|
1
|
|
|
<?php |
|
2
|
|
|
namespace Bookdown\Bookdown\Process\Resource; |
|
3
|
|
|
|
|
4
|
|
|
use Bookdown\Bookdown\Config\RootConfig; |
|
5
|
|
|
use Psr\Log\LoggerInterface; |
|
6
|
|
|
use Bookdown\Bookdown\Content\Page; |
|
7
|
|
|
use Bookdown\Bookdown\Fsio; |
|
8
|
|
|
use Bookdown\Bookdown\Process\ProcessInterface; |
|
9
|
|
|
use DomDocument; |
|
10
|
|
|
use DomNode; |
|
11
|
|
|
use DomXpath; |
|
12
|
|
|
|
|
13
|
|
|
class CopyImageProcess implements ProcessInterface |
|
14
|
|
|
{ |
|
15
|
|
|
/** |
|
16
|
|
|
* @var Page |
|
17
|
|
|
*/ |
|
18
|
|
|
protected $page; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* @var RootConfig |
|
22
|
|
|
*/ |
|
23
|
|
|
protected $config; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* @var string |
|
27
|
|
|
*/ |
|
28
|
|
|
protected $html; |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* @var DomDocument |
|
32
|
|
|
*/ |
|
33
|
|
|
protected $doc; |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* @var Fsio |
|
37
|
|
|
*/ |
|
38
|
|
|
protected $fsio; |
|
39
|
|
|
|
|
40
|
|
|
/** |
|
41
|
|
|
* @var LoggerInterface |
|
42
|
|
|
*/ |
|
43
|
|
|
protected $logger; |
|
44
|
|
|
|
|
45
|
3 |
|
public function __construct( |
|
46
|
|
|
LoggerInterface $logger, |
|
47
|
|
|
Fsio $fsio, |
|
48
|
|
|
RootConfig $config |
|
49
|
|
|
) { |
|
50
|
3 |
|
$this->logger = $logger; |
|
51
|
3 |
|
$this->fsio = $fsio; |
|
52
|
3 |
|
$this->config = $config; |
|
53
|
3 |
|
} |
|
54
|
|
|
|
|
55
|
2 |
|
public function __invoke(Page $page) |
|
56
|
|
|
{ |
|
57
|
2 |
|
$this->logger->info(" Processing copy images for {$page->getTarget()}"); |
|
58
|
|
|
|
|
59
|
2 |
|
$this->reset($page); |
|
60
|
|
|
|
|
61
|
2 |
|
$this->loadHtml(); |
|
62
|
2 |
|
if ($this->html) { |
|
63
|
2 |
|
$this->loadDomDocument(); |
|
64
|
2 |
|
$this->processImageNodes(); |
|
65
|
2 |
|
$this->saveHtml(); |
|
66
|
2 |
|
} |
|
67
|
2 |
|
} |
|
68
|
|
|
|
|
69
|
2 |
|
protected function reset(Page $page) |
|
70
|
|
|
{ |
|
71
|
2 |
|
$this->page = $page; |
|
72
|
2 |
|
$this->html = null; |
|
73
|
2 |
|
$this->doc = null; |
|
74
|
2 |
|
} |
|
75
|
|
|
|
|
76
|
2 |
|
protected function loadHtml() |
|
77
|
|
|
{ |
|
78
|
2 |
|
$this->html = $this->fsio->get($this->page->getTarget()); |
|
79
|
2 |
|
} |
|
80
|
|
|
|
|
81
|
2 |
|
protected function saveHtml() |
|
82
|
|
|
{ |
|
83
|
2 |
|
$this->fsio->put($this->page->getTarget(), $this->html); |
|
84
|
2 |
|
} |
|
85
|
|
|
|
|
86
|
2 |
|
protected function loadDomDocument() |
|
87
|
|
|
{ |
|
88
|
2 |
|
$this->doc = new DomDocument(); |
|
89
|
2 |
|
$this->doc->formatOutput = true; |
|
90
|
2 |
|
