|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* RankChecker class |
|
4
|
|
|
* |
|
5
|
|
|
* @package ShahariaAzam\WPRankChecker |
|
6
|
|
|
*/ |
|
7
|
|
|
|
|
8
|
|
|
namespace ShahariaAzam\WPRankChecker; |
|
9
|
|
|
|
|
10
|
|
|
use Exception; |
|
11
|
|
|
use Psr\Cache\CacheItemPoolInterface; |
|
12
|
|
|
use Psr\Http\Client\ClientExceptionInterface; |
|
13
|
|
|
use Psr\Http\Client\ClientInterface; |
|
14
|
|
|
use ShahariaAzam\HTTPClientSupport\Exception\FlexiHTTPException; |
|
15
|
|
|
use ShahariaAzam\HTTPClientSupport\HTTPSupport; |
|
16
|
|
|
use Symfony\Component\DomCrawler\Crawler; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* Class RankChecker |
|
20
|
|
|
* |
|
21
|
|
|
* @package ShahariaAzam\WPRankChecker |
|
22
|
|
|
*/ |
|
23
|
|
|
class RankChecker extends HTTPSupport |
|
24
|
|
|
{ |
|
25
|
|
|
/** |
|
26
|
|
|
* @var CacheItemPoolInterface |
|
27
|
|
|
*/ |
|
28
|
|
|
private $cache; |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* How many pages we need to crawl. |
|
32
|
|
|
* Currently WordPress displays 20 plugins in it's every search result page. |
|
33
|
|
|
* i.e: https://wordpress.org/plugins/search/mail/page/1 |
|
34
|
|
|
* |
|
35
|
|
|
* So we will crawl 5 pages to get 100 plugins for that keyword |
|
36
|
|
|
* |
|
37
|
|
|
* @var int |
|
38
|
|
|
*/ |
|
39
|
|
|
private $searchPageLimit; |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* Keyword that we need to check rank for |
|
43
|
|
|
* |
|
44
|
|
|
* @var |
|
45
|
|
|
*/ |
|
46
|
|
|
private $keyword; |
|
47
|
|
|
|
|
48
|
|
|
/** |
|
49
|
|
|
* @var string |
|
50
|
|
|
*/ |
|
51
|
|
|
private $pluginListsHTML; |
|
52
|
|
|
|
|
53
|
|
|
/** |
|
54
|
|
|
* @var PluginEntity[] |
|
55
|
|
|
*/ |
|
56
|
|
|
private $results; |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* RankChecker constructor. |
|
60
|
|
|
* |
|
61
|
|
|
* @param ClientInterface $httpClient |
|
62
|
|
|
* @param CacheItemPoolInterface $cache |
|
63
|
|
|
*/ |
|
64
|
|
|
public function __construct(ClientInterface $httpClient, CacheItemPoolInterface $cache = null) |
|
65
|
|
|
{ |
|
66
|
|
|
$this->setHttpClient($httpClient); |
|
67
|
|
|
$this->searchPageLimit = 5; |
|
68
|
|
|
$this->cache = $cache; |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
/** |
|
72
|
|
|
* @param int $searchPageLimit |
|
73
|
|
|
* @return RankChecker |
|
74
|
|
|
*/ |
|
75
|
|
|
public function setSearchPageLimit(int $searchPageLimit) |
|
76
|
|
|
{ |
|
77
|
|
|
$this->searchPageLimit = $searchPageLimit; |
|
78
|
|
|
return $this; |
|
79
|
|
|
} |
|
80
|
|
|
|
|
81
|
|
|
/** |
|
82
|
|
|
* @param mixed $keyword |
|
83
|
|
|
* @return RankChecker |
|
84
|
|
|
*/ |
|
85
|
|
|
public function setKeyword($keyword) |
|
86
|
|
|
{ |
|
87
|
|
|
$this->keyword = $keyword; |
|
88
|
|
|
return $this; |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
/** |
|
92
|
|
|
* @return RankChecker |
|
93
|
|
|
* @throws RankCheckerException |
|
94
|
|
|
*/ |
|
95
|
|
|
public function checkRanks() |
|
96
|
|
|
{ |
|
97
|
|
|
if (empty($this->keyword)) { |
|
98
|
|
|
throw new RankCheckerException("You didn't provide any keyword"); |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
|
|
for ($i = 1; $i <= $this->searchPageLimit; $i++) { |
|
102
|
|
|
$contents = $this->fetchPage($i); |
|
103
|
|
|
$dom = new Crawler($contents); |
|
104
|
|
|
$this->filterHTML($dom); |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
|
|
$this->parseResult(); |
|
108
|
|
|
|
|
109
|
|
|
return $this; |
|
110
|
|
|
} |
|
111
|
|
|
|
|
112
|
|
|
/** |
|
113
|
|
|
* @return PluginEntity[] |
|
114
|
|
|
*/ |
|
115
|
|
|
public function getResults() |
|
116
|
|
|
{ |
|
117
|
|
|
return $this->results; |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
|
|
/** |
|
121
|
|
|
* @param $slug |
|
122
|
|
|
* @return int|null |
|
123
|
|
|
*/ |
|
124
|
|
|
public function getRankBySlug($slug) |
|
125
|
|
|
{ |
|
126
|
|
|
$result = array_filter( |
|
127
|
|
|
$this->results, |
|
128
|
|
|
function (PluginEntity $entity) use ($slug) { |
