1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Goose\Modules\Extractors; |
4
|
|
|
|
5
|
|
|
use Goose\Article; |
6
|
|
|
use Goose\Traits\ArticleMutatorTrait; |
7
|
|
|
use Goose\Modules\AbstractModule; |
8
|
|
|
use Goose\Modules\ModuleInterface; |
9
|
|
|
use DOMWrap\Element; |
10
|
|
|
|
11
|
|
|
/** |
12
|
|
|
* Publish Date Extractor |
13
|
|
|
* |
14
|
|
|
* @package Goose\Modules\Extractors |
15
|
|
|
* @license http://www.apache.org/licenses/LICENSE-2.0 Apache License 2.0 |
16
|
|
|
*/ |
17
|
|
|
class PublishDateExtractor extends AbstractModule implements ModuleInterface { |
18
|
|
|
use ArticleMutatorTrait; |
19
|
|
|
|
20
|
|
|
/** |
21
|
|
|
* @param Article $article |
22
|
|
|
* |
23
|
|
|
* @return \DateTime |
24
|
|
|
*/ |
25
|
|
|
public function run(Article $article) { |
26
|
|
|
$this->article($article); |
27
|
|
|
|
28
|
|
|
$dt = null; |
|
|
|
|
29
|
|
|
|
30
|
|
|
$dt = $this->getDateFromSchemaOrg(); |
31
|
|
|
|
32
|
|
|
if (is_null($dt)) { |
33
|
|
|
$dt = $this->getDateFromOpenGraph(); |
34
|
|
|
} |
35
|
|
|
|
36
|
|
|
if (is_null($dt)) { |
37
|
|
|
$dt = $this->getDateFromURL(); |
38
|
|
|
} |
39
|
|
|
|
40
|
|
|
if (is_null($dt)) { |
41
|
|
|
$dt = $this->getDateFromDublinCore(); |
42
|
|
|
} |
43
|
|
|
|
44
|
|
|
if (is_null($dt)) { |
45
|
|
|
$dt = $this->getDateFromParsely(); |
46
|
|
|
} |
47
|
|
|
|
48
|
|
|
$article->setPublishDate($dt); |
49
|
|
|
} |
50
|
|
|
|
51
|
|
|
private function getDateFromURL() { |
52
|
|
|
// Determine date based on URL |
53
|
|
|
if (preg_match('@(?:[\d]{4})(?<delimiter>[/-])(?:[\d]{2})\k<delimiter>(?:[\d]{2})@U', $this->article()->getFinalUrl(), $matches)) { |
54
|
|
|
$dt = \DateTime::createFromFormat('Y' . $matches['delimiter'] . 'm' . $matches['delimiter'] . 'd', $matches[0]); |
55
|
|
|
$dt->setTime(0, 0, 0); |
56
|
|
|
|
57
|
|
|
if ($dt === false) { |
58
|
|
|
return null; |
59
|
|
|
} |
60
|
|
|
|
61
|
|
|
return $dt; |
62
|
|
|
} |
63
|
|
|
|
64
|
|
|
/** @todo Add more date detection methods */ |
65
|
|
|
|
66
|
|
|
return null; |
67
|
|
|
} |
68
|
|
|
|
69
|
|
|
/** |
70
|
|
|
* Check for and determine dates from Schema.org's datePublished property. |
71
|
|
|
* |
72
|
|
|
* Checks HTML tags (e.g. <meta>, <time>, etc.) and JSON-LD. |
73
|
|
|
* |
74
|
|
|
* @return \DateTime|null |
75
|
|
|
* |
76
|
|
|
* @see https://schema.org/datePublished |
77
|
|
|
*/ |
78
|
|
|
private function getDateFromSchemaOrg() { |
79
|
|
|
$dt = null; |
80
|
|
|
|
81
|
|
|
// Check for HTML tags (<meta>, <time>, etc.) |
82
|
|
|
$nodes = $this->article()->getRawDoc()->find('*[itemprop="datePublished"]'); |
83
|
|
|
|
84
|
|
|
/* @var $node Element */ |
85
|
|
|
foreach ($nodes as $node) { |
86
|
|
|
try { |
87
|
|
|
if ($node->hasAttribute('datetime')) { |
88
|
|
|
$dt = new \DateTime($node->getAttribute('datetime')); |
89
|
|
|
break; |
90
|
|
|
} |
91
|
|
|
if ($node->hasAttribute('content')) { |
92
|
|
|
$dt = new \DateTime($node->getAttribute('content')); |
93
|
|
|
break; |
94
|
|
|
} |
95
|
|
|
} |
96
|
|
|
catch (\Exception $e) { |
97
|
|
|
// Do nothing here in case the node has unrecognizable date information. |
98
|
|
|
} |
99
|
|
|
} |
100
|
|
|
|
101
|
|
|
if (!is_null($dt)) { |
102
|
|
|
return $dt; |
103
|
|
|
} |
104
|
|
|
|
105
|
|
|
// Check for JSON-LD |
106
|
|
|
$nodes = $this->article()->getRawDoc()->find('script[type="application/ld+json"]'); |
107
|
|
|
|
108
|
|
|
/* @var $node Element */ |
109
|
|
View Code Duplication |
foreach ($nodes as $node) { |
|
|
|
|
110
|
|
|
try { |
111
|
|
|
$json = json_decode($node->text()); |
112
|
|
|
if (isset($json->datePublished)) { |
113
|
|
|
$dt = new \DateTime($json->datePublished); |
114
|
|
|
break; |
115
|
|
|
} |
116
|
|
|
} |
117
|
|
|
catch (\Exception $e) { |
118
|
|
|
// Do nothing here in case the node has unrecognizable date information. |
119
|
|
|
} |
120
|
|
|
} |
121
|
|
|
|
122
|
|
|
return $dt; |
123
|
|
|
} |
124
|
|
|
|
125
|
|
|
/** |
126
|
|
|
* Check for and determine dates based on Dublin Core standards. |
127
|
|
|
* |
128
|
|
|
* @return \DateTime|null |
129
|
|
|
* |
130
|
|
|
* @see http://dublincore.org/documents/dcmi-terms/#elements-date |
131
|
|
|
* @see http://dublincore.org/documents/2000/07/16/usageguide/qualified-html.shtml |
132
|
|
|
*/ |
133
|
|
|
private function getDateFromDublinCore() { |
134
|
|
|
$dt = null; |
135
|
|
|
$nodes = $this->article()->getRawDoc()->find('*[name="dc.date"], *[name="dc.date.issued"], *[name="DC.date.issued"]'); |
136
|
|
|
|
137
|
|
|
/* @var $node Element */ |
138
|
|
View Code Duplication |
foreach ($nodes as $node) { |
|
|
|
|
139
|
|
|
try { |
140
|
|
|
if ($node->hasAttribute('content')) { |
141
|
|
|
$dt = new \DateTime($node->getAttribute('content')); |
142
|
|
|
break; |
143
|
|
|
} |
144
|
|
|
} |
145
|
|
|
catch (\Exception $e) { |
146
|
|
|
// Do nothing here in case the node has unrecognizable date information. |
147
|
|
|
} |
148
|
|
|
} |
149
|
|
|
|
150
|
|
|
if (!is_null($dt)) { |
151
|
|
|
return $dt; |
152
|
|
|
} |
153
|
|
|
|
154
|
|
|
return $dt; |
155
|
|
|
} |
156
|
|
|
|
157
|
|
|
/** |
158
|
|
|
* Check for and determine dates based on OpenGraph standards. |
159
|
|
|
* |
160
|
|
|
* @return \DateTime|null |
161
|
|
|
* |
162
|
|
|
* @see http://ogp.me/ |
163
|
|
|
* @see http://ogp.me/#type_article |
164
|
|
|
*/ |
165
|
|
|
private function getDateFromOpenGraph() { |
166
|
|
|
$dt = null; |
167
|
|
|
|
168
|
|
|
$og_data = $this->article()->getOpenGraph(); |
169
|
|
|
|
170
|
|
|
try { |
171
|
|
|
if (isset($og_data['published_time'])) { |
172
|
|
|
$dt = new \DateTime($og_data['published_time']); |
173
|
|
|
} |
174
|
|
|
if (is_null($dt) && isset($og_data['pubdate'])) { |
175
|
|
|
$dt = new \DateTime($og_data['pubdate']); |
176
|
|
|
} |
177
|
|
|
} |
178
|
|
|
catch (\Exception $e) { |
179
|
|
|
// Do nothing here in case the node has unrecognizable date information. |
180
|
|
|
} |
181
|
|
|
|
182
|
|
|
return $dt; |
183
|
|
|
} |
184
|
|
|
|
185
|
|
|
/** |
186
|
|
|
* Check for and determine dates based on Parsely metadata. |
187
|
|
|
* |
188
|
|
|
* Checks JSON-LD, <meta> tags and parsely-page. |
189
|
|
|
* |
190
|
|
|
* @return \DateTime|null |
191
|
|
|
* |
192
|
|
|
* @see https://www.parsely.com/help/integration/jsonld/ |
193
|
|
|
* @see https://www.parsely.com/help/integration/metatags/ |
194
|
|
|
* @see https://www.parsely.com/help/integration/ppage/ |
195
|
|
|
*/ |
196
|
|
|
private function getDateFromParsely() { |
197
|
|
|
$dt = null; |
198
|
|
|
|
199
|
|
|
// JSON-LD |
200
|
|
|
$nodes = $this->article()->getRawDoc()->find('script[type="application/ld+json"]'); |
201
|
|
|
|
202
|
|
|
/* @var $node Element */ |
203
|
|
View Code Duplication |
foreach ($nodes as $node) { |
|
|
|
|
204
|
|
|
try { |
205
|
|
|
$json = json_decode($node->text()); |
206
|
|
|
if (isset($json->dateCreated)) { |
207
|
|
|
$dt = new \DateTime($json->dateCreated); |
208
|
|
|
break; |
209
|
|
|
} |
210
|
|
|
} |
211
|
|
|
catch (\Exception $e) { |
212
|
|
|
// Do nothing here in case the node has unrecognizable date information. |
213
|
|
|
} |
214
|
|
|
} |
215
|
|
|
|
216
|
|
|
if (!is_null($dt)) { |
217
|
|
|
return $dt; |
218
|
|
|
} |
219
|
|
|
|
220
|
|
|
// <meta> tags |
221
|
|
|
$nodes = $this->article()->getRawDoc()->find('meta[name="parsely-pub-date"]'); |
222
|
|
|
|
223
|
|
|
/* @var $node Element */ |
224
|
|
View Code Duplication |
foreach ($nodes as $node) { |
|
|
|
|
225
|
|
|
try { |
226
|
|
|
if ($node->hasAttribute('content')) { |
227
|
|
|
$dt = new \DateTime($node->getAttribute('content')); |
228
|
|
|
break; |
229
|
|
|
} |
230
|
|
|
} |
231
|
|
|
catch (\Exception $e) { |
232
|
|
|
// Do nothing here in case the node has unrecognizable date information. |
233
|
|
|
} |
234
|
|
|
} |
235
|
|
|
|
236
|
|
|
if (!is_null($dt)) { |
237
|
|
|
return $dt; |
238
|
|
|
} |
239
|
|
|
|
240
|
|
|
// parsely-page |
241
|
|
|
$nodes = $this->article()->getRawDoc()->find('meta[name="parsely-page"]'); |
242
|
|
|
|
243
|
|
|
/* @var $node Element */ |
244
|
|
View Code Duplication |
foreach ($nodes as $node) { |
|
|
|
|
245
|
|
|
try { |
246
|
|
|
if ($node->hasAttribute('content')) { |
247
|
|
|
$json = json_decode($node->getAttribute('content')); |
248
|
|
|
if (isset($json->pub_date)) { |
249
|
|
|
$dt = new \DateTime($json->pub_date); |
250
|
|
|
break; |
251
|
|
|
} |
252
|
|
|
} |
253
|
|
|
} |
254
|
|
|
catch (\Exception $e) { |
255
|
|
|
// Do nothing here in case the node has unrecognizable date information. |
256
|
|
|
} |
257
|
|
|
} |
258
|
|
|
|
259
|
|
|
return $dt; |
260
|
|
|
} |
261
|
|
|
} |
262
|
|
|
|
This check looks for variable assignements that are either overwritten by other assignments or where the variable is not used subsequently.
Both the
$myVar
assignment in line 1 and the$higher
assignment in line 2 are dead. The first because$myVar
is never used and the second because$higher
is always overwritten for every possible time line.