1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* This file is part of dispositif/wikibot application (@github) |
4
|
|
|
* 2019/2020 © Philippe M. <[email protected]> |
5
|
|
|
* For the full copyright and MIT license information, please view the license file. |
6
|
|
|
*/ |
7
|
|
|
|
8
|
|
|
declare(strict_types=1); |
9
|
|
|
|
10
|
|
|
namespace App\Application; |
11
|
|
|
|
12
|
|
|
use App\Domain\Exceptions\ConfigException; |
13
|
|
|
use App\Domain\Models\Summary; |
14
|
|
|
use App\Infrastructure\Logger; |
15
|
|
|
use App\Infrastructure\PageListInterface; |
16
|
|
|
use App\Infrastructure\ServiceFactory; |
17
|
|
|
use Codedungeon\PHPCliColors\Color; |
18
|
|
|
use Exception; |
19
|
|
|
use Mediawiki\Api\MediawikiFactory; |
20
|
|
|
use Mediawiki\Api\UsageException; |
21
|
|
|
use Mediawiki\DataModel\EditInfo; |
22
|
|
|
use Psr\Log\LoggerInterface; |
23
|
|
|
use Throwable; |
24
|
|
|
|
25
|
|
|
abstract class AbstractBotTaskWorker |
26
|
|
|
{ |
27
|
|
|
const TASK_BOT_FLAG = false; |
28
|
|
|
const SLEEP_AFTER_EDITION = 60; |
29
|
|
|
const MINUTES_DELAY_AFTER_LAST_HUMAN_EDIT = 15; |
30
|
|
|
const CHECK_EDIT_CONFLICT = true; |
31
|
|
|
const ARTICLE_ANALYZED_FILENAME = __DIR__.'/resources/article_edited.txt'; |
32
|
|
|
const SKIP_LASTEDIT_BY_BOT = true; |
33
|
|
|
const SKIP_NOT_IN_MAIN_WIKISPACE = true; |
34
|
|
|
|
35
|
|
|
/** |
36
|
|
|
* @var PageListInterface |
37
|
|
|
*/ |
38
|
|
|
protected $pageListGenerator; |
39
|
|
|
/** |
40
|
|
|
* @var WikiBotConfig |
41
|
|
|
*/ |
42
|
|
|
protected $bot; |
43
|
|
|
/** |
44
|
|
|
* @var MediawikiFactory |
45
|
|
|
*/ |
46
|
|
|
protected $wiki; |
47
|
|
|
/** |
48
|
|
|
* @var WikiPageAction |
49
|
|
|
*/ |
50
|
|
|
protected $pageAction; |
51
|
|
|
protected $defaultTaskname; |
52
|
|
|
protected $titleTaskname; |
53
|
|
|
|
54
|
|
|
// todo move (modeAuto, maxLag) to BotConfig |
55
|
|
|
protected $modeAuto = false; |
56
|
|
|
protected $maxLag = 5; |
57
|
|
|
/** |
58
|
|
|
* @var Logger|LoggerInterface |
59
|
|
|
*/ |
60
|
|
|
protected $log; |
61
|
|
|
/** |
62
|
|
|
* array des articles déjà anal |
63
|
|
|
*/ |
64
|
|
|
protected $pastAnalyzed; |
65
|
|
|
/** |
66
|
|
|
* @var Summary |
67
|
|
|
*/ |
68
|
|
|
protected $summary; |
69
|
|
|
|
70
|
|
|
/** |
71
|
|
|
* Goo2ouvrageWorker constructor. |
72
|
|
|
* |
73
|
|
|
* @param WikiBotConfig $bot |
74
|
|
|
* @param MediawikiFactory $wiki |
75
|
|
|
* @param PageListInterface $pagesGen |
76
|
|
|
*/ |
77
|
|
|
public function __construct(WikiBotConfig $bot, MediawikiFactory $wiki, ?PageListInterface $pagesGen = null) |
78
|
|
|
{ |
79
|
|
|
$this->log = $bot->log; |
80
|
|
|
$this->wiki = $wiki; |
81
|
|
|
$this->bot = $bot; |
82
|
|
|
if ($pagesGen) { |
83
|
|
|
$this->pageListGenerator = $pagesGen; |
84
|
|
|
} |
85
|
|
|
$this->setUpInConstructor(); |
86
|
|
|
|
87
|
|
|
$this->defaultTaskname = $bot->taskName; |
88
|
|
|
|
89
|
|
|
$analyzed = @file(static::ARTICLE_ANALYZED_FILENAME, FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES); |
90
|
|
|
$this->pastAnalyzed = ($analyzed !== false) ? $analyzed : []; |
91
|
|
|
|
92
|
|
|
// @throw exception on "Invalid CSRF token" |
93
|
|
|
$this->run();//todo delete that and use (Worker)->run($duration) or process management |
94
|
|
|
} |
95
|
|
|
|
