|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace App\Http\Controllers\VirtualCrawler\SPOJ; |
|
4
|
|
|
|
|
5
|
|
|
use App\Http\Controllers\VirtualCrawler\CrawlerBase; |
|
6
|
|
|
use App\Models\ProblemModel; |
|
7
|
|
|
use KubAT\PhpSimple\HtmlDomParser; |
|
8
|
|
|
use Auth; |
|
9
|
|
|
use Requests; |
|
10
|
|
|
use Exception; |
|
11
|
|
|
|
|
12
|
|
|
class SPOJ extends CrawlerBase |
|
13
|
|
|
{ |
|
14
|
|
|
public $oid=10; |
|
15
|
|
|
private $imgi; |
|
16
|
|
|
/** |
|
17
|
|
|
* Initial |
|
18
|
|
|
* |
|
19
|
|
|
* @return Response |
|
|
|
|
|
|
20
|
|
|
*/ |
|
21
|
|
|
public function __construct($action='crawl_problem', $con='all', $cached=false) |
|
|
|
|
|
|
22
|
|
|
{ |
|
23
|
|
|
set_time_limit(0); // Pandora's box, engage! |
|
24
|
|
|
if ($action=='judge_level') { |
|
25
|
|
|
$this->judge_level(); |
|
26
|
|
|
} else if($con=="all"){ |
|
27
|
|
|
$this->getSpojProblem(); |
|
28
|
|
|
}else{ |
|
29
|
|
|
//TODO |
|
30
|
|
|
} |
|
31
|
|
|
} |
|
32
|
|
|
|
|
33
|
|
|
public function judge_level() |
|
34
|
|
|
{ |
|
35
|
|
|
// TODO |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
|
|
private static function find($pattern, $subject) |
|
39
|
|
|
{ |
|
40
|
|
|
if (preg_match($pattern, $subject, $matches)) { |
|
41
|
|
|
return $matches[1]; |
|
42
|
|
|
} |
|
43
|
|
|
return null; |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
private function cacheImage($dom) |
|
47
|
|
|
{ |
|
48
|
|
|
foreach ($dom->find('img') as $ele) { |
|
49
|
|
|
$url='https://www.spoj.com'.$ele->src; |
|
50
|
|
|
$res=Requests::get($url, ['Referer' => 'https://www.spoj.com']); |
|
51
|
|
|
$ext=['image/jpeg'=>'.jpg', 'image/png'=>'.png', 'image/gif'=>'.gif', 'image/bmp'=>'.bmp']; |
|
52
|
|
|
if (isset($res->headers['content-type'])) { |
|
53
|
|
|
$cext=$ext[$res->headers['content-type']]; |
|
54
|
|
|
} else { |
|
55
|
|
|
$pos=strpos($ele->src, '.'); |
|
56
|
|
|
if ($pos===false) { |
|
57
|
|
|
$cext=''; |
|
58
|
|
|
} else { |
|
59
|
|
|
$cext=substr($ele->src, $pos); |
|
60
|
|
|
} |
|
61
|
|
|
} |
|
62
|
|
|
$fn=$this->pro['index_id'].'_'.($this->imgi++).$cext; |
|
63
|
|
|
$dir=base_path("public/external/spoj/img"); |
|
64
|
|
|
if (!file_exists($dir)) { |
|
65
|
|
|
mkdir($dir, 0755, true); |
|
66
|
|
|
} |
|
67
|
|
|
file_put_contents(base_path("public/external/spoj/img/$fn"), $res->body); |
|
68
|
|
|
$ele->src='/external/spoj/img/'.$fn; |
|
69
|
|
|
} |
|
70
|
|
|
return $dom; |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
public function getProblemDetail($pid) { |
|
74
|
|
|
$cate = []; |
|
75
|
|
|
|
|
76
|
|
|
$url = "https://www.spoj.com/problems/".$pid; |
|
77
|
|
|
$res = Requests::get($url); |
|
78
|
|
|
if (strpos("<h1>Not Found</h1>",$res->body) !== false) { |
|
79
|
|
|
header('HTTP/1.1 404 Not Found'); |
|
80
|
|
|
die(); |
|
81
|
|
|
}else{ |
|
82
|
|
|
$this->pro['title'] = trim(self::find('/<h2 id="problem-name".* - (.*)<\/h2>/sU',$res->body)); |
|
83
|
|
|
$temp = $res->body; |
|
84
|
|
|
$temp = explode('<tr><td>Time limit:</td><td>', $temp)[1]; |
|
85
|
|
|
$this->pro['time_limit']=trim(explode('s </td></tr>', $temp)[0]); //as for html arrange |
|
86
|
|
|
$temp = $res->body; |
|
87
|
|
|
$temp = explode('<tr><td>Memory limit:</td><td>', $temp)[1]; |
|
88
|
|
|
$this->pro['memory_limit']=trim(explode('MB</td></tr>' , $temp)[0]); |
|
89
|
|
|
$temp = $res->body; |
|
90
|
|
|
$temp = explode('<tr><td>Languages:</td><td>', $temp)[1]; |
|
91
|
|
|
$this->pro['note'] = "Languages Limit:".explode('</td></tr>', $temp)[0]; |
|
92
|
|
|
|
|
93
|
|
|
if(strpos($res->body, "<tr><td>Resource:</td><td>")!==false) |
|
94
|
|
|
{ |
|
95
|
|
|
$temp = $res->body; |
|
96
|
|
|
$temp = explode('<tr><td>Resource:</td><td>', $temp)[1]; |
|
97
|
|
|
$this->pro['source'] = strip_tags(explode('</td></tr>', $temp)[0]); |
|
98
|
|
|
}else { |
|
99
|
|
|
$this->pro['source'] = "SPOJ-".$pid; |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
if(strpos($res->body,"No tags")===false) |
|
103
|
|
|
{ |
|
104
|
|
|
$temp = $res->body; |
|
105
|
|
|
$temp = explode('<div id="problem-tags" class="col-lg-12 text-center">', $temp)[1]; |
|
106
|
|
|
$temp = explode('</div>', $temp)[0]; |
|
107
|
|
|
$cat = explode('<a href="/problems/tag/', $temp); |
|
108
|
|
|
for($i = 1;$i < count($cat); $i++) |
|
|
|
|
|
|
109
|
|
|
{ |
|
110
|
|
|
$temp = $cat[$i]; |
|
111
|
|
|
$temp = explode('">', $temp)[0]; |
|
112
|
|
|
array_push($cate,$temp); |
|
113
|
|
|
} |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
$temp = $res->body; |
|
117
|
|
|
$temp = explode('<div id="problem-body">', $temp)[1]; |
|
118
|
|
|
$content = explode('<div class="text-center">', $temp)[0]; |
|
119
|
|
|
|
|
120
|
|
|
$this->pro['description'] = $this->cacheImage(HtmlDomParser::str_get_html(explode('<h3>Input</h3>', $content)[0], true, true, DEFAULT_TARGET_CHARSET, false)); |
|
121
|
|
|
$content = explode('<h3>Input</h3>', $content)[1]; |
|
122
|
|
|
$this->pro['input'] = explode('<h3>Output</h3>', $content)[0]; |
|
123
|
|
|
$content = explode('<h3>Output</h3>', $content)[1]; |
|
124
|
|
|
$tgis->pro['output'] = explode('<h3>Example</h3>', $content)[0]; |
|
|
|
|
|
|
125
|
|
|
$content = explode('<h3>Example</h3>', $content)[1]; |
|
|
|
|
|
|
126
|
|
|
|
|
127
|
|
|
//TODO: Get compiler of each problem. |
|
128
|
|
|
} |
|
129
|
|
|
} |
|
130
|
|
|
|
|
131
|
|
|
public function getSpojProblem() { |
|
132
|
|
|
$types = ['classical','challenge','partial','tutorial','riddle','basics']; |
|
133
|
|
|
|
|
134
|
|
|
foreach($types as $type) { |
|
135
|
|
|
$iterator = 0; |
|
136
|
|
|
$endPos = false; |
|
137
|
|
|
while($endPos == false) { |
|
|
|
|
|
|
138
|
|
|
$url = "https://www.spoj.com/problems/".$type."/sort=0,start=".$iterator*50; |
|
139
|
|
|
$res = Requests::get($url); |
|
140
|
|
|
$problemTable = explode('<td align="left">', $res->body); |
|
141
|
|
|
for($cnt = 1; $cnt < count($problemTable); $cnt++) { |
|
|
|
|
|
|
142
|
|
|
$problemLink = $problemTable[$cnt]; |
|
143
|
|
|
$problemLink = explode('<a href="/problems/', $problemLink)[1]; |
|
144
|
|
|
$pid=explode('">', $problemLink)[0]; |
|
145
|
|
|
|
|
146
|
|
|
$problemLink = $problemTable[$cnt]; |
|
147
|
|
|
$problemLink = explode('See the best solutions.">', $problemLink)[1]; |
|
148
|
|
|
$solved_count=explode('</a></td>', $problemLink)[0]; |
|
|
|
|
|
|
149
|
|
|
|
|
150
|
|
|
$problemLink = $problemTable[$cnt]; |
|
151
|
|
|
$index = self::find('/<td align="center">([\s]*?)<\/td>/',$problemLink); |
|
152
|
|
|
|
|
153
|
|
|
$this->pro['pcode'] = "SPOJ-".$pid; |
|
154
|
|
|
$this->pro['OJ'] = 10; |
|
155
|
|
|
$this->pro['contest_id'] = null; |
|
156
|
|
|
$this->pro['index_id'] = $index; |
|
157
|
|
|
$this->pro['origin'] = "https://www.spoj.com/problems/".$pid; |
|
158
|
|
|
$this->pro['input_type']='standard input'; |
|
159
|
|
|
$this->pro['output_type']='standard output'; |
|
160
|
|
|
$this->getProblemDetail($pid); |
|
161
|
|
|
|
|
162
|
|
|
$problem=$problemModel->pid($this->pro['pcode']); |
|
163
|
|
|
|
|
164
|
|
|
if ($problem) { |
|
165
|
|
|
$problemModel->clearTags($problem); |
|
166
|
|
|
$new_pid=$this->update_problem($this->oid); |
|
167
|
|
|
}else { |
|
168
|
|
|
$new_pid=$this->insert_problem($this->oid); |
|
169
|
|
|
} |
|
170
|
|
|
} |
|
171
|
|
|
if(count($table) < 50) $endPos = true; |
|
|
|
|
|
|
172
|
|
|
$iterator++; |
|
173
|
|
|
} |
|
174
|
|
|
} |
|
175
|
|
|
} |
|
176
|
|
|
} |