|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace App\Babel\Crawl; |
|
4
|
|
|
|
|
5
|
|
|
use App\Models\ProblemModel; |
|
6
|
|
|
use KubAT\PhpSimple\HtmlDomParser; |
|
7
|
|
|
use Auth; |
|
8
|
|
|
|
|
9
|
|
|
class CrawlerBase |
|
10
|
|
|
{ |
|
11
|
|
|
public $pro=[ |
|
12
|
|
|
'file'=> 0, |
|
13
|
|
|
'pcode'=>'', |
|
14
|
|
|
'solved_count'=>'', |
|
15
|
|
|
'time_limit'=>'', |
|
16
|
|
|
'memory_limit'=>'', |
|
17
|
|
|
'title'=>'', |
|
18
|
|
|
'OJ'=>'', |
|
19
|
|
|
'description'=>'', |
|
20
|
|
|
'input'=>'', |
|
21
|
|
|
'output'=>'', |
|
22
|
|
|
'note'=>'', |
|
23
|
|
|
'input_type'=>'', |
|
24
|
|
|
'output_type'=>'', |
|
25
|
|
|
'contest_id'=>'', |
|
26
|
|
|
'index_id'=>'', |
|
27
|
|
|
'origin'=>'', |
|
28
|
|
|
'source'=>'', |
|
29
|
|
|
'sample'=>[], |
|
30
|
|
|
'markdown'=>0, |
|
31
|
|
|
'tot_score'=>1, |
|
32
|
|
|
'partial'=>0, |
|
33
|
|
|
'special_compiler'=>null, |
|
34
|
|
|
]; |
|
35
|
|
|
|
|
36
|
|
|
public $data=null; |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* Initial |
|
40
|
|
|
* |
|
41
|
|
|
* @return Response |
|
|
|
|
|
|
42
|
|
|
*/ |
|
43
|
|
|
public function __construct() |
|
44
|
|
|
{ |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
public static function cmp($a, $b) |
|
48
|
|
|
{ |
|
49
|
|
|
return ($a[1]>$b[1]) ?-1 : 1; |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
private function _cacheImage($data) |
|
53
|
|
|
{ |
|
54
|
|
|
if(!isset($data["ori"]) || !isset($data["path"]) || !isset($data["baseurl"]) || !isset($data["space_deli"]) || !isset($data["cookie"])){ |
|
55
|
|
|
throw new Exception("data is not completely exist in cacheImage"); |
|
|
|
|
|
|
56
|
|
|
} |
|
57
|
|
|
$ori = $data["ori"]; |
|
58
|
|
|
$path = $data["path"]; |
|
59
|
|
|
$baseurl = $data["baseurl"]; |
|
60
|
|
|
$space_deli = $data["space_deli"]; |
|
61
|
|
|
$cookie = $data["cookie"]; |
|
62
|
|
|
|
|
63
|
|
|
$para["path"]=$path; |
|
|
|
|
|
|
64
|
|
|
$para["base"]=$baseurl; |
|
65
|
|
|
$para["trans"]=!$space_deli; |
|
66
|
|
|
$para["cookie"]=$cookie; |
|
67
|
|
|
|
|
68
|
|
|
if ($space_deli) { |
|
69
|
|
|
$reg="/< *im[a]?g[^>]*src *= *[\"\\']?([^\"\\' >]*)[^>]*>/si"; |
|
70
|
|
|
} else { |
|
71
|
|
|
$reg="/< *im[a]?g[^>]*src *= *[\"\\']?([^\"\\'>]*)[^>]*>/si"; |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
return preg_replace_callback($reg, function($matches) use ($para) { |
|
75
|
|
|
global $config; |
|
76
|
|
|
$url=trim($matches[1]); |
|
77
|
|
|
if (stripos($url, "http://")===false && stripos($url, "https://")===false) { |
|
78
|
|
|
if ($para["trans"]) { |
|
79
|
|
|
$url=str_replace(" ", "%20", $url); |
|
80
|
|
|
} |
|
81
|
|
|
$url=$para["base"].$url; |
|
82
|
|
|
} |
|
83
|
|
|
$name=basename($url); |
|
84
|
|
|
$name="images/".strtr($name, ":", "_"); |
|
85
|
|
|
$result=str_replace(trim($matches[1]), "online_Judges/spoj/images/".strtr(basename($url), ":", "_"), $matches[0]); |
