1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* An RFX object contains the parsed information for an RFX |
4
|
|
|
*/ |
5
|
|
|
|
6
|
|
|
declare(strict_types = 1); |
7
|
|
|
|
8
|
|
|
namespace App\Model; |
9
|
|
|
|
10
|
|
|
/** |
11
|
|
|
* This class contains information about a single RfX page. |
12
|
|
|
* @codeCoverageIgnore |
13
|
|
|
*/ |
14
|
|
|
class RFX extends Model |
15
|
|
|
{ |
16
|
|
|
/** @var array Data we parsed out of the page text */ |
17
|
|
|
private $data; |
18
|
|
|
|
19
|
|
|
/** @var array Duplicate voters */ |
20
|
|
|
private $duplicates; |
21
|
|
|
|
22
|
|
|
/** @var null|string Username of the user we're looking for. */ |
23
|
|
|
private $userLookingFor; |
24
|
|
|
|
25
|
|
|
/** @var string Section we found the user we're looking for */ |
26
|
|
|
private $userSectionFound; |
27
|
|
|
|
28
|
|
|
/** |
29
|
|
|
* Attempts to find a signature in $input using the default regex. |
30
|
|
|
* Returns matches. |
31
|
|
|
* |
32
|
|
|
* @param string $input The line we're looking for |
33
|
|
|
* @param array $matches Pointer to an array where we stash results |
34
|
|
|
* |
35
|
|
|
* @TODO: Make this cleaner |
36
|
|
|
* |
37
|
|
|
* @return int |
38
|
|
|
*/ |
39
|
|
|
protected function findSig(string $input, array &$matches): int |
40
|
|
|
{ |
41
|
|
|
//Supports User: and User talk: wikilinks, {{fullurl}}, |
42
|
|
|
// unsubstituted {{unsigned}}, unsubstituted {{unsigned2}}, |
43
|
|
|
// anything that looks like a custom sig template |
44
|
|
|
// TODO: Cross-wiki this sucker |
45
|
|
|
$regexp |
46
|
|
|
= //1: Normal [[User:XX]] and [[User talk:XX]] |
47
|
|
|
"/\[\[[Uu]ser(?:[\s_][Tt]alk)?\:([^\]\|\/]*)(?:\|[^\]]*)?\]\]" |
48
|
|
|
//2: {{fullurl}} and {{unsigned}} templates |
49
|
|
|
. "|\{\{(?:[Ff]ullurl\:[Uu]ser(?:[\s_][Tt]alk)?\:|" |
50
|
|
|
. "[Uu]nsigned\|)([^\}\|]*)(?:|[\|\}]*)?\}\}" |
51
|
|
|
//3: {{User:XX/sig}} templates |
52
|
|
|
. "|(?:\{\{)[Uu]ser(?:[\s_][Tt]alk)?\:([^\}\/\|]*)" |
53
|
|
|
//4: {{unsigned2|Date|XX}} templates |
54
|
|
|
. "|\{\{[Uu]nsigned2\|[^\|]*\|([^\}]*)\}\}" |
55
|
|
|
//5: [[User:XX/sig]] links (compromise measure) |
56
|
|
|
. "|(?:\[\[)[Uu]ser\:([^\]\/\|]*)\/[Ss]ig[\|\]]/"; |
57
|
|
|
|
58
|
|
|
return preg_match_all( |
59
|
|
|
$regexp, |
60
|
|
|
$input, |
61
|
|
|
$matches, |
62
|
|
|
PREG_OFFSET_CAPTURE |
63
|
|
|
); |
64
|
|
|
} |
65
|
|
|
|
66
|
|
|
/** |
67
|
|
|
* This function parses the wikitext and stores it within this function. |
68
|
|
|
* It's been split out to make this class testable |
69
|
|
|
* |
70
|
|
|
* @param array $sectionArray Section names that we're looking for |
71
|
|
|
* @param string $rawWikiText The text of the page we're parsing |
72
|
|
|
* @param string $dateRegexp Valid Regular Expression for the end date |
73
|
|
|
*/ |
74
|
|
|
private function setUp(array $sectionArray, string $rawWikiText, string $dateRegexp): void |
75
|
|
|
{ |
76
|
|
|
$this->data = []; |
77
|
|
|
|
78
|
|
|
$lines = explode("\n", $rawWikiText); |
79
|
|
|
|
80
|
|
|
$keys = join("|", $sectionArray); |
81
|
|
|
|
82
|
|
|
$lastSection = ""; |
83
|
|
|
|
84
|
|
|
foreach ($lines as $line) { |
85
|
|
|
if (preg_match("/={1,6}\s?($keys)\s?={1,6}/i", $line, $matches)) { |
86
|
|
|
$lastSection = strtolower($matches[1]); |
87
|
|
|
} elseif ("" == $lastSection |
88
|
|
|
&& preg_match( |
89
|
|
|
"/$dateRegexp/i", |
