|
1
|
|
|
<?php |
|
2
|
|
|
declare(strict_types = 1); |
|
3
|
|
|
|
|
4
|
|
|
namespace AppBundle\Model; |
|
5
|
|
|
|
|
6
|
|
|
/** |
|
7
|
|
|
* A Blame will search the given page for the given text and return the relevant revisions and authors. |
|
8
|
|
|
*/ |
|
9
|
|
|
class Blame extends Authorship |
|
10
|
|
|
{ |
|
11
|
|
|
/** @var string Text to search for. */ |
|
12
|
|
|
protected $query; |
|
13
|
|
|
|
|
14
|
|
|
/** @var array|null Matches, keyed by revision ID, each with keys 'edit' <Edit> and 'tokens' <string[]>. */ |
|
15
|
|
|
protected $matches; |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* Blame constructor. |
|
19
|
|
|
* @param Page $page The page to process. |
|
20
|
|
|
* @param string $query Text to search for. |
|
21
|
|
|
* @param string|null $target Either a revision ID or date in YYYY-MM-DD format. Null to use latest revision. |
|
22
|
|
|
*/ |
|
23
|
2 |
|
public function __construct(Page $page, string $query, ?string $target = null) |
|
24
|
|
|
{ |
|
25
|
2 |
|
parent::__construct($page, $target); |
|
26
|
|
|
|
|
27
|
2 |
|
$this->query = $query; |
|
28
|
2 |
|
} |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* Get the search query. |
|
32
|
|
|
* @return string |
|
33
|
|
|
*/ |
|
34
|
1 |
|
public function getQuery(): string |
|
35
|
|
|
{ |
|
36
|
1 |
|
return $this->query; |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
/** |
|
40
|
|
|
* Matches, keyed by revision ID, each with keys 'edit' <Edit> and 'tokens' <string[]>. |
|
41
|
|
|
* @return array|null |
|
42
|
|
|
*/ |
|
43
|
1 |
|
public function getMatches(): ?array |
|
44
|
|
|
{ |
|
45
|
1 |
|
return $this->matches; |
|
46
|
|
|
} |
|
47
|
|
|
|
|
48
|
|
|
/** |
|
49
|
|
|
* Strip out spaces, since they are not accounted for in the WikiWho API. |
|
50
|
|
|
* @return string |
|
51
|
|
|
*/ |
|
52
|
2 |
|
public function getTokenizedQuery(): string |
|
53
|
|
|
{ |
|
54
|
2 |
|
return strtolower(preg_replace('/\s*/m', '', $this->query)); |
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
|
|
/** |
|
58
|
|
|
* Get authorship attribution from the WikiWho API. |
|
59
|
|
|
* @see https://f-squared.org/wikiwho/ |
|
60
|
|
|
*/ |
|
61
|
1 |
|
public function prepareData(): void |
|
62
|
|
|
{ |
|
63
|
1 |
|
if (isset($this->matches)) { |
|
64
|
|
|
return; |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
// Set revision data. self::setRevisionData() returns null if there are errors. |
|
68
|
1 |
|
$revisionData = $this->getRevisionData(true); |
|
69
|
1 |
|
if (null === $revisionData) { |
|
70
|
|
|
return; |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
1 |
|
$matches = $this->searchTokens($revisionData['tokens']); |
|
74
|
|
|
|
|
75
|
|
|
// We want the results grouped by editor and revision ID. |
|
76
|
1 |
|
$this->matches = []; |
|
77
|
1 |
|
foreach ($matches as $match) { |
|
78
|
1 |
|
if (isset($this->matches[$match['id']])) { |
|
79
|
1 |
|
$this->matches[$match['id']]['tokens'][] = $match['token']; |
|
80
|
1 |
|
continue; |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
1 |
|
$edit = $this->getRepository()->getEditFromRevId($this->page, $match['id']); |
|
|
|
|
|
|
84
|
1 |
|
$this->matches[$match['id']] = [ |
|
85
|
1 |
|
'edit' => $edit, |
|
86
|
1 |
|
'tokens' => [$match['token']], |
|
87
|
|
|
]; |
|
88
|
|
|
} |
|
89
|
1 |
|
} |
|
90
|
|
|
|
|
91
|
|
|
/** |
|
92
|
|
|
* @param array $tokens |
|
93
|
|
|
* @return array |
|
94
|
|
|
*/ |
|
95
|
1 |
|
private function searchTokens(array $tokens): array |
|
96
|
|
|
{ |
|
97
|
1 |
|
$matchData = []; |
|
98
|
1 |
|
$matchDataSoFar = []; |
|
99
|
1 |
|
$matchSoFar = ''; |
|
100
|
|
|
|
|
101
|
1 |
|
foreach ($tokens as $token) { |
|
102
|
1 |
|
if (0 === strpos($this->getTokenizedQuery(), $matchSoFar.$token['str'])) { |
|
103
|
1 |
|
$matchSoFar .= $token['str']; |
|
104
|
1 |
|
$matchDataSoFar[] = [ |
|
105
|
1 |
|
'id' => $token['o_rev_id'], |
|
106
|
1 |
|
'editor' => $token['editor'], |
|
107
|
1 |
|
'token' => $token['str'], |
|
108
|
|
|
]; |
|
109
|
1 |
|
} elseif (!empty($matchSoFar)) { |
|
110
|
|
|
$matchDataSoFar = []; |
|
111
|
|
|
$matchSoFar = ''; |
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
1 |
|
if ($matchSoFar === $this->getTokenizedQuery()) { |
|
115
|
1 |
|
$matchData = array_merge($matchData, $matchDataSoFar); |
|
116
|
1 |
|
$matchDataSoFar = []; |
|
117
|
1 |
|
$matchSoFar = ''; |
|
118
|
|
|
} |
|
119
|
|
|
} |
|
120
|
|
|
|
|
121
|
|
|
// Full matches usually come last, but are the most relevant. |
|
122
|
1 |
|
return array_reverse($matchData); |
|
123
|
|
|
} |
|
124
|
|
|
} |
|
125
|
|
|
|
x-tools /
xtools
Loading history...