$this->doc->loadHtml(mb_convert_encoding($this->html, 'HTML-ENTITIES', 'UTF-8'), LIBXML_HTML_NODEFDTD); |
|
91
|
2 |
|
} |
|
92
|
|
|
|
|
93
|
2 |
|
protected function processImageNodes() |
|
94
|
|
|
{ |
|
95
|
2 |
|
$nodes = $this->getImageNodes(); |
|
96
|
2 |
|
$this->addImages($nodes); |
|
97
|
2 |
|
$this->setHtmlFromDomDocument(); |
|
98
|
2 |
|
} |
|
99
|
|
|
|
|
100
|
2 |
|
protected function getImageNodes() |
|
101
|
|
|
{ |
|
102
|
2 |
|
$xpath = new DomXpath($this->doc); |
|
103
|
2 |
|
$query = '//img'; |
|
104
|
2 |
|
return $xpath->query($query); |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
2 |
|
protected function addImages($nodes) |
|
108
|
|
|
{ |
|
109
|
2 |
|
foreach ($nodes as $node) { |
|
110
|
1 |
|
$this->addImage($node); |
|
111
|
2 |
|
} |
|
112
|
2 |
|
} |
|
113
|
|
|
|
|
114
|
1 |
|
protected function addImage(DomNode $node) |
|
115
|
|
|
{ |
|
116
|
1 |
|
if ($src = $this->downloadImage($node)) { |
|
117
|
1 |
|
$node->attributes->getNamedItem('src')->nodeValue = $src; |
|
118
|
1 |
|
} |
|
119
|
1 |
|
} |
|
120
|
|
|
|
|
121
|
1 |
|
protected function downloadImage(DomNode $node) |
|
122
|
|
|
{ |
|
123
|
1 |
|
$image = $node->attributes->getNamedItem('src')->nodeValue; |
|
124
|
|
|
|
|
125
|
|
|
# no image or absolute URI |
|
126
|
1 |
|
if (!$image || preg_match('#^(http(s)?|//)#', $image)) { |
|
127
|
1 |
|
return ''; |
|
128
|
|
|
} |
|
129
|
1 |
|
$imageName = basename($image); |
|
130
|
1 |
|
$originFile = dirname($this->page->getOrigin()) . '/' . ltrim($image, '/'); |
|
131
|
|
|
|
|
132
|
1 |
|
$dir = dirname($this->page->getTarget()) . '/img/'; |
|
133
|
1 |
|
$file = $dir . $imageName; |
|
134
|
|
|
|
|
135
|
1 |
|
if (!$this->fsio->isDir($dir)) { |
|
136
|
1 |
|
$this->fsio->mkdir($dir); |
|
137
|
1 |
|
} |
|
138
|
1 |
|
$this->fsio->put($file, $this->fsio->get($originFile)); |
|
139
|
1 |
|
return $this->config->getRootHref() . (str_replace($this->config->getTarget(), '', $dir)) . $imageName; |
|
140
|
|
|
} |
|
141
|
|
|
|
|
142
|
2 |
|
protected function setHtmlFromDomDocument() |
|
143
|
|
|
{ |
|
144
|
|
|
// retain the modified html |
|
145
|
2 |
|
$this->html = trim($this->doc->saveHtml($this->doc->documentElement)); |
|
146
|
|
|
|
|
147
|
|
|
// strip the html and body tags added by DomDocument |
|
148
|
2 |
|
$this->html = substr( |
|
149
|
2 |
|
$this->html, |
|
150
|
2 |
|
strlen('<html><body>'), |
|
151
|
2 |
|
-1 * strlen('</body></html>') |
|
152
|
2 |
|
); |
|
153
|
|
|
|
|
154
|
|
|
// still may be whitespace all about |
|
155
|
2 |
|
$this->html = trim($this->html) . PHP_EOL; |
|
156
|
2 |
|
} |
|
157
|
|
|
} |
|
158
|
|
|
|
bookdown /
Bookdown.Bookdown
