1
|
|
|
<?php |
2
|
|
|
/* For licensing terms, see /license.txt */ |
3
|
|
|
|
4
|
|
|
class CcEntities |
5
|
|
|
{ |
6
|
|
|
/** |
7
|
|
|
* Prepares convert for inclusion into XML. |
8
|
|
|
* |
9
|
|
|
* @param string $value |
10
|
|
|
* |
11
|
|
|
* @return string |
12
|
|
|
*/ |
13
|
|
|
public static function safexml($value) |
14
|
|
|
{ |
15
|
|
|
$result = htmlspecialchars(html_entity_decode($value, ENT_QUOTES, 'UTF-8'), |
16
|
|
|
ENT_NOQUOTES, |
17
|
|
|
'UTF-8', |
18
|
|
|
false); |
19
|
|
|
|
20
|
|
|
return $result; |
21
|
|
|
} |
22
|
|
|
|
23
|
|
|
public function loadXmlResource($path_to_file) |
24
|
|
|
{ |
25
|
|
|
$resource = new DOMDocument(); |
26
|
|
|
|
27
|
|
|
Cc1p3Convert::logAction('Load the XML resource file: '.$path_to_file); |
28
|
|
|
|
29
|
|
|
if (!$resource->load($path_to_file)) { |
30
|
|
|
Cc1p3Convert::logAction('Cannot load the XML resource file: '.$path_to_file, false); |
31
|
|
|
} |
32
|
|
|
|
33
|
|
|
return $resource; |
34
|
|
|
} |
35
|
|
|
|
36
|
|
|
public function updateSources($html, $rootPath = '') |
37
|
|
|
{ |
38
|
|
|
$document = $this->loadHtml($html); |
39
|
|
|
|
40
|
|
|
$tags = ['img' => 'src', 'a' => 'href']; |
41
|
|
|
|
42
|
|
|
foreach ($tags as $tag => $attribute) { |
43
|
|
|
$elements = $document->getElementsByTagName($tag); |
44
|
|
|
|
45
|
|
|
foreach ($elements as $element) { |
46
|
|
|
$attribute_value = $element->getAttribute($attribute); |
47
|
|
|
$protocol = parse_url($attribute_value, PHP_URL_SCHEME); |
48
|
|
|
|
49
|
|
|
if (empty($protocol)) { |
50
|
|
|
$attribute_value = str_replace("\$IMS-CC-FILEBASE\$", "", $attribute_value); |
51
|
|
|
$attribute_value = $this->fullPath($rootPath."/".$attribute_value, "/"); |
52
|
|
|
$attribute_value = "\$@FILEPHP@\$"."/".$attribute_value; |
53
|
|
|
} |
54
|
|
|
|
55
|
|
|
$element->setAttribute($attribute, $attribute_value); |
56
|
|
|
} |
57
|
|
|
} |
58
|
|
|
|
59
|
|
|
$html = $this->htmlInsidebody($document); |
60
|
|
|
|
61
|
|
|
return $html; |
62
|
|
|
} |
63
|
|
|
|
64
|
|
|
public function fullPath($path, $dir_sep = DIRECTORY_SEPARATOR) |
65
|
|
|
{ |
66
|
|
|
$token = '$IMS-CC-FILEBASE$'; |
67
|
|
|
$path = str_replace($token, '', $path); |
68
|
|
|
|
69
|
|
|
if (is_string($path) && ($path != '')) { |
70
|
|
|
$dot_dir = '.'; |
71
|
|
|
$up_dir = '..'; |
72
|
|
|
$length = strlen($path); |
73
|
|
|
$rtemp = trim($path); |
74
|
|
|
$start = strrpos($path, $dir_sep); |
75
|
|
|
$can_continue = ($start !== false); |
76
|
|
|
$result = $can_continue ? '' : $path; |
77
|
|
|
$rcount = 0; |
78
|
|
|
|
79
|
|
|
while ($can_continue) { |
80
|
|
|
$dir_part = ($start !== false) ? substr($rtemp, $start + 1, $length - $start) : $rtemp; |
