| commit | author | age | ||
| aa055c | 1 | <?php |
| T | 2 | |
| 3 | /* | |
| 4 | +-----------------------------------------------------------------------+ | |
| e019f2 | 5 | | This file is part of the Roundcube Webmail client | |
| 60226a | 6 | | Copyright (C) 2009-2012, The Roundcube Dev Team | |
| 7fe381 | 7 | | | |
| T | 8 | | Licensed under the GNU General Public License version 3 or | |
| 9 | | any later version with exceptions for skins & plugins. | | |
| 10 | | See the README file for a full license statement. | | |
| aa055c | 11 | | | |
| T | 12 | | PURPOSE: | |
| 13 | | Handle string replacements based on preg_replace_callback | | |
| 14 | +-----------------------------------------------------------------------+ | |
| 15 | | Author: Thomas Bruederli <roundcube@gmail.com> | | |
| 16 | +-----------------------------------------------------------------------+ | |
| 17 | */ | |
| 18 | ||
| 19 | /** | |
| 20 | * Helper class for string replacements based on preg_replace_callback | |
| 21 | * | |
| 9ab346 | 22 | * @package Framework |
| AM | 23 | * @subpackage Utils |
| aa055c | 24 | */ |
| T | 25 | class rcube_string_replacer |
| 26 | { | |
| 0d2144 | 27 | public static $pattern = '/##str_replacement\[([0-9]+)\]##/'; |
| AM | 28 | public $mailto_pattern; |
| 29 | public $link_pattern; | |
| 30 | private $values = array(); | |
| 1e3254 | 31 | private $options = array(); |
| aa055c | 32 | |
| 88ed23 | 33 | |
| 1e3254 | 34 | function __construct($options = array()) |
| 0d2144 | 35 | { |
| AM | 36 | // Simplified domain expression for UTF8 characters handling |
| 37 | // Support unicode/punycode in top-level domain part | |
| 16915e | 38 | $utf_domain = '[^?&@"\'\\/()<>\s\r\t\n]+\\.?([^\\x00-\\x2f\\x3b-\\x40\\x5b-\\x60\\x7b-\\x7f]{2,}|xn--[a-zA-Z0-9]{2,})'; |
| 0d2144 | 39 | $url1 = '.:;,'; |
| 0931a9 | 40 | $url2 = 'a-zA-Z0-9%=#$@+?!&\\/_~\\[\\]\\(\\){}\*-'; |
| 88ed23 | 41 | |
| 0d2144 | 42 | $this->link_pattern = "/([\w]+:\/\/|\W[Ww][Ww][Ww]\.|^[Ww][Ww][Ww]\.)($utf_domain([$url1]?[$url2]+)*)/"; |
| AM | 43 | $this->mailto_pattern = "/(" |
| 44 | ."[-\w!\#\$%&\'*+~\/^`|{}=]+(?:\.[-\w!\#\$%&\'*+~\/^`|{}=]+)*" // local-part | |
| 45 | ."@$utf_domain" // domain-part | |
| 46 | ."(\?[$url1$url2]+)?" // e.g. ?subject=test... | |
| 47 | .")/"; | |
| 1e3254 | 48 | |
| AM | 49 | $this->options = $options; |
| aa055c | 50 | } |
| 0ff554 | 51 | |
| 0d2144 | 52 | /** |
| AM | 53 | * Add a string to the internal list |
| 54 | * | |
| 55 | * @param string String value | |
| 56 | * @return int Index of value for retrieval | |
| 57 | */ | |
| 58 | public function add($str) | |
| 59 | { | |
| 60 | $i = count($this->values); | |
| 61 | $this->values[$i] = $str; | |
| 62 | return $i; | |
| a1013c | 63 | } |
| aa055c | 64 | |
| 0d2144 | 65 | /** |
| AM | 66 | * Build replacement string |
| 67 | */ | |
| 68 | public function get_replacement($i) | |
| 69 | { | |
| 70 | return '##str_replacement['.$i.']##'; | |
| 71 | } | |
| aa055c | 72 | |
| 0d2144 | 73 | /** |
| AM | 74 | * Callback function used to build HTML links around URL strings |
| 75 | * | |
| 76 | * @param array Matches result from preg_replace_callback | |
| 77 | * @return int Index of saved string value | |
| 78 | */ | |
| 79 | public function link_callback($matches) | |
| 80 | { | |
| 81 | $i = -1; | |
| 82 | $scheme = strtolower($matches[1]); | |
| aa055c | 83 | |
| 0d2144 | 84 | if (preg_match('!^(http|ftp|file)s?://!i', $scheme)) { |
| AM | 85 | $url = $matches[1] . $matches[2]; |
| 0ff554 | 86 | } |
| 0d2144 | 87 | else if (preg_match('/^(\W*)(www\.)$/i', $matches[1], $m)) { |
| AM | 88 | $url = $m[2] . $matches[2]; |
| 89 | $url_prefix = 'http://'; | |
| 90 | $prefix = $m[1]; | |
| 0ff554 | 91 | } |
| A | 92 | |
| 0d2144 | 93 | if ($url) { |
| AM | 94 | $suffix = $this->parse_url_brackets($url); |
| 1e3254 | 95 | $attrib = (array)$this->options['link_attribs']; |
| AM | 96 | $attrib['href'] = $url_prefix . $url; |
| 97 | ||
| 70229c | 98 | $i = $this->add(html::a($attrib, rcube::Q($url)) . $suffix); |
| 0d2144 | 99 | } |