96
|
|
|
protected function setUpInConstructor(): void |
97
|
|
|
{ |
98
|
|
|
// optional implementation |
99
|
|
|
} |
100
|
|
|
|
101
|
|
|
/** |
102
|
|
|
* @throws ConfigException |
103
|
|
|
* @throws Throwable |
104
|
|
|
* @throws UsageException |
105
|
|
|
*/ |
106
|
|
|
public function run() |
107
|
|
|
{ |
108
|
|
|
$titles = $this->getTitles(); |
109
|
|
|
echo date('d-m-Y H:i')." *** NEW WORKER ***\n"; |
110
|
|
|
foreach ($titles as $title) { |
111
|
|
|
$this->titleProcess($title); |
112
|
|
|
sleep(3); |
113
|
|
|
} |
114
|
|
|
} |
115
|
|
|
|
116
|
|
|
/** |
117
|
|
|
* @return array |
118
|
|
|
* @throws ConfigException |
119
|
|
|
*/ |
120
|
|
|
protected function getTitles(): array |
121
|
|
|
{ |
122
|
|
|
if ($this->pageListGenerator === null) { |
123
|
|
|
throw new ConfigException('Empty PageListGenerator'); |
124
|
|
|
} |
125
|
|
|
|
126
|
|
|
return $this->pageListGenerator->getPageTitles(); |
127
|
|
|
} |
128
|
|
|
|
129
|
|
|
/** |
130
|
|
|
* @param string $title |
131
|
|
|
* |
132
|
|
|
* @throws UsageException |
133
|
|
|
* @throws Throwable |
134
|
|
|
*/ |
135
|
|
|
protected function titleProcess(string $title): void |
136
|
|
|
{ |
137
|
|
|
echo "---------------------\n"; |
138
|
|
|
echo date('d-m-Y H:i').' '.Color::BG_CYAN." $title ".Color::NORMAL."\n"; |
139
|
|
|
sleep(1); |
140
|
|
|
|
141
|
|
|
if (in_array($title, $this->pastAnalyzed)) { |
142
|
|
|
echo "Skip : déjà analysé\n"; |
143
|
|
|
|
144
|
|
|
return; |
145
|
|
|
} |
146
|
|
|
|
147
|
|
|
$this->titleTaskname = $this->defaultTaskname; |
148
|
|
|
|
149
|
|
|
$text = $this->getText($title); |
150
|
|
|
if (static::SKIP_LASTEDIT_BY_BOT && $this->pageAction->getLastEditor() === getenv('BOT_NAME')) { |
151
|
|
|
echo "Skip : déjà édité par le bot\n"; |
152
|
|
|
|
153
|
|
|
return; |
154
|
|
|
} |
155
|
|
|
if (empty($text) || !$this->checkAllowedEdition($title, $text)) { |
156
|
|
|
return; |
157
|
|
|
} |
158
|
|
|
|
159
|
|
|
$this->summary = new Summary($this->defaultTaskname); |
160
|
|
|
$this->summary->setBotFlag(static::TASK_BOT_FLAG); |
161
|
|
|
|
162
|
|
|
$newText = $this->processDomain($title, $text); |
163
|
|
|
|
164
|
|
|
$this->memorizeAndSaveAnalyzedPage($title); |
165
|
|
|
|
166
|
|
|
if (empty($newText) || $newText === $text) { |
167
|
|
|
echo "Skip identique ou vide\n"; |
168
|
|
|
|
169
|
|
|
return; |
170
|
|
|
} |
171
|
|
|
|
172
|
|
|
if (!$this->modeAuto) { |
173
|
|
|
$ask = readline(Color::LIGHT_MAGENTA."*** ÉDITION ? [y/n/auto]".Color::NORMAL); |
174
|
|
|
if ('auto' === $ask) { |
175
|
|
|
$this->modeAuto = true; |
176
|
|
|
} elseif ('y' !== $ask) { |
177
|
|
|
return; |
178
|
|
|
} |
179
|
|
|
} |
180
|
|
|
|
181
|
|
|
$this->doEdition($newText); |
182
|
|
|
} |
183
|
|
|
|
184
|
|
|
/** |
185
|
|
|
* todo DI |
186
|
|
|
* |
187
|
|
|
* @param string $title |
188
|
|
|
* |
189
|
|
|
* @return string|null |
190
|
|
|
* @throws Exception |
191
|
|
|
* @throws Exception |
192
|
|
|
*/ |
193
|
|
|
protected function getText(string $title): ?string |
194
|
|
|
{ |
195
|
|
|
$this->pageAction = ServiceFactory::wikiPageAction($title); |
196
|
|
|
if (static::SKIP_NOT_IN_MAIN_WIKISPACE && $this->pageAction->getNs() !== 0) { |
197
|
|
|
throw new Exception("La page n'est pas dans Main (ns!==0)"); |
198
|
|
