|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace App\Http\Controllers\VirtualCrawler\HDU; |
|
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
|
|
|
|
|
13
|
|
|
class HDU extends CrawlerBase |
|
14
|
|
|
{ |
|
15
|
|
|
public $oid=8; |
|
16
|
|
|
private $con; |
|
17
|
|
|
private $imgi; |
|
18
|
|
|
|
|
19
|
|
|
/** |
|
20
|
|
|
* Initial |
|
21
|
|
|
* |
|
22
|
|
|
* @return Response |
|
|
|
|
|
|
23
|
|
|
*/ |
|
24
|
|
|
public function __construct($action='crawl_problem', $con='all', $cached=false) |
|
|
|
|
|
|
25
|
|
|
{ |
|
26
|
|
|
set_time_limit(0); // Pandora's box, engage! |
|
27
|
|
|
if ($action=='judge_level') { |
|
28
|
|
|
$this->judge_level(); |
|
29
|
|
|
} else { |
|
30
|
|
|
$this->crawler($con); |
|
31
|
|
|
} |
|
32
|
|
|
} |
|
33
|
|
|
|
|
34
|
|
|
public function judge_level() |
|
35
|
|
|
{ |
|
36
|
|
|
// TODO |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
private static function find($pattern, $subject) |
|
40
|
|
|
{ |
|
41
|
|
|
if (preg_match($pattern, $subject, $matches)) { |
|
42
|
|
|
return $matches[1]; |
|
43
|
|
|
} |
|
44
|
|
|
return null; |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
private function cacheImage($dom) |
|
48
|
|
|
{ |
|
49
|
|
|
foreach ($dom->find('img') as $ele) { |
|
50
|
|
|
$src=str_replace('../../..', '', $ele->src); |
|
51
|
|
|
if (strpos($src, '://')!==false) { |
|
52
|
|
|
$url=$src; |
|
|
|
|
|
|
53
|
|
|
} elseif ($src[0]=='/') { |
|
54
|
|
|
$url='http://acm.hdu.edu.cn'.$src; |
|
55
|
|
|
} else { |
|
56
|
|
|
$url='http://acm.hdu.edu.cn/'.$src; |
|
57
|
|
|
} |
|
58
|
|
|
$res=Requests::get($src, ['Referer' => 'http://acm.hdu.edu.cn']); |
|
59
|
|
|
$ext=['image/jpeg'=>'.jpg', 'image/png'=>'.png', 'image/gif'=>'.gif', 'image/bmp'=>'.bmp']; |
|
60
|
|
|
if (isset($res->headers['content-type'])) { |
|
61
|
|
|
$cext=$ext[$res->headers['content-type']]; |
|
62
|
|
|
} else { |
|
63
|
|
|
$pos=strpos($ele->src, '.'); |
|
64
|
|
|
if ($pos===false) { |
|
65
|
|
|
$cext=''; |
|
66
|
|
|
} else { |
|
67
|
|
|
$cext=substr($ele->src, $pos); |
|
68
|
|
|
} |
|
69
|
|
|
} |
|
70
|
|
|
$fn=$this->con.'_'.($this->imgi++).$cext; |
|
71
|
|
|
$dir=base_path("public/external/hdu/img"); |
|
72
|
|
|
if (!file_exists($dir)) { |
|
73
|
|
|
mkdir($dir, 0755, true); |
|
74
|
|
|
} |
|
75
|
|
|
file_put_contents(base_path("public/external/hdu/img/$fn"), $res->body); |
|
76
|
|
|
$ele->src='/external/hdu/img/'.$fn; |
|
77
|
|
|
} |
|
78
|
|
|
return $dom; |
|
79
|
|
|
} |
|
80
|
|
|
|
|
81
|
|
|
public function crawler($con) |
|
82
|
|
|
{ |
|
83
|
|
|
if($con == "all") { |
|
84
|
|
|
return ; |
|
85
|
|
|
} |
|
86
|
|
|
$this->con = $con; |
|
87
|
|
|
$this->imgi = 1; |
|
88
|
|
|
$problemModel = new ProblemModel(); |
|
89
|
|
|
$res = Requests::get("http://acm.hdu.edu.cn/showproblem.php?pid={$con}"); |
|
90
|
|
|
if (strpos("No such problem",$res->body) !== false) { |
|
91
|
|
|
header('HTTP/1.1 404 Not Found'); |
|
92
|
|
|
die(); |
|
93
|
|
|
} |
|
94
|
|
|
else if(strpos("Invalid Parameter.",$res->body) !== false) { |
|
95
|
|
|
header('HTTP/1.1 404 Not Found'); |
|
96
|
|
|
die(); |
|
97
|
|
|
} |
|
98
|
|
|
else { |
|
99
|
|
|
$this->pro['pcode'] = "HDU".$con; |
|
100
|
|
|
$this->pro['OJ'] = 1; |
|
101
|
|
|
$this->pro['contest_id'] = null; |
|
102
|
|
|
$this->pro['index_id'] = $con; |
|
103
|
|
|
$this->pro['origin'] = "http://acm.hdu.edu.cn/showproblem.php?pid={$con}"; |
|
104
|
|
|
$this->pro['title'] = self::find("/<h1 style='color:#1A5CC8'>([\s\S]*?)<\/h1>/",$res->body); |
|
105
|
|
|
$this->pro['time_limit'] = self::find('/Time Limit:.*\/(.*) MS/',$res->body); |
|
106
|
|
|
$this->pro['memory_limit'] = self::find('/Memory Limit:.*\/(.*) K/',$res->body); |
|
107
|
|
|
$this->pro['solved_count'] = self::find("/Accepted Submission(s): ([\d+]*?)/",$res->body); |
|
108
|
|
|
$this->pro['input_type']='standard input'; |
|
109
|
|
|
$this->pro['output_type']='standard output'; |
|
110
|
|
|
$this->pro['description'] = $this->cacheImage(self::find("/this->problem Description.*<div class=panel_content>(.*)<\/div><div class=panel_bottom>/sU",$res->body)); |
|
111
|
|
|
$this->pro['input'] = self::find("/<div class=panel_title align=left>Input.*<div class=panel_content>(.*)<\/div><div class=panel_bottom>/sU",$res->body); |
|
112
|
|
|
$this->pro['output'] = self::find("/<div class=panel_title align=left>Output.*<div class=panel_content>(.*)<\/div><div class=panel_bottom>/sU",$res->body); |
|
113
|
|
|
$this->pro['sample'] = []; |
|
114
|
|
|
$this->pro['sample']['sample_input'] = self::find("/<pre><div.*>(.*)<\/div><\/pre>/sU",$res->body); |
|
115
|
|
|
$this->pro['sample']['sample_output'] = self::find("/<div.*>Sample Output<\/div><div.*><pre><div.*>(.*)<\/div><\/pre><\/div>/sU",$res->body); |
|
116
|
|
|
$this->pro['note'] = self::find("/<i>Hint<\/i><\/div>(.*)<\/div><i style='font-size:1px'>/sU",$res->body); |
|
117
|
|
|
$this->pro['source'] = self::find("/<div class=panel_title align=left>Source<\/div> (.*)<div class=panel_bottom>/sU",$res->body); |
|
118
|
|
|
$this->pro['force_raw'] = 0; |
|
119
|
|
|
$problem=$problemModel->pid($this->pro['pcode']); |
|
120
|
|
|
|
|
121
|
|
|
if ($problem) { |
|
122
|
|
|
$problemModel->clearTags($problem); |
|
123
|
|
|
$new_pid=$this->update_problem($this->oid); |
|
|
|
|
|
|
124
|
|
|
} else { |
|
125
|
|
|
$new_pid=$this->insert_problem($this->oid); |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
|
|
// $problemModel->addTags($new_pid, $tag); // not present |
|
129
|
|
|
} |
|
130
|
|
|
} |
|
131
|
|
|
} |
|
132
|
|
|
|
ZsgsDesign /
NOJ
Loading history...