|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Jerodev\Diglett; |
|
4
|
|
|
|
|
5
|
|
|
use Symfony\Component\DomCrawler\Crawler; |
|
6
|
|
|
|
|
7
|
|
|
class Diglett |
|
8
|
|
|
{ |
|
9
|
|
|
/** |
|
10
|
|
|
* The Symfony DomCrawler to work with. |
|
11
|
|
|
* |
|
12
|
|
|
* @var Crawler |
|
13
|
|
|
*/ |
|
14
|
|
|
private $crawler; |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* The css selector parser |
|
18
|
|
|
* |
|
19
|
|
|
* @var CssFilterParser |
|
20
|
|
|
*/ |
|
21
|
|
|
private $cssFilterParser; |
|
22
|
|
|
|
|
23
|
|
|
/** |
|
24
|
|
|
* Create a diglett instance from a Symfony Crawler. |
|
25
|
|
|
* |
|
26
|
|
|
* @param Crawler |
|
27
|
|
|
* @param array $cssFilter An array of extra ICssFilterl classes to filter on |
|
28
|
|
|
*/ |
|
29
|
|
|
public function __construct(Crawler $crawler, array $cssFilters = []) |
|
30
|
|
|
{ |
|
31
|
|
|
$this->crawler = $crawler; |
|
32
|
|
|
$this->cssFilterParser = new CssFilterParser($cssFilters); |
|
33
|
|
|
} |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* Get the underlying crawler object |
|
37
|
|
|
* |
|
38
|
|
|
* @return Crawler |
|
39
|
|
|
*/ |
|
40
|
|
|
public function getCrawler(): Crawler |
|
41
|
|
|
{ |
|
42
|
|
|
return $this->crawler; |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
|
|
/** |
|
46
|
|
|
* Use special css selectors to filter on the current node collection |
|
47
|
|
|
* |
|
48
|
|
|
* @param string $selector |
|
49
|
|
|
* @return Crawler|null |
|
50
|
|
|
*/ |
|
51
|
|
|
public function filter(string $selector): ?Crawler |
|
52
|
|
|
{ |
|
53
|
|
|
$parsedSelector = $this->cssFilterParser->parse($selector); |
|
54
|
|
|
|
|
55
|
|
|
$crawler = $this->getCrawler(); |
|
56
|
|
|
foreach ($parsedSelector as $part) |
|
57
|
|
|
{ |
|
58
|
|
|
$crawler = $crawler->filter($part['selector']); |
|
59
|
|
|
|
|
60
|
|
|
foreach ($part['functions'] as $function) |
|
61
|
|
|
{ |
|
62
|
|
|
$crawler = $function->filterNodes($crawler); |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
if (empty($crawler) || $crawler->count() === 0) |
|
66
|
|
|
{ |
|
67
|
|
|
break; |
|
68
|
|
|
} |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
return $crawler; |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
|
|
75
|
|
|
/** |
|
76
|
|
|
* Use special css selectors to fetch several values |
|
77
|
|
|
* |
|
78
|
|
|
* @param array $selectors |
|
79
|
|
|
* @return array |
|
80
|
|
|
*/ |
|
81
|
|
|
public function getTexts(array $selectors): array |
|
82
|
|
|
{ |
|
83
|
|
|
$results = []; |
|
84
|
|
|
foreach ($selectors as $key => $value) |
|
85
|
|
|
{ |
|
86
|
|
|
$results[$key] = $this->getText($value); |
|
87
|
|
|
} |
|
88
|
|
|
|
|
89
|
|
|
return $results; |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
|
|
/** |
|
93
|
|
|
* Get the value for a single special css selector |
|
94
|
|
|
* |
|
95
|
|
|
* @param string $selector |
|
96
|
|
|
* @return string|null |
|
97
|
|
|
*/ |
|
98
|
|
|
public function getText(string $selector): ?string |
|
99
|
|
|
{ |
|
100
|
|
|
$attribute = null; |
|
101
|
|
|
$selector = preg_replace_callback( |
|
102
|
|
|
'/\{(.*?)\}$/', |
|
103
|
|
|
function ($matches) use (&$attribute) { |
|
104
|
|
|
$attribute = $matches[1] ?? null; |
|
105
|
|
|
}, |
|
106
|
|
|
$selector |
|
107
|
|
|
); |
|
108
|
|
|
|
|
109
|
|
|
$crawler = $this->filter($selector); |
|
110
|
|
|
if ($crawler === null) |
|
111
|
|
|
{ |
|
112
|
|
|
return null; |
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
|
|
return $attribute === null ? $crawler->text() : $crawler->attr($attribute); |
|
116
|
|
|
} |
|
117
|
|
|
} |
|
118
|
|
|
|
jerodev /
diglett