90
|
|
|
$line, |
91
|
|
|
$matches |
92
|
|
|
) |
93
|
|
|
) { |
94
|
|
|
$this->end = $matches[1]; |
95
|
|
|
} elseif ("" != $lastSection |
96
|
|
|
&& 0 === preg_match("/^\s*#?:.*/i", $line) |
97
|
|
|
) { |
98
|
|
|
$this->findSig($line, $matches); |
99
|
|
|
if (!isset($matches[1][0])) { |
100
|
|
|
continue; |
101
|
|
|
} |
102
|
|
|
$foundUser = trim($matches[1][0][0]); |
103
|
|
|
$this->data[$lastSection][] = $foundUser; |
104
|
|
|
if (strtolower($foundUser) === strtolower($this->userLookingFor)) { |
|
|
|
|
105
|
|
|
$this->userSectionFound = $lastSection; |
106
|
|
|
} |
107
|
|
|
} |
108
|
|
|
} |
109
|
|
|
|
110
|
|
|
$final = []; // initialize the final array |
111
|
|
|
$finalRaw = []; // Initialize the raw data array |
112
|
|
|
|
113
|
|
|
foreach (array_keys($this->data) as $key) { |
114
|
|
|
$finalRaw = array_merge($finalRaw, $this->data[$key]); |
115
|
|
|
} |
116
|
|
|
|
117
|
|
|
foreach ($finalRaw as $foundUsername) { |
118
|
|
|
$final[] = $foundUsername; // group all array's elements |
119
|
|
|
} |
120
|
|
|
|
121
|
|
|
$final = array_count_values($final); // find repetition and its count |
122
|
|
|
|
123
|
|
|
$final = array_diff($final, [1]); // remove single occurrences |
124
|
|
|
|
125
|
|
|
$this->duplicates = array_keys($final); |
126
|
|
|
} |
127
|
|
|
|
128
|
|
|
/** |
129
|
|
|
* RFX constructor. |
130
|
|
|
* |
131
|
|
|
* @param string $rawWikiText The text of the page we're parsing |
132
|
|
|
* @param array $sectionArray Section names that we're looking for |
133
|
|
|
* @param string $userNamespace Plain text of the user namespace |
134
|
|
|
* @param string $dateRegexp Valid Regular Expression for the end date |
135
|
|
|
* @param string|null $userLookingFor User we're trying to find. |
136
|
|
|
*/ |
137
|
|
|
public function __construct( |
138
|
|
|
string $rawWikiText, |
139
|
|
|
array $sectionArray = ["Support", "Oppose", "Neutral"], |
140
|
|
|
string $userNamespace = "User", |
|
|
|
|
141
|
|
|
string $dateRegexp = "final .*end(?:ing|ed)?(?: no earlier than)? (.*?)? \(UTC\)", |
142
|
|
|
?string $userLookingFor = null |
143
|
|
|
) { |
144
|
|
|
$this->userLookingFor = $userLookingFor; |
145
|
|
|
|
146
|
|
|
$this->setUp($sectionArray, $rawWikiText, $dateRegexp); |
147
|
|
|
} |
148
|
|
|
|
149
|
|
|
/** |
150
|
|
|
* Which section we found the user we're looking for. |
151
|
|
|
* |
152
|
|
|
* @return string |
153
|
|
|
*/ |
154
|
|
|
public function getUserSectionFound(): string |
155
|
|
|
{ |
156
|
|
|
return $this->userSectionFound; |
157
|
|
|
} |
158
|
|
|
|
159
|
|
|
/** |
160
|
|
|
* Returns data on the given section name. |
161
|
|
|
* |
162
|
|
|
* @param string $sectionName The section we're looking at |
163
|
|
|
* |
164
|
|
|
* @return array |
165
|
|
|
*/ |
166
|
|
|
public function getSection(string $sectionName): array |
167
|
|
|
{ |
168
|
|
|
$sectionName = strtolower($sectionName); |
169
|
|
|
if (!isset($this->data[$sectionName])) { |
170
|
|
|
return []; |
171
|
|
|
} else { |
172
|
|
|
return $this->data[$sectionName]; |
173
|
|
|
} |
174
|
|
|
} |
175
|
|
|
|
176
|
|
|
/** |
177
|
|
|
* Get an array of duplicate votes. |
178
|
|
|
* |
179
|
|
|
* @return array |
180
|
|
|
*/ |
181
|
|
|
public function getDuplicates(): array |
182
|
|
|
{ |
183
|
|
|
return $this->duplicates; |
184
|
|
|
} |
185
|
|
|
} |
186
|
|
|
|