchAll(); |
|
481
|
|
|
|
|
482
|
|
|
$terms = array_map(function ($entry) { |
|
483
|
|
|
return $entry['term']; |
|
484
|
|
|
}, $res); |
|
485
|
|
|
|
|
486
|
|
|
$errors = []; |
|
487
|
|
|
|
|
488
|
|
View Code Duplication |
if (!in_array('label', $terms)) { |
|
|
|
|
|
|
489
|
|
|
$errors[] = [ |
|
490
|
|
|
'prio' => 2, |
|
491
|
|
|
'name' => 'Wikidata', |
|
492
|
|
|
'notice' => "Label for language <em>$lang</em> is missing", // FIXME: i18n |
|
493
|
|
|
'explanation' => "See: <a target='_blank' " . |
|
494
|
|
|
"href='//www.wikidata.org/wiki/Help:Label'>Help:Label</a>", |
|
495
|
|
|
]; |
|
496
|
|
|
} |
|
497
|
|
View Code Duplication |
if (!in_array('description', $terms)) { |
|
|
|
|
|
|
498
|
|
|
$errors[] = [ |
|
499
|
|
|
'prio' => 3, |
|
500
|
|
|
'name' => 'Wikidata', |
|
501
|
|
|
'notice' => "Description for language <em>$lang</em> is missing", // FIXME: i18n |
|
502
|
|
|
'explanation' => "See: <a target='_blank' " . |
|
503
|
|
|
"href='//www.wikidata.org/wiki/Help:Description'>Help:Description</a>", |
|
504
|
|
|
]; |
|
505
|
|
|
} |
|
506
|
|
|
|
|
507
|
|
|
return $errors; |
|
508
|
|
|
} |
|
509
|
|
|
|
|
510
|
|
|
/** |
|
511
|
|
|
* Get the size of the diff. |
|
512
|
|
|
* @param int $revIndex The index of the revision within $this->pageHistory |
|
513
|
|
|
* @return int Size of the diff |
|
514
|
|
|
*/ |
|
515
|
|
|
private function getDiffSize($revIndex) |
|
516
|
|
|
{ |
|
517
|
|
|
$rev = $this->pageHistory[$revIndex]; |
|
518
|
|
|
|
|
519
|
|
|
if ($revIndex === 0) { |
|
520
|
|
|
return $rev['length']; |
|
521
|
|
|
} |
|
522
|
|
|
|
|
523
|
|
|
$lastRev = $this->pageHistory[$revIndex - 1]; |
|
524
|
|
|
|
|
525
|
|
|
// TODO: Remove once T101631 is resolved |
|
526
|
|
|
// Treat as zero change in size if length of previous edit is missing |
|
527
|
|
|
if ($lastRev['length'] === null) { |
|
528
|
|
|
return 0; |
|
529
|
|
|
} else { |
|
530
|
|
|
return $rev['length'] - $lastRev['length']; |
|
531
|
|
|
} |
|
532
|
|
|
} |
|
533
|
|
|
|
|
534
|
|
|
/** |
|
535
|
|
|
* Parse the revision history, which should be at $this->pageHistory |
|
536
|
|
|
* @return array Associative "master" array of metadata about the page |
|
537
|
|
|
*/ |
|
538
|
|
|
private function parseHistory() |
|
539
|
|
|
{ |
|
540
|
|
|
$revisionCount = $this->pageInfo['page']->getNumRevisions(); |
|
541
|
|
|
if ($revisionCount == 0) { |
|
542
|
|
|
// $this->error = "no records"; |
|
|
|
|
|
|
543
|
|
|
return; |
|
544
|
|
|
} |
|
545
|
|
|
|
|
546
|
|
|
$firstEdit = $this->pageInfo['firstEdit']; |
|
547
|
|
|
|
|
548
|
|
|
// Get UNIX timestamp of the first day of the month of the first edit |
|
549
|
|
|
// This is used as a comparison when building our array of per-month stats |
|
550
|
|
|
$firstEditMonth = mktime(0, 0, 0, (int) $firstEdit->getMonth(), 1, $firstEdit->getYear()); |
