1
|
|
|
<?php |
2
|
|
|
/* |
3
|
|
|
* This file is part of dispositif/wikibot application (@github) |
4
|
|
|
* 2019-2023 © Philippe M./Irønie <[email protected]> |
5
|
|
|
* For the full copyright and MIT license information, view the license file. |
6
|
|
|
*/ |
7
|
|
|
|
8
|
|
|
declare(strict_types=1); |
9
|
|
|
|
10
|
|
|
namespace App\Application; |
11
|
|
|
|
12
|
|
|
use App\Application\InfrastructurePorts\SMSInterface; |
13
|
|
|
use App\Domain\Exceptions\ConfigException; |
14
|
|
|
use App\Domain\Exceptions\StopActionException; |
15
|
|
|
use App\Domain\Utils\WikiTextUtil; |
16
|
|
|
use App\Infrastructure\Monitor\NullLogger; |
17
|
|
|
use DateInterval; |
18
|
|
|
use DateTime; |
19
|
|
|
use DateTimeImmutable; |
20
|
|
|
use DomainException; |
21
|
|
|
use Mediawiki\Api\MediawikiFactory; |
22
|
|
|
use Mediawiki\Api\UsageException; |
23
|
|
|
use Psr\Log\LoggerInterface; |
24
|
|
|
use Throwable; |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* Define wiki configuration of the bot. |
28
|
|
|
* See also .env file for parameters. |
29
|
|
|
* Page/Edit/Summary status are defined in Application/.../PageWorkStatus or Domain/OptiStatus |
30
|
|
|
*/ |
31
|
|
|
class WikiBotConfig |
32
|
|
|
{ |
33
|
|
|
public const VERSION = '2.2'; |
34
|
|
|
public const WATCHPAGE_FILENAME = __DIR__ . '/resources/watch_pages.json'; |
35
|
|
|
public const EXIT_ON_CHECK_WATCHPAGE = false; |
36
|
|
|
// do not stop if they play with {stop} on bot talk page |
37
|
|
|
public const BLACKLIST_EDITOR = ['OrlodrimBot']; |
38
|
|
|
// Use that timers config instead of worker config ? |
39
|
|
|
public const BOT_FLAG = false; |
40
|
|
|
public const MODE_AUTO = false; |
41
|
|
|
public const EXIT_ON_WIKIMESSAGE = true; |
42
|
|
|
public const EDIT_LAPS = 20; |
43
|
|
|
public const EDIT_LAPS_MANUAL = 20; |
44
|
|
|
public const EDIT_LAPS_AUTOBOT = 60; |
45
|
|
|
public const EDIT_LAPS_FLAGBOT = 8; |
46
|
|
|
public const TALK_STOP_CHECK_INTERVAL = 'PT2M'; |
47
|
|
|
public const TALK_PAGE_PREFIX = 'Discussion_utilisateur:'; |
48
|
|
|
public const SLEEP_BEFORE_STOP_TALKPAGE = 30; |
49
|
|
|
public const DELAY_MINUTES_AFTER_HUMAN_EDIT = 15; |
50
|
|
|
|
51
|
|
|
protected const BOT_DAY = '2306'; // 23 june |
52
|
|
|
|
53
|
|
|
protected string $taskName = 'Amélioration'; |
54
|
|
|
/** |
55
|
|
|
* @var LoggerInterface |
56
|
|
|
*/ |
57
|
|
|
protected $log; |
58
|
|
|
/** |
59
|
|
|
* @var DateTimeImmutable |
60
|
|
|
*/ |
61
|
|
|
protected $lastCheckStopDate; |
62
|
|
|
protected SMSInterface|null $SMSClient; |
63
|
|
|
protected $mediawikiFactory; |
64
|
|
|
protected ?string $gitCommitHash = null; |
65
|
|
|
|
66
|
|
|
public function __construct(MediawikiFactory $mediawikiFactory, ?LoggerInterface $logger = null, ?SMSInterface $SMSClient = null) |
67
|
|
|
{ |
68
|
|
|
$this->log = $logger ?? new NullLogger(); |
69
|
|
|
ini_set('user_agent', getenv('USER_AGENT')); |
70
|
|
