1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* @package s9e\TextFormatter |
5
|
|
|
* @copyright Copyright (c) 2010-2018 The s9e Authors |
6
|
|
|
* @license http://www.opensource.org/licenses/mit-license.php The MIT License |
7
|
|
|
*/ |
8
|
|
|
namespace s9e\TextFormatter\Plugins\MediaEmbed; |
9
|
|
|
|
10
|
|
|
use InvalidArgumentException; |
11
|
|
|
use RuntimeException; |
12
|
|
|
use s9e\TextFormatter\Configurator\Items\Regexp; |
13
|
|
|
use s9e\TextFormatter\Configurator\Items\Tag; |
14
|
|
|
use s9e\TextFormatter\Configurator\JavaScript\Dictionary; |
15
|
|
|
use s9e\TextFormatter\Plugins\ConfiguratorBase; |
16
|
|
|
use s9e\TextFormatter\Plugins\MediaEmbed\Configurator\Collections\CachedDefinitionCollection; |
17
|
|
|
use s9e\TextFormatter\Plugins\MediaEmbed\Configurator\TemplateBuilder; |
18
|
|
|
|
19
|
|
|
class Configurator extends ConfiguratorBase |
20
|
|
|
{ |
21
|
|
|
/** |
22
|
|
|
* @var array List of filters that are explicitly allowed in attribute definitions |
23
|
|
|
*/ |
24
|
|
|
public $allowedFilters = ['stripslashes', 'urldecode']; |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* @var bool Whether to create the MEDIA BBCode |
28
|
|
|
*/ |
29
|
|
|
protected $createMediaBBCode = true; |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* @var Configurator\Collections\SiteDefinitionCollection Default sites |
33
|
|
|
*/ |
34
|
|
|
public $defaultSites; |
35
|
|
|
|
36
|
|
|
/** |
37
|
|
|
* {@inheritdoc} |
38
|
|
|
*/ |
39
|
|
|
protected $quickMatch = '://'; |
40
|
|
|
|
41
|
|
|
/** |
42
|
|
|
* {@inheritdoc} |
43
|
|
|
*/ |
44
|
|
|
protected $regexp = '/\\bhttps?:\\/\\/[^["\'\\s]+/Si'; |
45
|
|
|
|
46
|
|
|
/** |
47
|
|
|
* @var array Configured sites |
48
|
|
|
*/ |
49
|
|
|
protected $sites = []; |
50
|
|
|
|
51
|
|
|
/** |
52
|
|
|
* @var string Name of the tag used to handle embeddable URLs |
53
|
|
|
*/ |
54
|
|
|
protected $tagName = 'MEDIA'; |
55
|
|
|
|
56
|
|
|
/** |
57
|
|
|
* @var TemplateBuilder |
58
|
|
|
*/ |
59
|
|
|
protected $templateBuilder; |
60
|
|
|
|
61
|
|
|
/** |
62
|
|
|
* {@inheritdoc} |
63
|
|
|
*/ |
64
|
24 |
|
protected function setUp() |
65
|
|
|
{ |
66
|
24 |
|
$this->defaultSites = new CachedDefinitionCollection; |
67
|
24 |
|
$this->templateBuilder = new TemplateBuilder; |
68
|
|
|
|
69
|
|
|
// Create a MEDIA tag |
70
|
24 |
|
$this->createMediaTag(); |
71
|
|
|
|
72
|
|
|
// Create a [MEDIA] BBCode if applicable |
73
|
24 |
|
if ($this->createMediaBBCode) |
74
|
|
|
{ |
75
|
23 |
|
$this->configurator->BBCodes->set($this->tagName, ['contentAttributes' => ['url']]); |
76
|
|
|
} |
77
|
24 |
|
} |
78
|
|
|
|
79
|
|
|
/** |
80
|
|
|
* {@inheritdoc} |
81
|
|
|
*/ |
82
|
5 |
|
public function asConfig() |
83
|
|
|
{ |
84
|
5 |
|
if (empty($this->sites)) |
85
|
|
|
{ |
86
|
1 |
|
return; |
87
|
|
|
} |
88
|
|
|
|
89
|
|
|
return [ |
90
|
4 |
|
'quickMatch' => $this->quickMatch, |
91
|
4 |
|
'regexp' => $this->regexp, |
92
|
4 |
|
'tagName' => $this->tagName |
93
|
|
|
]; |
94
|
|
|
} |
95
|
|
|
|
96
|
|
|
/** |
97
|
|
|
* Add a media site |
98
|
|
|
* |
99
|
|
|
* @param string $siteId Site's ID |
100
|
|
|
* @param array $siteConfig Site's config |
101
|
|
|
* @return Tag Tag created for this site |
102
|
|
|
*/ |
103
|
20 |
|
public function add($siteId, array $siteConfig = null) |
104
|
|
|
{ |
105
|
|
|
// Normalize or retrieve the site definition |
106
|
20 |
|
$siteId = $this->normalizeId($siteId); |
107
|
19 |
|
if (isset($siteConfig)) |
108
|
|
|
{ |
109
|
15 |
|
$siteConfig = $this->defaultSites->normalizeValue($siteConfig); |
110
|
|
|
} |
111
|
|
|
else |
112
|
|
|
{ |
113
|
4 |
|
$siteConfig = $this->defaultSites->get($siteId); |
114
|
|
|
} |
115
|
18 |
|
$siteConfig['extract'] = $this->convertRegexps($siteConfig['extract']); |
116
|
18 |
|
$siteConfig['scrape'] = $this->convertScrapes($siteConfig['scrape']); |
117
|
|
|
|
118
|
|
|
// Check the safety of attribute filters |
119
|
18 |
|
$this->checkAttributeFilters($siteConfig['attributes']); |
120
|
|
|
|
121
|
|
|
// Create the tag for this site |
122
|
17 |
|
$tag = new Tag([ |
123
|
17 |
|
'attributes' => $this->getAttributesConfig($siteConfig), |
124
|
|
|
'rules' => [ |
125
|
17 |
|
'allowChild' => 'URL', |
126
|
|
|
'autoClose' => true, |
127
|
17 |
|
'denyChild' => [$siteId, $this->tagName] |
128
|
|
|
], |
129
|
17 |
|
'template' => $this->templateBuilder->build($siteId, $siteConfig) |
130
|
|
|
]); |
131
|
|
|
|
132
|
17 |
|
$this->configurator->templateNormalizer->normalizeTag($tag); |
133
|
17 |
|
$this->configurator->templateChecker->checkTag($tag); |
134
|
16 |
|
$this->configurator->tags->add($siteId, $tag); |
135
|
16 |
|
$this->sites[$siteId] = $siteConfig; |
136
|
|
|
|
137
|
16 |
|
return $tag; |
138
|
|
|
} |
139
|
|
|
|
140
|
|
|
/** |
141
|
|
|
* {@inheritdoc} |
142
|
|
|
*/ |
143
|
2 |
|
public function finalize() |
144
|
|
|
{ |
145
|
2 |
|
$hosts = []; |
146
|
2 |
|
$sites = []; |
147
|
2 |
|
foreach ($this->sites as $siteId => $siteConfig) |
148
|
|
|
{ |
149
|
1 |
|
foreach ($siteConfig['host'] as $host) |
150
|
|
|
{ |
151
|
1 |
|
$hosts[$host] = $siteId; |
152
|
|
|
} |
153
|
1 |
|
$sites[$siteId] = [$siteConfig['extract'], $siteConfig['scrape']]; |
154
|
|
|
} |
155
|
|
|
|
156
|
2 |
|
$this->configurator->registeredVars['MediaEmbed.hosts'] = new Dictionary($hosts); |
157
|
2 |
|
$this->configurator->registeredVars['MediaEmbed.sites'] = new Dictionary($sites); |
158
|
2 |
|
} |
159
|
|
|
|
160
|
|
|
/** |
161
|
|
|
* Check the safety of given attributes |
162
|
|
|
* |
163
|
|
|
* @param array $attributes |
164
|
|
|
* @return void |
165
|
|
|
*/ |
166
|
18 |
|
protected function checkAttributeFilters(array $attributes) |
167
|
|
|
{ |
168
|
18 |
|
foreach ($attributes as $attrConfig) |
169
|
|
|
{ |
170
|
6 |
|
if (empty($attrConfig['filterChain'])) |
171
|
|
|
{ |
172
|
3 |
|
continue; |
173
|
|
|
} |
174
|
5 |
|
foreach ($attrConfig['filterChain'] as $filter) |
175
|
|
|
{ |
176
|
5 |
|
if (substr($filter, 0, 1) !== '#' && !in_array($filter, $this->allowedFilters, true)) |
177
|
|
|
{ |
178
|
5 |
|
throw new RuntimeException("Filter '$filter' is not allowed in media sites"); |
179
|
|
|
} |
180
|
|
|
} |
181
|
|
|
} |
182
|
17 |
|
} |
183
|
|
|
|
184
|
|
|
/** |
185
|
|
|
* Convert given regexp to a [regexp, map] pair |
186
|
|
|
* |
187
|
|
|
* @param string $regexp Original regexp |
188
|
|
|
* @return array [regexp, [list of captures' names]] |
189
|
|
|
*/ |
190
|
16 |
|
protected function convertRegexp($regexp) |
191
|
|
|
{ |
192
|
16 |
|
$regexp = new Regexp($regexp); |
193
|
|
|
|
194
|
16 |
|
return [$regexp, $regexp->getCaptureNames()]; |
195
|
|
|
} |
196
|
|
|
|
197
|
|
|
/** |
198
|
|
|
* Convert a list of regexps |
199
|
|
|
* |
200
|
|
|
* @param string[] $regexps Original list |
201
|
|
|
* @return array[] Converted list |
202
|
|
|
*/ |
203
|
18 |
|
protected function convertRegexps(array $regexps) |
204
|
|
|
{ |
205
|
18 |
|
return array_map([$this, 'convertRegexp'], $regexps); |
206
|
|
|
} |
207
|
|
|
|
208
|
|
|
/** |
209
|
|
|
* Convert all regexps in a scraping config |
210
|
|
|
* |
211
|
|
|
* @param array $config Original config |
212
|
|
|
* @return array Converted config |
213
|
|
|
*/ |
214
|
4 |
|
protected function convertScrapeConfig(array $config) |
215
|
|
|
{ |
216
|
4 |
|
$config['extract'] = $this->convertRegexps($config['extract']); |
217
|
4 |
|
$config['match'] = $this->convertRegexps($config['match']); |
218
|
|
|
|
219
|
4 |
|
return $config; |
220
|
|
|
} |
221
|
|
|
|
222
|
|
|
/** |
223
|
|
|
* Convert all regexps in a list of scraping configs |
224
|
|
|
* |
225
|
|
|
* @param array[] $scrapes Original config |
226
|
|
|
* @return array[] Converted config |
227
|
|
|
*/ |
228
|
18 |
|
protected function convertScrapes(array $scrapes) |
229
|
|
|
{ |
230
|
18 |
|
return array_map([$this, 'convertScrapeConfig'], $scrapes); |
231
|
|
|
} |
232
|
|
|
|
233
|
|
|
/** |
234
|
|
|
* Create the default MEDIA tag |
235
|
|
|
* |
236
|
|
|
* @return void |
237
|
|
|
*/ |
238
|
24 |
|
protected function createMediaTag() |
239
|
|
|
{ |
240
|
24 |
|
$tag = $this->configurator->tags->add($this->tagName); |
241
|
|
|
|
242
|
|
|
// This tag should not need to be closed and should not contain itself |
243
|
24 |
|
$tag->rules->autoClose(); |
244
|
24 |
|
$tag->rules->denyChild($this->tagName); |
245
|
|
|
|
246
|
|
|
// Empty this tag's filter chain and add our tag filter |
247
|
24 |
|
$tag->filterChain->clear(); |
248
|
24 |
|
$tag->filterChain |
249
|
24 |
|
->append(__NAMESPACE__ . '\\Parser::filterTag') |
250
|
24 |
|
->resetParameters() |
251
|
24 |
|
->addParameterByName('tag') |
252
|
24 |
|
->addParameterByName('parser') |
253
|
24 |
|
->addParameterByName('MediaEmbed.hosts') |
254
|
24 |
|
->addParameterByName('MediaEmbed.sites') |
255
|
24 |
|
->addParameterByName('cacheDir') |
256
|
24 |
|
->setJS(file_get_contents(__DIR__ . '/Parser/tagFilter.js')); |
257
|
24 |
|
} |
258
|
|
|
|
259
|
|
|
/** |
260
|
|
|
* Return the list of named captures from a list of [regexp, map] pairs |
261
|
|
|
* |
262
|
|
|
* @param array[] $regexps List of [regexp, map] pairs |
263
|
|
|
* @return string[] |
264
|
|
|
*/ |
265
|
17 |
|
protected function getAttributeNamesFromRegexps(array $regexps) |
266
|
|
|
{ |
267
|
17 |
|
$attrNames = []; |
268
|
17 |
|
foreach ($regexps as list($regexp, $map)) |
269
|
|
|
{ |
270
|
15 |
|
$attrNames += array_flip(array_filter($map)); |
271
|
|
|
} |
272
|
|
|
|
273
|
17 |
|
return $attrNames; |
274
|
|
|
} |
275
|
|
|
|
276
|
|
|
/** |
277
|
|
|
* Get the attributes config for given site config |
278
|
|
|
* |
279
|
|
|
* @param array $siteConfig Site's config |
280
|
|
|
* @return array Map of [attrName => attrConfig] |
281
|
|
|
*/ |
282
|
17 |
|
protected function getAttributesConfig(array $siteConfig) |
283
|
|
|
{ |
284
|
17 |
|
$attrNames = $this->getAttributeNamesFromRegexps($siteConfig['extract']); |
285
|
17 |
|
foreach ($siteConfig['scrape'] as $scrapeConfig) |
286
|
|
|
{ |
287
|
4 |
|
$attrNames += $this->getAttributeNamesFromRegexps($scrapeConfig['extract']); |
288
|
|
|
} |
289
|
|
|
|
290
|
17 |
|
$attributes = $siteConfig['attributes'] + array_fill_keys(array_keys($attrNames), []); |
291
|
17 |
|
foreach ($attributes as &$attrConfig) |
292
|
|
|
{ |
293
|
15 |
|
$attrConfig += ['required' => false]; |
294
|
|
|
} |
295
|
17 |
|
unset($attrConfig); |
296
|
|
|
|
297
|
17 |
|
return $attributes; |
298
|
|
|
} |
299
|
|
|
|
300
|
|
|
/** |
301
|
|
|
* Validate and normalize a site ID |
302
|
|
|
* |
303
|
|
|
* @param string $siteId |
304
|
|
|
* @return string |
305
|
|
|
*/ |
306
|
20 |
|
protected function normalizeId($siteId) |
307
|
|
|
{ |
308
|
20 |
|
$siteId = strtolower($siteId); |
309
|
|
|
|
310
|
20 |
|
if (!preg_match('(^[a-z0-9]+$)', $siteId)) |
311
|
|
|
{ |
312
|
1 |
|
throw new InvalidArgumentException('Invalid site ID'); |
313
|
|
|
} |
314
|
|
|
|
315
|
19 |
|
return $siteId; |
316
|
|
|
} |
317
|
|
|
} |