1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Sesame\Crawler; |
4
|
|
|
|
5
|
|
|
use Sesame\Model\Article; |
6
|
|
|
use \Symfony\Component\DomCrawler\Crawler as DomCrawler; |
7
|
|
|
|
8
|
|
|
/** |
9
|
|
|
* Class ArticleCrawler |
10
|
|
|
* |
11
|
|
|
* @package Sesame\Crawler |
12
|
|
|
*/ |
13
|
|
|
class ArticleCrawler extends Crawler |
14
|
|
|
{ |
15
|
|
|
|
16
|
|
|
/** |
17
|
|
|
* ArticleCrawler constructor. |
18
|
|
|
*/ |
19
|
|
|
public function __construct() |
20
|
|
|
{ |
21
|
|
|
parent::__construct(); |
22
|
|
|
} |
23
|
|
|
|
24
|
|
|
/** |
25
|
|
|
* crawlArticle |
26
|
|
|
* |
27
|
|
|
* @param string $responseBody |
28
|
|
|
* @param bool $crawlVariations |
29
|
|
|
* @return Article |
30
|
|
|
*/ |
31
|
|
|
public function crawlArticle(string $responseBody, bool $crawlVariations): Article |
|
|
|
|
32
|
|
|
{ |
33
|
|
|
$articleCrawler = new DomCrawler($responseBody); |
34
|
|
|
$article = new Article(); |
35
|
|
|
|
36
|
|
|
// extract name of the article |
37
|
|
|
$nameHtml = $articleCrawler->filter('h1.product-name'); |
38
|
|
|
$article->setName($nameHtml->text()); |
39
|
|
|
|
40
|
|
|
// extract ammount of total orders |
41
|
|
|
$article->setOrders($this->extractOrders($articleCrawler)); |
42
|
|
|
|
43
|
|
|
// extract amount of articles in stock |
44
|
|
|
$article->setStock($this->extractStock($articleCrawler)); |
45
|
|
|
|
46
|
|
|
// extract rating information |
47
|
|
|
$article->setRating($this->extractRating($articleCrawler)); |
|
|
|
|
48
|
|
|
$article->setRatingAmount($this->extractRatingAmount($articleCrawler)); |
49
|
|
|
|
50
|
|
|
// extract seller information |
51
|
|
|
$article->setSeller($this->extractSeller($articleCrawler)); |
52
|
|
|
$article->setSellerLocation($this->extractSellerLocation($articleCrawler)); |
53
|
|
|
$article->setSellerSince($this->extractSellerSince($articleCrawler)); |
54
|
|
|
//$article->setSellerFeedback($this->extractSellerFeedback($articleCrawler)); |
55
|
|
|
|
56
|
|
|
// extract product properties |
57
|
|
|
$article->setProperties($this->extractProperties($articleCrawler)); |
58
|
|
|
|
59
|
|
|
// extract description |
60
|
|
|
$article->setDescription($this->extractDescription($articleCrawler)); |
61
|
|
|
|
62
|
|
|
// extract price information |
63
|
|
|
$article->setPrice($this->extractPrice($articleCrawler)); |
64
|
|
|
$article->setPriceDiscount($this->extractPriceDiscount($articleCrawler)); |
65
|
|
|
$article->setPriceCurrency($this->extractPriceCurrency($articleCrawler)); |
66
|
|
|
|
67
|
|
|
return $article; |
68
|
|
|
} |
69
|
|
|
|
70
|
|
|
/** |
71
|
|
|
* @param DomCrawler $articleCrawler |
72
|
|
|
* @return string |
73
|
|
|
*/ |
74
|
|
|
protected function extractDescription(DomCrawler $articleCrawler): string |
75
|
|
|
{ |
76
|
|
|
$descriptionHtml = $articleCrawler->filter('#j-product-description'); |
77
|
|
|
$descriptionString = $descriptionHtml->text(); |
78
|
|
|
|
79
|
|
|
return $descriptionString; |
80
|
|
|
} |
81
|
|
|
|
82
|
|
|
/** |
83
|
|
|
* @param DomCrawler $articleCrawler |
84
|
|
|
