|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Chay22\RecSelMeter; |
|
4
|
|
|
|
|
5
|
|
|
use Chay22\RecSelMeter\Connection; |
|
6
|
|
|
|
|
7
|
|
|
class Parser extends Connection |
|
8
|
|
|
{ |
|
9
|
|
|
/** |
|
10
|
|
|
* @var $xpath |
|
11
|
|
|
*/ |
|
12
|
|
|
protected $xpath; |
|
13
|
|
|
|
|
14
|
|
|
/** |
|
15
|
|
|
* Retrieve amount product sold |
|
16
|
|
|
* |
|
17
|
|
|
* @var $sold |
|
18
|
|
|
*/ |
|
19
|
|
|
public $sold; |
|
20
|
|
|
|
|
21
|
|
|
/** |
|
22
|
|
|
* @var $cod |
|
23
|
|
|
*/ |
|
24
|
|
|
public $cod = false; |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* Latest bump attempt time |
|
28
|
|
|
* |
|
29
|
|
|
* @var $bump |
|
30
|
|
|
*/ |
|
31
|
|
|
public $bump; |
|
32
|
|
|
|
|
33
|
1 |
|
function __construct($url) |
|
|
|
|
|
|
34
|
|
|
{ |
|
35
|
1 |
|
parent::__construct($url); |
|
36
|
1 |
|
date_default_timezone_set('Asia/Jakarta'); |
|
37
|
1 |
|
libxml_use_internal_errors(true); |
|
38
|
|
|
|
|
39
|
1 |
|
if ($this->header['http_code'] != 200 ) { |
|
40
|
|
|
throw new \Exception("URL not found!"); |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
1 |
|
$this->xpath = $this->document(new \DOMDocument); |
|
44
|
1 |
|
} |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* Creates DOMXPath object from content |
|
48
|
|
|
* |
|
49
|
|
|
* @return object |
|
50
|
|
|
*/ |
|
51
|
1 |
|
protected function document(\DOMDocument $doc) |
|
52
|
|
|
{ |
|
53
|
1 |
|
$doc->loadHTML($this->content); |
|
54
|
1 |
|
return new \DOMXPath($doc); |
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
|
|
/** |
|
58
|
|
|
* Find element from content with xpath expression |
|
59
|
|
|
* |
|
60
|
|
|
* @param string $query xpath query expression |
|
61
|
|
|
* @return bool(false)|DOMNodeList |
|
|
|
|
|
|
62
|
|
|
* @see http://php.net/manual/en/domxpath.query.php |
|
63
|
|
|
*/ |
|
64
|
6 |
|
public function find($query) |
|
65
|
|
|
{ |
|
66
|
6 |
|
return $this->xpath->query($query); |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
/** |
|
70
|
|
|
* Fetch all result into array |
|
71
|
|
|
* |
|
72
|
|
|
* @return array |
|
73
|
|
|
*/ |
|
74
|
1 |
|
public function fetch() |
|
75
|
|
|
{ |
|
76
|
1 |
|
$product = $this->product(); |
|
77
|
|
|
|
|
78
|
|
|
return [ |
|
79
|
1 |
|
'username' => $this->username(), |
|
80
|
1 |
|
'user_id' => $this->userID(), |
|
81
|
1 |
|
'rank' => $this->rank(), |
|
82
|
1 |
|
'feedback' => $this->feedback(), |
|
83
|
1 |
|
'feedback_percent' => $this->feedbackPercent(), |
|
84
|
1 |
|
'join_date' => $this->joinDate(), |
|
85
|
1 |
|
'verified' => $this->verifiedSeller(), |
|
86
|
1 |
|
'published_at' => $this->threadPublished(), |
|
87
|
1 |
|
'image_count' => $this->countImage(), |
|
88
|
1 |
|
'sold' => $product['sold'], |
|
89
|
1 |
|
'cod' => $product['cod'], |
|
90
|
1 |
|
'last_bump' => $product['bump'], |
|
91
|
1 |
|
]; |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
/** |
|
95
|
|
|
* Get username |
|
96
|
|
|
