application/LinkUtils.php

   1 <?php
   2
   3 /**
   4  * Get cURL callback function for CURLOPT_WRITEFUNCTION
   5  *
   6  * @param string $charset     to extract from the downloaded page (reference)
   7  * @param string $title       to extract from the downloaded page (reference)
   8  * @param string $curlGetInfo Optionnaly overrides curl_getinfo function
   9  *
  10  * @return Closure
  11  */
  12 function get_curl_download_callback(&$charset, &$title, $curlGetInfo = 'curl_getinfo')
  13 {
  14     /**
  15      * cURL callback function for CURLOPT_WRITEFUNCTION (called during the download).
  16      *
  17      * While downloading the remote page, we check that the HTTP code is 200 and content type is 'html/text'
  18      * Then we extract the title and the charset and stop the download when it's done.
  19      *
  20      * @param resource $ch   cURL resource
  21      * @param string   $data chunk of data being downloaded
  22      *
  23      * @return int|bool length of $data or false if we need to stop the download
  24      */
  25     return function(&$ch, $data) use ($curlGetInfo, &$charset, &$title) {
  26         $responseCode = $curlGetInfo($ch, CURLINFO_RESPONSE_CODE);
  27         if (!empty($responseCode) && $responseCode != 200) {
  28             return false;
  29         }
  30         $contentType = $curlGetInfo($ch, CURLINFO_CONTENT_TYPE);
  31         if (!empty($contentType) && strpos($contentType, 'text/html') === false) {
  32             return false;
  33         }
  34         if (empty($charset)) {
  35             $charset = header_extract_charset($contentType);
  36         }
  37         if (empty($charset)) {
  38             $charset = html_extract_charset($data);
  39         }
  40         if (empty($title)) {
  41             $title = html_extract_title($data);
  42         }
  43         // We got everything we want, stop the download.
  44         if (!empty($responseCode) && !empty($contentType) && !empty($charset) && !empty($title)) {
  45             return false;
  46         }
  47
  48         return strlen($data);
  49     };
  50 }
  51
  52 /**
  53  * Extract title from an HTML document.
  54  *
  55  * @param string $html HTML content where to look for a title.
  56  *
  57  * @return bool|string Extracted title if found, false otherwise.
  58  */
  59 function html_extract_title($html)
  60 {
  61     if (preg_match('!<title.*?>(.*?)</title>!is', $html, $matches)) {
  62         return trim(str_replace("\n", '', $matches[1]));
  63     }
  64     return false;
  65 }
  66
  67 /**
  68  * Extract charset from HTTP header if it's defined.
  69  *
  70  * @param string $header HTTP header Content-Type line.
  71  *
  72  * @return bool|string Charset string if found (lowercase), false otherwise.
  73  */
  74 function header_extract_charset($header)
  75 {
  76     preg_match('/charset="?([^; ]+)/i', $header, $match);
  77     if (! empty($match[1])) {
  78         return strtolower(trim($match[1]));
  79     }
  80
  81     return false;
  82 }
  83
  84 /**
  85  * Extract charset HTML content (tag <meta charset>).
  86  *
  87  * @param string $html HTML content where to look for charset.
  88  *
  89  * @return bool|string Charset string if found, false otherwise.
  90  */
  91 function html_extract_charset($html)
  92 {
  93     // Get encoding specified in HTML header.
  94     preg_match('#<meta .*charset=["\']?([^";\'>/]+)["\']? */?>#Usi', $html, $enc);
  95     if (!empty($enc[1])) {
  96         return strtolower($enc[1]);
  97     }
  98
  99     return false;
 100 }
 101
 102 /**
 103  * Count private links in given linklist.
 104  *
 105  * @param array|Countable $links Linklist.
 106  *
 107  * @return int Number of private links.
 108  */
 109 function count_private($links)
 110 {
 111     $cpt = 0;
 112     foreach ($links as $link) {
 113         if ($link['private']) {
 114             $cpt += 1;
 115         }
 116     }
 117
 118     return $cpt;
 119 }
 120
 121 /**
 122  * In a string, converts URLs to clickable links.
 123  *
 124  * @param string $text       input string.
 125  * @param string $redirector if a redirector is set, use it to gerenate links.
 126  * @param bool   $urlEncode  Use `urlencode()` on the URL after the redirector or not.
 127  *
 128  * @return string returns $text with all links converted to HTML links.
 129  *
 130  * @see Function inspired from http://www.php.net/manual/en/function.preg-replace.php#85722
 131  */
 132 function text2clickable($text, $redirector = '', $urlEncode = true)
 133 {
 134     $regex = '!(((?:https?|ftp|file)://|apt:|magnet:)\S+[a-z0-9\(\)]/?)!si';
 135
 136     if (empty($redirector)) {
 137         return preg_replace($regex, '<a href="$1">$1</a>', $text);
 138     }
 139     // Redirector is set, urlencode the final URL.
 140     return preg_replace_callback(
 141         $regex,
 142         function ($matches) use ($redirector, $urlEncode) {
 143             $url = $urlEncode ? urlencode($matches[1]) : $matches[1];
 144             return '<a href="' . $redirector . $url .'">'. $matches[1] .'</a>';
 145         },
 146         $text
 147     );
 148 }
 149
 150 /**
 151  * Auto-link hashtags.
 152  *
 153  * @param string $description Given description.
 154  * @param string $indexUrl    Root URL.
 155  *
 156  * @return string Description with auto-linked hashtags.
 157  */
 158 function hashtag_autolink($description, $indexUrl = '')
 159 {
 160     /*
 161      * To support unicode: http://stackoverflow.com/a/35498078/1484919
 162      * \p{Pc} - to match underscore
 163      * \p{N} - numeric character in any script
 164      * \p{L} - letter from any language
 165      * \p{Mn} - any non marking space (accents, umlauts, etc)
 166      */
 167     $regex = '/(^|\s)#([\p{Pc}\p{N}\p{L}\p{Mn}]+)/mui';
 168     $replacement = '$1<a href="'. $indexUrl .'?addtag=$2" title="Hashtag $2">#$2</a>';
 169     return preg_replace($regex, $replacement, $description);
 170 }
 171
 172 /**
 173  * This function inserts &nbsp; where relevant so that multiple spaces are properly displayed in HTML
 174  * even in the absence of <pre>  (This is used in description to keep text formatting).
 175  *
 176  * @param string $text input text.
 177  *
 178  * @return string formatted text.
 179  */
 180 function space2nbsp($text)
 181 {
 182     return preg_replace('/(^| ) /m', '$1&nbsp;', $text);
 183 }
 184
 185 /**
 186  * Format Shaarli's description
 187  *
 188  * @param string $description shaare's description.
 189  * @param string $redirector  if a redirector is set, use it to gerenate links.
 190  * @param bool   $urlEncode  Use `urlencode()` on the URL after the redirector or not.
 191  * @param string $indexUrl    URL to Shaarli's index.
 192
 193  * @return string formatted description.
 194  */
 195 function format_description($description, $redirector = '', $urlEncode = true, $indexUrl = '') {
 196     return nl2br(space2nbsp(hashtag_autolink(text2clickable($description, $redirector, $urlEncode), $indexUrl)));
 197 }
 198
 199 /**
 200  * Generate a small hash for a link.
 201  *
 202  * @param DateTime $date Link creation date.
 203  * @param int      $id   Link ID.
 204  *
 205  * @return string the small hash generated from link data.
 206  */
 207 function link_small_hash($date, $id)
 208 {
 209     return smallHash($date->format(LinkDB::LINK_DATE_FORMAT) . $id);
 210 }