|
1
|
|
|
<?php
|
|
2
|
|
|
|
|
3
|
|
|
namespace Swader\Diffbot\Entity;
|
|
4
|
|
|
|
|
5
|
|
|
use Swader\Diffbot\Abstracts\Entity;
|
|
6
|
|
|
|
|
7
|
|
|
class Post extends Entity
|
|
8
|
|
|
{
|
|
9
|
|
|
|
|
10
|
67 |
|
public function __construct(array $data)
|
|
11
|
|
|
{
|
|
12
|
67 |
|
if (class_exists('\Carbon\Carbon')) {
|
|
13
|
67 |
|
$format = 'D, d M o H:i:s e';
|
|
14
|
67 |
|
\Carbon\Carbon::setToStringFormat($format);
|
|
15
|
67 |
|
}
|
|
16
|
|
|
|
|
17
|
67 |
|
parent::__construct($data);
|
|
18
|
67 |
|
}
|
|
19
|
|
|
|
|
20
|
|
|
/**
|
|
21
|
|
|
* Should always return "post"
|
|
22
|
|
|
* @return string
|
|
23
|
|
|
*/
|
|
24
|
1 |
|
public function getType()
|
|
25
|
|
|
{
|
|
26
|
1 |
|
return $this->data['type'];
|
|
27
|
|
|
}
|
|
28
|
|
|
|
|
29
|
|
|
/**
|
|
30
|
|
|
* Alias for getLang()
|
|
31
|
|
|
* @see getLang()
|
|
32
|
|
|
* @return string
|
|
33
|
|
|
*/
|
|
34
|
1 |
|
public function getHumanLanguage()
|
|
35
|
|
|
{
|
|
36
|
1 |
|
return $this->getLang();
|
|
37
|
|
|
}
|
|
38
|
|
|
|
|
39
|
|
|
/**
|
|
40
|
|
|
* Returns the human language of the page as determined by Diffbot when looking at content.
|
|
41
|
|
|
* The code returned is a two-character ISO 639-1 code: http://en.wikipedia.org/wiki/List_of_ISO_639-1_codes
|
|
42
|
|
|
* @return string
|
|
43
|
|
|
*/
|
|
44
|
1 |
|
public function getLang()
|
|
45
|
|
|
{
|
|
46
|
1 |
|
return $this->data['humanLanguage'];
|
|
47
|
|
|
}
|
|
48
|
|
|
|
|
49
|
|
|
/**
|
|
50
|
|
|
* Returns plaintext version of article (no HTML) as parsed by Diffbot.
|
|
51
|
|
|
* Only the content is returned, the text in the surrounding (layout etc) elements is ignored.
|
|
52
|
|
|
* @return string
|
|
53
|
|
|
*/
|
|
54
|
1 |
|
public function getText()
|
|
55
|
|
|
{
|
|
56
|
1 |
|
return $this->data['text'];
|
|
57
|
|
|
}
|
|
58
|
|
|
|
|
59
|
|
|
/**
|
|
60
|
|
|
* Returns full HTML of the article's content - only the content, not the surrounding layout HTML.
|
|
61
|
|
|
* @return string
|
|
62
|
|
|
*/
|
|
63
|
1 |
|
public function getHtml()
|
|
64
|
|
|
{
|
|
65
|
1 |
|
return $this->data['html'];
|
|
66
|
|
|
}
|
|
67
|
|
|
|
|
68
|
|
|
/**
|
|
69
|
|
|
* Returns date as per http://www.w3.org/Protocols/rfc2616/rfc2616-sec3.html#sec3.3
|
|
70
|
|
|
* Example date: "Wed, 18 Dec 2013 00:00:00 GMT"
|
|
71
|
|
|
* This will be a Carbon (https://github.com/briannesbitt/Carbon) instance if Carbon is installed.
|
|
72
|
|
|
* @return \Carbon\Carbon | string
|
|
73
|
|
|
*/
|
|
74
|
1 |
|
public function getDate()
|
|
75
|
|
|
{
|
|
76
|
|
|
|
|
77
|
1 |
|
return (class_exists('\Carbon\Carbon')) ?
