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