|
1
|
|
|
<?php |
|
2
|
|
|
/* |
|
3
|
|
|
* This file is part of the feed-io package. |
|
4
|
|
|
* |
|
5
|
|
|
* (c) Alexandre Debril <[email protected]> |
|
6
|
|
|
* |
|
7
|
|
|
* For the full copyright and license information, please view the LICENSE |
|
8
|
|
|
* file that was distributed with this source code. |
|
9
|
|
|
*/ |
|
10
|
|
|
|
|
11
|
|
|
namespace FeedIo; |
|
12
|
|
|
|
|
13
|
|
|
use DOMDocument; |
|
14
|
|
|
use FeedIo\Feed\ItemInterface; |
|
15
|
|
|
use FeedIo\Feed\NodeInterface; |
|
16
|
|
|
use FeedIo\Parser\MissingFieldsException; |
|
17
|
|
|
use FeedIo\Parser\UnsupportedFormatException; |
|
18
|
|
|
use Psr\Log\LoggerInterface; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* Parses a DOM document if its format matches the parser's standard |
|
22
|
|
|
* |
|
23
|
|
|
* Depends on : |
|
24
|
|
|
* - FeedIo\StandardAbstract |
|
25
|
|
|
* - Psr\Log\LoggerInterface |
|
26
|
|
|
* |
|
27
|
|
|
*/ |
|
28
|
|
|
class Parser |
|
29
|
|
|
{ |
|
30
|
|
|
|
|
31
|
|
|
/** |
|
32
|
|
|
* @var \Psr\Log\LoggerInterface |
|
33
|
|
|
*/ |
|
34
|
|
|
protected $logger; |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* @var array[FilterInterface] |
|
38
|
|
|
*/ |
|
39
|
|
|
protected $filters = array(); |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* @var StandardAbstract |
|
43
|
|
|
*/ |
|
44
|
|
|
protected $standard; |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* @param StandardAbstract $standard |
|
48
|
|
|
* @param LoggerInterface $logger |
|
49
|
|
|
*/ |
|
50
|
34 |
|
public function __construct(StandardAbstract $standard, LoggerInterface $logger) |
|
51
|
|
|
{ |
|
52
|
34 |
|
$this->standard = $standard; |
|
53
|
34 |
|
$this->logger = $logger; |
|
54
|
34 |
|
} |
|
55
|
|
|
|
|
56
|
|
|
/** |
|
57
|
|
|
* @return StandardAbstract |
|
58
|
|
|
*/ |
|
59
|
21 |
|
public function getStandard() |
|
60
|
|
|
{ |
|
61
|
21 |
|
return $this->standard; |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
/** |
|
65
|
|
|
* @param $tagName |
|
66
|
|
|
* @return bool |
|
67
|
|
|
*/ |
|
68
|
7 |
|
public function isItem($tagName) |
|
69
|
|
|
{ |
|
70
|
7 |
|
return (strtolower($this->standard->getItemNodeName()) === strtolower($tagName)); |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
/** |
|
74
|
|
|
* @param FilterInterface $filter |
|
75
|
|
|
* @return $this |
|
76
|
|
|
*/ |
|
77
|
10 |
|
public function addFilter(FilterInterface $filter) |
|
78
|
|
|
{ |
|
79
|
10 |
|
$this->filters[] = $filter; |
|
80
|
|
|
|
|
81
|
10 |
|
return $this; |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
/** |
|
85
|
|
|
* @param DOMDocument $document |
|
86
|
|
|
* @param FeedInterface $feed |
|
87
|
8 |
|
* @return \FeedIo\FeedInterface |
|
88
|
|
|
* @throws Parser\MissingFieldsException |
|
89
|
8 |
|
* @throws Parser\UnsupportedFormatException |
|
90
|
1 |
|
*/ |
|
91
|
|
|
public function parse(DOMDocument $document, FeedInterface $feed) |
|
92
|
|
|
{ |
|
93
|
8 |
|
if (!$this->standard->canHandle($document)) { |
|
94
|
|
|
throw new UnsupportedFormatException('this is not a supported format'); |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
$this->checkBodyStructure($document, $this->standard->getMandatoryFields()); |
|
98
|
|
|
$element = $this->standard->getMainElement($document); |
|
99
|
|
|
|
|
100
|
|
|
$this->parseNode($feed, $element, $this->standard->getFeedRuleSet()); |
