|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Ttskch\Esa; |
|
4
|
|
|
|
|
5
|
|
|
use Symfony\Component\DomCrawler\Crawler; |
|
6
|
|
|
use Symfony\Component\Routing\Generator\UrlGeneratorInterface; |
|
7
|
|
|
|
|
8
|
|
|
class HtmlHandler |
|
9
|
|
|
{ |
|
10
|
|
|
/** |
|
11
|
|
|
* @var Crawler |
|
12
|
|
|
*/ |
|
13
|
|
|
private $crawler; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* @var UrlGeneratorInterface |
|
17
|
|
|
*/ |
|
18
|
|
|
private $urlGenerator; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* @var EmojiManager |
|
22
|
|
|
*/ |
|
23
|
|
|
private $emojiManager; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* @var string |
|
27
|
|
|
*/ |
|
28
|
|
|
private $teamName; |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* @param array $replacements |
|
32
|
|
|
*/ |
|
33
|
|
|
public function __construct(Crawler $crawler, UrlGeneratorInterface $urlGenerator, EmojiManager $emojiManager, $teamName) |
|
34
|
|
|
{ |
|
35
|
|
|
$this->crawler = $crawler; |
|
36
|
|
|
$this->urlGenerator = $urlGenerator; |
|
37
|
|
|
$this->emojiManager = $emojiManager; |
|
38
|
|
|
$this->teamName = $teamName; |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* @param string $html |
|
43
|
|
|
* @return $this |
|
44
|
|
|
*/ |
|
45
|
|
|
public function initialize($html) |
|
46
|
|
|
{ |
|
47
|
|
|
$this->crawler->clear(); |
|
48
|
|
|
$this->crawler->addHtmlContent($html); |
|
49
|
|
|
|
|
50
|
|
|
return $this; |
|
51
|
|
|
} |
|
52
|
|
|
|
|
53
|
|
|
/** |
|
54
|
|
|
* @return string |
|
55
|
|
|
*/ |
|
56
|
|
|
public function dumpHtml() |
|
57
|
|
|
{ |
|
58
|
|
|
$this->ensureInitialized(); |
|
59
|
|
|
|
|
60
|
|
|
return $this->crawler->html(); |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
/** |
|
64
|
|
|
* Replace links to other post with links to see the post on esaba. |
|
65
|
|
|
* |
|
66
|
|
|
* @param string $routeName |
|
67
|
|
|
* @param string $routeVariableName |
|
68
|
|
|
*/ |
|
69
|
|
|
public function replacePostUrls($routeName, $routeVariableName) |
|
70
|
|
|
{ |
|
71
|
|
|
$backReferenceNumberForPostId = null; |
|
72
|
|
|
$pattern = $this->getPostUrlPattern($backReferenceNumberForPostId); |
|
73
|
|
|
$walker = $this->getATagWalkerForPostUrls($pattern, $backReferenceNumberForPostId, $routeName, $routeVariableName); |
|
74
|
|
|
|
|
75
|
|
|
$this->replaceATagWithWalker($pattern, $walker); |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
|
|
/** |
|
79
|
|
|
* Disable @mention links. |
|
80
|
|
|
*/ |
|
81
|
|
|
public function disableMentionLinks() |
|
82
|
|
|
{ |
|
83
|
|
|
$pattern = $this->getMentionLinkPattern(); |
|
84
|
|
|
$walker = $this->getATagWalkerForMentionLinks($pattern); |
|
85
|
|
|
|
|
86
|
|
|
$this->replaceATagWithWalker($pattern, $walker); |
|
87
|
|
|
} |
|
88
|
|
|
|
|
89
|
|
|
/** |
|
90
|
|
|
* Replace <a> tag href values for specified regexp pattern with closure returns map of ['pattern' => regexp pattern, 'replacement' => replacement]. |
|
91
|
|
|
* |
|
92
|
|
|
* @param string $pattern |
|
93
|
|
|
* @param \Closure $walker |
|
94
|
|
|
*/ |
|
95
|
|
|
public function replaceATagWithWalker($pattern, \Closure $walker) |
|
96
|
|
|
{ |
|
97
|
|
|
$this->ensureInitialized(); |
|
98
|
|
|
|
|
99
|
|
|
