1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace TreeHouse\IoBundle\Scrape\Modifier\Item\Mapper; |
4
|
|
|
|
5
|
|
|
use Symfony\Component\DomCrawler\Crawler; |
6
|
|
|
use Symfony\Component\HttpFoundation\ParameterBag; |
7
|
|
|
use TreeHouse\Feeder\Modifier\Item\Mapper\MapperInterface; |
8
|
|
|
|
9
|
|
|
class NodeMapper implements MapperInterface, CrawlerAwareInterface |
10
|
|
|
{ |
11
|
|
|
/** |
12
|
|
|
* @var Crawler |
13
|
|
|
*/ |
14
|
|
|
protected $crawler; |
15
|
|
|
|
16
|
|
|
/** |
17
|
|
|
* @var array |
18
|
|
|
*/ |
19
|
|
|
protected $mapping = []; |
20
|
|
|
|
21
|
|
|
/** |
22
|
|
|
* @var string[] |
23
|
|
|
*/ |
24
|
|
|
protected $filters = []; |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* @var callable[] |
28
|
|
|
*/ |
29
|
|
|
protected $extractors = []; |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* @param array $mapping |
33
|
|
|
*/ |
34
|
24 |
|
public function __construct(array $mapping) |
35
|
|
|
{ |
36
|
24 |
|
foreach ($mapping as $name => $selector) { |
37
|
14 |
|
if (is_string($selector)) { |
38
|
8 |
|
$extractor = [$this, 'extractHtml']; |
39
|
10 |
|
} elseif (is_array($selector) && sizeof($selector) === 2) { |
40
|
8 |
|
list($selector, $extractor) = $selector; |
41
|
|
|
} else { |
42
|
2 |
|
throw new \InvalidArgumentException('A mapping value must be either a string or array<string, callable>'); |
43
|
|
|
} |
44
|
|
|
|
45
|
12 |
|
$this->addMapping($name, $selector, $extractor); |
46
|
|
|
} |
47
|
20 |
|
} |
48
|
|
|
|
49
|
|
|
/** |
50
|
|
|
* @return array |
51
|
|
|
*/ |
52
|
6 |
|
public function getMapping() |
53
|
|
|
{ |
54
|
6 |
|
return $this->mapping; |
55
|
|
|
} |
56
|
|
|
|
57
|
|
|
/** |
58
|
|
|
* @param string $name |
59
|
|
|
* @param string $selector |
60
|
|
|
* @param string|callable $extractor |
61
|
|
|
*/ |
62
|
20 |
|
public function addMapping($name, $selector, $extractor = 'extractHtml') |
63
|
|
|
{ |
64
|
20 |
|
if (is_string($extractor) && method_exists($this, $extractor)) { |
65
|
8 |
|
$extractor = [$this, $extractor]; |
66
|
|
|
} |
67
|
|
|
|
68
|
20 |
|
if (!is_callable($extractor)) { |
69
|
4 |
|
throw new \InvalidArgumentException( |
70
|
4 |
|
sprintf('The extractor of a mapping must be a callable, but got %s', json_encode($extractor)) |
71
|
|
|
); |
72
|
|
|
} |
73
|
|
|
|
74
|
16 |
|
$this->mapping[$name] = $selector; |
75
|
16 |
|
$this->extractors[$name] = $extractor; |
76
|
16 |
|
$this->filters[$name] = 'filter'; |
77
|
|
|
|
78
|
16 |
|
if (preg_match('~^//~', $selector) || strpos($selector, 'descendant-or-self::') === 0) { |
79
|
|
|
$this->filters[$name] = 'filterXpath'; |
80
|
|
|
} |
81
|
16 |
|
} |
82
|
|
|
|
83
|
|
|
/** |
84
|
|
|
* @inheritdoc |
85
|
|
|
*/ |
86
|
4 |
|
public function setCrawler(Crawler $crawler) |
87
|
|
|
{ |
88
|
4 |
|
$this->crawler = $crawler; |
89
|
4 |
|
} |
90
|
|
|
|
91
|
|
|
/** |
92
|
|
|
* @param ParameterBag $item |
93
|
|
|
* |
94
|
|
|
* @throws \LogicException |
95
|
|
|
* |
96
|
|
|
* @return ParameterBag |
97
|
|
|
*/ |
98
|
6 |
|
public function map(ParameterBag $item) |
99
|
|
|
{ |
100
|
6 |
|
if (null === $this->crawler) { |
101
|
2 |
|
throw new \LogicException('setCrawler() should be called before map()'); |
102
|
|
|
} |
103
|
|
|
|
104
|
4 |
|
foreach ($this->mapping as $name => $selector) { |
105
|
4 |
|
$filter = $this->filters[$name]; |
106
|
4 |
|
$extractor = $this->extractors[$name]; |
107
|
|
|
|
108
|
|
|
/** @var Crawler $node */ |
109
|
4 |
|
$node = $this->crawler->$filter($selector); |
110
|
|
|
|
111
|
4 |
|
if ($node->count() === 0) { |
112
|
|
|
$value = null; |
113
|
|
|
} else { |
114
|
4 |
|
$value = call_user_func($extractor, $name, $node); |
115
|
|
|
} |
116
|
|
|
|
117
|
4 |
|
$item->set($name, $value); |
118
|
|
|
} |
119
|
|
|
|
120
|
4 |
|
return $item; |
121
|
|
|
} |
122
|
|
|
|
123
|
|
|
/** |
124
|
|
|
* @param string $field |
125
|
|
|
* @param Crawler $node |
126
|
|
|
* |
127
|
|
|
* @return string |
128
|
|
|
*/ |
129
|
2 |
|
protected function extractHtml($field, Crawler $node) |
130
|
|
|
{ |
131
|
2 |
|
return $node->html(); |
132
|
|
|
} |
133
|
|
|
|
134
|
|
|
/** |
135
|
|
|
* @param string $field |
136
|
|
|
* @param Crawler $node |
137
|
|
|
* |
138
|
|
|
* @return string |
139
|
|
|
*/ |
140
|
2 |
|
protected function extractText($field, Crawler $node) |
|
|
|
|
141
|
|
|
{ |
142
|
2 |
|
return $node->text(); |
143
|
|
|
} |
144
|
|
|
} |
145
|
|
|
|
This check looks from parameters that have been defined for a function or method, but which are not used in the method body.