|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace PiedWeb\SeoPocketCrawler; |
|
4
|
|
|
|
|
5
|
|
|
use League\Uri\Http; |
|
6
|
|
|
use League\Uri\UriInfo; |
|
7
|
|
|
|
|
8
|
|
|
class Url |
|
9
|
|
|
{ |
|
10
|
|
|
private static $autoIncrement = 1; |
|
11
|
|
|
|
|
12
|
|
|
public $id; |
|
13
|
|
|
public $discovered; |
|
14
|
|
|
public $uri; |
|
15
|
|
|
public $click; |
|
16
|
|
|
public $pagerank; |
|
17
|
|
|
public $inboundlinks; |
|
18
|
|
|
public $inboundlinks_nofollow = 0; |
|
19
|
|
|
public $can_be_crawled; |
|
20
|
|
|
public $indexable; |
|
21
|
|
|
public $mime_type; |
|
22
|
|
|
public $links; |
|
23
|
|
|
public $links_duplicate; |
|
24
|
|
|
public $links_self; |
|
25
|
|
|
public $links_internal; |
|
26
|
|
|
public $links_sub; |
|
27
|
|
|
public $links_external; |
|
28
|
|
|
public $word_count; |
|
29
|
|
|
public $load_time; |
|
30
|
|
|
public $size; |
|
31
|
|
|
public $title; |
|
32
|
|
|
public $updated_at; |
|
33
|
|
|
//public $kws; |
|
34
|
|
|
//public $h1; |
|
35
|
|
|
//public $breadcrumb_level; |
|
36
|
|
|
//public $breadcrumb_first; |
|
37
|
|
|
//public $breadcrumb_text; |
|
38
|
|
|
|
|
39
|
|
|
public function __construct($url, $click = null) |
|
40
|
|
|
{ |
|
41
|
|
|
$this->id = $this->getId(); |
|
42
|
|
|
$this->uri = substr($url, strlen(UriInfo::getOrigin(Http::createFromString($url)))); //! |
|
|
|
|
|
|
43
|
|
|
$this->updated_at = date('Ymd'); |
|
44
|
|
|
$this->inboundlinks = 0; |
|
45
|
|
|
$this->click = $click; |
|
46
|
|
|
} |
|
47
|
|
|
|
|
48
|
|
|
public function getId() |
|
49
|
|
|
{ |
|
50
|
|
|
if (null === $this->id) { |
|
51
|
|
|
$this->id = self::$autoIncrement; |
|
52
|
|
|
++self::$autoIncrement; |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
|
|
return $this->id; |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
public function setDiscovered(int $discovered) |
|
59
|
|
|
{ |
|
60
|
|
|
$this->discovered = $discovered; |
|
61
|
|
|
|
|
62
|
|
|
return $this; |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
public function setMimeType(string $mimeType) |
|
66
|
9 |
|
{ |
|
67
|
|
|
$this->mime_type = 'text/html' == $mimeType ? 1 : $mimeType; |
|
68
|
9 |
|
|
|
69
|
9 |
|
return $this; |
|
70
|
9 |
|
} |
|
71
|
9 |
|
|
|
72
|
9 |
|
/** |
|
73
|
9 |
|
public function setH1(string $h1) |
|
74
|
|
|
{ |
|
75
|
9 |
|
$this->h1 = $this->title == $h1 ? '=' : $h1; |
|
76
|
|
|
|
|
77
|
9 |
|
return $this; |
|
78
|
9 |
|
}/**/ |
|
79
|
9 |
|
public function setId($id) |
|
80
|
|
|
{ |
|
81
|
|
|
$this->id = intval($id); |
|
82
|
9 |
|
} |
|
83
|
|
|
|
|
84
|
|
|
public function getDiscovered() |
|
85
|
9 |
|
{ |
|
86
|
|
|
return $this->discovered; |
|