|
} |
199
|
|
|
|
200
|
|
|
return $this->pageAction->getText(); |
201
|
|
|
} |
202
|
|
|
|
203
|
|
|
/** |
204
|
|
|
* todo distinguer 2 methodes : ban temporaire et permanent (=> logAnalyzed) |
205
|
|
|
* Controle droit d'edition. |
206
|
|
|
* |
207
|
|
|
* @param string $title |
208
|
|
|
* @param string $text |
209
|
|
|
* |
210
|
|
|
* @return bool |
211
|
|
|
* @throws UsageException |
212
|
|
|
*/ |
213
|
|
|
protected function checkAllowedEdition(string $title, string $text): bool |
214
|
|
|
{ |
215
|
|
|
$this->bot->checkStopOnTalkpage(true); |
216
|
|
|
|
217
|
|
|
if (WikiBotConfig::isEditionRestricted($text)) { |
218
|
|
|
echo "SKIP : protection/3R/travaux.\n"; |
219
|
|
|
|
220
|
|
|
return false; |
221
|
|
|
} |
222
|
|
|
if ($this->bot->minutesSinceLastEdit($title) < static::MINUTES_DELAY_AFTER_LAST_HUMAN_EDIT) { |
223
|
|
|
echo "SKIP : édition humaine dans les dernières ".static::MINUTES_DELAY_AFTER_LAST_HUMAN_EDIT." minutes.\n"; |
224
|
|
|
|
225
|
|
|
return false; |
226
|
|
|
} |
227
|
|
|
if (preg_match('#{{ ?En-tête label ?\| ?AdQ#i', $text)) { |
228
|
|
|
echo "SKIP : AdQ.\n"; // BA ?? |
229
|
|
|
|
230
|
|
|
return false; |
231
|
|
|
} |
232
|
|
|
|
233
|
|
|
return true; |
234
|
|
|
} |
235
|
|
|
|
236
|
|
|
/** |
237
|
|
|
* return $newText for editing |
238
|
|
|
* |
239
|
|
|
* @param string $title |
240
|
|
|
* @param string $text |
241
|
|
|
* |
242
|
|
|
* @return string|null |
243
|
|
|
*/ |
244
|
|
|
abstract protected function processDomain(string $title, string $text): ?string; |
245
|
|
|
|
246
|
|
|
protected function doEdition(string $newText): void |
247
|
|
|
{ |
248
|
|
|
$summaryText = $this->generateSummaryText(); |
249
|
|
|
|
250
|
|
|
try { |
251
|
|
|
$result = $this->pageAction->editPage( |
252
|
|
|
$newText, |
253
|
|
|
new EditInfo($summaryText, $this->summary->isMinorFlag(), $this->summary->isBotFlag(), $this->maxLag), |
254
|
|
|
static::CHECK_EDIT_CONFLICT |
255
|
|
|
); |
256
|
|
|
} catch (Throwable $e) { |
257
|
|
|
if (preg_match('#Invalid CSRF token#', $e->getMessage())) { |
258
|
|
|
throw new \Exception('Invalid CSRF token'); |
259
|
|
|
} |
260
|
|
|
|
261
|
|
|
// If not a critical edition error |
262
|
|
|
// example : Wiki Conflict : Page has been edited after getText() |
263
|
|
|
$this->log->warning($e->getMessage()); |
264
|
|
|
|
265
|
|
|
return; |
266
|
|
|
} |
267
|
|
|
|
268
|
|
|
dump($result); |
269
|
|
|
echo "Sleep ".(string)static::SLEEP_AFTER_EDITION."\n"; |
270
|
|
|
sleep(static::SLEEP_AFTER_EDITION); |
271
|
|
|
} |
272
|
|
|
|
273
|
|
|
/** |
274
|
|
|
* @param string $title |
275
|
|
|
*/ |
276
|
|
|
private function memorizeAndSaveAnalyzedPage(string $title): void |
277
|
|
|
{ |
278
|
|
|
if (!in_array($title, $this->pastAnalyzed)) { |
279
|
|
|
$this->pastAnalyzed[] = $title; |
280
|
|
|
@file_put_contents(static::ARTICLE_ANALYZED_FILENAME, $title.PHP_EOL, FILE_APPEND); |
|
|
|
|
281
|
|
|
sleep(1); |
282
|
|
|
} |
283
|
|
|
} |
284
|
|
|
|
285
|
|
|
/** |
286
|
|
|
* @return string |
287
|
|
|
*/ |
288
|
|
|
protected function generateSummaryText(): string |
289
|
|
|
{ |
290
|
|
|
return $this->summary->serialize(); |
291
|
|
|
} |
292
|
|
|
} |
293
|
|
|
|
If you suppress an error, we recommend checking for the error condition explicitly: