|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace App\Http\Controllers\VirtualCrawler\Vijos; |
|
4
|
|
|
|
|
5
|
|
|
use App\Http\Controllers\VirtualCrawler\CrawlerBase; |
|
6
|
|
|
use App\Models\ProblemModel; |
|
7
|
|
|
use KubAT\PhpSimple\HtmlDomParser; |
|
8
|
|
|
use Auth,Requests,Exception; |
|
9
|
|
|
|
|
10
|
|
|
class Vijos extends CrawlerBase |
|
11
|
|
|
{ |
|
12
|
|
|
public $oid=5; |
|
13
|
|
|
private $con, $imgi; |
|
|
|
|
|
|
14
|
|
|
/** |
|
15
|
|
|
* Initial |
|
16
|
|
|
* |
|
17
|
|
|
* @return Response |
|
|
|
|
|
|
18
|
|
|
*/ |
|
19
|
|
|
public function __construct($action = 'crawl_problem', $con = 'all', $cached = false) |
|
|
|
|
|
|
20
|
|
|
{ |
|
21
|
|
|
set_time_limit(0); // Pandora's box, engage! |
|
22
|
|
|
if ($action=='judge_level') { |
|
23
|
|
|
$this->judge_level(); |
|
24
|
|
|
} else { |
|
25
|
|
|
$this->crawling($con); |
|
26
|
|
|
} |
|
27
|
|
|
} |
|
28
|
|
|
|
|
29
|
|
|
public function judge_level() |
|
30
|
|
|
{ |
|
31
|
|
|
// TODO |
|
32
|
|
|
} |
|
33
|
|
|
|
|
34
|
|
|
public function crawling($con) |
|
35
|
|
|
{ |
|
36
|
|
|
if ($con == 'all') { |
|
37
|
|
|
// TODO |
|
38
|
|
|
return; |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
|
|
try { |
|
42
|
|
|
$dom = HtmlDomParser::file_get_html('https://vijos.org/p/'.$con, false, null, 0, -1, true, true, DEFAULT_TARGET_CHARSET, false); |
|
43
|
|
|
} |
|
44
|
|
|
catch (Exception $e) { |
|
45
|
|
|
if (strpos($e->getMessage(), '404 Not Found') !== false) { |
|
46
|
|
|
header('HTTP/1.1 404 Not Found'); |
|
47
|
|
|
die(); |
|
48
|
|
|
} |
|
49
|
|
|
if (strpos($e->getMessage(), '403 Forbidden') !== false) { |
|
50
|
|
|
header('HTTP/1.1 403 Forbidden'); |
|
51
|
|
|
die(); |
|
52
|
|
|
} |
|
53
|
|
|
throw $e; |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
$mainDiv = $dom->find(".section__body", 0); |
|
57
|
|
|
|
|
58
|
|
|
$eles = $mainDiv->children(); |
|
59
|
|
|
array_push($eles, null); |
|
60
|
|
|
$this->pro['description'] = null; |
|
61
|
|
|
$this->pro['input'] = null; |
|
62
|
|
|
$this->pro['output'] = null; |
|
63
|
|
|
$this->pro['sample'] = []; |
|
64
|
|
|
$this->pro['note'] = null; |
|
65
|
|
|
$this->pro['sampleDesc'] = null; |
|
66
|
|
|
$this->pro['limit'] = null; |
|
67
|
|
|
$patterns = [ |
|
68
|
|
|
'description' => '<h1>描述</h1>', |
|
69
|
|
|
'_format' => '<h1>格式</h1>', |
|
70
|
|
|
'input' => '<h2>输入格式</h2>', |
|
71
|
|
|
'output' => '<h2>输出格式</h2>', |
|
72
|
|
|
'_sample' => '/^<h1>样例\d+<\/h1>$/u', |
|
73
|
|
|
'__sampleInput' => '/^<h2>样例输入\d+<\/h2>$/u', |
|
74
|
|
|
'__sampleOutput' => '/^<h2>样例输出\d+<\/h2>$/u', |
|
75
|
|
|
'limit' => '<h1>限制</h1>', |
|
76
|
|
|
'note' => '<h1>提示</h1>', |
|
77
|
|
|
'sampleDesc' => '/<h1>样例(说明|解释)<\/h1>|<h2>样例说明1<\/h2>/', // P2036 has <h2>样例说明1</h2> |
|
78
|
|
|
'source' => '<h1>来源</h1>', |
|
79
|
|
|
]; |
|
80
|
|
|
$lastPart = ''; |
|
81
|
|
|
$content = ''; |
|
82
|
|
|
$cursample = []; |
|
83
|
|
|
foreach ($eles as $ele) { |
|
84
|
|
|
$html = $ele ? $ele->outertext : null; |
|
85
|
|
|
$match = !$ele; |
|
86
|
|
|
if (!$match) { |
|
87
|
|
|
foreach ($patterns as $key=>$value) { |
|
88
|
|
|
if ($value[0] != '/' && $html == $value || $value[0] == '/' && preg_match($value, $html)) { |
|
|
|
|
|
|
89
|
|
|
$match = $key; |
|
90
|
|
|
break; |
|
91
|
|
|
} |
|
92
|
|
|
} |
|
93
|
|
|
} |
|
94
|
|
|
