application/LinkUtils.php

   1 <?php
   2
   3 /**
   4  * Extract title from an HTML document.
   5  *
   6  * @param string $html HTML content where to look for a title.
   7  *
   8  * @return bool|string Extracted title if found, false otherwise.
   9  */
  10 function html_extract_title($html)
  11 {
  12     if (preg_match('!<title.*?>(.*?)</title>!is', $html, $matches)) {
  13         return trim(str_replace("\n", '', $matches[1]));
  14     }
  15     return false;
  16 }
  17
  18 /**
  19  * Determine charset from downloaded page.
  20  * Priority:
  21  *   1. HTTP headers (Content type).
  22  *   2. HTML content page (tag <meta charset>).
  23  *   3. Use a default charset (default: UTF-8).
  24  *
  25  * @param array  $headers           HTTP headers array.
  26  * @param string $htmlContent       HTML content where to look for charset.
  27  * @param string $defaultCharset    Default charset to apply if other methods failed.
  28  *
  29  * @return string Determined charset.
  30  */
  31 function get_charset($headers, $htmlContent, $defaultCharset = 'utf-8')
  32 {
  33     if ($charset = headers_extract_charset($headers)) {
  34         return $charset;
  35     }
  36
  37     if ($charset = html_extract_charset($htmlContent)) {
  38         return $charset;
  39     }
  40
  41     return $defaultCharset;
  42 }
  43
  44 /**
  45  * Extract charset from HTTP headers if it's defined.
  46  *
  47  * @param array $headers HTTP headers array.
  48  *
  49  * @return bool|string Charset string if found (lowercase), false otherwise.
  50  */
  51 function headers_extract_charset($headers)
  52 {
  53     if (! empty($headers['Content-Type']) && strpos($headers['Content-Type'], 'charset=') !== false) {
  54         preg_match('/charset="?([^; ]+)/i', $headers['Content-Type'], $match);
  55         if (! empty($match[1])) {
  56             return strtolower(trim($match[1]));
  57         }
  58     }
  59
  60     return false;
  61 }
  62
  63 /**
  64  * Extract charset HTML content (tag <meta charset>).
  65  *
  66  * @param string $html HTML content where to look for charset.
  67  *
  68  * @return bool|string Charset string if found, false otherwise.
  69  */
  70 function html_extract_charset($html)
  71 {
  72     // Get encoding specified in HTML header.
  73     preg_match('#<meta .*charset=["\']?([^";\'>/]+)["\']? */?>#Usi', $html, $enc);
  74     if (!empty($enc[1])) {
  75         return strtolower($enc[1]);
  76     }
  77
  78     return false;
  79 }
  80
  81 /**
  82  * Count private links in given linklist.
  83  *
  84  * @param array|Countable $links Linklist.
  85  *
  86  * @return int Number of private links.
  87  */
  88 function count_private($links)
  89 {
  90     $cpt = 0;
  91     foreach ($links as $link) {
  92         if ($link['private']) {
  93             $cpt += 1;
  94         }
  95     }
  96
  97     return $cpt;
  98 }
  99
 100 /**
 101  * In a string, converts URLs to clickable links.
 102  *
 103  * @param string $text       input string.
 104  * @param string $redirector if a redirector is set, use it to gerenate links.
 105  * @param bool   $urlEncode  Use `urlencode()` on the URL after the redirector or not.
 106  *
 107  * @return string returns $text with all links converted to HTML links.
 108  *
 109  * @see Function inspired from http://www.php.net/manual/en/function.preg-replace.php#85722
 110  */
 111 function text2clickable($text, $redirector = '', $urlEncode = true)
 112 {
 113     $regex = '!(((?:https?|ftp|file)://|apt:|magnet:)\S+[a-z0-9\(\)]/?)!si';
 114
 115     if (empty($redirector)) {
 116         return preg_replace($regex, '<a href="$1">$1</a>', $text);
 117     }
 118     // Redirector is set, urlencode the final URL.
 119     return preg_replace_callback(
 120         $regex,
 121         function ($matches) use ($redirector, $urlEncode) {
 122             $url = $urlEncode ? urlencode($matches[1]) : $matches[1];
 123             return '<a href="' . $redirector . $url .'">'. $matches[1] .'</a>';
 124         },
 125         $text
 126     );
 127 }
 128
 129 /**
 130  * Auto-link hashtags.
 131  *
 132  * @param string $description Given description.
 133  * @param string $indexUrl    Root URL.
 134  *
 135  * @return string Description with auto-linked hashtags.
 136  */
 137 function hashtag_autolink($description, $indexUrl = '')
 138 {
 139     /*
 140      * To support unicode: http://stackoverflow.com/a/35498078/1484919
 141      * \p{Pc} - to match underscore
 142      * \p{N} - numeric character in any script
 143      * \p{L} - letter from any language
 144      * \p{Mn} - any non marking space (accents, umlauts, etc)
 145      */
 146     $regex = '/(^|\s)#([\p{Pc}\p{N}\p{L}\p{Mn}]+)/mui';
 147     $replacement = '$1<a href="'. $indexUrl .'?addtag=$2" title="Hashtag $2">#$2</a>';
 148     return preg_replace($regex, $replacement, $description);
 149 }
 150
 151 /**
 152  * This function inserts &nbsp; where relevant so that multiple spaces are properly displayed in HTML
 153  * even in the absence of <pre>  (This is used in description to keep text formatting).
 154  *
 155  * @param string $text input text.
 156  *
 157  * @return string formatted text.
 158  */
 159 function space2nbsp($text)
 160 {
 161     return preg_replace('/(^| ) /m', '$1&nbsp;', $text);
 162 }
 163
 164 /**
 165  * Format Shaarli's description
 166  *
 167  * @param string $description shaare's description.
 168  * @param string $redirector  if a redirector is set, use it to gerenate links.
 169  * @param bool   $urlEncode  Use `urlencode()` on the URL after the redirector or not.
 170  * @param string $indexUrl    URL to Shaarli's index.
 171
 172  * @return string formatted description.
 173  */
 174 function format_description($description, $redirector = '', $urlEncode = true, $indexUrl = '') {
 175     return nl2br(space2nbsp(hashtag_autolink(text2clickable($description, $redirector, $urlEncode), $indexUrl)));
 176 }
 177
 178 /**
 179  * Generate a small hash for a link.
 180  *
 181  * @param DateTime $date Link creation date.
 182  * @param int      $id   Link ID.
 183  *
 184  * @return string the small hash generated from link data.
 185  */
 186 function link_small_hash($date, $id)
 187 {
 188     return smallHash($date->format(LinkDB::LINK_DATE_FORMAT) . $id);
 189 }