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         $cpt = $link['private'] == true ? $cpt + 1 : $cpt;
  93     }
  94
  95     return $cpt;
  96 }
  97
  98 /**
  99  * In a string, converts URLs to clickable links.
 100  *
 101  * @param string $text       input string.
 102  * @param string $redirector if a redirector is set, use it to gerenate links.
 103  *
 104  * @return string returns $text with all links converted to HTML links.
 105  *
 106  * @see Function inspired from http://www.php.net/manual/en/function.preg-replace.php#85722
 107  */
 108 function text2clickable($text, $redirector = '')
 109 {
 110     $regex = '!(((?:https?|ftp|file)://|apt:|magnet:)\S+[[:alnum:]]/?)!si';
 111
 112     if (empty($redirector)) {
 113         return preg_replace($regex, '<a href="$1">$1</a>', $text);
 114     }
 115     // Redirector is set, urlencode the final URL.
 116     return preg_replace_callback(
 117         $regex,
 118         function ($matches) use ($redirector) {
 119             return '<a href="' . $redirector . urlencode($matches[1]) .'">'. $matches[1] .'</a>';
 120         },
 121         $text
 122     );
 123 }
 124
 125 /**
 126  * Auto-link hashtags.
 127  *
 128  * @param string $description Given description.
 129  * @param string $indexUrl    Root URL.
 130  *
 131  * @return string Description with auto-linked hashtags.
 132  */
 133 function hashtag_autolink($description, $indexUrl = '')
 134 {
 135     /*
 136      * To support unicode: http://stackoverflow.com/a/35498078/1484919
 137      * \p{Pc} - to match underscore
 138      * \p{N} - numeric character in any script
 139      * \p{L} - letter from any language
 140      * \p{Mn} - any non marking space (accents, umlauts, etc)
 141      */
 142     $regex = '/(^|\s)#([\p{Pc}\p{N}\p{L}\p{Mn}]+)/mui';
 143     $replacement = '$1<a href="'. $indexUrl .'?addtag=$2" title="Hashtag $2">#$2</a>';
 144     return preg_replace($regex, $replacement, $description);
 145 }
 146
 147 /**
 148  * This function inserts &nbsp; where relevant so that multiple spaces are properly displayed in HTML
 149  * even in the absence of <pre>  (This is used in description to keep text formatting).
 150  *
 151  * @param string $text input text.
 152  *
 153  * @return string formatted text.
 154  */
 155 function space2nbsp($text)
 156 {
 157     return preg_replace('/(^| ) /m', '$1&nbsp;', $text);
 158 }
 159
 160 /**
 161  * Format Shaarli's description
 162  *
 163  * @param string $description shaare's description.
 164  * @param string $redirector  if a redirector is set, use it to gerenate links.
 165  * @param string $indexUrl    URL to Shaarli's index.
 166  *
 167  * @return string formatted description.
 168  */
 169 function format_description($description, $redirector = '', $indexUrl = '') {
 170     return nl2br(space2nbsp(hashtag_autolink(text2clickable($description, $redirector), $indexUrl)));
 171 }
 172
 173 /**
 174  * Generate a small hash for a link.
 175  *
 176  * @param DateTime $date Link creation date.
 177  * @param int      $id   Link ID.
 178  *
 179  * @return string the small hash generated from link data.
 180  */
 181 function link_small_hash($date, $id)
 182 {
 183     return smallHash($date->format(LinkDB::LINK_DATE_FORMAT) . $id);
 184 }