81
|
|
|
$can_continue = ($dir_part !== false); |
82
|
|
|
|
83
|
|
|
if ($can_continue) { |
84
|
|
|
if ($dir_part != $dot_dir) { |
85
|
|
|
if ($dir_part == $up_dir) { |
86
|
|
|
$rcount++; |
87
|
|
|
} else { |
88
|
|
|
if ($rcount > 0) { |
89
|
|
|
$rcount--; |
90
|
|
|
} else { |
91
|
|
|
$result = ($result == '') ? $dir_part : $dir_part.$dir_sep.$result; |
92
|
|
|
} |
93
|
|
|
} |
94
|
|
|
} |
95
|
|
|
$rtemp = substr($path, 0, $start); |
96
|
|
|
$start = strrpos($rtemp, $dir_sep); |
97
|
|
|
$can_continue = (($start !== false) || (strlen($rtemp) > 0)); |
98
|
|
|
} |
99
|
|
|
} |
100
|
|
|
} |
101
|
|
|
|
102
|
|
|
return $result; |
|
|
|
|
103
|
|
|
} |
104
|
|
|
|
105
|
|
|
public function includeTitles($html) |
106
|
|
|
{ |
107
|
|
|
$document = $this->loadHtml($html); |
108
|
|
|
|
109
|
|
|
$images = $document->getElementsByTagName('img'); |
110
|
|
|
|
111
|
|
|
foreach ($images as $image) { |
112
|
|
|
$src = $image->getAttribute('src'); |
113
|
|
|
$alt = $image->getAttribute('alt'); |
114
|
|
|
$title = $image->getAttribute('title'); |
115
|
|
|
|
116
|
|
|
$filename = pathinfo($src); |
117
|
|
|
$filename = $filename['filename']; |
118
|
|
|
|
119
|
|
|
$alt = empty($alt) ? $filename : $alt; |
120
|
|
|
$title = empty($title) ? $filename : $title; |
121
|
|
|
|
122
|
|
|
$image->setAttribute('alt', $alt); |
123
|
|
|
$image->setAttribute('title', $title); |
124
|
|
|
} |
125
|
|
|
|
126
|
|
|
$html = $this->htmlInsidebody($document); |
127
|
|
|
|
128
|
|
|
return $html; |
129
|
|
|
} |
130
|
|
|
|
131
|
|
|
public function getExternalXml($identifier) |
132
|
|
|
{ |
133
|
|
|
$xpath = Cc1p3Convert::newxPath(Cc1p3Convert::$manifest, Cc1p3Convert::$namespaces); |
134
|
|
|
|
135
|
|
|
$files = $xpath->query('/imscc:manifest/imscc:resources/imscc:resource[@identifier="'. |
136
|
|
|
$identifier.'"]/imscc:file/@href'); |
137
|
|
|
|
138
|
|
|
if (empty($files)) { |
139
|
|
|
$response = ''; |
140
|
|
|
} else { |
141
|
|
|
$response = $files->item(0)->nodeValue; |
142
|
|
|
} |
143
|
|
|
|
144
|
|
|
return $response; |
145
|
|
|
} |
146
|
|
|
|
147
|
|
|
public function generateRandomString($length = 6) |
148
|
|
|
{ |
149
|
|
|
$response = ''; |
150
|
|
|
$source = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ'; |
151
|
|
|
|
152
|
|
|
if ($length > 0) { |
153
|
|
|
$response = ''; |
154
|
|
|
$source = str_split($source, 1); |
155
|
|
|
|
156
|
|
|
for ($i = 1; $i <= $length; $i++) { |
157
|
|
|
$num = mt_rand(1, count($source)); |
158
|
|
|
$response .= $source[$num - 1]; |
159
|
|
|
} |
160
|
|
|
} |
161
|
|
|
|
162
|
|
|
return $response; |
163
|
|
|
} |
164
|
|
|
|
165
|
|
|
public function truncateText($text, $max, $remove_html) |
