| commit | author | age | ||
| ed132e | 1 | <?php |
| T | 2 | |
| 3 | /* | |
| 4 | +-----------------------------------------------------------------------+ | |
| 5 | | program/include/rcube_vcard.php | | |
| 6 | | | | |
| e019f2 | 7 | | This file is part of the Roundcube Webmail client | |
| f5e7b3 | 8 | | Copyright (C) 2008-2009, The Roundcube Dev Team | |
| ed132e | 9 | | Licensed under the GNU GPL | |
| T | 10 | | | |
| 11 | | PURPOSE: | | |
| 12 | | Logical representation of a vcard address record | | |
| 13 | +-----------------------------------------------------------------------+ | |
| 14 | | Author: Thomas Bruederli <roundcube@gmail.com> | | |
| 15 | +-----------------------------------------------------------------------+ | |
| 16 | ||
| 1d786c | 17 | $Id$ |
| ed132e | 18 | |
| T | 19 | */ |
| 20 | ||
| 21 | ||
| 22 | /** | |
| 23 | * Logical representation of a vcard-based address record | |
| 24 | * Provides functions to parse and export vCard data format | |
| 25 | * | |
| 26 | * @package Addressbook | |
| 27 | * @author Thomas Bruederli <roundcube@gmail.com> | |
| 28 | */ | |
| 29 | class rcube_vcard | |
| 30 | { | |
| 6bdb61 | 31 | private static $values_decoded = false; |
| 0dbac3 | 32 | private $raw = array( |
| T | 33 | 'FN' => array(), |
| 34 | 'N' => array(array('','','','','')), | |
| 35 | ); | |
| ed132e | 36 | |
| T | 37 | public $business = false; |
| 38 | public $displayname; | |
| 39 | public $surname; | |
| 40 | public $firstname; | |
| 41 | public $middlename; | |
| 42 | public $nickname; | |
| 43 | public $organization; | |
| 44 | public $notes; | |
| 45 | public $email = array(); | |
| 46 | ||
| 47 | ||
| 48 | /** | |
| 49 | * Constructor | |
| 50 | */ | |
| 6bdb61 | 51 | public function __construct($vcard = null, $charset = RCMAIL_CHARSET, $detect = false) |
| ed132e | 52 | { |
| T | 53 | if (!empty($vcard)) |
| 6bdb61 | 54 | $this->load($vcard, $charset, $detect); |
| ed132e | 55 | } |
| T | 56 | |
| 57 | ||
| 58 | /** | |
| 59 | * Load record from (internal, unfolded) vcard 3.0 format | |
| 60 | * | |
| 61 | * @param string vCard string to parse | |
| 6bdb61 | 62 | * @param string Charset of string values |
| T | 63 | * @param boolean True if loading a 'foreign' vcard and extra heuristics for charset detection is required |
| ed132e | 64 | */ |
| 6bdb61 | 65 | public function load($vcard, $charset = RCMAIL_CHARSET, $detect = false) |
| ed132e | 66 | { |
| 6bdb61 | 67 | self::$values_decoded = false; |
| ed132e | 68 | $this->raw = self::vcard_decode($vcard); |
| 6bdb61 | 69 | |
| 5570ad | 70 | // resolve charset parameters |
| 6bdb61 | 71 | if ($charset == null) { |
| T | 72 | $this->raw = self::charset_convert($this->raw); |
| 73 | } | |
| 74 | // vcard has encoded values and charset should be detected | |
| 75 | else if ($detect && self::$values_decoded && | |
| 76 | ($detected_charset = self::detect_encoding(self::vcard_encode($this->raw))) && $detected_charset != RCMAIL_CHARSET) { | |
| 77 | $this->raw = self::charset_convert($this->raw, $detected_charset); | |
| 78 | } | |
| ed132e | 79 | |
| T | 80 | // find well-known address fields |
| 5570ad | 81 | $this->displayname = $this->raw['FN'][0][0]; |
| ed132e | 82 | $this->surname = $this->raw['N'][0][0]; |
| T | 83 | $this->firstname = $this->raw['N'][0][1]; |
| 84 | $this->middlename = $this->raw['N'][0][2]; | |
| 5570ad | 85 | $this->nickname = $this->raw['NICKNAME'][0][0]; |
| T | 86 | $this->organization = $this->raw['ORG'][0][0]; |
| 87 | $this->business = ($this->raw['X-ABSHOWAS'][0][0] == 'COMPANY') || (join('', (array)$this->raw['N'][0]) == '' && !empty($this->organization)); | |
| ed132e | 88 | |
| T | 89 | foreach ((array)$this->raw['EMAIL'] as $i => $raw_email) |
| bb8781 | 90 | $this->email[$i] = is_array($raw_email) ? $raw_email[0] : $raw_email; |
| ed132e | 91 | |
| T | 92 | // make the pref e-mail address the first entry in $this->email |
| 93 | $pref_index = $this->get_type_index('EMAIL', 'pref'); | |
| 94 | if ($pref_index > 0) { | |
| 95 | $tmp = $this->email[0]; | |
| 96 | $this->email[0] = $this->email[$pref_index]; | |
| 97 | $this->email[$pref_index] = $tmp; | |
| 98 | } | |
| c56f1f | 99 | |
| A | 100 | // make sure displayname is not empty (required by RFC2426) |
| 101 | if (!strlen($this->displayname)) { | |
| 102 | // the same method is used in steps/mail/addcontact.inc | |
| 103 | $this->displayname = ucfirst(preg_replace('/[\.\-]/', ' ', | |
| 104 | substr($this->email[0], 0, strpos($this->email[0], '@')))); | |
| 105 | } | |
| ed132e | 106 | } |
| T | 107 | |
| 108 | ||
| 109 | /** | |
| 110 | * Convert the data structure into a vcard 3.0 string | |
| 111 | */ | |
| 112 | public function export() | |
| 113 | { | |
| 114 | return self::rfc2425_fold(self::vcard_encode($this->raw)); | |
| 115 | } | |
| 116 | ||
| 117 | ||
| 118 | /** | |
| 119 | * Setter for address record fields | |
| 120 | * | |
| 121 | * @param string Field name | |
| 122 | * @param string Field value | |
| 123 | * @param string Section name | |
| 124 | */ | |
| 0dbac3 | 125 | public function set($field, $value, $section = 'HOME') |
| ed132e | 126 | { |
| T | 127 | switch ($field) { |
| 128 | case 'name': | |
| 129 | case 'displayname': | |
| 5570ad | 130 | $this->raw['FN'][0][0] = $value; |
| ed132e | 131 | break; |
| T | 132 | |
| 133 | case 'firstname': | |
| 134 | $this->raw['N'][0][1] = $value; | |
| 135 | break; | |
| 136 | ||
| 137 | case 'surname': | |
| 138 | $this->raw['N'][0][0] = $value; | |
| 139 | break; | |
| 140 | ||
| 141 | case 'nickname': | |
| 5570ad | 142 | $this->raw['NICKNAME'][0][0] = $value; |
| ed132e | 143 | break; |
| T | 144 | |
| 145 | case 'organization': | |
| 5570ad | 146 | $this->raw['ORG'][0][0] = $value; |
| ed132e | 147 | break; |
| T | 148 | |
| 149 | case 'email': | |
| 150 | $index = $this->get_type_index('EMAIL', $section); | |
| 151 | if (!is_array($this->raw['EMAIL'][$index])) { | |
| 152 | $this->raw['EMAIL'][$index] = array(0 => $value, 'type' => array('INTERNET', $section, 'pref')); | |
| 153 | } | |
| 154 | else { | |
| 155 | $this->raw['EMAIL'][$index][0] = $value; | |
| 156 | } | |
| 157 | break; | |
| 158 | } | |
| 159 | } | |
| 160 | ||
| 161 | ||
| 162 | /** | |
| 163 | * Find index with the '$type' attribute | |
| 164 | * | |
| 165 | * @param string Field name | |
| 166 | * @return int Field index having $type set | |
| 167 | */ | |
| 168 | private function get_type_index($field, $type = 'pref') | |
| 169 | { | |
| 170 | $result = 0; | |
| 171 | if ($this->raw[$field]) { | |
| 172 | foreach ($this->raw[$field] as $i => $data) { | |
| 173 | if (is_array($data['type']) && in_array_nocase('pref', $data['type'])) | |
| 174 | $result = $i; | |
| 175 | } | |
| 176 | } | |
| 177 | ||
| 178 | return $result; | |
| 179 | } | |
| 5570ad | 180 | |
| T | 181 | |
| 182 | /** | |
| 183 | * Convert a whole vcard (array) to UTF-8. | |
| 6bdb61 | 184 | * If $force_charset is null, each member value that has a charset parameter will be converted |
| 5570ad | 185 | */ |
| 6bdb61 | 186 | private static function charset_convert($card, $force_charset = null) |
| 5570ad | 187 | { |
| T | 188 | foreach ($card as $key => $node) { |
| 189 | foreach ($node as $i => $subnode) { | |
| 6bdb61 | 190 | if (is_array($subnode) && (($charset = $force_charset) || ($subnode['charset'] && ($charset = $subnode['charset'][0])))) { |
| 5570ad | 191 | foreach ($subnode as $j => $value) { |
| T | 192 | if (is_numeric($j) && is_string($value)) |
| 193 | $card[$key][$i][$j] = rcube_charset_convert($value, $charset); | |
| 194 | } | |
| 195 | unset($card[$key][$i]['charset']); | |
| 196 | } | |
| 197 | } | |
| 198 | } | |
| 199 | ||
| 200 | return $card; | |
| 201 | } | |
| ed132e | 202 | |
| T | 203 | |
| 204 | /** | |
| 205 | * Factory method to import a vcard file | |
| 206 | * | |
| 207 | * @param string vCard file content | |
| 208 | * @return array List of rcube_vcard objects | |
| 209 | */ | |
| 210 | public static function import($data) | |
| 211 | { | |
| 212 | $out = array(); | |
| 213 | ||
| 5570ad | 214 | // check if charsets are specified (usually vcard version < 3.0 but this is not reliable) |
| T | 215 | if (preg_match('/charset=/i', substr($data, 0, 2048))) |
| 216 | $charset = null; | |
| ed132e | 217 | // detect charset and convert to utf-8 |
| 5570ad | 218 | else if (($charset = self::detect_encoding($data)) && $charset != RCMAIL_CHARSET) { |
| T | 219 | $data = rcube_charset_convert($data, $charset); |
| 6fa87f | 220 | $data = preg_replace(array('/^[\xFE\xFF]{2}/', '/^\xEF\xBB\xBF/', '/^\x00+/'), '', $data); // also remove BOM |
| 5570ad | 221 | $charset = RCMAIL_CHARSET; |
| ed132e | 222 | } |
| T | 223 | |
| 224 | $vcard_block = ''; | |
| 225 | $in_vcard_block = false; | |
| 226 | ||
| 227 | foreach (preg_split("/[\r\n]+/", $data) as $i => $line) { | |
| 228 | if ($in_vcard_block && !empty($line)) | |
| 229 | $vcard_block .= $line . "\n"; | |
| 230 | ||
| 928bca | 231 | $line = trim($line); |
| A | 232 | |
| 233 | if (preg_match('/^END:VCARD$/i', $line)) { | |
| ed132e | 234 | // parse vcard |
| 6bdb61 | 235 | $obj = new rcube_vcard(self::cleanup($vcard_block), $charset, true); |
| ed132e | 236 | if (!empty($obj->displayname)) |
| T | 237 | $out[] = $obj; |
| 238 | ||
| 239 | $in_vcard_block = false; | |
| 240 | } | |
| 928bca | 241 | else if (preg_match('/^BEGIN:VCARD$/i', $line)) { |
| ed132e | 242 | $vcard_block = $line . "\n"; |
| T | 243 | $in_vcard_block = true; |
| 244 | } | |
| 245 | } | |
| 246 | ||
| 247 | return $out; | |
| 248 | } | |
| 249 | ||
| 250 | ||
| 251 | /** | |
| 252 | * Normalize vcard data for better parsing | |
| 253 | * | |
| 254 | * @param string vCard block | |
| 255 | * @return string Cleaned vcard block | |
| 256 | */ | |
| 257 | private static function cleanup($vcard) | |
| 258 | { | |
| 259 | // Convert special types (like Skype) to normal type='skype' classes with this simple regex ;) | |
| 260 | $vcard = preg_replace( | |
| 261 | '/item(\d+)\.(TEL|URL)([^:]*?):(.*?)item\1.X-ABLabel:(?:_\$!<)?([\w-() ]*)(?:>!\$_)?./s', | |
| 262 | '\2;type=\5\3:\4', | |
| 263 | $vcard); | |
| 264 | ||
| 265 | // Remove cruft like item1.X-AB*, item1.ADR instead of ADR, and empty lines | |
| 266 | $vcard = preg_replace(array('/^item\d*\.X-AB.*$/m', '/^item\d*\./m', "/\n+/"), array('', '', "\n"), $vcard); | |
| 267 | ||
| b58f11 | 268 | // if N doesn't have any semicolons, add some |
| T | 269 | $vcard = preg_replace('/^(N:[^;\R]*)$/m', '\1;;;;', $vcard); |
| ed132e | 270 | |
| T | 271 | return $vcard; |
| 272 | } | |
| 273 | ||
| 478c7c | 274 | private static function rfc2425_fold_callback($matches) |
| A | 275 | { |
| 276 | return ":\n ".rtrim(chunk_split($matches[1], 72, "\n ")); | |
| 277 | } | |
| ed132e | 278 | |
| T | 279 | private static function rfc2425_fold($val) |
| 280 | { | |
| 5e6815 | 281 | return preg_replace_callback('/:([^\n]{72,})/', array('self', 'rfc2425_fold_callback'), $val) . "\n"; |
| ed132e | 282 | } |
| T | 283 | |
| 284 | ||
| 285 | /** | |
| 286 | * Decodes a vcard block (vcard 3.0 format, unfolded) | |
| 287 | * into an array structure | |
| 288 | * | |
| 289 | * @param string vCard block to parse | |
| 290 | * @return array Raw data structure | |
| 291 | */ | |
| 292 | private static function vcard_decode($vcard) | |
| 293 | { | |
| 294 | // Perform RFC2425 line unfolding | |
| 295 | $vcard = preg_replace(array("/\r/", "/\n\s+/"), '', $vcard); | |
| 296 | ||
| b58f11 | 297 | $lines = preg_split('/\r?\n/', $vcard); |
| ed132e | 298 | $data = array(); |
| b58f11 | 299 | |
| T | 300 | for ($i=0; $i < count($lines); $i++) { |
| 301 | if (!preg_match('/^([^\\:]*):(.+)$/', $lines[$i], $line)) | |
| 302 | continue; | |
| ed132e | 303 | |
| b58f11 | 304 | // convert 2.1-style "EMAIL;internet;home:" to 3.0-style "EMAIL;TYPE=internet;TYPE=home:" |
| T | 305 | if (($data['VERSION'][0] == "2.1") && preg_match('/^([^;]+);([^:]+)/', $line[1], $regs2) && !preg_match('/^TYPE=/i', $regs2[2])) { |
| 306 | $line[1] = $regs2[1]; | |
| 307 | foreach (explode(';', $regs2[2]) as $prop) | |
| 308 | $line[1] .= ';' . (strpos($prop, '=') ? $prop : 'TYPE='.$prop); | |
| ed132e | 309 | } |
| T | 310 | |
| cc97ea | 311 | if (!preg_match('/^(BEGIN|END)$/i', $line[1]) && preg_match_all('/([^\\;]+);?/', $line[1], $regs2)) { |
| 93af15 | 312 | $entry = array(); |
| cc97ea | 313 | $field = strtoupper($regs2[1][0]); |
| b58f11 | 314 | |
| T | 315 | foreach($regs2[1] as $attrid => $attr) { |
| 316 | if ((list($key, $value) = explode('=', $attr)) && $value) { | |
| 5570ad | 317 | $value = trim($value); |
| b58f11 | 318 | if ($key == 'ENCODING') { |
| 93af15 | 319 | // add next line(s) to value string if QP line end detected |
| 23a2ee | 320 | while ($value == 'QUOTED-PRINTABLE' && preg_match('/=$/', $lines[$i])) |
| b58f11 | 321 | $line[2] .= "\n" . $lines[++$i]; |
| T | 322 | |
| 323 | $line[2] = self::decode_value($line[2], $value); | |
| 324 | } | |
| 325 | else | |
| 326 | $entry[strtolower($key)] = array_merge((array)$entry[strtolower($key)], (array)self::vcard_unquote($value, ',')); | |
| 327 | } | |
| 328 | else if ($attrid > 0) { | |
| 93af15 | 329 | $entry[$key] = true; // true means attr without =value |
| b58f11 | 330 | } |
| T | 331 | } |
| 332 | ||
| 93af15 | 333 | $entry = array_merge($entry, (array)self::vcard_unquote($line[2])); |
| 5570ad | 334 | $data[$field][] = $entry; |
| b58f11 | 335 | } |
| ed132e | 336 | } |
| b58f11 | 337 | |
| T | 338 | unset($data['VERSION']); |
| ed132e | 339 | return $data; |
| T | 340 | } |
| 341 | ||
| 342 | ||
| 343 | /** | |
| 344 | * Split quoted string | |
| 345 | * | |
| 346 | * @param string vCard string to split | |
| 347 | * @param string Separator char/string | |
| 348 | * @return array List with splitted values | |
| 349 | */ | |
| 350 | private static function vcard_unquote($s, $sep = ';') | |
| 351 | { | |
| 352 | // break string into parts separated by $sep, but leave escaped $sep alone | |
| 353 | if (count($parts = explode($sep, strtr($s, array("\\$sep" => "\007")))) > 1) { | |
| 354 | foreach($parts as $s) { | |
| 355 | $result[] = self::vcard_unquote(strtr($s, array("\007" => "\\$sep")), $sep); | |
| 356 | } | |
| 357 | return $result; | |
| 358 | } | |
| 359 | else { | |
| 360 | return strtr($s, array("\r" => '', '\\\\' => '\\', '\n' => "\n", '\,' => ',', '\;' => ';', '\:' => ':')); | |
| bb8781 | 361 | } |
| T | 362 | } |
| 363 | ||
| 364 | ||
| 365 | /** | |
| 366 | * Decode a given string with the encoding rule from ENCODING attributes | |
| 367 | * | |
| 368 | * @param string String to decode | |
| 369 | * @param string Encoding type (quoted-printable and base64 supported) | |
| 370 | * @return string Decoded 8bit value | |
| 371 | */ | |
| 372 | private static function decode_value($value, $encoding) | |
| 373 | { | |
| 374 | switch (strtolower($encoding)) { | |
| 375 | case 'quoted-printable': | |
| 6bdb61 | 376 | self::$values_decoded = true; |
| bb8781 | 377 | return quoted_printable_decode($value); |
| T | 378 | |
| 379 | case 'base64': | |
| 6bdb61 | 380 | self::$values_decoded = true; |
| bb8781 | 381 | return base64_decode($value); |
| T | 382 | |
| 383 | default: | |
| 384 | return $value; | |
| ed132e | 385 | } |
| T | 386 | } |
| 387 | ||
| 388 | ||
| 389 | /** | |
| 390 | * Encodes an entry for storage in our database (vcard 3.0 format, unfolded) | |
| 391 | * | |
| 392 | * @param array Raw data structure to encode | |
| 393 | * @return string vCard encoded string | |
| 394 | */ | |
| 395 | static function vcard_encode($data) | |
| 396 | { | |
| 397 | foreach((array)$data as $type => $entries) { | |
| 398 | /* valid N has 5 properties */ | |
| b58f11 | 399 | while ($type == "N" && is_array($entries[0]) && count($entries[0]) < 5) |
| ed132e | 400 | $entries[0][] = ""; |
| T | 401 | |
| 402 | foreach((array)$entries as $entry) { | |
| 403 | $attr = ''; | |
| 404 | if (is_array($entry)) { | |
| 405 | $value = array(); | |
| 406 | foreach($entry as $attrname => $attrvalues) { | |
| 407 | if (is_int($attrname)) | |
| 408 | $value[] = $attrvalues; | |
| 409 | elseif ($attrvalues === true) | |
| 93af15 | 410 | $attr .= ";$attrname"; // true means just tag, not tag=value, as in PHOTO;BASE64:... |
| ed132e | 411 | else { |
| T | 412 | foreach((array)$attrvalues as $attrvalue) |
| 413 | $attr .= ";$attrname=" . self::vcard_quote($attrvalue, ','); | |
| 414 | } | |
| 415 | } | |
| 416 | } | |
| 417 | else { | |
| 418 | $value = $entry; | |
| 419 | } | |
| 420 | ||
| 421 | $vcard .= self::vcard_quote($type) . $attr . ':' . self::vcard_quote($value) . "\n"; | |
| 422 | } | |
| 423 | } | |
| 424 | ||
| 0dbac3 | 425 | return "BEGIN:VCARD\nVERSION:3.0\n{$vcard}END:VCARD"; |
| ed132e | 426 | } |
| T | 427 | |
| 428 | ||
| 429 | /** | |
| 430 | * Join indexed data array to a vcard quoted string | |
| 431 | * | |
| 432 | * @param array Field data | |
| 433 | * @param string Separator | |
| 434 | * @return string Joined and quoted string | |
| 435 | */ | |
| 436 | private static function vcard_quote($s, $sep = ';') | |
| 437 | { | |
| 438 | if (is_array($s)) { | |
| 439 | foreach($s as $part) { | |
| 440 | $r[] = self::vcard_quote($part, $sep); | |
| 441 | } | |
| 442 | return(implode($sep, (array)$r)); | |
| 443 | } | |
| 444 | else { | |
| 445 | return strtr($s, array('\\' => '\\\\', "\r" => '', "\n" => '\n', ';' => '\;', ':' => '\:')); | |
| 446 | } | |
| 447 | } | |
| 448 | ||
| 449 | ||
| 450 | /** | |
| 451 | * Returns UNICODE type based on BOM (Byte Order Mark) | |
| 452 | * | |
| 453 | * @param string Input string to test | |
| 454 | * @return string Detected encoding | |
| 455 | */ | |
| 456 | private static function detect_encoding($string) | |
| 457 | { | |
| 458 | if (substr($string, 0, 4) == "\0\0\xFE\xFF") return 'UTF-32BE'; // Big Endian | |
| 459 | if (substr($string, 0, 4) == "\xFF\xFE\0\0") return 'UTF-32LE'; // Little Endian | |
| 460 | if (substr($string, 0, 2) == "\xFE\xFF") return 'UTF-16BE'; // Big Endian | |
| 461 | if (substr($string, 0, 2) == "\xFF\xFE") return 'UTF-16LE'; // Little Endian | |
| 462 | if (substr($string, 0, 3) == "\xEF\xBB\xBF") return 'UTF-8'; | |
| 463 | ||
| 6fa87f | 464 | // use mb_detect_encoding() |
| T | 465 | $encodings = array('UTF-8', 'ISO-8859-1', 'ISO-8859-2', 'ISO-8859-3', |
| 466 | 'ISO-8859-4', 'ISO-8859-5', 'ISO-8859-6', 'ISO-8859-7', 'ISO-8859-8', 'ISO-8859-9', | |
| 467 | 'ISO-8859-10', 'ISO-8859-13', 'ISO-8859-14', 'ISO-8859-15', 'ISO-8859-16', | |
| 468 | 'WINDOWS-1252', 'WINDOWS-1251', 'BIG5', 'GB2312'); | |
| 469 | ||
| ae9124 | 470 | if (function_exists('mb_detect_encoding') && ($enc = mb_detect_encoding($string, $encodings))) |
| abdc58 | 471 | return $enc; |
| A | 472 | |
| ed132e | 473 | // No match, check for UTF-8 |
| T | 474 | // from http://w3.org/International/questions/qa-forms-utf-8.html |
| 475 | if (preg_match('/\A( | |
| 476 | [\x09\x0A\x0D\x20-\x7E] | |
| 477 | | [\xC2-\xDF][\x80-\xBF] | |
| 478 | | \xE0[\xA0-\xBF][\x80-\xBF] | |
| 479 | | [\xE1-\xEC\xEE\xEF][\x80-\xBF]{2} | |
| 480 | | \xED[\x80-\x9F][\x80-\xBF] | |
| 481 | | \xF0[\x90-\xBF][\x80-\xBF]{2} | |
| 482 | | [\xF1-\xF3][\x80-\xBF]{3} | |
| 483 | | \xF4[\x80-\x8F][\x80-\xBF]{2} | |
| 484 | )*\z/xs', substr($string, 0, 2048))) | |
| 485 | return 'UTF-8'; | |
| 486 | ||
| ae9124 | 487 | return rcmail::get_instance()->config->get('default_charset', 'ISO-8859-1'); # fallback to Latin-1 |
| ed132e | 488 | } |
| T | 489 | |
| 490 | } | |
| 491 | ||
| 492 | ||
