| 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(); | |
| 848410 | 37 | protected $noword = '[^\w@.#-]'; |
| aa055c | 38 | |
| 88ed23 | 39 | |
| 1e3254 | 40 | function __construct($options = array()) |
| 0d2144 | 41 | { |
| AM | 42 | // Simplified domain expression for UTF8 characters handling |
| 43 | // Support unicode/punycode in top-level domain part | |
| 16915e | 44 | $utf_domain = '[^?&@"\'\\/()<>\s\r\t\n]+\\.?([^\\x00-\\x2f\\x3b-\\x40\\x5b-\\x60\\x7b-\\x7f]{2,}|xn--[a-zA-Z0-9]{2,})'; |
| 0d2144 | 45 | $url1 = '.:;,'; |
| 001d33 | 46 | $url2 = 'a-zA-Z0-9%=#$@+?|!&\\/_~\\[\\]\\(\\){}\*\x80-\xFE-'; |
| 88ed23 | 47 | |
| 848410 | 48 | // Supported link prefixes |
| AM | 49 | $link_prefix = "([\w]+:\/\/|{$this->noword}[Ww][Ww][Ww]\.|^[Ww][Ww][Ww]\.)"; |
| 50 | ||
| c91d49 | 51 | $this->options = $options; |
| AM | 52 | $this->linkref_index = '/\[([^\]#]+)\](:?\s*##str_replacement_(\d+)##)/'; |
| 53 | $this->linkref_pattern = '/\[([^\]#]+)\]/'; | |
| 848410 | 54 | $this->link_pattern = "/$link_prefix($utf_domain([$url1]*[$url2]+)*)/"; |
| c91d49 | 55 | $this->mailto_pattern = "/(" |
| 0d2144 | 56 | ."[-\w!\#\$%&\'*+~\/^`|{}=]+(?:\.[-\w!\#\$%&\'*+~\/^`|{}=]+)*" // local-part |
| AM | 57 | ."@$utf_domain" // domain-part |
| 58 | ."(\?[$url1$url2]+)?" // e.g. ?subject=test... | |
| 59 | .")/"; | |
| aa055c | 60 | } |
| 0ff554 | 61 | |
| 0d2144 | 62 | /** |
| AM | 63 | * Add a string to the internal list |
| 64 | * | |
| 65 | * @param string String value | |
| 66 | * @return int Index of value for retrieval | |
| 67 | */ | |
| 68 | public function add($str) | |
| 69 | { | |
| 70 | $i = count($this->values); | |
| 71 | $this->values[$i] = $str; | |
| 72 | return $i; | |
| a1013c | 73 | } |
| aa055c | 74 | |
| 0d2144 | 75 | /** |
| AM | 76 | * Build replacement string |
| 77 | */ | |
| 78 | public function get_replacement($i) | |
| 79 | { | |
| d1abd8 | 80 | return '##str_replacement_' . $i . '##'; |
| 0d2144 | 81 | } |
| aa055c | 82 | |
| 0d2144 | 83 | /** |
| AM | 84 | * Callback function used to build HTML links around URL strings |
| 85 | * | |
| 86 | * @param array Matches result from preg_replace_callback | |
| 87 | * @return int Index of saved string value | |
| 88 | */ | |
| 89 | public function link_callback($matches) | |
| 90 | { | |
| 91 | $i = -1; | |
| 92 | $scheme = strtolower($matches[1]); | |
| aa055c | 93 | |
| 0d2144 | 94 | if (preg_match('!^(http|ftp|file)s?://!i', $scheme)) { |
| AM | 95 | $url = $matches[1] . $matches[2]; |
| 0ff554 | 96 | } |
| 848410 | 97 | else if (preg_match("/^({$this->noword}*)(www\.)$/i", $matches[1], $m)) { |
| 0d2144 | 98 | $url = $m[2] . $matches[2]; |
| AM | 99 | $url_prefix = 'http://'; |
| 100 | $prefix = $m[1]; | |
| 0ff554 | 101 | } |
| A | 102 | |
| 0d2144 | 103 | if ($url) { |
| AM | 104 | $suffix = $this->parse_url_brackets($url); |
| 1e3254 | 105 | $attrib = (array)$this->options['link_attribs']; |
| AM | 106 | $attrib['href'] = $url_prefix . $url; |
| 107 | ||
| 70229c | 108 | $i = $this->add(html::a($attrib, rcube::Q($url)) . $suffix); |
| e480ca | 109 | $this->urls[$i] = $attrib['href']; |
| 0d2144 | 110 | } |
| AM | 111 | |
| 112 | // Return valid link for recognized schemes, otherwise | |
| 113 | // return the unmodified string for unrecognized schemes. | |
| 70229c | 114 | return $i >= 0 ? $prefix . $this->get_replacement($i) : $matches[0]; |
| e480ca | 115 | } |
| TB | 116 | |
| 117 | /** | |
| 118 | * Callback to add an entry to the link index | |
| 119 | */ | |
| 120 | public function linkref_addindex($matches) | |
| 121 | { | |
| 122 | $key = $matches[1]; | |
| c856b7 | 123 | $this->linkrefs[$key] = $this->urls[$matches[3]]; |
| e480ca | 124 | |
| c856b7 | 125 | return $this->get_replacement($this->add('['.$key.']')) . $matches[2]; |
| e480ca | 126 | } |
| TB | 127 | |
| 128 | /** | |
| 129 | * Callback to replace link references with real links | |
| 130 | */ | |
| 131 | public function linkref_callback($matches) | |
| 132 | { | |
| 133 | $i = 0; | |
| c856b7 | 134 | if ($url = $this->linkrefs[$matches[1]]) { |
| e480ca | 135 | $attrib = (array)$this->options['link_attribs']; |
| TB | 136 | $attrib['href'] = $url; |
| 137 | $i = $this->add(html::a($attrib, rcube::Q($matches[1]))); | |