| AM | 100 | |
| 101 | // Return valid link for recognized schemes, otherwise | |
| 102 | // return the unmodified string for unrecognized schemes. | |
| 70229c | 103 | return $i >= 0 ? $prefix . $this->get_replacement($i) : $matches[0]; |
| 0ff554 | 104 | } |
| A | 105 | |
| 0d2144 | 106 | /** |
| AM | 107 | * Callback function used to build mailto: links around e-mail strings |
| 108 | * | |
| 109 | * @param array Matches result from preg_replace_callback | |
| 110 | * @return int Index of saved string value | |
| 111 | */ | |
| 112 | public function mailto_callback($matches) | |
| 113 | { | |
| 114 | $href = $matches[1]; | |
| 115 | $suffix = $this->parse_url_brackets($href); | |
| 116 | $i = $this->add(html::a('mailto:' . $href, rcube::Q($href)) . $suffix); | |
| 0ff554 | 117 | |
| 0d2144 | 118 | return $i >= 0 ? $this->get_replacement($i) : ''; |
| AM | 119 | } |
| 120 | ||
| 121 | /** | |
| 122 | * Look up the index from the preg_replace matches array | |
| 123 | * and return the substitution value. | |
| 124 | * | |
| 125 | * @param array Matches result from preg_replace_callback | |
| 126 | * @return string Value at index $matches[1] | |
| 127 | */ | |
| 128 | public function replace_callback($matches) | |
| 129 | { | |
| 130 | return $this->values[$matches[1]]; | |
| 131 | } | |
| 132 | ||
| 133 | /** | |
| 134 | * Replace all defined (link|mailto) patterns with replacement string | |
| 135 | * | |
| 136 | * @param string $str Text | |
| 137 | * | |
| 138 | * @return string Text | |
| 139 | */ | |
| 140 | public function replace($str) | |
| 141 | { | |
| 142 | // search for patterns like links and e-mail addresses | |
| 143 | $str = preg_replace_callback($this->link_pattern, array($this, 'link_callback'), $str); | |
| 144 | $str = preg_replace_callback($this->mailto_pattern, array($this, 'mailto_callback'), $str); | |
| 145 | ||
| 146 | return $str; | |
| 147 | } | |
| 148 | ||
| 149 | /** | |
| 150 | * Replace substituted strings with original values | |
| 151 | */ | |
| 152 | public function resolve($str) | |
| 153 | { | |
| 154 | return preg_replace_callback(self::$pattern, array($this, 'replace_callback'), $str); | |
| 155 | } | |
| 156 | ||
| 157 | /** | |
| 158 | * Fixes bracket characters in URL handling | |
| 159 | */ | |
| 160 | public static function parse_url_brackets(&$url) | |
| 161 | { | |
| 162 | // #1487672: special handling of square brackets, | |
| 163 | // URL regexp allows [] characters in URL, for example: | |
| 164 | // "http://example.com/?a[b]=c". However we need to handle | |
| 165 | // properly situation when a bracket is placed at the end | |
| 166 | // of the link e.g. "[http://example.com]" | |
| 0931a9 | 167 | // Yes, this is not perfect handles correctly only paired characters |
| AM | 168 | // but it should work for common cases |
| 169 | ||
| 0d2144 | 170 | if (preg_match('/(\\[|\\])/', $url)) { |
| AM | 171 | $in = false; |
| 172 | for ($i=0, $len=strlen($url); $i<$len; $i++) { | |
| 173 | if ($url[$i] == '[') { | |
| 174 | if ($in) | |
| 175 | break; | |
| 176 | $in = true; | |
| 177 | } | |
| 178 | else if ($url[$i] == ']') { | |
| 179 | if (!$in) | |
| 180 | break; | |
| 181 | $in = false; | |
| 182 | } | |
| 183 | } | |
| 184 | ||
| 185 | if ($i < $len) { | |
| 186 | $suffix = substr($url, $i); | |
| 187 | $url = substr($url, 0, $i); | |
| 188 | } | |
| 189 | } | |
| 190 | ||
| 0931a9 | 191 | // Do the same for parentheses |
| AM | 192 | if (preg_match('/(\\(|\\))/', $url)) { |
| 193 | $in = false; | |
| 194 | for ($i=0, $len=strlen($url); $i<$len; $i++) { | |
| 195 | if ($url[$i] == '(') { | |
| 196 | if ($in) | |
| 197 | break; | |
| 198 | $in = true; | |
| 199 | } | |
| 200 | else if ($url[$i] == ')') { | |
| 201 | if (!$in) | |
| 202 | break; | |
| 203 | $in = false; | |
| 204 | } | |
| 205 | } | |
| 206 | ||
| 207 | if ($i < $len) { | |
| 208 | $suffix = substr($url, $i); | |
| 209 | $url = substr($url, 0, $i); | |
| 210 | } | |
| 211 | } | |
| 212 | ||
| 0d2144 | 213 | return $suffix; |
| AM | 214 | } |
| 0ff554 | 215 | } |