|
551
|
|
|
|
|
552
|
|
|
$lastEdit = $this->pageInfo['lastEdit']; |
|
553
|
|
|
$secondLastEdit = $revisionCount === 1 ? $lastEdit : $this->pageHistory[ $revisionCount - 2 ]; |
|
|
|
|
|
|
554
|
|
|
|
|
555
|
|
|
// Now we can start our master array. This one will be HUGE! |
|
556
|
|
|
$data = [ |
|
557
|
|
|
'general' => [ |
|
558
|
|
|
'max_add' => $firstEdit, |
|
559
|
|
|
'max_del' => $firstEdit, |
|
560
|
|
|
'editor_count' => 0, |
|
561
|
|
|
'anon_count' => 0, |
|
562
|
|
|
'minor_count' => 0, |
|
563
|
|
|
'count_history' => ['day' => 0, 'week' => 0, 'month' => 0, 'year' => 0], |
|
564
|
|
|
'current_size' => $this->pageHistory[$revisionCount-1]['length'], |
|
565
|
|
|
'textshares' => [], |
|
566
|
|
|
'textshare_total' => 0, |
|
567
|
|
|
'automated_count' => 0, |
|
568
|
|
|
'revert_count' => 0, |
|
569
|
|
|
'added' => 0, |
|
570
|
|
|
], |
|
571
|
|
|
'max_edits_per_month' => 0, // for bar chart in "Month counts" section |
|
572
|
|
|
'editors' => [], |
|
573
|
|
|
'anons' => [], |
|
574
|
|
|
'year_count' => [], |
|
575
|
|
|
'tools' => [], |
|
576
|
|
|
]; |
|
577
|
|
|
|
|
578
|
|
|
// restore existing general data |
|
579
|
|
|
$data['general'] = array_merge($data['general'], $this->pageInfo['general']); |
|
580
|
|
|
|
|
581
|
|
|
// And now comes the logic for filling said master array |
|
582
|
|
|
foreach ($this->pageHistory as $i => $rev) { |
|
583
|
|
|
$edit = new Edit($this->pageInfo['page'], $rev); |
|
|
|
|
|
|
584
|
|
|
$diffSize = $this->getDiffSize($i); |
|
|
|
|
|
|
585
|
|
|
$username = htmlspecialchars($rev['username']); |
|
586
|
|
|
|
|
587
|
|
|
// Sometimes, with old revisions (2001 era), the revisions from 2002 come before 2001 |
|
588
|
|
|
if ($edit->getTimestamp() < $firstEdit->getTimestamp()) { |
|
589
|
|
|
$firstEdit = $edit; |
|
590
|
|
|
} |
|
591
|
|
|
|
|
592
|
|
|
// Fill in the blank arrays for the year and 12 months |
|
593
|
|
|
if (!isset($data['year_count'][$edit->getYear()])) { |
|
594
|
|
|
$data['year_count'][$edit->getYear()] = [ |
|
595
|
|
|
'all' => 0, |
|
596
|
|
|
'minor' => 0, |
|
597
|
|
|
'anon' => 0, |
|
598
|
|
|
'automated' => 0, |
|
599
|
|
|
'size' => 0, // keep track of the size by the end of the year |
|
600
|
|
|
'events' => [], |
|
601
|
|
|
'months' => [], |
|
602
|
|
|
]; |
|
603
|
|
|
|
|
604
|
|
|
for ($i = 1; $i <= 12; $i++) { |
|
605
|
|
|
$timeObj = mktime(0, 0, 0, $i, 1, $edit->getYear()); |
|
606
|
|
|
|
|
607
|
|
|
// don't show zeros for months before the first edit or after the current month |
|
608
|
|
|
if ($timeObj < $firstEditMonth || $timeObj > strtotime('last day of this month')) { |
|
609
|
|
|
continue; |
|
610
|
|
|
} |
|
611
|
|
|
|
|
612
|
|
|
$data['year_count'][$edit->getYear()]['months'][sprintf('%02d', $i)] = [ |
