githubFork/roundcubemail.git

thomascube

2010-04-01 1d773d71414316e0b9836a15c35576593427ee21

commit \| author \| age
ed132e	1	<?php
T	2
	3	/*
	4	+-----------------------------------------------------------------------+
	5	\| program/include/rcube_vcard.php \|
	6	\| \|
	7	\| This file is part of the RoundCube Webmail client \|
cbbef3	8	\| Copyright (C) 2008-2009, RoundCube Dev. - Switzerland \|
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	{
0dbac3	31	private $raw = array(
T	32	'FN' => array(),
	33	'N' => array(array('','','','','')),
	34	);
ed132e	35
T	36	public $business = false;
	37	public $displayname;
	38	public $surname;
	39	public $firstname;
	40	public $middlename;
	41	public $nickname;
	42	public $organization;
	43	public $notes;
	44	public $email = array();
	45
	46
	47	/**
	48	* Constructor
	49	*/
5570ad	50	public function __construct($vcard = null, $charset = RCMAIL_CHARSET)
ed132e	51	{
T	52	if (!empty($vcard))
5570ad	53	$this->load($vcard, $charset);
ed132e	54	}
T	55
	56
	57	/**
	58	* Load record from (internal, unfolded) vcard 3.0 format
	59	*
	60	* @param string vCard string to parse
	61	*/
5570ad	62	public function load($vcard, $charset = RCMAIL_CHARSET)
ed132e	63	{
T	64	$this->raw = self::vcard_decode($vcard);
5570ad	65
T	66	// resolve charset parameters
	67	if ($charset == null)
	68	$this->raw = $this->charset_convert($this->raw);
ed132e	69
T	70	// find well-known address fields
5570ad	71	$this->displayname = $this->raw['FN'][0][0];
ed132e	72	$this->surname = $this->raw['N'][0][0];
T	73	$this->firstname = $this->raw['N'][0][1];
	74	$this->middlename = $this->raw['N'][0][2];
5570ad	75	$this->nickname = $this->raw['NICKNAME'][0][0];
T	76	$this->organization = $this->raw['ORG'][0][0];
	77	$this->business = ($this->raw['X-ABSHOWAS'][0][0] == 'COMPANY') \|\| (join('', (array)$this->raw['N'][0]) == '' && !empty($this->organization));
ed132e	78
T	79	foreach ((array)$this->raw['EMAIL'] as $i => $raw_email)
bb8781	80	$this->email[$i] = is_array($raw_email) ? $raw_email[0] : $raw_email;
ed132e	81
T	82	// make the pref e-mail address the first entry in $this->email
	83	$pref_index = $this->get_type_index('EMAIL', 'pref');
	84	if ($pref_index > 0) {
	85	$tmp = $this->email[0];
	86	$this->email[0] = $this->email[$pref_index];
	87	$this->email[$pref_index] = $tmp;
	88	}
	89	}
	90
	91
	92	/**
	93	* Convert the data structure into a vcard 3.0 string
	94	*/
	95	public function export()
	96	{
	97	return self::rfc2425_fold(self::vcard_encode($this->raw));
	98	}
	99
	100
	101	/**
	102	* Setter for address record fields
	103	*
	104	* @param string Field name
	105	* @param string Field value
	106	* @param string Section name
	107	*/
0dbac3	108	public function set($field, $value, $section = 'HOME')
ed132e	109	{
T	110	switch ($field) {
	111	case 'name':
	112	case 'displayname':
5570ad	113	$this->raw['FN'][0][0] = $value;
ed132e	114	break;
T	115
	116	case 'firstname':
	117	$this->raw['N'][0][1] = $value;
	118	break;
	119
	120	case 'surname':
	121	$this->raw['N'][0][0] = $value;
	122	break;
	123
	124	case 'nickname':
5570ad	125	$this->raw['NICKNAME'][0][0] = $value;
ed132e	126	break;
T	127
	128	case 'organization':
5570ad	129	$this->raw['ORG'][0][0] = $value;
ed132e	130	break;
T	131
	132	case 'email':
	133	$index = $this->get_type_index('EMAIL', $section);
	134	if (!is_array($this->raw['EMAIL'][$index])) {
	135	$this->raw['EMAIL'][$index] = array(0 => $value, 'type' => array('INTERNET', $section, 'pref'));
	136	}
	137	else {
	138	$this->raw['EMAIL'][$index][0] = $value;
	139	}
	140	break;
	141	}
	142	}
	143
	144
	145	/**
	146	* Find index with the '$type' attribute
	147	*
	148	* @param string Field name
	149	* @return int Field index having $type set
	150	*/
	151	private function get_type_index($field, $type = 'pref')
	152	{
	153	$result = 0;
	154	if ($this->raw[$field]) {
	155	foreach ($this->raw[$field] as $i => $data) {
	156	if (is_array($data['type']) && in_array_nocase('pref', $data['type']))
	157	$result = $i;
	158	}
	159	}
	160
	161	return $result;
	162	}
5570ad	163
T	164
	165	/**
	166	* Convert a whole vcard (array) to UTF-8.