* @return int |
85
|
|
|
*/ |
86
|
|
|
protected function extractOrders(DomCrawler $articleCrawler): int |
87
|
|
|
{ |
88
|
|
|
$ordersHtml = $articleCrawler->filter('#j-order-num'); |
89
|
|
|
$ordersString = $ordersHtml->text(); |
90
|
|
|
$ordersStringParts = explode(' ', $ordersString); |
91
|
|
|
|
92
|
|
|
return (int) $ordersStringParts[0]; |
93
|
|
|
} |
94
|
|
|
|
95
|
|
|
/** |
96
|
|
|
* @param DomCrawler $articleCrawler |
97
|
|
|
* @return float |
98
|
|
|
*/ |
99
|
|
|
protected function extractPrice(DomCrawler $articleCrawler): float |
100
|
|
|
{ |
101
|
|
|
$priceHtml = $articleCrawler->filter('#j-sku-price'); |
102
|
|
|
$priceString = $priceHtml->text(); |
103
|
|
|
|
104
|
|
|
return (float) $priceString; |
105
|
|
|
} |
106
|
|
|
|
107
|
|
|
/** |
108
|
|
|
* @param DomCrawler $articleCrawler |
109
|
|
|
* @return string |
110
|
|
|
*/ |
111
|
|
|
protected function extractPriceCurrency(DomCrawler $articleCrawler): string |
112
|
|
|
{ |
113
|
|
|
$priceCurrencyHtml = $articleCrawler->filter('.p-price-content .p-symbol'); |
114
|
|
|
$priceCurrencyString = $priceCurrencyHtml->attr('content'); |
115
|
|
|
|
116
|
|
|
return $priceCurrencyString; |
|
|
|
|
117
|
|
|
} |
118
|
|
|
|
119
|
|
|
/** |
120
|
|
|
* @param DomCrawler $articleCrawler |
121
|
|
|
* @return float |
122
|
|
|
*/ |
123
|
|
|
protected function extractPriceDiscount(DomCrawler $articleCrawler): float |
124
|
|
|
{ |
125
|
|
|
$priceDiscountHtml = $articleCrawler->filter('#j-sku-discount-price'); |
126
|
|
|
$priceDiscountString = $priceDiscountHtml->text(); |
127
|
|
|
|
128
|
|
|
return (float) $priceDiscountString; |
129
|
|
|
} |
130
|
|
|
|
131
|
|
|
/** |
132
|
|
|
* @param DomCrawler $articleCrawler |
133
|
|
|
* @return array |
134
|
|
|
*/ |
135
|
|
|
protected function extractProperties(DomCrawler $articleCrawler): array |
136
|
|
|
{ |
137
|
|
|
$properties = array(); |
138
|
|
|
$propertyElements = $articleCrawler->filter('.product-property-list .property-item'); |
139
|
|
|
foreach ($propertyElements as $propertyElement) { |
140
|
|
|
$propertyCrawler = new DomCrawler($propertyElement); |
141
|
|
|
$propertyTitle = substr($propertyCrawler->filter('.propery-title')->text(), 0, -1); |
142
|
|
|
$propertyValue = $propertyCrawler->filter('.propery-des')->text(); |
143
|
|
|
if (array_key_exists($propertyTitle, $properties)) { |
144
|
|
|
$properties[$propertyTitle] .= '-' . $propertyValue; |
145
|
|
|
} else { |
146
|
|
|
$properties[$propertyTitle] = $propertyValue; |
147
|
|
|
} |
148
|
|
|
} |
149
|
|
|
|
150
|
|
|
return $properties; |
151
|
|
|
} |
152
|
|
|
|
153
|
|
|
/** |
154
|
|
|
* @param DomCrawler $articleCrawler |
155
|
|
|
* @return float |
156
|
|
|
*/ |
157
|
|
|
protected function extractRating(DomCrawler $articleCrawler): float |
158
|
|
|
{ |
159
|
|
|
$ratingHtml = $articleCrawler->filter('#j-customer-reviews-trigger .percent-num'); |
160
|
|
|
$ratingString = $ratingHtml->text(); |
161
|
|
|
$ratingStringParts = explode(' ', $ratingString); |
162
|
|
|
|
163
|
|
|
return (float) $ratingStringParts[0]; |