* |
|
97
|
|
|
* @return string |
|
98
|
|
|
*/ |
|
99
|
2 |
|
public function username() |
|
100
|
|
|
{ |
|
101
|
2 |
|
return $this->find('//meta[@name="author"]') |
|
102
|
2 |
|
->item(0) |
|
103
|
2 |
|
->getAttribute('content'); |
|
104
|
|
|
} |
|
105
|
|
|
|
|
106
|
|
|
/** |
|
107
|
|
|
* Get user id |
|
108
|
|
|
* |
|
109
|
|
|
* @return int |
|
110
|
|
|
*/ |
|
111
|
2 |
|
public function userID() |
|
112
|
|
|
{ |
|
113
|
2 |
|
$query = $this->find( |
|
114
|
|
|
'//div[contains(@class,"seller-detail-info")]' . |
|
115
|
|
|
'/span[contains(@class,"username")]/a' |
|
116
|
2 |
|
)->item(0) |
|
117
|
2 |
|
->getAttribute('href'); |
|
118
|
2 |
|
$query = explode('/', $query); |
|
119
|
2 |
|
$query = end($query); |
|
120
|
|
|
|
|
121
|
2 |
|
return $query + 0; |
|
122
|
|
|
} |
|
123
|
|
|
|
|
124
|
|
|
/** |
|
125
|
|
|
* Amount of image of thread |
|
126
|
|
|
* |
|
127
|
|
|
* @return int |
|
128
|
|
|
*/ |
|
129
|
2 |
|
public function countImage() |
|
130
|
|
|
{ |
|
131
|
2 |
|
return $this->find( |
|
132
|
|
|
'//div[contains(@id,"carousel-thumb")]' . |
|
133
|
|
|
'/div[contains(@class,"thumbnail")]' |
|
134
|
2 |
|
)->length; |
|
135
|
|
|
} |
|
136
|
|
|
|
|
137
|
|
|
/** |
|
138
|
|
|
* Get user rank |
|
139
|
|
|
* |
|
140
|
|
|
* @return string |
|
141
|
|
|
*/ |
|
142
|
1 |
|
public function rank() |
|
143
|
|
|
{ |
|
144
|
1 |
|
return $this->find( |
|
145
|
|
|
'//div[contains(@class,"seller-detail-info")]' . |
|
146
|
|
|
'/span[contains(@class,"rank")]' |
|
147
|
1 |
|
)->item(0) |
|
148
|
1 |
|
->nodeValue; |
|
149
|
|
|
} |
|
150
|
|
|
|
|
151
|
|
|
/** |
|
152
|
|
|
* Get amount of feedback |
|
153
|
|
|
* |
|
154
|
|
|
* @return int |
|
155
|
|
|
*/ |
|
156
|
1 |
|
public function feedback() |
|
157
|
|
|
{ |
|
158
|
1 |
|
return $this->find( |
|
159
|
|
|
'//div[contains(@class,"seller-detail-info")]' . |
|
160
|
|
|
'/span[contains(@class,"feedback")]' |
|
161
|
1 |
|
)->item(0) |
|
162
|
1 |
|
->getElementsByTagName('a') |
|
163
|
1 |
|
->item(0) |
|
164
|
1 |
|
->textContent + 0; |
|
165
|
|
|
} |
|
166
|
|
|
|
|
167
|
|
|
/** |
|
168
|
|
|
* Get feedback precentage |
|
169
|
|
|
* |
|
170
|
|
|
* @return int |
|
171
|
|
|
*/ |
|
172
|
1 |
|
public function feedbackPercent() |
|
173
|
|
|
{ |
|
174
|
1 |
|
$query = $this->find( |
|
175
|
|
|
'//div[contains(@class,"seller-detail-info")]' . |
|
176
|
|
|
'/span[contains(@class,"feedback")]' |
|
177
|
1 |
|
)->item(0) |
|
178
|
1 |
|
->getElementsByTagName('strong') |
|
179
|
1 |
|
->item(0) |
|
180
|
1 |
|
->nodeValue; |
|
181
|
|
|
|
|
182
|
1 |
|
return str_replace('%', '', $query) + 0; |
|
183
|
|
|
} |
|
184
|
|
|
|
|
185
|
|
|
/** |
|
186
|
|
|
* Retrieve user join date |
|
187
|
|
|
* |
|
188
|
|
|
* @return string date |
|
189
|
|
|
*/ |
|
190
|
1 |
|
public function joinDate() |
|
191
|
|
|
{ |
|
192
|
1 |
|
$query = $this->find( |
|
193
|
|
|
'//div[contains(@class,"seller-detail-info")]' . |
|
194
|
|
|
'/span[not(@class)]' |