87
|
9 |
|
} |
|
88
|
|
|
|
|
89
|
9 |
|
public function getUri() |
|
90
|
|
|
{ |
|
91
|
|
|
return $this->uri; |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
public function setUri($uri) |
|
95
|
|
|
{ |
|
96
|
|
|
$this->uri = $uri; |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
|
|
public function getClick() |
|
100
|
|
|
{ |
|
101
|
|
|
return $this->click; |
|
102
|
|
|
} |
|
103
|
|
|
|
|
104
|
|
|
public function setClick($click) |
|
105
|
|
|
{ |
|
106
|
|
|
$this->click = (int) $click; |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
|
|
public function getPagerank() |
|
110
|
|
|
{ |
|
111
|
|
|
return $this->pagerank; |
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
|
|
public function setPagerank($pagerank) |
|
115
|
|
|
{ |
|
116
|
|
|
$this->pagerank = (float) $pagerank; |
|
117
|
|
|
} |
|
118
|
|
|
|
|
119
|
|
|
public function getInboundlinks() |
|
120
|
|
|
{ |
|
121
|
|
|
return $this->inboundlinks; |
|
122
|
|
|
} |
|
123
|
|
|
|
|
124
|
|
|
public function setInboundlinks($inboundlinks) |
|
125
|
|
|
{ |
|
126
|
|
|
$this->inboundlinks = (int) $inboundlinks; |
|
127
|
|
|
} |
|
128
|
|
|
|
|
129
|
|
|
public function getInboundlinksNofollow() |
|
130
|
|
|
{ |
|
131
|
|
|
return $this->inboundlinks_nofollow; |
|
132
|
|
|
} |
|
133
|
|
|
|
|
134
|
|
|
public function setInboundlinksNofollow($inboundlinks_nofollow) |
|
135
|
|
|
{ |
|
136
|
|
|
$this->inboundlinks_nofollow = (int) $inboundlinks_nofollow; |
|
137
|
|
|
} |
|
138
|
|
|
|
|
139
|
|
|
public function getCanBeCrawled() |
|
140
|
|
|
{ |
|
141
|
|
|
return $this->can_be_crawled; |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
|
|
public function setCanBeCrawled($can_be_crawled) |
|
145
|
|
|
{ |
|
146
|
|
|
$this->can_be_crawled = (bool) $can_be_crawled; |
|
147
|
|
|
} |
|
148
|
|
|
|
|
149
|
|
|
public function getIndexable() |
|
150
|
|
|
{ |
|
151
|
|
|
return $this->indexable; |
|
152
|
|
|
} |
|
153
|
|
|
|
|
154
|
|
|
public function setIndexable($indexable) |
|
155
|
|
|
{ |
|
156
|
|
|
$this->indexable = (int) $indexable; |
|
157
|
|
|
} |
|
158
|
|
|
|
|
159
|
|
|
public function getMimeType() |
|
160
|
|
|
{ |
|
161
|
|
|
return $this->mime_type; |
|
162
|
|
|
} |
|
163
|
|
|
|
|
164
|
|
|
public function getLinks() |
|
165
|
|
|
{ |
|
166
|
|
|
return $this->links; |
|
167
|
|
|
} |
|
168
|
|
|
|
|
169
|
|
|
public function setLinks($links) |
|
170
|
|
|
{ |
|
171
|
|
|
$this->links = (int) $links; |
|
172
|
|
|
} |
|
173
|
|
|
|
|
174
|
|
|
public function getLinksDuplicate() |
|
175
|
|
|
{ |
|
176
|
|
|
return (int) $this->links_duplicate; |
|
177
|
|
|
} |
|
178
|
|
|
|
|
179
|
|
|
public function setLinksDuplicate($links_duplicate) |
|
180
|
|
|
{ |
|
181
|
|
|
$this->links_duplicate = (int) $links_duplicate; |
|
182
|
|
|
} |
|
183
|
|
|
|
|
184
|
|
|
public function getLinksSelf() |
|
185
|
|
|
{ |
|
186
|
|
|
return $this->links_self; |
|
187
|
|
|
} |
|
188
|
|
|
|
|
189
|
|
|
public function setLinksSelf($links_self) |
|
190
|
|
|
{ |
|
191
|
|
|
$this->links_self = (int) $links_self; |
|
192
|
|
|
} |
|
193
|
|
|
|
|
194
|
|
|
public function getLinksInternal() |
|
195
|
|
|
{ |
|
196
|
|
|
return $this->links_internal; |
|
197
|
|
|
} |
|
198
|
|
|
|
|
199
|
|
|
public function setLinksInternal($links_internal) |
|
200
|
|
|
{ |
|
201
|
|
|
$this->links_internal = (int) $links_internal; |
|
202
|
|
|
} |
|
203
|
|
|
|
|
204
|
|
|
public function getLinksSub() |
|
205
|
|
|
{ |
|
206
|
|
|
return $this->links_sub; |
|
207
|
|
|
} |
|
208
|
|
|
|
|
209
|
|
|
public function setLinksSub($links_sub) |
|
210
|
|
|
{ |
|
211
|
|
|
$this->links_sub = (int) $links_sub; |
|
212
|
|
|
} |
|
213
|
|
|
|
|
214
|
|
|
public function getLinksExternal() |
|
215
|
|
|
{ |
|
216
|
|
|
return $this->links_external; |
|
217
|
|
|
} |
|
218
|
|
|
|
|
219
|
|
|
public function setLinksExternal($links_external) |
|
220
|
|
|
{ |
|
221
|
|
|
$this->links_external = (int) $links_external; |
|
222
|
|
|
} |
|
223
|
|
|
|
|
224
|
|
|
public function getWordCount() |
|
225
|
|
|
{ |
|
226
|
|
|
return $this->word_count; |
|
227
|
|
|
} |
|
228
|
|
|
|
|
229
|
|
|
public function setWordCount($word_count) |
|
230
|
|
|
{ |
|
231
|
|
|
$this->word_count = (int) $word_count; |
|
232
|
|
|
} |
|
233
|
|
|
|
|
234
|
|
|
public function getLoadTime() |
|
235
|
|
|
{ |
|
236
|
|
|
return $this->load_time; |
|
237
|
|
|
} |
|
238
|
|
|
|
|
239
|
|
|
public function setLoadTime($load_time) |
|
240
|
|
|
{ |
|
241
|
|
|
$this->load_time = (int) $load_time; |
|
242
|
|
|
} |
|
243
|
|
|
|
|
244
|
|
|
public function getSize() |
|
245
|
|
|
{ |
|
246
|
|
|
return $this->size; |
|
247
|
|
|
} |
|
248
|
|
|
|
|
249
|
|
|
public function setSize($size) |
|
250
|
|
|
{ |
|
251
|
|
|
$this->size = (int) $size; |
|
252
|
|
|
} |
|
253
|
|
|
|
|
254
|
|
|
public function getTitle() |
|
255
|
|
|
{ |
|
256
|
|
|
return $this->title; |
|
257
|
|
|
} |
|
258
|
|
|
|
|
259
|
|
|
public function setTitle($title) |
|
260
|
|
|
{ |
|
261
|
|
|
$this->title = $title; |
|
262
|
|
|
} |
|
263
|
|
|
|
|
264
|
|
|
public function getUpdatedAt() |
|
265
|
|
|
{ |
|
266
|
|
|
return $this->updated_at; |
|
267
|
|
|
} |
|
268
|
|
|
|
|
269
|
|
|
public function setUpdatedAt($updated_at) |
|
270
|
|
|
{ |
|
271
|
|
|
$this->updated_at = $updated_at; |
|
272
|
|
|
} |
|
273
|
|
|
} |
|
274
|
|
|
|