164
|
|
|
} |
165
|
|
|
|
166
|
|
|
/** |
167
|
|
|
* @param DomCrawler $articleCrawler |
168
|
|
|
* @return int |
169
|
|
|
*/ |
170
|
|
|
protected function extractRatingAmount(DomCrawler $articleCrawler): int |
171
|
|
|
{ |
172
|
|
|
$ratingAmountHtml = $articleCrawler->filter('#j-customer-reviews-trigger .rantings-num'); |
173
|
|
|
$ratingAmountString = $ratingAmountHtml->text(); |
174
|
|
|
$ratingAmountStringParts = explode(' ', $ratingAmountString); |
175
|
|
|
|
176
|
|
|
return (int) str_replace('(', '', $ratingAmountStringParts[0]); |
177
|
|
|
} |
178
|
|
|
|
179
|
|
|
/** |
180
|
|
|
* @param DomCrawler $articleCrawler |
181
|
|
|
* @return string |
182
|
|
|
*/ |
183
|
|
|
protected function extractSeller(DomCrawler $articleCrawler): string |
184
|
|
|
{ |
185
|
|
|
$sellerHtml = $articleCrawler->filter('#j-store-info-wrap .store-lnk'); |
186
|
|
|
$sellerString = $sellerHtml->text(); |
187
|
|
|
|
188
|
|
|
return $sellerString; |
189
|
|
|
} |
190
|
|
|
|
191
|
|
|
/** |
192
|
|
|
* @param DomCrawler $articleCrawler |
193
|
|
|
* @return float |
194
|
|
|
*/ |
195
|
|
|
protected function extractSellerFeedback(DomCrawler $articleCrawler): float |
196
|
|
|
{ |
197
|
|
|
$sellerFeedbackHtml = $articleCrawler->filter('.seller-score-feedback span a'); |
198
|
|
|
$sellerFeedbackString = str_replace('%', '', $sellerFeedbackHtml->text()); |
199
|
|
|
|
200
|
|
|
return (float) $sellerFeedbackString; |
201
|
|
|
} |
202
|
|
|
|
203
|
|
|
/** |
204
|
|
|
* @param DomCrawler $articleCrawler |
205
|
|
|
* @return string |
206
|
|
|
*/ |
207
|
|
|
protected function extractSellerLocation(DomCrawler $articleCrawler): string |
208
|
|
|
{ |
209
|
|
|
$sellerLocationHtml = $articleCrawler->filter('#j-store-info-wrap .store-address'); |
210
|
|
|
$sellerLocationString = trim($sellerLocationHtml->text()); |
211
|
|
|
|
212
|
|
|
return $sellerLocationString; |
213
|
|
|
} |
214
|
|
|
|
215
|
|
|
/** |
216
|
|
|
* @param DomCrawler $articleCrawler |
217
|
|
|
* @return int |
218
|
|
|
*/ |
219
|
|
|
protected function extractSellerSince(DomCrawler $articleCrawler): int |
220
|
|
|
{ |
221
|
|
|
$sellerSinceHtml = $articleCrawler->filter('.store-open-time span'); |
222
|
|
|
$sellerSinceString = $sellerSinceHtml->text(); |
223
|
|
|
$sellerSinceStringParts = explode(' ', $sellerSinceString); |
224
|
|
|
|
225
|
|
|
return (int) $sellerSinceStringParts[0]; |
226
|
|
|
} |
227
|
|
|
|
228
|
|
|
/** |
229
|
|
|
* @param DomCrawler $articleCrawler |
230
|
|
|
* @return int |
231
|
|
|
*/ |
232
|
|
|
protected function extractStock(DomCrawler $articleCrawler): int |
233
|
|
|
{ |
234
|
|
|
$stockAmountHtml = $articleCrawler->filter('#j-sell-stock-num'); |
235
|
|
|
$stockAmountString = $stockAmountHtml->text(); |
236
|
|
|
$stockAmountStringParts = explode(' ', $stockAmountString); |
237
|
|
|
|
238
|
|
|
return (int) $stockAmountStringParts[0]; |
239
|
|
|
} |
240
|
|
|
} |
241
|
|
|
|
This check looks for parameters that have been defined for a function or method, but which are not used in the method body.