|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace PhpEpub; |
|
6
|
|
|
|
|
7
|
|
|
class Parser |
|
8
|
|
|
{ |
|
9
|
33 |
|
public function __construct(private readonly XmlParser $xmlParser = new XmlParser()) |
|
10
|
|
|
{ |
|
11
|
33 |
|
} |
|
12
|
|
|
|
|
13
|
|
|
/** |
|
14
|
|
|
* Parse the EPUB file structure. |
|
15
|
|
|
* |
|
16
|
|
|
* @param string $directory The directory containing the extracted EPUB contents. |
|
17
|
|
|
*/ |
|
18
|
26 |
|
public function parse(string $directory): string |
|
19
|
|
|
{ |
|
20
|
|
|
// Validate mimetype |
|
21
|
26 |
|
$this->validateMimetype($directory); |
|
22
|
|
|
|
|
23
|
24 |
|
$containerPath = $directory . DIRECTORY_SEPARATOR . 'META-INF' . DIRECTORY_SEPARATOR . 'container.xml'; |
|
24
|
|
|
|
|
25
|
24 |
|
$opfPath = $this->extractOpfPath($containerPath); |
|
26
|
|
|
|
|
27
|
23 |
|
$this->validateOpf($directory . DIRECTORY_SEPARATOR . $opfPath); |
|
28
|
|
|
|
|
29
|
22 |
|
return $opfPath; |
|
30
|
|
|
} |
|
31
|
|
|
|
|
32
|
|
|
/** |
|
33
|
|
|
* Validates the mimetype file. |
|
34
|
|
|
*/ |
|
35
|
26 |
|
private function validateMimetype(string $directory): void |
|
36
|
|
|
{ |
|
37
|
26 |
|
$mimetypePath = $directory . DIRECTORY_SEPARATOR . 'mimetype'; |
|
38
|
|
|
|
|
39
|
26 |
|
if (! file_exists($mimetypePath)) { |
|
40
|
1 |
|
throw new Exception('Missing mimetype file: ' . $mimetypePath); |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
25 |
|
$mimetype = file_get_contents($mimetypePath); |
|
44
|
|
|
|
|
45
|
25 |
|
if ($mimetype === false || trim($mimetype) !== 'application/epub+zip') { |
|
46
|
1 |
|
throw new Exception('Invalid mimetype content: ' . $mimetypePath); |
|
47
|
|
|
} |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
/** |
|
51
|
|
|
* Extract the OPF path from container |
|
52
|
|
|
*/ |
|
53
|
24 |
|
private function extractOpfPath(string $containerPath): string |
|
54
|
|
|
{ |
|
55
|
24 |
|
$xml = $this->xmlParser->parse($containerPath); |
|
56
|
|
|
|
|
57
|
23 |
|
$namespaces = $xml->getNamespaces(true); |
|
58
|
|
|
|
|
59
|
23 |
|
$containerNamespace = $namespaces[''] ?? null; |
|
60
|
|
|
|
|
61
|
23 |
|
if ($containerNamespace === null) { |
|
62
|
|
|
throw new Exception('No container namespace found in container.xml'); |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
23 |
|
$xml->registerXPathNamespace('ns', $containerNamespace); |
|
66
|
|
|
|
|
67
|
23 |
|
$rootfiles = $xml->xpath('//ns:rootfile'); |
|
68
|
|
|
|
|
69
|
23 |
|
if ($rootfiles === false || $rootfiles === null) { |
|
70
|
|
|
throw new Exception('No rootfile found in container.xml'); |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
23 |
|
$rootfile = $rootfiles[0]; // Get the first rootfile node |
|
74
|
|
|
|
|
75
|
23 |
|
$opfPath = (string) $rootfile['full-path']; |
|
76
|
|
|
|
|
77
|
23 |
|
if ($opfPath === '') { |
|
78
|
|
|
throw new Exception('Missing full-path attribute in rootfile element'); |
|
79
|
|
|
} |
|
80
|
|
|
|
|
81
|
23 |
|
return $opfPath; |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
/** |
|
85
|
|
|
* Validates the OPF file and checks for the presence of the NCX file. |
|
86
|
|
|
*/ |
|
87
|
23 |
|
private function validateOpf(string $opfPath): void |
|
88
|
|
|
{ |
|
89
|
23 |
|
$xml = $this->xmlParser->parse($opfPath); |
|
90
|
|
|
|
|
91
|
22 |
|
$namespaces = $xml->getNamespaces(true); |
|
92
|
|
|
|
|
93
|
22 |
|
$containerNamespace = $namespaces[''] ?? null; |
|
94
|
|
|
|
|
95
|
22 |
|
if ($containerNamespace === null) { |
|
96
|
|
|
throw new Exception('No container namespace found in container.xml'); |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
22 |
|
$xml->registerXPathNamespace('opf', $containerNamespace); |
|
100
|
|
|
|
|
101
|
22 |
|
$items = $xml->xpath('//opf:manifest/opf:item'); |
|
102
|
|
|
|
|
103
|
22 |
|
if ($items === false || $items === null) { |
|
104
|
|
|
throw new Exception('Missing manifest in OPF file'); |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
22 |
|
$ncxItem = null; |
|
108
|
22 |
|
foreach ($items as $item) { |
|
109
|
22 |
|
if ((string) $item['media-type'] === 'application/x-dtbncx+xml') { |
|
110
|
15 |
|
$ncxItem = (string) $item['href']; |
|
111
|
15 |
|
break; |
|
112
|
|
|
} |
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
22 |
|
if ($ncxItem !== null) { |
|
116
|
15 |
|
$ncxPath = dirname($opfPath) . DIRECTORY_SEPARATOR . $ncxItem; |
|
117
|
15 |
|
$this->validateNcx($ncxPath); |
|
118
|
|
|
} |
|
119
|
|
|
} |
|
120
|
|
|
|
|
121
|
|
|
/** |
|
122
|
|
|
* Validates the NCX file. |
|
123
|
|
|
*/ |
|
124
|
15 |
|
private function validateNcx(string $ncxPath): void |
|
125
|
|
|
{ |
|
126
|
15 |
|
$xml = $this->xmlParser->parse($ncxPath); |
|
127
|
|
|
|
|
128
|
15 |
|
$namespaces = $xml->getNamespaces(true); |
|
129
|
|
|
|
|
130
|
15 |
|
$navMap = $xml->children($namespaces[''])->navMap; |
|
131
|
|
|
|
|
132
|
15 |
|
if (! $navMap) { |
|
133
|
|
|
throw new Exception('Missing navMap in NCX file'); |
|
134
|
|
|
} |
|
135
|
|
|
} |
|
136
|
|
|
} |
|
137
|
|
|
|
indy2kro /
php-epub