$targetATags = $this->crawler->filter('a')->reduce($this->getATagReducer($pattern)); |
|
100
|
|
|
$replacements = $targetATags->each($walker); |
|
101
|
|
|
$replacements = array_combine(array_column($replacements, 'pattern'), array_column($replacements, 'replacement')); |
|
102
|
|
|
|
|
103
|
|
|
$this->replaceHtml($replacements); |
|
104
|
|
|
} |
|
105
|
|
|
|
|
106
|
|
|
/** |
|
107
|
|
|
* @param string $backReferenceNumberForPostId For returning position of post id in regexp pattern. |
|
108
|
|
|
* @return string |
|
109
|
|
|
*/ |
|
110
|
|
|
public function getPostUrlPattern(&$backReferenceNumberForPostId) |
|
111
|
|
|
{ |
|
112
|
|
|
$backReferenceNumberForPostId = 3; |
|
113
|
|
|
|
|
114
|
|
|
return sprintf('#^((https?:)?//%s\.esa\.io)?/posts/(\d+)(/|/edit/?)?$#', $this->teamName); |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
/** |
|
118
|
|
|
* @return string |
|
119
|
|
|
*/ |
|
120
|
|
|
public function getMentionLinkPattern() |
|
121
|
|
|
{ |
|
122
|
|
|
return '#/members/([^\'"]+)#'; |
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
|
|
/** |
|
126
|
|
|
* Return closure reduces ATags Crawler with regexp pattern for href value. |
|
127
|
|
|
* |
|
128
|
|
|
* @param string $pattern |
|
129
|
|
|
* @return \Closure |
|
130
|
|
|
*/ |
|
131
|
|
|
public function getATagReducer($pattern) |
|
132
|
|
|
{ |
|
133
|
|
|
$reducer = function (Crawler $node) use ($pattern) { |
|
134
|
|
|
preg_match($pattern, $node->attr('href'), $matches); |
|
135
|
|
|
|
|
136
|
|
|
return boolval($matches); |
|
137
|
|
|
}; |
|
138
|
|
|
|
|
139
|
|
|
return $reducer; |
|
140
|
|
|
} |
|
141
|
|
|
|
|
142
|
|
|
/** |
|
143
|
|
|
* Return closure returns map of ['pattern' => regexp pattern, 'replacement' => replacement] for href value of post urls. |
|
144
|
|
|
* |
|
145
|
|
|
* @param string $pattern |
|
146
|
|
|
* @param int $backReferenceNumberForPostId |
|
147
|
|
|
* @param string $routeName |
|
148
|
|
|
* @param string $routeVariableName |
|
149
|
|
|
* @return \Closure |
|
150
|
|
|
*/ |
|
151
|
|
|
public function getATagWalkerForPostUrls($pattern, $backReferenceNumberForPostId, $routeName, $routeVariableName) |
|
152
|
|
|
{ |
|
153
|
|
|
$that = $this; |
|
154
|
|
|
|
|
155
|
|
|
$walker = function (Crawler $node) use ($pattern, $backReferenceNumberForPostId, $routeName, $routeVariableName, $that) { |
|
156
|
|
|
preg_match($pattern, $node->attr('href'), $matches); |
|
157
|
|
|
$href = $matches[0]; |
|
158
|
|
|
$postId = $matches[$backReferenceNumberForPostId]; |
|
159
|
|
|
|
|
160
|
|
|
$pattern = sprintf('/href=(\'|")%s\1/', str_replace('/', '\/', $href)); |
|
161
|
|
|
$replacement = sprintf('href="%s"', $that->urlGenerator->generate($routeName, [$routeVariableName => $postId])); |
|
162
|
|
|
|
|
163
|
|
|
return [ |
|
164
|
|
|
'pattern' => $pattern, |
|
165
|
|
|
'replacement' => $replacement, |
|
166
|
|
|
]; |
|
167
|
|
|
}; |
|
168
|
|
|
|
|
169
|
|
|
return $walker; |
|
170
|
|
|
} |
|
171
|
|
|
|
|
172
|
|
|
/** |
|
173
|
|
|
* Return closure returns map of ['pattern' => regexp pattern, 'replacement' => replacement] for href value of mention links. |
|
174
|
|
|
* |
|
175
|
|
|
* @param string $pattern |
|
176
|
|
|
* @return \Closure |
|
177
|
|
|
*/ |
|
178
|
|
|
public function getATagWalkerForMentionLinks($pattern) |
|
179
|
|
|
{ |
|
180
|
|
|
$walker = function (Crawler $node) use ($pattern) { |
|
181
|
|
|
preg_match($pattern, $node->attr('href'), $matches); |
|
182
|
|
|
$href = $matches[0]; |
|
183
|
|
|
|
|
184
|
|
|
$pattern = sprintf('/href=(\'|")%s\1/', str_replace('/', '\/', $href)); |
|
185
|
|
|
$replacement = ''; |
|
186
|
|
|
|
|
187
|
|
|
return [ |
|
188
|
|
|
'pattern' => $pattern, |
|
189
|
|
|
'replacement' => $replacement, |
|
190
|
|
|
]; |
|
191
|
|
|
}; |
|
192
|
|
|
|
|
193
|
|
|
return $walker; |
|
194
|
|
|
} |
|
195
|
|
|
|
|
196
|
|
|
/** |
|
197
|
|
|
* Replace emoji codes with img tags. |
|
198
|
|
|
*/ |
|
199
|
|
|
public function replaceEmojiCodes() |
|
200
|
|
|
{ |
|
201
|
|
|
$this->ensureInitialized(); |
|
202
|
|
|
|
|
203
|
|
|
$html = $this->crawler->html(); |
|
204
|
|
|
|
|
205
|
|
|
preg_match_all('/:(\w+):/', $html, $matches); |
|
206
|
|
|
|
|
207
|
|
|
for ($i = 0; $i < count($matches[0]); $i++) { |
|
|
|
|
|
|
208
|
|
|
$code = $matches[0][$i]; |
|
209
|
|
|
$replacement = sprintf('<img src="%s" class="emoji" title="%s" alt="%s">', $this->emojiManager->getImageUrl($matches[1][$i]), $code, $code); |
|
210
|
|
|
|
|
211
|
|
|
$html = str_replace($code, $replacement, $html); |
|
212
|
|
|
} |
|
213
|
|
|
|
|
214
|
|
|
$this->initialize($html); |
|
215
|
|
|
} |
|
216
|
|
|
|
|
217
|
|
|
/** |
|
218
|
|
|
* @param array $replacements map of [regexp pattern => replacement]. |
|
219
|
|
|
*/ |
|
220
|
|
|
public function replaceHtml(array $replacements) |
|
221
|
|
|
{ |
|
222
|
|
|
$this->ensureInitialized(); |
|
223
|
|
|
|
|
224
|
|
|
$html = $this->crawler->html(); |
|
225
|
|
|
|
|
226
|
|
|
foreach ($replacements as $pattern => $replacement) { |
|
227
|
|
|
$html = preg_replace($pattern, $replacement, $html); |
|
228
|
|
|
} |
|
229
|
|
|
|
|
230
|
|
|
$this->initialize($html); |
|
231
|
|
|
} |
|
232
|
|
|
|
|
233
|
|
|
/** |
|
234
|
|
|
* Return map of ['id' => id, 'text' => text] of headings as TOC. |
|
235
|
|
|
* |
|
236
|
|
|
* @return array |
|
237
|
|
|
*/ |
|
238
|
|
|
public function getToc() |
|
239
|
|
|
{ |
|
240
|
|
|
$this->ensureInitialized(); |
|
241
|
|
|
|
|
242
|
|
|
$toc = $this->crawler->filter('h1, h2, h3')->each($this->getWalkerForToc()); |
|
243
|
|
|
|
|
244
|
|
|
return $toc; |
|
245
|
|
|
} |
|
246
|
|
|
|
|
247
|
|
|
/** |
|
248
|
|
|
* Return closure returns map of ['id' => id, 'text' => text] of h tags. |
|
249
|
|
|
* |
|
250
|
|
|
* @return \Closure |
|
251
|
|
|
*/ |
|
252
|
|
|
public function getWalkerForToc() |
|
253
|
|
|
{ |
|
254
|
|
|
$walker = function (Crawler $node) { |
|
255
|
|
|
return [ |
|
256
|
|
|
'id' => $node->attr('id'), |
|
257
|
|
|
// 'text' => $node->text(), |
|
258
|
|
|
'text' => preg_replace('/^\s*>\s*/', '', $node->text()), // workaround... |
|
259
|
|
|
]; |
|
260
|
|
|
}; |
|
261
|
|
|
|
|
262
|
|
|
return $walker; |
|
263
|
|
|
} |
|
264
|
|
|
|
|
265
|
|
|
private function ensureInitialized() |
|
266
|
|
|
{ |
|
267
|
|
|
if (!$this->crawler->count()) { |
|
268
|
|
|
throw new \LogicException('Initialize before using.'); |
|
269
|
|
|
} |
|
270
|
|
|
} |
|
271
|
|
|
} |
|
272
|
|
|
|
If the size of the collection does not change during the iteration, it is generally a good practice to compute it beforehand, and not on each iteration: