| 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 | { | |
| d1abd8 | 27 | public static $pattern = '/##str_replacement_(\d+)##/'; |
| 0d2144 | 28 | public $mailto_pattern; |
| AM | 29 | public $link_pattern; |
| c856b7 | 30 | public $linkref_index; |
| TB | 31 | public $linkref_pattern; |
| e480ca | 32 | |
| c91d49 | 33 | protected $values = array(); |
| AM | 34 | protected $options = array(); |
| 35 | protected $linkrefs = array(); | |
| 36 | protected $urls = array(); | |
| aa055c | 37 | |
| 88ed23 | 38 | |
| 1e3254 | 39 | function __construct($options = array()) |
| 0d2144 | 40 | { |
| AM | 41 | // Simplified domain expression for UTF8 characters handling |
| 42 | // Support unicode/punycode in top-level domain part | |
| 16915e | 43 | $utf_domain = '[^?&@"\'\\/()<>\s\r\t\n]+\\.?([^\\x00-\\x2f\\x3b-\\x40\\x5b-\\x60\\x7b-\\x7f]{2,}|xn--[a-zA-Z0-9]{2,})'; |
| 0d2144 | 44 | $url1 = '.:;,'; |
| 001d33 | 45 | $url2 = 'a-zA-Z0-9%=#$@+?|!&\\/_~\\[\\]\\(\\){}\*\x80-\xFE-'; |
| 88ed23 | 46 | |
| c91d49 | 47 | $this->options = $options; |
| AM | 48 | $this->linkref_index = '/\[([^\]#]+)\](:?\s*##str_replacement_(\d+)##)/'; |
| 49 | $this->linkref_pattern = '/\[([^\]#]+)\]/'; | |
| 50 | $this->link_pattern = "/([\w]+:\/\/|\W[Ww][Ww][Ww]\.|^[Ww][Ww][Ww]\.)($utf_domain([$url1]*[$url2]+)*)/"; | |
| 51 | $this->mailto_pattern = "/(" | |
| 0d2144 | 52 | ."[-\w!\#\$%&\'*+~\/^`|{}=]+(?:\.[-\w!\#\$%&\'*+~\/^`|{}=]+)*" // local-part |
| AM | 53 | ."@$utf_domain" // domain-part |
| 54 | ."(\?[$url1$url2]+)?" // e.g. ?subject=test... | |
| 55 | .")/"; | |
| aa055c | 56 | } |
| 0ff554 | 57 | |
| 0d2144 | 58 | /** |
| AM | 59 | * Add a string to the internal list |
| 60 | * | |
| 61 | * @param string String value | |
| 62 | * @return int Index of value for retrieval | |
| 63 | */ | |
| 64 | public function add($str) | |
| 65 | { | |
| 66 | $i = count($this->values); | |
| 67 | $this->values[$i] = $str; | |
| 68 | return $i; | |
| a1013c | 69 | } |
| aa055c | 70 | |
| 0d2144 | 71 | /** |
| AM | 72 | * Build replacement string |
| 73 | */ | |
| 74 | public function get_replacement($i) | |
| 75 | { | |
| d1abd8 | 76 | return '##str_replacement_' . $i . '##'; |
| 0d2144 | 77 | } |
| aa055c | 78 | |
| 0d2144 | 79 | /** |
| AM | 80 | * Callback function used to build HTML links around URL strings |
| 81 | * | |
| 82 | * @param array Matches result from preg_replace_callback | |
| 83 | * @return int Index of saved string value | |
| 84 | */ | |
| 85 | public function link_callback($matches) | |
| 86 | { | |
| 87 | $i = -1; | |
| 88 | $scheme = strtolower($matches[1]); | |
| aa055c | 89 | |
| 0d2144 | 90 | if (preg_match('!^(http|ftp|file)s?://!i', $scheme)) { |
| AM | 91 | $url = $matches[1] . $matches[2]; |
| 0ff554 | 92 | } |
| 0d2144 | 93 | else if (preg_match('/^(\W*)(www\.)$/i', $matches[1], $m)) { |
| AM | 94 | $url = $m[2] . $matches[2]; |
| 95 | $url_prefix = 'http://'; | |
| 96 | $prefix = $m[1]; | |
| 0ff554 | 97 | } |
| A | 98 | |
| 0d2144 | 99 | if ($url) { |
| AM | 100 | $suffix = $this->parse_url_brackets($url); |
| 1e3254 | 101 | $attrib = (array)$this->options['link_attribs']; |
| AM | 102 | $attrib['href'] = $url_prefix . $url; |
| 103 | ||
| 70229c | 104 | $i = $this->add(html::a($attrib, rcube::Q($url)) . $suffix); |
| e480ca | 105 | $this->urls[$i] = $attrib['href']; |
| 0d2144 | 106 | } |
| AM | 107 | |
| 108 | // Return valid link for recognized schemes, otherwise | |
| 109 | // return the unmodified string for unrecognized schemes. | |
| 70229c | 110 | return $i >= 0 ? $prefix . $this->get_replacement($i) : $matches[0]; |
| e480ca | 111 | } |
| TB | 112 | |
| 113 | /** | |
| 114 | * Callback to add an entry to the link index | |
| 115 | */ | |
| 116 | public function linkref_addindex($matches) | |
| 117 | { | |
| 118 | $key = $matches[1]; | |
| c856b7 | 119 | $this->linkrefs[$key] = $this->urls[$matches[3]]; |
| e480ca | 120 | |
| c856b7 | 121 | return $this->get_replacement($this->add('['.$key.']')) . $matches[2]; |
| e480ca | 122 | } |
| TB | 123 | |
| 124 | /** | |
| 125 | * Callback to replace link references with real links | |
| 126 | */ | |
| 127 | public function linkref_callback($matches) | |
| 128 | { | |
| 129 | $i = 0; | |
| c856b7 | 130 | if ($url = $this->linkrefs[$matches[1]]) { |
| e480ca | 131 | $attrib = (array)$this->options['link_attribs']; |
