|
1
|
|
|
<?php |
|
2
|
|
|
abstract class FeedItem_Common extends FeedItem { |
|
3
|
|
|
protected $elem; |
|
4
|
|
|
protected $xpath; |
|
5
|
|
|
protected $doc; |
|
6
|
|
|
|
|
7
|
|
|
public function __construct($elem, $doc, $xpath) { |
|
8
|
|
|
$this->elem = $elem; |
|
9
|
|
|
$this->xpath = $xpath; |
|
10
|
|
|
$this->doc = $doc; |
|
11
|
|
|
|
|
12
|
|
|
try { |
|
13
|
|
|
|
|
14
|
|
|
$source = $elem->getElementsByTagName("source")->item(0); |
|
15
|
|
|
|
|
16
|
|
|
// we don't need <source> element |
|
17
|
|
|
if ($source) { |
|
18
|
|
|
$elem->removeChild($source); |
|
19
|
|
|
} |
|
20
|
|
|
} catch (DOMException $e) { |
|
21
|
|
|
// |
|
22
|
|
|
} |
|
23
|
|
|
} |
|
24
|
|
|
|
|
25
|
|
|
public function get_element() { |
|
26
|
|
|
return $this->elem; |
|
27
|
|
|
} |
|
28
|
|
|
|
|
29
|
|
|
public function get_author() { |
|
30
|
|
|
$author = $this->elem->getElementsByTagName("author")->item(0); |
|
31
|
|
|
|
|
32
|
|
|
if ($author) { |
|
33
|
|
|
$name = $author->getElementsByTagName("name")->item(0); |
|
34
|
|
|
|
|
35
|
|
|
if ($name) { |
|
36
|
|
|
return clean($name->nodeValue); |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
$email = $author->getElementsByTagName("email")->item(0); |
|
40
|
|
|
|
|
41
|
|
|
if ($email) { |
|
42
|
|
|
return clean($email->nodeValue); |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
|
|
if ($author->nodeValue) { |
|
46
|
|
|
return clean($author->nodeValue); |
|
47
|
|
|
} |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
$author_elems = $this->xpath->query("dc:creator", $this->elem); |
|
51
|
|
|
$authors = []; |
|
52
|
|
|
|
|
53
|
|
|
foreach ($author_elems as $author) { |
|
54
|
|
|
array_push($authors, clean($author->nodeValue)); |
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
|
|
return implode(", ", $authors); |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
|
|
public function get_comments_url() { |
|
61
|
|
|
//RSS only. Use a query here to avoid namespace clashes (e.g. with slash). |
|
62
|
|
|
//might give a wrong result if a default namespace was declared (possible with XPath 2.0) |
|
63
|
|
|
$com_url = $this->xpath->query("comments", $this->elem)->item(0); |
|
64
|
|
|
|
|
65
|
|
|
if ($com_url) { |
|
66
|
|
|
return clean($com_url->nodeValue); |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
//Atom Threading Extension (RFC 4685) stuff. Could be used in RSS feeds, so it's in common. |
|
70
|
|
|
//'text/html' for type is too restrictive? |
|
71
|
|
|
$com_url = $this->xpath->query("atom:link[@rel='replies' and contains(@type,'text/html')]/@href", $this->elem)->item(0); |
|
72
|
|
|
|
|
73
|
|
|
if ($com_url) { |
|
74
|
|
|
return clean($com_url->nodeValue); |
|
75
|
|
|
} |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
|
|
public function get_comments_count() { |
|
79
|
|
|
//also query for ATE stuff here |
|
80
|
|
|
$query = "slash:comments|thread:total|atom:link[@rel='replies']/@thread:count"; |
|
81
|
|
|
$comments = $this->xpath->query($query, $this->elem)->item(0); |
|
82
|
|
|
|
|
83
|
|
|
if ($comments) { |
|
84
|
|
|
return clean($comments->nodeValue); |
|
85
|
|
|
} |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
// this is common for both Atom and RSS types and deals with various media: elements |
|
89
|
|
|
public function get_enclosures() { |
|
90
|
|
|
$encs = []; |
|
91
|
|
|
|
|
92
|
|
|
$enclosures = $this->xpath->query("media:content", $this->elem); |
|
93
|
|
|
|
|
94
|
|
|
foreach ($enclosures as $enclosure) { |
|
95
|
|
|
$enc = new FeedEnclosure(); |
|
96
|
|
|
|
|
97
|
|
|
$enc->type = clean($enclosure->getAttribute("type")); |
|
98
|
|
|
$enc->link = clean($enclosure->getAttribute("url")); |
|
99
|
|
|
$enc->length = clean($enclosure->getAttribute("length")); |
|
100
|
|
|
$enc->height = clean($enclosure->getAttribute("height")); |
|
101
|
|
|
$enc->width = clean($enclosure->getAttribute("width")); |