|
613
|
|
|
'all' => 0, |
|
614
|
|
|
'minor' => 0, |
|
615
|
|
|
'anon' => 0, |
|
616
|
|
|
'automated' => 0, |
|
617
|
|
|
]; |
|
618
|
|
|
} |
|
619
|
|
|
} |
|
620
|
|
|
|
|
621
|
|
|
// Increment year and month counts for all edits |
|
622
|
|
|
$data['year_count'][$edit->getYear()]['all']++; |
|
623
|
|
|
$data['year_count'][$edit->getYear()]['months'][$edit->getMonth()]['all']++; |
|
624
|
|
|
$data['year_count'][$edit->getYear()]['size'] = (int) $rev['length']; |
|
625
|
|
|
|
|
626
|
|
|
$editsThisMonth = $data['year_count'][$edit->getYear()]['months'][$edit->getMonth()]['all']; |
|
627
|
|
|
if ($editsThisMonth > $data['max_edits_per_month']) { |
|
628
|
|
|
$data['max_edits_per_month'] = $editsThisMonth; |
|
629
|
|
|
} |
|
630
|
|
|
|
|
631
|
|
|
// Fill in various user stats |
|
632
|
|
|
if (!isset($data['editors'][$username])) { |
|
633
|
|
|
$data['general']['editor_count']++; |
|
634
|
|
|
$data['editors'][$username] = [ |
|
635
|
|
|
'all' => 0, |
|
636
|
|
|
'minor' => 0, |
|
637
|
|
|
'minor_percentage' => 0, |
|
638
|
|
|
'first' => date('Y-m-d, H:i', strtotime($rev['timestamp'])), |
|
639
|
|
|
'first_id' => $rev['id'], |
|
640
|
|
|
'last' => null, |
|
641
|
|
|
'atbe' => null, |
|
642
|
|
|
'added' => 0, |
|
643
|
|
|
'sizes' => [], |
|
644
|
|
|
'urlencoded' => rawurlencode($rev['username']), |
|
645
|
|
|
]; |
|
646
|
|
|
} |
|
647
|
|
|
|
|
648
|
|
|
// Increment user counts |
|
649
|
|
|
$data['editors'][$username]['all']++; |
|
650
|
|
|
$data['editors'][$username]['last'] = date('Y-m-d, H:i', strtotime($rev['timestamp'])); |
|
651
|
|
|
$data['editors'][$username]['last_id'] = $rev['id']; |
|
652
|
|
|
|
|
653
|
|
|
// Store number of KB added with this edit |
|
654
|
|
|
$data['editors'][$username]['sizes'][] = $rev['length'] / 1024; |
|
655
|
|
|
|
|
656
|
|
|
// check if it was a revert |
|
657
|
|
|
if ($this->aeh->isRevert($rev['comment'])) { |
|
658
|
|
|
$data['general']['revert_count']++; |
|
659
|
|
|
} else { |
|
660
|
|
|
// edit was NOT a revert |
|
661
|
|
|
|
|
662
|
|
|
if ($edit->getSize() > 0) { |
|
663
|
|
|
$data['general']['added'] += $edit->getSize(); |
|
664
|
|
|
$data['editors'][$username]['added'] += $edit->getSize(); |
|
665
|
|
|
} |
|
666
|
|
|
|
|
667
|
|
|
// determine if the next revision was a revert |
|
668
|
|
|
$nextRevision = isset($this->pageHistory[$i + 1]) ? $this->pageHistory[$i + 1] : null; |
|
669
|
|
|
$nextRevisionIsRevert = $nextRevision && |
|
670
|
|
|
$this->getDiffSize($i + 1) === -$edit->getSize() && |
|
671
|
|
|
$this->aeh->isRevert($nextRevision['comment']); |
|
672
|
|
|
|
|
673
|
|
|
// don't count this edit as content removal if the next edit reverted it |
|
674
|
|
|
if (!$nextRevisionIsRevert && $edit->getSize() < $data['general']['max_del']->getSize()) { |
