]>
Commit | Line | Data |
---|---|---|
1557cefb A |
1 | <?php |
2 | ||
336a28fa | 3 | use Shaarli\Bookmark\Bookmark; |
f24896b2 | 4 | |
1557cefb | 5 | /** |
d65342e3 | 6 | * Extract title from an HTML document. |
1557cefb | 7 | * |
d65342e3 | 8 | * @param string $html HTML content where to look for a title. |
1557cefb | 9 | * |
d65342e3 | 10 | * @return bool|string Extracted title if found, false otherwise. |
1557cefb | 11 | */ |
d65342e3 | 12 | function html_extract_title($html) |
1557cefb | 13 | { |
d65342e3 A |
14 | if (preg_match('!<title.*?>(.*?)</title>!is', $html, $matches)) { |
15 | return trim(str_replace("\n", '', $matches[1])); | |
1557cefb | 16 | } |
d65342e3 | 17 | return false; |
1557cefb A |
18 | } |
19 | ||
20 | /** | |
d65342e3 | 21 | * Extract charset from HTTP header if it's defined. |
1557cefb | 22 | * |
d65342e3 | 23 | * @param string $header HTTP header Content-Type line. |
1557cefb A |
24 | * |
25 | * @return bool|string Charset string if found (lowercase), false otherwise. | |
26 | */ | |
d65342e3 | 27 | function header_extract_charset($header) |
1557cefb | 28 | { |
1ea09a1b | 29 | preg_match('/charset=["\']?([^; "\']+)/i', $header, $match); |
d65342e3 A |
30 | if (! empty($match[1])) { |
31 | return strtolower(trim($match[1])); | |
1557cefb A |
32 | } |
33 | ||
34 | return false; | |
35 | } | |
36 | ||
37 | /** | |
38 | * Extract charset HTML content (tag <meta charset>). | |
39 | * | |
40 | * @param string $html HTML content where to look for charset. | |
41 | * | |
42 | * @return bool|string Charset string if found, false otherwise. | |
43 | */ | |
44 | function html_extract_charset($html) | |
45 | { | |
46 | // Get encoding specified in HTML header. | |
ce7b0b64 | 47 | preg_match('#<meta .*charset=["\']?([^";\'>/]+)["\']? */?>#Usi', $html, $enc); |
1557cefb A |
48 | if (!empty($enc[1])) { |
49 | return strtolower($enc[1]); | |
50 | } | |
51 | ||
52 | return false; | |
53 | } | |
141a86c5 | 54 | |
6a487252 A |
55 | /** |
56 | * Extract meta tag from HTML content in either: | |
57 | * - OpenGraph: <meta property="og:[tag]" ...> | |
58 | * - Meta tag: <meta name="[tag]" ...> | |
59 | * | |
60 | * @param string $tag Name of the tag to retrieve. | |
61 | * @param string $html HTML content where to look for charset. | |
62 | * | |
63 | * @return bool|string Charset string if found, false otherwise. | |
64 | */ | |
65 | function html_extract_tag($tag, $html) | |
66 | { | |
67 | $propertiesKey = ['property', 'name', 'itemprop']; | |
68 | $properties = implode('|', $propertiesKey); | |
2cd0509b A |
69 | // We need a OR here to accept either 'property=og:noquote' or 'property="og:unrelated og:my-tag"' |
70 | $orCondition = '["\']?(?:og:)?'. $tag .'["\']?|["\'][^\'"]*?(?:og:)?' . $tag . '[^\'"]*?[\'"]'; | |
6a487252 | 71 | // Try to retrieve OpenGraph image. |
2cd0509b | 72 | $ogRegex = '#<meta[^>]+(?:'. $properties .')=(?:'. $orCondition .')[^>]*content=["\'](.*?)["\'].*?>#'; |
6a487252 A |
73 | // If the attributes are not in the order property => content (e.g. Github) |
74 | // New regex to keep this readable... more or less. | |
2cd0509b | 75 | $ogRegexReverse = '#<meta[^>]+content=["\'](.*?)["\'][^>]+(?:'. $properties .')=(?:'. $orCondition .').*?>#'; |
6a487252 A |
76 | |
77 | if (preg_match($ogRegex, $html, $matches) > 0 | |
78 | || preg_match($ogRegexReverse, $html, $matches) > 0 | |
79 | ) { | |
80 | return $matches[1]; | |
81 | } | |
82 | ||
83 | return false; | |
84 | } | |
85 | ||
141a86c5 | 86 | /** |
336a28fa | 87 | * In a string, converts URLs to clickable bookmarks. |
9ccca401 A |
88 | * |
89 | * @param string $text input string. | |
9ccca401 | 90 | * |
336a28fa | 91 | * @return string returns $text with all bookmarks converted to HTML bookmarks. |
9ccca401 A |
92 | * |
93 | * @see Function inspired from http://www.php.net/manual/en/function.preg-replace.php#85722 | |
94 | */ | |
520d2957 | 95 | function text2clickable($text) |
9ccca401 | 96 | { |
601faf97 | 97 | $regex = '!(((?:https?|ftp|file)://|apt:|magnet:)\S+[a-z0-9\(\)]/?)!si'; |
520d2957 | 98 | return preg_replace($regex, '<a href="$1">$1</a>', $text); |
9ccca401 A |
99 | } |
100 | ||
101 | /** | |
102 | * Auto-link hashtags. | |
103 | * | |
104 | * @param string $description Given description. | |
105 | * @param string $indexUrl Root URL. | |
106 | * | |
107 | * @return string Description with auto-linked hashtags. | |
108 | */ | |
109 | function hashtag_autolink($description, $indexUrl = '') | |
110 | { | |
111 | /* | |
112 | * To support unicode: http://stackoverflow.com/a/35498078/1484919 | |
113 | * \p{Pc} - to match underscore | |
114 | * \p{N} - numeric character in any script | |
115 | * \p{L} - letter from any language | |
116 | * \p{Mn} - any non marking space (accents, umlauts, etc) | |
117 | */ | |
118 | $regex = '/(^|\s)#([\p{Pc}\p{N}\p{L}\p{Mn}]+)/mui'; | |
03340c18 | 119 | $replacement = '$1<a href="'. $indexUrl .'./add-tag/$2" title="Hashtag $2">#$2</a>'; |
9ccca401 A |
120 | return preg_replace($regex, $replacement, $description); |
121 | } | |
122 | ||
123 | /** | |
124 | * This function inserts where relevant so that multiple spaces are properly displayed in HTML | |
125 | * even in the absence of <pre> (This is used in description to keep text formatting). | |
126 | * | |
127 | * @param string $text input text. | |
128 | * | |
129 | * @return string formatted text. | |
130 | */ | |
131 | function space2nbsp($text) | |
132 | { | |
133 | return preg_replace('/(^| ) /m', '$1 ', $text); | |
134 | } | |
135 | ||
136 | /** | |
137 | * Format Shaarli's description | |
138 | * | |
139 | * @param string $description shaare's description. | |
7af9a418 | 140 | * @param string $indexUrl URL to Shaarli's index. |
fd08b50a | 141 | |
9ccca401 A |
142 | * @return string formatted description. |
143 | */ | |
520d2957 | 144 | function format_description($description, $indexUrl = '') |
f211e417 | 145 | { |
520d2957 | 146 | return nl2br(space2nbsp(hashtag_autolink(text2clickable($description), $indexUrl))); |
9ccca401 | 147 | } |
d592daea A |
148 | |
149 | /** | |
150 | * Generate a small hash for a link. | |
151 | * | |
152 | * @param DateTime $date Link creation date. | |
153 | * @param int $id Link ID. | |
154 | * | |
155 | * @return string the small hash generated from link data. | |
156 | */ | |
157 | function link_small_hash($date, $id) | |
158 | { | |
336a28fa | 159 | return smallHash($date->format(Bookmark::LINK_DATE_FORMAT) . $id); |
d592daea | 160 | } |
a8e7da01 A |
161 | |
162 | /** | |
163 | * Returns whether or not the link is an internal note. | |
164 | * Its URL starts by `?` because it's actually a permalink. | |
165 | * | |
166 | * @param string $linkUrl | |
167 | * | |
168 | * @return bool true if internal note, false otherwise. | |
169 | */ | |
170 | function is_note($linkUrl) | |
171 | { | |
172 | return isset($linkUrl[0]) && $linkUrl[0] === '?'; | |
173 | } |