|
195
|
1 |
|
)->item(0) |
|
196
|
1 |
|
->nodeValue; |
|
197
|
1 |
|
$query = str_replace('join: ', '', strtolower($query)); |
|
198
|
1 |
|
$date = new \DateTime($query); |
|
199
|
|
|
|
|
200
|
1 |
|
return $date->format('Y-m-d H:i:s'); |
|
201
|
|
|
} |
|
202
|
|
|
|
|
203
|
|
|
/** |
|
204
|
|
|
* Check whether verified seller |
|
205
|
|
|
* |
|
206
|
|
|
* @return bool |
|
207
|
|
|
*/ |
|
208
|
2 |
|
public function verifiedSeller() |
|
209
|
|
|
{ |
|
210
|
2 |
|
if ($this->find('//div[contains(@class,"vsl-badge")]')->length > 0) { |
|
211
|
|
|
return true; |
|
212
|
|
|
} |
|
213
|
|
|
|
|
214
|
2 |
|
return false; |
|
215
|
|
|
} |
|
216
|
|
|
|
|
217
|
|
|
/** |
|
218
|
|
|
* Find date of thread started |
|
219
|
|
|
* |
|
220
|
|
|
* @return string date thread started |
|
221
|
|
|
*/ |
|
222
|
2 |
|
public function threadPublished() |
|
223
|
|
|
{ |
|
224
|
2 |
|
$query = $this->find( |
|
225
|
|
|
'//div[contains(@class,"user-details")]/time' |
|
226
|
2 |
|
)->item(0) |
|
227
|
2 |
|
->getAttribute('datetime'); |
|
228
|
2 |
|
$date = new \DateTime($query); |
|
229
|
|
|
|
|
230
|
2 |
|
return $date->format('Y-m-d H:i:s'); |
|
231
|
|
|
} |
|
232
|
|
|
|
|
233
|
|
|
/** |
|
234
|
|
|
* Return product specified entities |
|
235
|
|
|
* |
|
236
|
|
|
* @return array sold, bump, cod |
|
237
|
|
|
*/ |
|
238
|
1 |
|
public function product() |
|
239
|
|
|
{ |
|
240
|
1 |
|
$query = $this->find( |
|
241
|
|
|
'//div[contains(@class,"item-attributes")]/table/tr' |
|
242
|
1 |
|
); |
|
243
|
1 |
|
foreach ($query as $value) { |
|
244
|
|
|
$nodeValue = strtolower($value->nodeValue); |
|
245
|
|
|
$titleEntity = trim(explode(':', $nodeValue)[0]); |
|
246
|
|
|
$content = explode(':', $nodeValue)[1]; |
|
247
|
|
|
|
|
248
|
|
|
if($titleEntity === 'terjual') { |
|
249
|
|
|
$content = trim($content); |
|
250
|
|
|
$this->sold = explode(' ', $content)[0] + 0; |
|
251
|
|
|
} |
|
252
|
|
|
|
|
253
|
|
|
if($titleEntity === 'lokasi') { |
|
254
|
|
|
if(strpos($content, 'bisa cod') !== false) { |
|
255
|
|
|
$this->cod = true; |
|
256
|
|
|
} |
|
257
|
|
|
} |
|
258
|
|
|
|
|
259
|
|
|
if($titleEntity === 'last sundul') { |
|
260
|
|
|
$product['bump'] = trim($content); |
|
|
|
|
|
|
261
|
|
|
if (strpos(trim($content), 'ago') !== false) { |
|
262
|
|
|
$bumpTime = explode(' ', trim($content)); |
|
263
|
|
|
$date = new \DateTime; |
|
264
|
|
|
$date->modify('- '.$bumpTime[0] . ' '. $bumpTime[1]); |
|
265
|
|
|
|
|
266
|
|
|
$this->bump = $date->format('Y-m-d H:i:s'); |
|
267
|
|
|
} |
|
268
|
|
|
} |
|
269
|
|
|
|
|
270
|
1 |
|
} |
|
271
|
|
|
|
|
272
|
|
|
return [ |
|
273
|
1 |
|
'sold' => $this->sold, |
|
274
|
1 |
|
'cod' => $this->cod, |
|
275
|
1 |
|
'bump' => $this->bump, |
|
276
|
1 |
|
]; |
|
277
|
|
|
} |
|
278
|
|
|
} |
|
279
|
|
|
|
Adding explicit visibility (
private,protected, orpublic) is generally recommend to communicate to other developers how, and from where this method is intended to be used.