|
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
|
|
|
|
|
11
|
|
|
namespace App\Application; |
|
12
|
|
|
|
|
13
|
|
|
|
|
14
|
|
|
use App\Domain\Utils\WikiTextUtil; |
|
15
|
|
|
use Codedungeon\PHPCliColors\Color; |
|
16
|
|
|
use Exception; |
|
17
|
|
|
|
|
18
|
|
|
abstract class RefBotWorker extends AbstractBotTaskWorker |
|
19
|
|
|
{ |
|
20
|
|
|
public const TASK_BOT_FLAG = false; |
|
21
|
|
|
public const MAX_REFS_PROCESSED_IN_ARTICLE = 30; |
|
22
|
|
|
|
|
23
|
|
|
protected $warning = false; |
|
24
|
|
|
|
|
25
|
|
|
public function hasWarning(): bool |
|
26
|
|
|
{ |
|
27
|
|
|
return (bool)$this->warning; |
|
28
|
|
|
} |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* @inheritDoc |
|
32
|
|
|
* @throws Exception |
|
33
|
|
|
*/ |
|
34
|
|
|
protected function processWithDomainWorker(string $title, string $text): ?string |
|
35
|
|
|
{ |
|
36
|
|
|
return $this->processText($text); |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
/** |
|
40
|
|
|
* @param $text |
|
41
|
|
|
* |
|
42
|
|
|
* @return string|string[] |
|
43
|
|
|
* @throws Exception |
|
44
|
|
|
*/ |
|
45
|
|
|
public function processText($text) |
|
46
|
|
|
{ |
|
47
|
|
|
$refs = WikiTextUtil::extractRefsAndListOfLinks($text); |
|
48
|
|
|
if ($refs === []) { |
|
49
|
|
|
$this->log->debug('empty extractRefsAndListOfLinks'); |
|
50
|
|
|
|
|
51
|
|
|
return $text; |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
// Avoid memory leak problem : bot limited to N refs in an article |
|
55
|
|
|
$refs = array_slice($refs, 0, self::MAX_REFS_PROCESSED_IN_ARTICLE, true); |
|
56
|
|
|
|
|
57
|
|
|
foreach ($refs as $ref) { |
|
58
|
|
|
$refContent = WikiTextUtil::stripFinalPoint(trim($ref[1])); |
|
59
|
|
|
|
|
60
|
|
|
$newRefContent = $this->processRefContent($refContent); |
|
61
|
|
|
|
|
62
|
|
|
$text = $this->replaceRefInText($ref, $newRefContent, $text); |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
return $text; |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
public abstract function processRefContent($refContent): string; |
|
69
|
|
|
|
|
70
|
|
|
protected function replaceRefInText(array $ref, string $replace, string $text) |
|
71
|
|
|
{ |
|
72
|
|
|
// Pas de changement |
|
73
|
|
|
if (WikiTextUtil::stripFinalPoint(trim($replace)) === WikiTextUtil::stripFinalPoint(trim($ref[1]))) { |
|
74
|
|
|
return $text; |
|
75
|
|
|
} |
|
76
|
|
|
$replace = $this->addFinalPeriod($ref[0], $replace); |
|
77
|
|
|
$result = str_replace($ref[1], $replace, $ref[0]); |
|
78
|
|
|
$this->printDiff($ref[0], $result); |
|
79
|
|
|
|
|
80
|
|
|
return str_replace($ref[0], $result, $text); |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
/** |
|
84
|
|
|
* Add a final period '.' before an eventual '</ref>' |
|
85
|
|
|
*/ |
|
86
|
|
|
protected function addFinalPeriod($ref, string $replace): string |
|
87
|
|
|
{ |
|
88
|
|
|
if (preg_match('#</ref>#', $ref)) { |
|
89
|
|
|
$replace .= '.'; |
|
90
|
|
|
} |
|
91
|
|
|
return $replace; |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
protected function printDiff(string $before, string $after, string $level = 'debug'): void |
|
95
|
|
|
{ |
|
96
|
|
|
$this->log->log($level, sprintf("%s--%s %s\n", Color::BG_LIGHT_RED, Color::NORMAL, $before)); |
|
97
|
|
|
$this->log->log($level, sprintf("%s--%s %s\n", Color::BG_LIGHT_GREEN, Color::NORMAL, $after)); |
|
98
|
|
|
} |
|
99
|
|
|
} |
|
100
|
|
|
|
Dispositif /
Wikibot