|
|
78
|
1 |
|
new \Carbon\Carbon($this->data['date'], 'GMT') :
|
|
79
|
1 |
|
$this->data['date'];
|
|
80
|
|
|
}
|
|
81
|
|
|
|
|
82
|
|
|
/**
|
|
83
|
|
|
* Returns the full name of the author, as signed on the article's page
|
|
84
|
|
|
* @return string | null
|
|
85
|
|
|
*/
|
|
86
|
1 |
|
public function getAuthor()
|
|
87
|
|
|
{
|
|
88
|
1 |
|
return (isset($this->data['author'])) ? $this->data['author'] : null;
|
|
89
|
|
|
}
|
|
90
|
|
|
|
|
91
|
|
|
/**
|
|
92
|
|
|
* Returns the url of the author - their profile URL if available
|
|
93
|
|
|
* @return string | null
|
|
94
|
|
|
*/
|
|
95
|
1 |
|
public function getAuthorUrl()
|
|
96
|
|
|
{
|
|
97
|
1 |
|
return (isset($this->data['authorUrl'])) ? $this->data['authorUrl'] : null;
|
|
98
|
|
|
}
|
|
99
|
|
|
|
|
100
|
|
|
/**
|
|
101
|
|
|
* The array returned will contain all tags that Diffbot's AI concluded match the content
|
|
102
|
|
|
*
|
|
103
|
|
|
* Note that these are *not* the meta tags as defined by the author, but machine learned ones.
|
|
104
|
|
|
* The format of the array is:
|
|
105
|
|
|
*
|
|
106
|
|
|
* [
|
|
107
|
|
|
* [
|
|
108
|
|
|
* "id": 133907,
|
|
109
|
|
|
* "count": 3,
|
|
110
|
|
|
* "prevalence": 0.3103448275862069,
|
|
111
|
|
|
* "label": "Apache HTTP Server",
|
|
112
|
|
|
* "type": "Http://www.ontologydesignpatterns.org/ont/dul/DUL.owl#InformationEntity",
|
|
113
|
|
|
* "uri": "http://dbpedia.org/resource/Apache_HTTP_Server"
|
|
114
|
|
|
* ],
|
|
115
|
|
|
* [
|
|
116
|
|
|
* "id": 208652,
|
|
117
|
|
|
* "count": 5,
|
|
118
|
|
|
* "prevalence": 0.5172413793103449,
|
|
119
|
|
|
* "label": "PHP",
|
|
120
|
|
|
* "type": "Http://www.ontologydesignpatterns.org/ont/dul/DUL.owl#InformationEntity",
|
|
121
|
|
|
* "uri": "http://dbpedia.org/resource/PHP"
|
|
122
|
|
|
* ]
|
|
123
|
|
|
* ]
|
|
124
|
|
|
*
|
|
125
|
|
|
* @return array
|
|
126
|
|
|
*/
|
|
127
|
1 |
|
public function getTags()
|
|
128
|
|
|
{
|
|
129
|
1 |
|
return (isset($this->data['tags'])) ? $this->data['tags'] : [];
|
|
130
|
|
|
}
|
|
131
|
|
|
|
|
132
|
|
|
/**
|
|
133
|
|
|
* Returns the sentiment score of the analyzed article text, a value randing from
|
|
134
|
|
|
* -1.0 (very negative) to 1.0 (very positive).
|
|
135
|
|
|
* @return float|null
|
|
136
|
|
|
*/
|
|
137
|
1 |
|
public function getSentiment()
|
|
138
|
|
|
{
|
|
139
|
1 |
|
return (isset($this->data['sentiment'])) ? $this->data['sentiment'] : null;
|
|
140
|
|
|
}
|
|
141
|
|
|
|
|
142
|
|
|
/**
|
|
143
|
|
|
* Returns the number of upvotes on a post or 0 if none or unavailable
|
|
144
|
|
|
* @return int
|
|
145
|
|
|
*/
|
|
146
|
1 |
|
public function getVotes()
|
|
147
|
|
|
{
|
|
148
|
1 |
|
return (isset($this->data['votes'])) ? $this->data['votes'] : null;
|
|
149
|
|
|
}
|
|
150
|
|
|
|
|
151
|
|
|
/**
|
|
152
|
|
|
* Returns the ID of the post (usually the ordinary number of the post in
|
|
153
|
|
|
* the list of all posts, starting with 0 for the first one
|
|
154
|
|
|
* @return int
|
|
155
|
|
|
*/
|
|
156
|
1 |
|
public function getId()
|
|
157
|
|
|
{
|
|
158
|
1 |
|
return (isset($this->data['id'])) ? $this->data['id'] : 1;
|
|
159
|
|
|
}
|
|
160
|
|
|
|
|
161
|
|
|
/**
|
|
162
|
|
|
* If the post is a reply, this is the ID of the post it replies to
|
|
163
|
|
|
* @return int
|
|
164
|
|
|
*/
|
|
165
|
1 |
|
public function getParentId()
|
|
166
|
|
|
{
|
|
167
|
1 |
|
return (isset($this->data['parentId'])) ? $this->data['parentId'] : null;
|
|
168
|
|
|
}
|
|
169
|
|
|
|
|
170
|
|
|
/**
|
|
171
|
|
|
* Returns an array of images found in the page's content.
|
|
172
|
|
|
*
|
|
173
|
|
|
* Note that this (tries) to ignore content-unrelated images like ads arounds the page, etc.
|
|
174
|
|
|
* The format of the array will be:
|
|
175
|
|
|
*
|
|
176
|
|
|
* [
|
|
177
|
|
|
* {
|
|
178
|
|
|
* "height": 808,
|
|
179
|
|
|
* "diffbotUri": "image|3|-543943368",
|
|
180
|
|
|
* "naturalHeight": 808,
|
|
181
|
|
|
* "width": 717,
|
|
182
|
|
|
* "primary": true,
|
|
183
|
|
|
* "naturalWidth": 717,
|
|
184
|
|
|
* "url": "https://example.com/image1.png"
|
|
185
|
|
|
* },
|
|
186
|
|
|
* {
|
|
187
|
|
|
* "height": 506,
|
|
188
|
|
|
* "diffbotUri": "image|3|-844014913",
|
|
189
|
|
|
* "naturalHeight": 506,
|
|
190
|
|
|
* "width": 715,
|
|
191
|
|
|
* "naturalWidth": 715,
|
|
192
|
|
|
* "url": "https://example.com/image1.jpeg"
|
|
193
|
|
|
* }
|
|
194
|
|
|
* ]
|
|
195
|
|
|
*
|
|
196
|
|
|
* @return array
|
|
197
|
|
|
*/
|
|
198
|
1 |
|
public function getImages()
|
|
199
|
|
|
{
|
|
200
|
1 |
|
return (isset($this->data['images'])) ? $this->data['images'] : [];
|
|
201
|
|
|
}
|
|
202
|
|
|
|
|
203
|
|
|
/**
|
|
204
|
|
|
* Returns the URL which was crawled
|
|
205
|
|
|
* @return string
|
|
206
|
|
|
*/
|
|
207
|
1 |
|
public function getPageUrl()
|
|
208
|
|
|
{
|
|
209
|
1 |
|
return $this->data['pageUrl'];
|
|
210
|
|
|
}
|
|
211
|
|
|
|
|
212
|
|
|
/**
|
|
213
|
|
|
* An internal identifier for Diffbot, used for indexing in their databases
|
|
214
|
|
|
* @return string
|
|
215
|
|
|
*/
|
|
216
|
1 |
|
public function getDiffbotUri()
|
|
217
|
|
|
{
|
|
218
|
1 |
|
return $this->data['diffbotUri'];
|
|
219
|
|
|
}
|
|
220
|
|
|
} |
Swader /
diffbot-php-client