if (!$lastPart) { |
|
95
|
|
|
if ($match) $lastPart = $match; |
|
96
|
|
|
continue; |
|
97
|
|
|
} |
|
98
|
|
|
if ($match) { |
|
99
|
|
|
if ($lastPart[0] != '_') { |
|
100
|
|
|
$this->pro[$lastPart] = $content; |
|
101
|
|
|
$content = ''; |
|
102
|
|
|
} else if ($lastPart == '__sampleOutput') { // Assume output always follows by input |
|
103
|
|
|
array_push($this->pro['sample'], $cursample); |
|
104
|
|
|
$cursample = []; |
|
105
|
|
|
} |
|
106
|
|
|
$lastPart = $match; |
|
107
|
|
|
} else { |
|
108
|
|
|
if ($lastPart[1] != '_') { |
|
109
|
|
|
if ($lastPart != 'source') $content .= $html; |
|
110
|
|
|
else $content .= $ele->innertext; |
|
111
|
|
|
} else { // Code |
|
112
|
|
|
$code = trim($ele->find('code', 0)->innertext); |
|
113
|
|
|
if ($lastPart == '__sampleInput') { if (isset($cursample['sampleInput'])) die($con); } |
|
|
|
|
|
|
114
|
|
|
else { if (isset($cursample['sampleOutput'])) die($con); } |
|
|
|
|
|
|
115
|
|
|
if (count($ele->children()) != 1) die($con); |
|
|
|
|
|
|
116
|
|
|
if ($lastPart == '__sampleInput') $cursample['sample_input'] = $code; |
|
117
|
|
|
else $cursample['sample_output'] = $code; |
|
118
|
|
|
} |
|
119
|
|
|
} |
|
120
|
|
|
if (!$ele) break; |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
$this->pro['time_limit'] = 1000; |
|
124
|
|
|
$this->pro['memory_limit'] = 262144; |
|
125
|
|
|
if ($this->pro['sampleDesc']) { |
|
126
|
|
|
$this->pro['note'] = '<h3>样例说明</h3>'.$this->pro['sampleDesc'].$this->pro['note']; |
|
|
|
|
|
|
127
|
|
|
} |
|
128
|
|
|
if ($this->pro['limit']) { |
|
129
|
|
|
$this->pro['note'] = $this->pro['limit'].$this->pro['note']; |
|
130
|
|
|
$this->pro['time_limit'] = 0; |
|
131
|
|
|
$this->pro['memory_limit'] = 0; |
|
132
|
|
|
} |
|
133
|
|
|
|
|
134
|
|
|
$this->pro['pcode'] = 'VIJ'.$con; |
|
135
|
|
|
$this->pro['OJ'] = $this->oid; |
|
136
|
|
|
$this->pro['contest_id'] = null; |
|
137
|
|
|
$this->pro['index_id'] = $con; |
|
138
|
|
|
$this->pro['origin'] = 'https://vijos.org/p/'.$con; |
|
139
|
|
|
$this->pro['title'] = $dom->find('.section__header', 0)->find('h1', 0)->innertext; |
|
140
|
|
|
$this->pro['input_type'] = 'standard input'; |
|
141
|
|
|
$this->pro['output_type'] = 'standard output'; |
|
142
|
|
|
|
|
143
|
|
|
$this->pro['markdown'] = 0; |
|
144
|
|
|
$this->pro['tot_score'] = 100; |
|
145
|
|
|
$this->pro["partial"] = 1; |
|
146
|
|
|
$this->pro['source'] = 'Vijos'; // Force Override |
|
147
|
|
|
|
|
148
|
|
|
$info = $dom->find(".horizontal", 0); |
|
149
|
|
|
preg_match('/<dt>已通过<\/dt>[\s\S]*<dd>(\d+)<\/dd>/', $info->innertext, $match); |
|
150
|
|
|
$this->pro['solved_count'] = $match[1]; |
|
151
|
|
|
|
|
152
|
|
|
$problemModel=new ProblemModel(); |
|
153
|
|
|
$problem=$problemModel->pid($this->pro['pcode']); |
|
154
|
|
|
|
|
155
|
|
|
if ($problem) { |
|
156
|
|
|
$problemModel->clearTags($problem); |
|
157
|
|
|
$new_pid=$this->update_problem($this->oid); |
|
158
|
|
|
} else { |
|
159
|
|
|
$new_pid=$this->insert_problem($this->oid); |
|
160
|
|
|
} |
|
161
|
|
|
|
|
162
|
|
|
$tags = $info->find('.hasjs--hide', 0); |
|
163
|
|
|
if ($tags) { |
|
164
|
|
|
foreach ($tags->find('a') as $tag) { |
|
165
|
|
|
$problemModel->addTags($new_pid, $tag->innertext); |
|
166
|
|
|
} |
|
167
|
|
|
} |
|
168
|
|
|
} |
|
169
|
|
|
} |
|
170
|
|
|
|
ZsgsDesign /
NOJ
Loading history...