|
101
|
|
|
|
|
102
|
|
|
return $feed; |
|
103
|
8 |
|
} |
|
104
|
|
|
|
|
105
|
8 |
|
/** |
|
106
|
1 |
|
* @param DOMDocument $document |
|
107
|
|
|
* @param array $mandatoryFields |
|
108
|
|
|
* @return $this |
|
109
|
7 |
|
* @throws MissingFieldsException |
|
110
|
7 |
|
*/ |
|
111
|
7 |
|
public function checkBodyStructure(DOMDocument $document, array $mandatoryFields) |
|
112
|
|
|
{ |
|
113
|
7 |
|
$errors = array(); |
|
114
|
|
|
|
|
115
|
7 |
|
$element = $document->documentElement; |
|
116
|
|
|
foreach ($mandatoryFields as $field) { |
|
117
|
|
|
$list = $element->getElementsByTagName($field); |
|
118
|
|
|
if (0 === $list->length) { |
|
119
|
|
|
$errors[] = $field; |
|
120
|
|
|
} |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
if (!empty($errors)) { |
|
124
|
9 |
|
$message = "missing mandatory field(s) : ".implode(',', $errors); |
|
125
|
|
|
$this->logger->warning($message); |
|
126
|
9 |
|
throw new MissingFieldsException($message); |
|
127
|
|
|
} |
|
128
|
9 |
|
|
|
129
|
9 |
|
return $this; |
|
130
|
2 |
|
} |
|
131
|
2 |
|
|
|
132
|
2 |
|
/** |
|
133
|
|
|
* @param NodeInterface $item |
|
134
|
|
|
* @param \DOMElement $element |
|
135
|
|
|
* @param RuleSet $ruleSet |
|
136
|
9 |
|
* @return NodeInterface |
|
137
|
1 |
|
*/ |
|
138
|
1 |
|
public function parseNode(NodeInterface $item, \DOMElement $element, RuleSet $ruleSet) |
|
139
|
1 |
|
{ |
|
140
|
|
|
foreach ($element->childNodes as $node) { |
|
141
|
|
|
if ($node instanceof \DOMElement) { |
|
142
|
8 |
|
$this->handleNode($item, $node, $ruleSet); |
|
143
|
|
|
} |
|
144
|
|
|
} |
|
145
|
|
|
|
|
146
|
|
|
return $item; |
|
147
|
|
|
} |
|
148
|
|
|
|
|
149
|
|
|
/** |
|
150
|
|
|
* @param NodeInterface $item |
|
151
|
8 |
|
* @param \DOMElement $node |
|
152
|
|
|
* @param RuleSet $ruleSet |
|
153
|
8 |
|
* @return $this |
|
154
|
7 |
|
*/ |
|
155
|
7 |
|
protected function handleNode(NodeInterface $item, \DOMElement $node, RuleSet $ruleSet) |
|
156
|
|
|
{ |
|
157
|
|
|
if ($this->isItem($node->tagName) && $item instanceof FeedInterface) { |
|
158
|
|
|
$newItem = $this->parseNode($item->newItem(), $node, $this->standard->getItemRuleSet()); |
|
159
|
8 |
|
$this->addValidItem($item, $newItem); |
|
160
|
|
|
} else { |
|
161
|
|
|
$rule = $ruleSet->get($node->tagName); |
|
162
|
|
|
$rule->setProperty($item, $node); |
|
163
|
|
|
} |
|
164
|
|
|
|
|
165
|
|
|
return $this; |
|
166
|
|
|
} |
|
167
|
|
|
|
|
168
|
7 |
|
/** |
|
169
|
|
|
* @param FeedInterface $feed |
|
170
|
7 |
|
* @param NodeInterface $item |
|
171
|
6 |
|
* @return $this |
|
172
|
6 |
|
*/ |
|
173
|
|
|
public function addValidItem(FeedInterface $feed, NodeInterface $item) |
|
174
|
7 |
|
{ |
|
175
|
7 |
|
if ($item instanceof ItemInterface && $this->isValid($item)) { |
|
176
|
|
|
$feed->add($item); |
|
177
|
|
|
} |
|
178
|
7 |
|
|
|
179
|
|
|
return $this; |
|
180
|
|
|
} |
|
181
|
|
|
|
|
182
|
|
|
/** |
|
183
|
|
|
* @param ItemInterface $item |
|
184
|
|
|
* @return bool |
|
185
|
|
|
*/ |
|
186
|
6 |
|
public function isValid(ItemInterface $item) |
|
187
|
|
|
{ |
|
188
|
6 |
|
foreach ($this->filters as $filter) { |
|
189
|
6 |
|
if (!$filter->isValid($item)) { |
|
190
|
|
|
return false; |
|
191
|
|
|
} |
|
192
|
6 |
|
} |
|
193
|
|
|
|
|
194
|
|
|
return true; |
|
195
|
|
|
} |
|
196
|
|
|
} |
|
197
|
|
|
|
alexdebril /
feed-io