| TB | 132 | $attrib['href'] = $url; |
| 133 | $i = $this->add(html::a($attrib, rcube::Q($matches[1]))); | |
| 134 | } | |
| 135 | ||
| 136 | return $i > 0 ? '['.$this->get_replacement($i).']' : $matches[0]; | |
| 0ff554 | 137 | } |
| A | 138 | |
| 0d2144 | 139 | /** |
| AM | 140 | * Callback function used to build mailto: links around e-mail strings |
| 141 | * | |
| 142 | * @param array Matches result from preg_replace_callback | |
| 143 | * @return int Index of saved string value | |
| 144 | */ | |
| 145 | public function mailto_callback($matches) | |
| 146 | { | |
| 147 | $href = $matches[1]; | |
| 148 | $suffix = $this->parse_url_brackets($href); | |
| 149 | $i = $this->add(html::a('mailto:' . $href, rcube::Q($href)) . $suffix); | |
| 0ff554 | 150 | |
| 0d2144 | 151 | return $i >= 0 ? $this->get_replacement($i) : ''; |
| AM | 152 | } |
| 153 | ||
| 154 | /** | |
| 155 | * Look up the index from the preg_replace matches array | |
| 156 | * and return the substitution value. | |
| 157 | * | |
| 158 | * @param array Matches result from preg_replace_callback | |
| 159 | * @return string Value at index $matches[1] | |
| 160 | */ | |
| 161 | public function replace_callback($matches) | |
| 162 | { | |
| 163 | return $this->values[$matches[1]]; | |
| 164 | } | |
| 165 | ||
| 166 | /** | |
| 167 | * Replace all defined (link|mailto) patterns with replacement string | |
| 168 | * | |
| 169 | * @param string $str Text | |
| 170 | * | |
| 171 | * @return string Text | |
| 172 | */ | |
| 173 | public function replace($str) | |
| 174 | { | |
| 175 | // search for patterns like links and e-mail addresses | |
| 176 | $str = preg_replace_callback($this->link_pattern, array($this, 'link_callback'), $str); | |
| 177 | $str = preg_replace_callback($this->mailto_pattern, array($this, 'mailto_callback'), $str); | |
| e480ca | 178 | // resolve link references |
| TB | 179 | $str = preg_replace_callback($this->linkref_index, array($this, 'linkref_addindex'), $str); |
| 180 | $str = preg_replace_callback($this->linkref_pattern, array($this, 'linkref_callback'), $str); | |
| 0d2144 | 181 | |
| AM | 182 | return $str; |
| 183 | } | |
| 184 | ||
| 185 | /** | |
| 186 | * Replace substituted strings with original values | |
| 187 | */ | |
| 188 | public function resolve($str) | |
| 189 | { | |
| 190 | return preg_replace_callback(self::$pattern, array($this, 'replace_callback'), $str); | |
| 191 | } | |
| 192 | ||
| 193 | /** | |
| 194 | * Fixes bracket characters in URL handling | |
| 195 | */ | |
| 196 | public static function parse_url_brackets(&$url) | |
| 197 | { | |
| 198 | // #1487672: special handling of square brackets, | |
| 199 | // URL regexp allows [] characters in URL, for example: | |
| 200 | // "http://example.com/?a[b]=c". However we need to handle | |
| 201 | // properly situation when a bracket is placed at the end | |
| 202 | // of the link e.g. "[http://example.com]" | |
| 0931a9 | 203 | // Yes, this is not perfect handles correctly only paired characters |
| AM | 204 | // but it should work for common cases |
| 205 | ||
| 0d2144 | 206 | if (preg_match('/(\\[|\\])/', $url)) { |
| AM | 207 | $in = false; |
| 208 | for ($i=0, $len=strlen($url); $i<$len; $i++) { | |
| 209 | if ($url[$i] == '[') { | |
| 210 | if ($in) | |
| 211 | break; | |
| 212 | $in = true; | |
| 213 | } | |
| 214 | else if ($url[$i] == ']') { | |
| 215 | if (!$in) | |
| 216 | break; | |
| 217 | $in = false; | |
| 218 | } | |
| 219 | } | |
| 220 | ||
| 221 | if ($i < $len) { | |
| 222 | $suffix = substr($url, $i); | |
| 223 | $url = substr($url, 0, $i); | |
| 224 | } | |
| 225 | } | |
| 226 | ||
| 0931a9 | 227 | // Do the same for parentheses |
| AM | 228 | if (preg_match('/(\\(|\\))/', $url)) { |
| 229 | $in = false; | |
| 230 | for ($i=0, $len=strlen($url); $i<$len; $i++) { | |
| 231 | if ($url[$i] == '(') { | |
| 232 | if ($in) | |
| 233 | break; | |
| 234 | $in = true; | |
| 235 | } | |
| 236 | else if ($url[$i] == ')') { | |
| 237 | if (!$in) | |
| 238 | break; | |
| 239 | $in = false; | |
| 240 | } | |
| 241 | } | |
| 242 | ||
| 243 | if ($i < $len) { | |
| 244 | $suffix = substr($url, $i); | |
| 245 | $url = substr($url, 0, $i); | |
| 246 | } | |
| 247 | } | |
| 248 | ||
| 0d2144 | 249 | return $suffix; |
| AM | 250 | } |
| 0ff554 | 251 | } |