|
129
|
|
|
return $entity->getSlug() === $slug; |
|
130
|
|
|
} |
|
131
|
|
|
); |
|
132
|
|
|
|
|
133
|
|
|
if (empty($result)) { |
|
134
|
|
|
return null; |
|
135
|
|
|
} |
|
136
|
|
|
|
|
137
|
|
|
return array_keys($result)[0]; |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
|
|
/** |
|
141
|
|
|
* @param int $page |
|
142
|
|
|
* @return string |
|
143
|
|
|
* @throws RankCheckerException |
|
144
|
|
|
*/ |
|
145
|
|
|
private function fetchPage($page = 1) |
|
146
|
|
|
{ |
|
147
|
|
|
$resultPageUrl = sprintf("https://wordpress.org/plugins/search/%s/page/%d", $this->keyword, $page); |
|
148
|
|
|
|
|
149
|
|
|
try { |
|
150
|
|
|
$response = $this->httpRequest('GET', $resultPageUrl); |
|
151
|
|
|
} catch (ClientExceptionInterface $e) { |
|
152
|
|
|
throw new RankCheckerException($e->getMessage(), $e->getCode(), $e->getPrevious()); |
|
153
|
|
|
} |
|
154
|
|
|
|
|
155
|
|
|
if ($response->getStatusCode() !== 200) { |
|
156
|
|
|
throw new RankCheckerException("Failed to fetch " . $resultPageUrl, $response->getStatusCode()); |
|
157
|
|
|
} |
|
158
|
|
|
|
|
159
|
|
|
$responseBody = (string) $response->getBody(); |
|
160
|
|
|
if (empty($responseBody)) { |
|
161
|
|
|
throw new RankCheckerException("Empty response from " . $resultPageUrl, $response->getStatusCode()); |
|
162
|
|
|
} |
|
163
|
|
|
|
|
164
|
|
|
return (string) $response->getBody(); |
|
165
|
|
|
} |
|
166
|
|
|
|
|
167
|
|
|
/** |
|
168
|
|
|
* @param Crawler $dom |
|
169
|
|
|
* @return RankChecker |
|
170
|
|
|
* @throws Exception |
|
171
|
|
|
*/ |
|
172
|
|
|
private function filterHTML(Crawler $dom) |
|
173
|
|
|
{ |
|
174
|
|
|
$pc = $dom->filter('article.plugin-card'); |
|
175
|
|
|
if ($pc->count() < 1) { |
|
176
|
|
|
throw new RankCheckerException(".plugin-card HTML node couldn't be found in the DOM tree"); |
|
177
|
|
|
} |
|
178
|
|
|
|
|
179
|
|
|
$pc->each( |
|
180
|
|
|
function (Crawler $crawler) { |
|
181
|
|
|
$this->pluginListsHTML .= "<div class='plugin-card'>".$crawler->html()."</div>"; |
|
182
|
|
|
} |
|
183
|
|
|
); |
|
184
|
|
|
|
|
185
|
|
|
return $this; |
|
186
|
|
|
} |
|
187
|
|
|
|
|
188
|
|
|
/** |
|
189
|
|
|
* @return RankChecker |
|
190
|
|
|
*/ |
|
191
|
|
|
private function parseResult() |
|
192
|
|
|
{ |
|
193
|
|
|
$pcDom = new Crawler($this->pluginListsHTML); |
|
194
|
|
|
|
|
195
|
|
|
$pluginCard = $pcDom->filter('.plugin-card'); |
|
196
|
|
|
|
|
197
|
|
|
$rank = 1; |
|
198
|
|
|
|
|
199
|
|
|
$pluginCard->each( |
|
200
|
|
|
function (Crawler $crawler) use (&$rank) { |
|
201
|
|
|
$titleDom = $crawler->filter('h3.entry-title'); |
|
202
|
|
|
if (!empty($titleDom)) { |
|
203
|
|
|
$title = $titleDom->eq(0); |
|
204
|
|
|
} |
|
205
|
|
|
|
|
206
|
|
|
$linkDom = $titleDom->filter('a'); |
|
207
|
|
|
if (!empty($linkDom)) { |
|
208
|
|
|
$link = $linkDom->attr('href'); |
|
209
|
|
|
} |
|
210
|
|
|
|
|
211
|
|
|
$pluginRatingDom = $crawler->filter('.plugin-rating')->eq(0); |
|
212
|
|
|
$pluginRating = $pluginRatingDom->filter('.wporg-ratings')->eq(0)->attr('data-rating'); |
|
213
|
|
|
|
|
214
|
|
|
$summaryExcerpt = $crawler->filter('.entry-excerpt')->text(); |
|
215
|
|
|
$author = $crawler->filter('.plugin-author')->text(); |
|
216
|
|
|
|
|
217
|
|
|
$plugin = new PluginEntity(); |
|
218
|
|
|
$plugin->setTitle($title->text()); |
|
|
|
|
|
|
219
|
|
|
$plugin->setUrl($link); |
|
|
|
|
|
|
220
|
|
|
|
|
221
|
|
|
$re = '/\/plugins\/(.*)\//'; |
|
222
|
|
|
|
|
223
|
|
|
preg_match($re, $plugin->getUrl(), $matches, PREG_OFFSET_CAPTURE, 1); |
|
224
|
|
|
if (!empty($matches)) { |
|
225
|
|
|
$plugin->setSlug($matches[1][0]); |
|
226
|
|
|
} |
|
227
|
|
|
|
|
228
|
|
|
$plugin->setRating((float) $pluginRating); |
|
229
|
|
|
$plugin->setSummary(trim($summaryExcerpt)); |
|
230
|
|
|
$plugin->setAuthor(trim($author)); |
|
231
|
|
|
|
|
232
|
|
|
|
|
233
|
|
|
$this->results[$rank] = $plugin; |
|
234
|
|
|
$rank++; |
|
235
|
|
|
} |
|
236
|
|
|
); |
|
237
|
|
|
|
|
238
|
|
|
return $this; |
|
239
|
|
|
} |
|
240
|
|
|
} |
|
241
|
|
|
|