166
|
|
|
{ |
167
|
|
|
if ($max > 10) { |
168
|
|
|
$text = substr($text, 0, ($max - 6)).' [...]'; |
169
|
|
|
} else { |
170
|
|
|
$text = substr($text, 0, $max); |
171
|
|
|
} |
172
|
|
|
|
173
|
|
|
$text = $remove_html ? strip_tags($text) : $text; |
174
|
|
|
|
175
|
|
|
return $text; |
176
|
|
|
} |
177
|
|
|
|
178
|
|
|
protected function prepareContent($content) |
179
|
|
|
{ |
180
|
|
|
$result = $content; |
181
|
|
|
if (empty($result)) { |
182
|
|
|
return ''; |
183
|
|
|
} |
184
|
|
|
$encoding = null; |
185
|
|
|
$xml_error = new LibxmlErrorsMgr(); |
186
|
|
|
$dom = new DOMDocument(); |
187
|
|
|
$dom->validateOnParse = false; |
188
|
|
|
$dom->strictErrorChecking = false; |
189
|
|
|
if ($dom->loadHTML($content)) { |
190
|
|
|
$encoding = $dom->xmlEncoding; |
191
|
|
|
} |
192
|
|
|
if (empty($encoding)) { |
193
|
|
|
$encoding = mb_detect_encoding($content, 'auto', true); |
194
|
|
|
} |
195
|
|
|
if (!empty($encoding) && !mb_check_encoding($content, 'UTF-8')) { |
196
|
|
|
$result = mb_convert_encoding($content, 'UTF-8', $encoding); |
197
|
|
|
} |
198
|
|
|
|
199
|
|
|
// See if we can strip off body tag and anything outside of it. |
200
|
|
|
foreach (['body', 'html'] as $tagname) { |
201
|
|
|
$regex = str_replace('##', $tagname, "/<##[^>]*>(.+)<\/##>/is"); |
202
|
|
|
if (preg_match($regex, $result, $matches)) { |
203
|
|
|
$result = $matches[1]; |
204
|
|
|
break; |
205
|
|
|
} |
206
|
|
|
} |
207
|
|
|
|
208
|
|
|
return $result; |
209
|
|
|
} |
210
|
|
|
|
211
|
|
|
/** |
212
|
|
|
* @param string $html |
213
|
|
|
* |
214
|
|
|
* @return DOMDocument |
215
|
|
|
*/ |
216
|
|
|
private function loadHtml($html) |
217
|
|
|
{ |
218
|
|
|
// Need to make sure that the html passed has charset meta tag. |
219
|
|
|
$metatag = '<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />'; |
220
|
|
|
if (strpos($html, $metatag) === false) { |
221
|
|
|
$html = '<html><head>'.$metatag.'</head><body>'.$html.'</body></html>'; |
222
|
|
|
} |
223
|
|
|
|
224
|
|
|
$document = new DOMDocument(); |
225
|
|
|
@$document->loadHTML($html); |
|
|
|
|
226
|
|
|
|
227
|
|
|
return $document; |
228
|
|
|
} |
229
|
|
|
|
230
|
|
|
/** |
231
|
|
|
* @param DOMDocument $domdocument |
232
|
|
|
* |
233
|
|
|
* @return string |
234
|
|
|
*/ |
235
|
|
|
private function htmlInsidebody($domdocument) |
236
|
|
|
{ |
237
|
|
|
$html = ''; |
238
|
|
|
$bodyitems = $domdocument->getElementsByTagName('body'); |
239
|
|
|
if ($bodyitems->length > 0) { |
240
|
|
|
$body = $bodyitems->item(0); |
241
|
|
|
$html = str_ireplace(['<body>', '</body>'], '', $body->C14N()); |
242
|
|
|
} |
243
|
|
|
|
244
|
|
|
return $html; |
|
|
|
|
245
|
|
|
} |
246
|
|
|
} |
247
|
|
|
|