|
86
|
|
|
$ch=curl_init(); |
|
87
|
|
|
curl_setopt($ch, CURLOPT_URL, $url); |
|
|
|
|
|
|
88
|
|
|
curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false); |
|
89
|
|
|
curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true); |
|
90
|
|
|
if ($para["cookie"]!="") { |
|
91
|
|
|
curl_setopt($ch, CURLOPT_COOKIEFILE, $para["cookie"]); |
|
92
|
|
|
} |
|
93
|
|
|
curl_setopt($ch, CURLOPT_HEADER, 0); |
|
94
|
|
|
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true); |
|
95
|
|
|
$content=curl_exec($ch); |
|
|
|
|
|
|
96
|
|
|
curl_close($ch); |
|
|
|
|
|
|
97
|
|
|
$fp=fopen($name, "wb"); |
|
98
|
|
|
fwrite($fp, $content); |
|
|
|
|
|
|
99
|
|
|
fclose($fp); |
|
|
|
|
|
|
100
|
|
|
return $result; |
|
101
|
|
|
}, $ori); |
|
102
|
|
|
} |
|
103
|
|
|
|
|
104
|
|
|
public function procInfo($data) |
|
105
|
|
|
{ |
|
106
|
|
|
if(isset($data["path"])) $path = $data["path"]; else throw new Exception("path is not exist in data"); |
|
107
|
|
|
if(isset($data["baseurl"])) $baseurl = $data["baseurl"]; else throw new Exception("baseurl is not exist in data"); |
|
108
|
|
|
if(isset($data["space_deli"])) $space_deli = $data["space_deli"]; else $space_deli = true; |
|
109
|
|
|
if(isset($data["cookie"])) $cookie = $data["cookie"]; else $cookie = ""; |
|
110
|
|
|
|
|
111
|
|
|
$this->pro["description"]=$this->_cacheImage($this->pro["description"], $path, $baseurl, $space_deli, $cookie); |
|
|
|
|
|
|
112
|
|
|
$this->pro["input"]=$this->_cacheImage($this->pro["input"], $path, $baseurl, $space_deli, $cookie); |
|
113
|
|
|
$this->pro["output"]=$this->_cacheImage($this->pro["output"], $path, $baseurl, $space_deli, $cookie); |
|
114
|
|
|
$this->pro["note"]=$this->_cacheImage($this->pro["note"], $path, $baseurl, $space_deli, $cookie); |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
public function getUrl($url) |
|
118
|
|
|
{ |
|
119
|
|
|
$ch=curl_init(); |
|
120
|
|
|
curl_setopt($ch, CURLOPT_URL, $url); |
|
|
|
|
|
|
121
|
|
|
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true); |
|
122
|
|
|
curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false); |
|
123
|
|
|
curl_setopt($ch, CURLOPT_HEADER, 0); |
|
124
|
|
|
curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true); |
|
125
|
|
|
$content=curl_exec($ch); |
|
|
|
|
|
|
126
|
|
|
curl_close($ch); |
|
|
|
|
|
|
127
|
|
|
return $content; |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
|
|
public function insertProblem($oid=2) |
|
|
|
|
|
|
131
|
|
|
{ |
|
132
|
|
|
$problemModel=new ProblemModel(); |
|
133
|
|
|
return $problemModel->insertProblem($this->pro); |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
|
|
public function updateProblem($oid=2) |
|
|
|
|
|
|
137
|
|
|
{ |
|
138
|
|
|
$problemModel=new ProblemModel(); |
|
139
|
|
|
return $problemModel->updateProblem($this->pro); |
|
140
|
|
|
} |
|
141
|
|
|
} |
|
142
|
|
|
|