|
675
|
|
|
$data['general']['max_del'] = $edit; |
|
676
|
|
|
} |
|
677
|
|
|
|
|
678
|
|
|
// FIXME: possibly remove this |
|
679
|
|
|
if ($edit->getLength() > 0) { |
|
680
|
|
|
// keep track of added content |
|
681
|
|
|
$data['general']['textshare_total'] += $edit->getLength(); |
|
682
|
|
|
if (!isset($data['textshares'][$username]['all'])) { |
|
683
|
|
|
$data['textshares'][$username]['all'] = 0; |
|
684
|
|
|
} |
|
685
|
|
|
$data['textshares'][$username]['all'] += $edit->getLength(); |
|
686
|
|
|
} |
|
687
|
|
|
|
|
688
|
|
|
if ($edit->getSize() > $data['general']['max_add']->getSize()) { |
|
689
|
|
|
$data['general']['max_add'] = $edit; |
|
690
|
|
|
} |
|
691
|
|
|
} |
|
692
|
|
|
|
|
693
|
|
|
if ($edit->isAnon()) { |
|
694
|
|
|
if (!isset($rev['rev_user']['anons'][$username])) { |
|
695
|
|
|
$data['general']['anon_count']++; |
|
696
|
|
|
} |
|
697
|
|
|
// Anonymous, increase counts |
|
698
|
|
|
$data['anons'][] = $username; |
|
699
|
|
|
$data['year_count'][$edit->getYear()]['anon']++; |
|
700
|
|
|
$data['year_count'][$edit->getYear()]['months'][$edit->getMonth()]['anon']++; |
|
701
|
|
|
} |
|
702
|
|
|
|
|
703
|
|
|
if ($edit->isMinor()) { |
|
704
|
|
|
// Logged in, increase counts |
|
705
|
|
|
$data['general']['minor_count']++; |
|
706
|
|
|
$data['year_count'][$edit->getYear()]['minor']++; |
|
707
|
|
|
$data['year_count'][$edit->getYear()]['months'][$edit->getMonth()]['minor']++; |
|
708
|
|
|
$data['editors'][$username]['minor']++; |
|
709
|
|
|
} |
|
710
|
|
|
|
|
711
|
|
|
$automatedTool = $this->aeh->getTool($rev['comment']); |
|
712
|
|
|
if ($automatedTool) { |
|
713
|
|
|
$data['general']['automated_count']++; |
|
714
|
|
|
$data['year_count'][$edit->getYear()]['automated']++; |
|
715
|
|
|
$data['year_count'][$edit->getYear()]['months'][$edit->getMonth()]['automated']++; |
|
716
|
|
|
|
|
717
|
|
|
if (!isset($data['tools'][$automatedTool])) { |
|
718
|
|
|
$data['tools'][$automatedTool] = [ |
|
719
|
|
|
'count' => 1, |
|
720
|
|
|
'link' => $this->aeh->getTools()[$automatedTool]['link'], |
|
721
|
|
|
]; |
|
722
|
|
|
} else { |
|
723
|
|
|
$data['tools'][$automatedTool]['count']++; |
|
724
|
|
|
} |
|
725
|
|
|
} |
|
726
|
|
|
|
|
727
|
|
|
// Increment "edits per <time>" counts |
|
728
|
|
View Code Duplication |
if (strtotime($rev['timestamp']) > strtotime('-1 day')) { |
|
|
|
|
|
|
729
|
|
|
$data['general']['count_history']['day']++; |
|
730
|
|
|
} |
|
731
|
|
View Code Duplication |
if (strtotime($rev['timestamp']) > strtotime('-1 week')) { |
|
|
|
|
|
|
732
|
|
|
$data['general']['count_history']['week']++; |
|
733
|
|
|
} |
|
734
|
|
View Code Duplication |
if (strtotime($rev['timestamp']) > strtotime('-1 month')) { |
|
|
|
|
|
|
735
|
|
|
$data['general']['count_history']['month']++; |
|
736
|
|