|
102
|
|
|
|
|
103
|
|
|
$medium = clean($enclosure->getAttribute("medium")); |
|
104
|
|
|
if (!$enc->type && $medium) { |
|
105
|
|
|
$enc->type = strtolower("$medium/generic"); |
|
106
|
|
|
} |
|
107
|
|
|
|
|
108
|
|
|
$desc = $this->xpath->query("media:description", $enclosure)->item(0); |
|
109
|
|
|
if ($desc) { |
|
110
|
|
|
$enc->title = clean($desc->nodeValue); |
|
111
|
|
|
} |
|
112
|
|
|
|
|
113
|
|
|
array_push($encs, $enc); |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
$enclosures = $this->xpath->query("media:group", $this->elem); |
|
117
|
|
|
|
|
118
|
|
|
foreach ($enclosures as $enclosure) { |
|
119
|
|
|
$enc = new FeedEnclosure(); |
|
120
|
|
|
|
|
121
|
|
|
$content = $this->xpath->query("media:content", $enclosure)->item(0); |
|
122
|
|
|
|
|
123
|
|
|
if ($content) { |
|
124
|
|
|
$enc->type = clean($content->getAttribute("type")); |
|
125
|
|
|
$enc->link = clean($content->getAttribute("url")); |
|
126
|
|
|
$enc->length = clean($content->getAttribute("length")); |
|
127
|
|
|
$enc->height = clean($content->getAttribute("height")); |
|
128
|
|
|
$enc->width = clean($content->getAttribute("width")); |
|
129
|
|
|
|
|
130
|
|
|
$medium = clean($content->getAttribute("medium")); |
|
131
|
|
|
if (!$enc->type && $medium) { |
|
132
|
|
|
$enc->type = strtolower("$medium/generic"); |
|
133
|
|
|
} |
|
134
|
|
|
|
|
135
|
|
|
$desc = $this->xpath->query("media:description", $content)->item(0); |
|
136
|
|
|
if ($desc) { |
|
137
|
|
|
$enc->title = clean($desc->nodeValue); |
|
138
|
|
|
} else { |
|
139
|
|
|
$desc = $this->xpath->query("media:description", $enclosure)->item(0); |
|
140
|
|
|
if ($desc) { |
|
141
|
|
|
$enc->title = clean($desc->nodeValue); |
|
142
|
|
|
} |
|
143
|
|
|
} |
|
144
|
|
|
|
|
145
|
|
|
array_push($encs, $enc); |
|
146
|
|
|
} |
|
147
|
|
|
} |
|
148
|
|
|
|
|
149
|
|
|
$enclosures = $this->xpath->query("media:thumbnail", $this->elem); |
|
150
|
|
|
|
|
151
|
|
|
foreach ($enclosures as $enclosure) { |
|
152
|
|
|
$enc = new FeedEnclosure(); |
|
153
|
|
|
|
|
154
|
|
|
$enc->type = "image/generic"; |
|
155
|
|
|
$enc->link = clean($enclosure->getAttribute("url")); |
|
156
|
|
|
$enc->height = clean($enclosure->getAttribute("height")); |
|
157
|
|
|
$enc->width = clean($enclosure->getAttribute("width")); |
|
158
|
|
|
|
|
159
|
|
|
array_push($encs, $enc); |
|
160
|
|
|
} |
|
161
|
|
|
|
|
162
|
|
|
return $encs; |
|
163
|
|
|
} |
|
164
|
|
|
|
|
165
|
|
|
public function count_children($node) { |
|
166
|
|
|
return $node->getElementsByTagName("*")->length; |
|
167
|
|
|
} |
|
168
|
|
|
|
|
169
|
|
|
public function subtree_or_text($node) { |
|
170
|
|
|
if ($this->count_children($node) == 0) { |
|
171
|
|
|
return $node->nodeValue; |
|
172
|
|
|
} else { |
|
173
|
|
|
return $node->c14n(); |
|
174
|
|
|
} |
|
175
|
|
|
} |
|
176
|
|
|
|
|
177
|
|
|
public static function normalize_categories($cats) { |
|
178
|
|
|
|
|
179
|
|
|
$tmp = []; |
|
180
|
|
|
|
|
181
|
|
|
foreach ($cats as $rawcat) { |
|
182
|
|
|
$tmp = array_merge($tmp, explode(",", $rawcat)); |
|
183
|
|
|
} |
|
184
|
|
|
|
|
185
|
|
|
$tmp = array_map(function($srccat) { |
|
186
|
|
|
$cat = clean(trim(mb_strtolower($srccat))); |
|
187
|
|
|
|
|
188
|
|
|
// we don't support numeric tags |
|
189
|
|
|
if (is_numeric($cat)) { |
|
190
|
|
|
$cat = 't:'.$cat; |
|
191
|
|
|
} |
|
192
|
|
|
|
|
193
|
|
|
$cat = preg_replace('/[,\'\"]/', "", $cat); |
|
194
|
|
|
|
|
195
|
|
|
if (DB_TYPE == "mysql") { |
|
|
|
|
|
|
196
|
|
|
$cat = preg_replace('/[\x{10000}-\x{10FFFF}]/u', "\xEF\xBF\xBD", $cat); |
|
197
|
|
|
} |
|
198
|
|
|
|
|
199
|
|
|
if (mb_strlen($cat) > 250) { |
|
200
|
|
|
$cat = mb_substr($cat, 0, 250); |
|
201
|
|
|
} |
|
202
|
|
|
|
|
203
|
|
|
return $cat; |
|
204
|
|
|
}, $tmp); |
|
205
|
|
|
|
|
206
|
|
|
asort($tmp); |
|
207
|
|
|
|
|
208
|
|
|
return array_unique($tmp); |
|
209
|
|
|
} |
|
210
|
|
|
} |
|
211
|
|
|
|