| 138 | } | |
| 139 | ||
| 140 | return $i > 0 ? '['.$this->get_replacement($i).']' : $matches[0]; | |
| 0ff554 | 141 | } |
| A | 142 | |
| 0d2144 | 143 | /** |
| AM | 144 | * Callback function used to build mailto: links around e-mail strings |
| 145 | * | |
| 146 | * @param array Matches result from preg_replace_callback | |
| 147 | * @return int Index of saved string value | |
| 148 | */ | |
| 149 | public function mailto_callback($matches) | |
| 150 | { | |
| 151 | $href = $matches[1]; | |
| 152 | $suffix = $this->parse_url_brackets($href); | |
| 153 | $i = $this->add(html::a('mailto:' . $href, rcube::Q($href)) . $suffix); | |
| 0ff554 | 154 | |
| 0d2144 | 155 | return $i >= 0 ? $this->get_replacement($i) : ''; |
| AM | 156 | } |
| 157 | ||
| 158 | /** | |
| 159 | * Look up the index from the preg_replace matches array | |
| 160 | * and return the substitution value. | |
| 161 | * | |
| 162 | * @param array Matches result from preg_replace_callback | |
| 163 | * @return string Value at index $matches[1] | |
| 164 | */ | |
| 165 | public function replace_callback($matches) | |
| 166 | { | |
| 167 | return $this->values[$matches[1]]; | |
| 168 | } | |
| 169 | ||
| 170 | /** | |
| 171 | * Replace all defined (link|mailto) patterns with replacement string | |
| 172 | * | |
| 173 | * @param string $str Text | |
| 174 | * | |
| 175 | * @return string Text | |
| 176 | */ | |
| 177 | public function replace($str) | |
| 178 | { | |
| 179 | // search for patterns like links and e-mail addresses | |
| 180 | $str = preg_replace_callback($this->link_pattern, array($this, 'link_callback'), $str); | |
| 181 | $str = preg_replace_callback($this->mailto_pattern, array($this, 'mailto_callback'), $str); | |
| e480ca | 182 | // resolve link references |
| TB | 183 | $str = preg_replace_callback($this->linkref_index, array($this, 'linkref_addindex'), $str); |
| 184 | $str = preg_replace_callback($this->linkref_pattern, array($this, 'linkref_callback'), $str); | |
| 0d2144 | 185 | |
| AM | 186 | return $str; |
| 187 | } | |
| 188 | ||
| 189 | /** | |
| 190 | * Replace substituted strings with original values | |
| 191 | */ | |
| 192 | public function resolve($str) | |
| 193 | { | |
| 194 | return preg_replace_callback(self::$pattern, array($this, 'replace_callback'), $str); | |
| 195 | } | |
| 196 | ||
| 197 | /** | |
| 198 | * Fixes bracket characters in URL handling | |
| 199 | */ | |
| 200 | public static function parse_url_brackets(&$url) | |
| 201 | { | |
| 202 | // #1487672: special handling of square brackets, | |
| 203 | // URL regexp allows [] characters in URL, for example: | |
| 204 | // "http://example.com/?a[b]=c". However we need to handle | |
| 205 | // properly situation when a bracket is placed at the end | |
| 206 | // of the link e.g. "[http://example.com]" | |
| 0931a9 | 207 | // Yes, this is not perfect handles correctly only paired characters |
| AM | 208 | // but it should work for common cases |
| 209 | ||
| 0d2144 | 210 | if (preg_match('/(\\[|\\])/', $url)) { |
| AM | 211 | $in = false; |
| 212 | for ($i=0, $len=strlen($url); $i<$len; $i++) { | |
| 213 | if ($url[$i] == '[') { | |
| 214 | if ($in) | |
| 215 | break; | |
| 216 | $in = true; | |
| 217 | } | |
| 218 | else if ($url[$i] == ']') { | |
| 219 | if (!$in) | |
| 220 | break; | |
| 221 | $in = false; | |
| 222 | } | |
| 223 | } | |
| 224 | ||
| 225 | if ($i < $len) { | |
| 226 | $suffix = substr($url, $i); | |
| 227 | $url = substr($url, 0, $i); | |
| 228 | } | |
| 229 | } | |
| 230 | ||
| 0931a9 | 231 | // Do the same for parentheses |
| AM | 232 | if (preg_match('/(\\(|\\))/', $url)) { |
| 233 | $in = false; | |
| 234 | for ($i=0, $len=strlen($url); $i<$len; $i++) { | |
| 235 | if ($url[$i] == '(') { | |
| 236 | if ($in) | |
| 237 | break; | |
| 238 | $in = true; | |
| 239 | } | |
| 240 | else if ($url[$i] == ')') { | |
| 241 | if (!$in) | |
| 242 | break; | |
| 243 | $in = false; | |
| 244 | } | |
| 245 | } | |
| 246 | ||
| 247 | if ($i < $len) { | |
| 248 | $suffix = substr($url, $i); | |
| 249 | $url = substr($url, 0, $i); | |
| 250 | } | |
| 251 | } | |
| 252 | ||
| 0d2144 | 253 | return $suffix; |
| AM | 254 | } |
| 0ff554 | 255 | } |