|
} |
|
737
|
|
View Code Duplication |
if (strtotime($rev['timestamp']) > strtotime('-1 year')) { |
|
|
|
|
|
|
738
|
|
|
$data['general']['count_history']['year']++; |
|
739
|
|
|
} |
|
740
|
|
|
} |
|
741
|
|
|
|
|
742
|
|
|
// add percentages |
|
743
|
|
|
$data['general']['minor_percentage'] = round( |
|
744
|
|
|
($data['general']['minor_count'] / $revisionCount) * 100, |
|
745
|
|
|
1 |
|
746
|
|
|
); |
|
747
|
|
|
$data['general']['anon_percentage'] = round( |
|
748
|
|
|
($data['general']['anon_count'] / $revisionCount) * 100, |
|
749
|
|
|
1 |
|
750
|
|
|
); |
|
751
|
|
|
|
|
752
|
|
|
// other general statistics |
|
753
|
|
|
$dateFirst = $firstEdit->getTimestamp(); |
|
754
|
|
|
$dateLast = $lastEdit->getTimestamp(); |
|
755
|
|
|
$data['general']['datetime_first_edit'] = $dateFirst; |
|
756
|
|
|
$data['general']['datetime_last_edit'] = $dateLast; |
|
757
|
|
|
$interval = date_diff($dateLast, $dateFirst, true); |
|
758
|
|
|
|
|
759
|
|
|
$data['totaldays'] = $interval->format('%a'); |
|
760
|
|
|
$data['general']['average_days_per_edit'] = round($data['totaldays'] / $revisionCount, 1); |
|
761
|
|
|
$editsPerDay = $data['totaldays'] |
|
762
|
|
|
? $revisionCount / ($data['totaldays'] / (365 / 12 / 24)) |
|
763
|
|
|
: 0; |
|
764
|
|
|
$data['general']['edits_per_day'] = round($editsPerDay, 1); |
|
765
|
|
|
$editsPerMonth = $data['totaldays'] |
|
766
|
|
|
? $revisionCount / ($data['totaldays'] / (365 / 12)) |
|
767
|
|
|
: 0; |
|
768
|
|
|
$data['general']['edits_per_month'] = round($editsPerMonth, 1); |
|
769
|
|
|
$editsPerYear = $data['totaldays'] |
|
770
|
|
|
? $revisionCount / ($data['totaldays'] / 365) |
|
771
|
|
|
: 0; |
|
772
|
|
|
$data['general']['edits_per_year'] = round($editsPerYear, 1); |
|
773
|
|
|
$data['general']['edits_per_editor'] = round($revisionCount / count($data['editors']), 1); |
|
774
|
|
|
|
|
775
|
|
|
// If after processing max_del is positive, no edit actually removed text, so unset this value |
|
776
|
|
|
if ($data['general']['max_del']->getSize() > 0) { |
|
777
|
|
|
unset($data['general']['max_del']); |
|
778
|
|
|
} |
|
779
|
|
|
|
|
780
|
|
|
// Various sorts |
|
781
|
|
|
arsort($data['editors']); |
|
782
|
|
|
arsort($data['textshares']); |
|
783
|
|
|
arsort($data['tools']); |
|
784
|
|
|
ksort($data['year_count']); |
|
785
|
|
|
|
|
786
|
|
|
return $data; |
|
787
|
|
|
} |
|
788
|
|
|
} |
|
789
|
|
|
|
Let’s take a look at an example:
In the above example, the authenticate() method works fine as long as you just pass instances of MyUser. However, if you now also want to pass a different implementation of User which does not have a getDisplayName() method, the code will break.
Available Fixes
Change the type-hint for the parameter:
Add an additional type-check:
Add the method to the interface: