|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
/* |
|
6
|
|
|
* This file is part of the Superdesk Web Publisher Core Bundle. |
|
7
|
|
|
* |
|
8
|
|
|
* Copyright 2020 Sourcefabric z.ú. and contributors. |
|
9
|
|
|
* |
|
10
|
|
|
* For the full copyright and license information, please see the |
|
11
|
|
|
* AUTHORS and LICENSE files distributed with this source code. |
|
12
|
|
|
* |
|
13
|
|
|
* @copyright 2020 Sourcefabric z.ú |
|
14
|
|
|
* @license http://www.superdesk.org/license |
|
15
|
|
|
*/ |
|
16
|
|
|
|
|
17
|
|
|
namespace SWP\Bundle\CoreBundle\AppleNews\Converter; |
|
18
|
|
|
|
|
19
|
|
|
use SWP\Bundle\CoreBundle\AppleNews\Component\Body; |
|
20
|
|
|
use SWP\Bundle\CoreBundle\AppleNews\Component\EmbedWebVideo; |
|
21
|
|
|
use SWP\Bundle\CoreBundle\AppleNews\Component\FacebookPost; |
|
22
|
|
|
use SWP\Bundle\CoreBundle\AppleNews\Component\Figure; |
|
23
|
|
|
use SWP\Bundle\CoreBundle\AppleNews\Component\Heading; |
|
24
|
|
|
use SWP\Bundle\CoreBundle\AppleNews\Component\Instagram; |
|
25
|
|
|
use SWP\Bundle\CoreBundle\AppleNews\Component\Quote; |
|
26
|
|
|
use SWP\Bundle\CoreBundle\AppleNews\Component\Tweet; |
|
27
|
|
|
|
|
28
|
|
|
final class ArticleBodyToComponentsConverter |
|
29
|
|
|
{ |
|
30
|
|
|
public function convert(string $body): array |
|
31
|
|
|
{ |
|
32
|
|
|
if ('' === $body) { |
|
33
|
|
|
return []; |
|
34
|
|
|
} |
|
35
|
|
|
|
|
36
|
|
|
$document = new \DOMDocument(); |
|
37
|
|
|
libxml_use_internal_errors(true); |
|
38
|
|
|
$document->loadHTML('<?xml encoding="UTF-8">'.self::stripHtmlTags($body)); |
|
39
|
|
|
$document->encoding = 'UTF-8'; |
|
40
|
|
|
libxml_clear_errors(); |
|
41
|
|
|
|
|
42
|
|
|
/** @var \DOMNodeList $body */ |
|
43
|
|
|
if (!($body = $document->getElementsByTagName('body')->item(0))) { |
|
44
|
|
|
throw new \InvalidArgumentException('Invalid HTML was provided'); |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
$components = []; |
|
48
|
|
|
foreach ($body->childNodes as $node) { |
|
49
|
|
|
switch ($node->nodeName) { |
|
50
|
|
|
case 'h1': |
|
51
|
|
|
case 'h2': |
|
52
|
|
|
case 'h3': |
|
53
|
|
|
case 'h4': |
|
54
|
|
|
case 'h5': |
|
55
|
|
|
case 'h6': |
|
56
|
|
|
if ('' !== $node->textContent) { |
|
57
|
|
|
$level = substr($node->nodeName, 1); |
|
58
|
|
|
$components[] = new Heading($node->textContent, (int) $level, 'fullMarginAboveHalfBelowLayout'); |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
break; |
|
62
|
|
|
case 'p': |
|
63
|
|
|
if ('' !== $node->textContent) { |
|
64
|
|
|
if ($node->getElementsByTagName('img')->length > 0) { |
|
65
|
|
|
$components[] = $this->createFigureComponent($node); |
|
66
|
|
|
|
|
67
|
|
|
break; |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
$bodyHtml = $node->ownerDocument->saveHTML($node); |
|
71
|
|
|
$components[] = new Body($bodyHtml, 'marginBetweenComponents'); |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
break; |
|
75
|
|
|
|
|
76
|
|
|
case 'figure': |
|
77
|
|
|
$components[] = $this->createFigureComponent($node); |
|
78
|
|
|
|
|
79
|
|
|
break; |
|
80
|
|
|
case 'div': |
|
81
|
|
|