|
$this->SMSClient = $SMSClient; |
71
|
|
|
$this->mediawikiFactory = $mediawikiFactory; |
72
|
|
|
} |
73
|
|
|
|
74
|
|
|
/** |
75
|
|
|
* Detect wiki-templates restricting the edition on a frwiki page. |
76
|
|
|
*/ |
77
|
|
|
public static function isEditionTemporaryRestrictedOnWiki(?string $text, ?string $botName = null): bool |
78
|
|
|
{ |
79
|
|
|
return empty($text) |
80
|
|
|
|| preg_match('#{{Formation#i', $text) > 0 |
81
|
|
|
|| preg_match('#{{En travaux#i', $text) > 0 |
82
|
|
|
|| preg_match('#{{En cours#i', $text) > 0 |
83
|
|
|
|| preg_match('#{{Protection#i', $text) > 0 |
84
|
|
|
|| preg_match('#\{\{(R3R|Règle des 3 révocations|travaux|en travaux|en cours|formation)#i', $text) > 0 |
85
|
|
|
|| self::isNoBotTag($text, $botName); |
86
|
|
|
} |
87
|
|
|
|
88
|
|
|
/** |
89
|
|
|
* Detect {{nobots}}, {{bots|deny=all}}, {{bots|deny=MyBot,BobBot}}. |
90
|
|
|
* Relevant out of the "main" wiki-namespace (talk pages, etc). |
91
|
|
|
*/ |
92
|
1 |
|
protected static function isNoBotTag(string $text, ?string $botName = null): bool |
93
|
|
|
{ |
94
|
1 |
|
$text = WikiTextUtil::removeHTMLcomments($text); |
95
|
|
|
$botName = $botName ?: self::getBotName(); |
96
|
|
|
$denyReg = (empty($botName)) ? '' : |
97
|
1 |
|
'|\{\{bots ?\| ?(optout|deny)\=[^\}]*' . preg_quote($botName, '#') . '[^\}]*\}\}'; |
98
|
1 |
|
return preg_match('#({{nobots}}|{{bots ?\| ?(optout|deny) ?= ?all ?}}' . $denyReg . ')#i', $text) > 0; |
99
|
1 |
|
} |
100
|
|
|
|
101
|
|
|
/** |
102
|
|
|
* @throws ConfigException |
103
|
|
|
*/ |
104
|
|
|
public static function getBotName(): string |
105
|
1 |
|
{ |
106
|
|
|
if (empty(getenv('BOT_NAME'))) { |
107
|
|
|
throw new ConfigException('BOT_NAME is not defined.'); |
108
|
|
|
} |
109
|
|
|
return getenv('BOT_NAME') ?? ''; |
110
|
|
|
} |
111
|
|
|
|
112
|
|
|
protected static function getBotOwner() |
113
|
|
|
{ |
114
|
|
|
return getenv('BOT_OWNER'); |
115
|
|
|
} |
116
|
|
|
|
117
|
|
|
public function getTaskName(): string |
118
|
|
|
{ |
119
|
|
|
return $this->taskName; |
120
|
|
|
} |
121
|
|
|
|
122
|
|
|
public function setTaskName(string $taskName): WikiBotConfig |
123
|
|
|
{ |
124
|
|
|
$this->taskName = $taskName; |
125
|
|
|
return $this; |
126
|
|
|
} |
127
|
|
|
|
128
|
|
|
public function getCurrentGitCommitHash(): ?string |
129
|
|
|
{ |
130
|
|
|
if ($this->gitCommitHash) { |
131
|
|
|
return $this->gitCommitHash; |
132
|
|
|
} |
133
|
|
|
$path = __DIR__ . '/../../.git/'; |
134
|
|
|
if (!file_exists($path)) { |
135
|
|
|
return null; |
136
|
|
|
} |
137
|
|
|
$head = trim(substr(file_get_contents($path . 'HEAD'), 4)); |
138
|
|
|
$hash = trim(file_get_contents(sprintf($path . $head))); |
139
|
|
|
$this->gitCommitHash = $hash; // cached |
140
|
|
|
|
141
|
|
|
return $hash; |
142
|
|
|
} |
143
|
|
|
|
144
|
|
|
public function getLogger(): LoggerInterface |
145
|
|
|
{ |
146
|
|
|
return $this->log; |
147
|
|
|
} |
148
|
|
|
|
149
|
|
|
/** |
150
|
|
|
* Throws Exception if "{{stop}}" or "STOP" on talk page. |
151
|
|
|
* @throws StopActionException |
152
|
|
|
*/ |
153
|
|
|
public function checkStopOnTalkpageOrException(?bool $botTalk = false): void |
154
|
|
|
{ |
155
|
|
|
if ($this->isLastCheckStopDateRecent()) { |
156
|
|
|
return; |
157
|
|
|
} |
158
|
|
|
$this->sleepAndExitOnBotDay(); |
159
|
|
|
|
160
|
|
|
// don't catch Exception (stop process if error) |
161
|
|
|
$pageAction = $this->getWikiBotPageAction(); |
162
|
|
|
$text = $pageAction->getText() ?? ''; |
163
|
|
|
$lastEditor = $pageAction->getLastEditor() ?? 'unknown'; |
164
|
|
|
|
165
|
|
|
if (preg_match('#({{stop}}|{{Stop}}|STOP)#', $text) > 0) { |
166
|
|
|
echo date('Y-m-d H:i'); |
167
|
|
|
echo sprintf("\n*** STOP ON TALK PAGE BY %s ***\n\n", $lastEditor); |
168
|
|
|
sleep(self::SLEEP_BEFORE_STOP_TALKPAGE); |
169
|
|
|
|
170
|
|
|
$this->sendSMSandFunnyTalk($lastEditor, $botTalk); |
171
|
|
|
|
172
|
|
|
throw new StopActionException(); |
173
|
|
|
} |
174
|
|
|
|
175
|
|
|
$this->lastCheckStopDate = new DateTimeImmutable(); |
176
|
|
|
} |
177
|
|
|
|
178
|
|
|
/** |
179
|
|
|
* Exit script on annual bot day. |
180
|
|
|
*/ |
181
|
|
|
public function sleepAndExitOnBotDay(): void |
182
|
|
|
{ |
183
|
|
|
if (date('dm') === self::BOT_DAY) { |
184
|
|
|
echo "BOT DAY ! Sleep before exit. \n"; |
185
|
|
|
sleep(60 * 60); |
186
|
|
|
exit; |
|
|
|
|
187
|
|
|
} |
188
|
|
|
} |
189
|
|
|
|
190
|
|
|
protected function isLastCheckStopDateRecent(): bool |
191
|
|
|
{ |
192
|
|
|
$now = new DateTimeImmutable(); |
193
|
|
|
$stopInterval = new DateInterval(self::TALK_STOP_CHECK_INTERVAL); |
194
|
|
|
|
195
|
|
|
return $this->lastCheckStopDate instanceof DateTimeImmutable |
196
|
|
|
&& $now < DateTime::createFromImmutable($this->lastCheckStopDate)->add($stopInterval); |
197
|
|
|
} |
198
|
|
|
|
199
|
|
|
/** |
200
|
|
|
* @throws UsageException |
201
|
|
|
*/ |
202
|
|
|
protected function getWikiBotPageAction(): WikiPageAction |
203
|
|
|
{ |
204
|
|
|
return new WikiPageAction($this->mediawikiFactory, $this->getBotTalkPageTitle()); |
205
|
|
|
} |
206
|
|
|
|
207
|
|
|
protected function getBotTalkPageTitle(): string |
208
|
|
|
{ |
209
|
|
|
return self::TALK_PAGE_PREFIX . $this::getBotName(); |
210
|
|
|
} |
211
|
|
|
|
212
|
|
|
protected function sendSMSandFunnyTalk(string $lastEditor, ?bool $botTalk): void |
213
|
|
|
{ |
214
|
|
|
$this->sendSMS($lastEditor); |
215
|
|
|
|
216
|
|
|
if ($botTalk) { |
217
|
|
|
$this->talkWithBot(); |
218
|
|
|
} |
219
|
|
|
} |
220
|
|
|
|
221
|
|
|
protected function sendSMS(string $lastEditor): bool |
222
|
|
|
{ |
223
|
|
|
if ($this->SMSClient instanceof SMSInterface) { |
224
|
|
|
try { |
225
|
|
|
return $this->SMSClient->send(sprintf('%s {stop} by %s', $this::getBotName(), $lastEditor)); |
226
|
|
|
} catch (Throwable) { |
227
|
|
|
return false; |
228
|
|
|
} |
229
|
|
|
} |
230
|
|
|
|
231
|
|
|
return false; |
232
|
|
|
} |
233
|
|
|
|
234
|
|
|
protected function talkWithBot(): bool |
235
|
|
|
{ |
236
|
|
|
if ($this instanceof TalkBotConfig) { |
237
|
|
|
try { |
238
|
|
|
return $this->botTalk(); |
239
|
|
|
} catch (Throwable) { |
240
|
|
|
// do nothing |
241
|
|
|
} |
242
|
|
|
} |
243
|
|
|
|
244
|
|
|
return false; |
245
|
|
|
} |
246
|
4 |
|
|
247
|
|
|
/** |
248
|
4 |
|
* Is there a new message on the discussion page of the bot (or owner) ? |
249
|
4 |
|
* @throws ConfigException |
250
|
|
|
*/ |
251
|
4 |
|
public function checkWatchPages() |
252
|
2 |
|
{ |
253
|
|
|
foreach ($this->getWatchPages() as $title => $lastTime) { |
254
|
|
|
$pageTime = $this->getTimestamp($title); |
255
|
2 |
|
|
256
|
|
|
// the page has been edited since last check ? |
257
|
|
|
if (!$pageTime || $pageTime !== $lastTime) { |
258
|
|
|
echo sprintf("WATCHPAGE '%s' has been edited since %s.\n", $title, $lastTime); |
259
|
|
|
|
260
|
|
|
// Ask? Mettre à jour $watchPages ? |
261
|
|
|
echo "Replace with $title => '$pageTime'"; |
262
|
|
|
|
263
|
|
|
$this->checkExitOnWatchPage(); |
264
|
|
|
} |
265
|
6 |
|
} |
266
|
|
|
} |
267
|
6 |
|
|
268
|
5 |
|
/** |
269
|
6 |
|
* @throws ConfigException |
270
|
|
|
*/ |
271
|
4 |
|
protected function getWatchPages(): array |
272
|
|
|
{ |
273
|
|
|
if (!file_exists(static::WATCHPAGE_FILENAME)) { |
274
|
2 |
|
throw new ConfigException('No watchpage file found.'); |
275
|
|
|
} |
276
|
|
|
|
277
|
|
|
try { |
278
|
|
|
$json = file_get_contents(static::WATCHPAGE_FILENAME); |
279
|
|
|
$array = json_decode($json, true, 512, JSON_THROW_ON_ERROR); |
280
|
|
|
} catch (Throwable) { |
281
|
|
|
throw new ConfigException('Watchpage file malformed.'); |
282
|
|
|
} |
283
|
|
|
|
284
|
|
|
return $array; |
285
|
|
|
} |
286
|
|
|
|
287
|
|
|
protected function getTimestamp(string $title): ?string |
288
|
|
|
{ |
289
|
|
|
$page = new WikiPageAction($this->mediawikiFactory, $title); |
290
|
|
|
|
291
|
|
|
return $page->page->getRevisions()->getLatest()->getTimestamp(); |
292
|
|
|
} |
293
|
|
|
|
294
|
|
|
protected function checkExitOnWatchPage(): void |
295
|
|
|
{ |
296
|
|
|
if (static::EXIT_ON_CHECK_WATCHPAGE) { |
297
|
|
|
echo "EXIT_ON_CHECK_WATCHPAGE\n"; |
298
|
|
|
|
299
|
|
|
throw new DomainException('exit from check watchpages'); |
300
|
|
|
} |
301
|
|
|
} |
302
|
|
|
|
303
|
|
|
/** |
304
|
|
|
* How many minutes since last edit ? Do not to disturb human editors ! |
305
|
|
|
* @return int minutes |
306
|
|
|
*/ |
307
|
|
|
public function minutesSinceLastEdit(string $title): int |
308
|
|
|
{ |
309
|
|
|
$time = $this->getTimestamp($title); // 2011-09-02T16:31:13Z |
310
|
|
|
|
311
|
|
|
return (int)round((time() - strtotime($time)) / 60); |
312
|
|
|
} |
313
|
|
|
} |
314
|
|
|
|
In general, usage of exit should be done with care and only when running in a scripting context like a CLI script.