if (!$node->hasAttribute('class')) { |
|
82
|
|
|
break; |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
|
|
$iframeElement = $node->getElementsByTagName('iframe') |
|
86
|
|
|
->item(0); |
|
87
|
|
|
|
|
88
|
|
|
if (null !== $iframeElement) { |
|
89
|
|
|
$webVideoUrl = $iframeElement->getAttribute('src'); |
|
90
|
|
|
$url = str_replace('\"', '', $webVideoUrl); |
|
91
|
|
|
if (false !== strpos($url, 'twitter.com')) { |
|
92
|
|
|
$parsedUrl = parse_url($url); |
|
93
|
|
|
parse_str($parsedUrl['query'], $url); |
|
94
|
|
|
|
|
95
|
|
|
$components[] = new Tweet($url['url']); |
|
96
|
|
|
} elseif (false !== strpos($url, 'iframe.ly')) { |
|
97
|
|
|
$parsedUrl = parse_url($url); |
|
98
|
|
|
parse_str($parsedUrl['query'], $url); |
|
99
|
|
|
|
|
100
|
|
|
$components[] = new FacebookPost($url['url']); |
|
101
|
|
|
} elseif (false !== strpos($url, 'facebook.com')) { |
|
102
|
|
|
$parsedUrl = parse_url($url); |
|
103
|
|
|
parse_str($parsedUrl['query'], $url); |
|
104
|
|
|
if ($this->isValidFacebookPostUrl($url['href'])) { |
|
105
|
|
|
$components[] = new FacebookPost($url['href']); |
|
106
|
|
|
} |
|
107
|
|
|
} elseif (false !== strpos($url, 'youtube.com') || false !== strpos($url, 'vimeo.com')) { |
|
108
|
|
|
$components[] = new EmbedWebVideo($url); |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
break; |
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
|
|
$instagramElement = $node->getElementsByTagName('blockquote') |
|
115
|
|
|
->item(0); |
|
116
|
|
|
|
|
117
|
|
|
if (null !== $instagramElement) { |
|
118
|
|
|
$instagramUrl = $instagramElement->getAttribute('data-instgrm-permalink'); |
|
119
|
|
|
$url = str_replace('\"', '', $instagramUrl); |
|
120
|
|
|
$components[] = new Instagram($url); |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
break; |
|
124
|
|
|
|
|
125
|
|
|
case 'blockquote': |
|
126
|
|
|
if ('' !== $node->textContent) { |
|
127
|
|
|
$components[] = new Quote('“'.$node->textContent.'”'); |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
|
|
break; |
|
131
|
|
|
} |
|
132
|
|
|
} |
|
133
|
|
|
|
|
134
|
|
|
return $components; |
|
135
|
|
|
} |
|
136
|
|
|
|
|
137
|
|
|
public static function stripHtmlTags(string $html): string |
|
138
|
|
|
{ |
|
139
|
|
|
return preg_replace('/<script.*>.*<\/script>/isU', '', $html); |
|
140
|
|
|
} |
|
141
|
|
|
|
|
142
|
|
|
private function isValidFacebookPostUrl(string $url): bool |
|
143
|
|
|
{ |
|
144
|
|
|
preg_match('/^https:\/\/www\.facebook\.com\/(photo(\.php|s)|permalink\.php|[^\/]+\/(activity|posts))[\/?].*$/', $url, $matches); |
|
145
|
|
|
|
|
146
|
|
|
return !empty($matches); |
|
147
|
|
|
} |
|
148
|
|
|
|
|
149
|
|
|
private function createFigureComponent(\DOMElement $node): Figure |
|
150
|
|
|
{ |
|
151
|
|
|
$src = $node->getElementsByTagName('img') |
|
152
|
|
|
->item(0) |
|
153
|
|
|
->getAttribute('src'); |
|
154
|
|
|
|
|
155
|
|
|
$caption = $node->getElementsByTagName('figcaption') |
|
156
|
|
|
->item(0) |
|
157
|
|
|
->textContent; |
|
158
|
|
|
|
|
159
|
|
|
return new Figure($src, $caption); |
|
160
|
|
|
} |
|
161
|
|
|
} |
|
162
|
|
|
|
superdesk /
web-publisher