	167	* Each member value that has a charset parameter will be converted.
	168	*/
	169	private function charset_convert($card)
	170	{
	171	foreach ($card as $key => $node) {
	172	foreach ($node as $i => $subnode) {
	173	if (is_array($subnode) && $subnode['charset'] && ($charset = $subnode['charset'][0])) {
	174	foreach ($subnode as $j => $value) {
	175	if (is_numeric($j) && is_string($value))
	176	$card[$key][$i][$j] = rcube_charset_convert($value, $charset);
	177	}
	178	unset($card[$key][$i]['charset']);
	179	}
	180	}
	181	}
	182
	183	return $card;
	184	}
ed132e	185
T	186
	187	/**
	188	* Factory method to import a vcard file
	189	*
	190	* @param string vCard file content
	191	* @return array List of rcube_vcard objects
	192	*/
	193	public static function import($data)
	194	{
	195	$out = array();
	196
5570ad	197	// check if charsets are specified (usually vcard version < 3.0 but this is not reliable)
T	198	if (preg_match('/charset=/i', substr($data, 0, 2048)))
	199	$charset = null;
ed132e	200	// detect charset and convert to utf-8
5570ad	201	else if (($charset = self::detect_encoding($data)) && $charset != RCMAIL_CHARSET) {
T	202	$data = rcube_charset_convert($data, $charset);
6fa87f	203	$data = preg_replace(array('/^[\xFE\xFF]{2}/', '/^\xEF\xBB\xBF/', '/^\x00+/'), '', $data); // also remove BOM
5570ad	204	$charset = RCMAIL_CHARSET;
ed132e	205	}
T	206
	207	$vcard_block = '';
	208	$in_vcard_block = false;
	209
	210	foreach (preg_split("/[\r\n]+/", $data) as $i => $line) {
	211	if ($in_vcard_block && !empty($line))
	212	$vcard_block .= $line . "\n";
	213
	214	if (trim($line) == 'END:VCARD') {
	215	// parse vcard
5570ad	216	$obj = new rcube_vcard(self::cleanup($vcard_block), $charset);
ed132e	217	if (!empty($obj->displayname))
T	218	$out[] = $obj;
	219
	220	$in_vcard_block = false;
	221	}
	222	else if (trim($line) == 'BEGIN:VCARD') {
	223	$vcard_block = $line . "\n";
	224	$in_vcard_block = true;
	225	}
	226	}
	227
	228	return $out;
	229	}
	230
	231
	232	/**
	233	* Normalize vcard data for better parsing
	234	*
	235	* @param string vCard block
	236	* @return string Cleaned vcard block
	237	*/
	238	private static function cleanup($vcard)
	239	{
	240	// Convert special types (like Skype) to normal type='skype' classes with this simple regex ;)
	241	$vcard = preg_replace(
	242	'/item(\d+)\.(TEL\|URL)([^:]?):(.?)item\1.X-ABLabel:(?:_\$!<)?([\w-() ]*)(?:>!\$_)?./s',
	243	'\2;type=\5\3:\4',
	244	$vcard);
	245
	246	// Remove cruft like item1.X-AB*, item1.ADR instead of ADR, and empty lines
	247	$vcard = preg_replace(array('/^item\d\.X-AB.$/m', '/^item\d*\./m', "/\n+/"), array('', '', "\n"), $vcard);
	248
b58f11	249	// if N doesn't have any semicolons, add some
T	250	$vcard = preg_replace('/^(N:[^;\R]*)$/m', '\1;;;;', $vcard);
ed132e	251
T	252	return $vcard;
	253	}
	254
478c7c	255	private static function rfc2425_fold_callback($matches)
A	256	{
	257	return ":\n ".rtrim(chunk_split($matches[1], 72, "\n "));
	258	}
ed132e	259
T	260	private static function rfc2425_fold($val)
	261	{
5e6815	262	return preg_replace_callback('/:([^\n]{72,})/', array('self', 'rfc2425_fold_callback'), $val) . "\n";
ed132e	263	}
T	264
	265
	266	/**
	267	* Decodes a vcard block (vcard 3.0 format, unfolded)
	268	* into an array structure
	269	*
	270	* @param string vCard block to parse
	271	* @return array Raw data structure
	272	*/
	273	private static function vcard_decode($vcard)
	274	{
	275	// Perform RFC2425 line unfolding
	276	$vcard = preg_replace(array("/\r/", "/\n\s+/"), '', $vcard);
	277
b58f11	278	$lines = preg_split('/\r?\n/', $vcard);
ed132e	279	$data = array();
b58f11	280
T	281	for ($i=0; $i < count($lines); $i++) {
	282	if (!preg_match('/^([^\\:]*):(.+)$/', $lines[$i], $line))
	283	continue;
ed132e	284
b58f11	285	// convert 2.1-style "EMAIL;internet;home:" to 3.0-style "EMAIL;TYPE=internet;TYPE=home:"
T	286	if (($data['VERSION'][0] == "2.1") && preg_match('/^([^;]+);([^:]+)/', $line[1], $regs2) && !preg_match('/^TYPE=/i', $regs2[2])) {
	287	$line[1] = $regs2[1];
	288	foreach (explode(';', $regs2[2]) as $prop)
	289	$line[1] .= ';' . (strpos($prop, '=') ? $prop : 'TYPE='.$prop);
ed132e	290	}
T	291
cc97ea	292	if (!preg_match('/^(BEGIN\|END)$/i', $line[1]) && preg_match_all('/([^\\;]+);?/', $line[1], $regs2)) {
93af15	293	$entry = array();
cc97ea	294	$field = strtoupper($regs2[1][0]);
b58f11	295
T	296	foreach($regs2[1] as $attrid => $attr) {
	297	if ((list($key, $value) = explode('=', $attr)) && $value) {
5570ad	298	$value = trim($value);
b58f11	299	if ($key == 'ENCODING') {
93af15	300	// add next line(s) to value string if QP line end detected
23a2ee	301	while ($value == 'QUOTED-PRINTABLE' && preg_match('/=$/', $lines[$i]))
b58f11	302	$line[2] .= "\n" . $lines[++$i];
T	303
	304	$line[2] = self::decode_value($line[2], $value);
	305	}
	306	else
	307	$entry[strtolower($key)] = array_merge((array)$entry[strtolower($key)], (array)self::vcard_unquote($value, ','));
	308	}
	309	else if ($attrid > 0) {
93af15	310	$entry[$key] = true; // true means attr without =value
b58f11	311	}
T	312	}
	313
93af15	314	$entry = array_merge($entry, (array)self::vcard_unquote($line[2]));
5570ad	315	$data[$field][] = $entry;
b58f11	316	}
ed132e	317	}
b58f11	318
T	319	unset($data['VERSION']);
ed132e	320	return $data;
T	321	}
	322
	323
	324	/**
	325	* Split quoted string
	326	*
	327	* @param string vCard string to split
	328	* @param string Separator char/string
	329	* @return array List with splitted values
	330	*/
	331	private static function vcard_unquote($s, $sep = ';')
	332	{
	333	// break string into parts separated by $sep, but leave escaped $sep alone
	334	if (count($parts = explode($sep, strtr($s, array("\\$sep" => "\007")))) > 1) {
	335	foreach($parts as $s) {
	336	$result[] = self::vcard_unquote(strtr($s, array("\007" => "\\$sep")), $sep);
	337	}
	338	return $result;
	339	}
	340	else {
	341	return strtr($s, array("\r" => '', '\\\\' => '\\', '\n' => "\n", '\,' => ',', '\;' => ';', '\:' => ':'));
bb8781	342	}
T	343	}
	344
	345
	346	/**
	347	* Decode a given string with the encoding rule from ENCODING attributes
	348	*
	349	* @param string String to decode
	350	* @param string Encoding type (quoted-printable and base64 supported)
	351	* @return string Decoded 8bit value
	352	*/
	353	private static function decode_value($value, $encoding)
	354	{
	355	switch (strtolower($encoding)) {
	356	case 'quoted-printable':
	357	return quoted_printable_decode($value);
	358
	359	case 'base64':
	360	return base64_decode($value);
	361
	362	default:
	363	return $value;
ed132e	364	}
T	365	}
	366
	367
	368	/**
	369	* Encodes an entry for storage in our database (vcard 3.0 format, unfolded)
	370	*
	371	* @param array Raw data structure to encode
	372	* @return string vCard encoded string
	373	*/
	374	static function vcard_encode($data)
	375	{
	376	foreach((array)$data as $type => $entries) {
	377	/* valid N has 5 properties */
b58f11	378	while ($type == "N" && is_array($entries[0]) && count($entries[0]) < 5)
ed132e	379	$entries[0][] = "";
T	380
	381	foreach((array)$entries as $entry) {
	382	$attr = '';
	383	if (is_array($entry)) {
	384	$value = array();
	385	foreach($entry as $attrname => $attrvalues) {
	386	if (is_int($attrname))
	387	$value[] = $attrvalues;
	388	elseif ($attrvalues === true)
93af15	389	$attr .= ";$attrname"; // true means just tag, not tag=value, as in PHOTO;BASE64:...
ed132e	390	else {
T	391	foreach((array)$attrvalues as $attrvalue)
	392	$attr .= ";$attrname=" . self::vcard_quote($attrvalue, ',');
	393	}
	394	}
	395	}
	396	else {
	397	$value = $entry;
	398	}
	399
	400	$vcard .= self::vcard_quote($type) . $attr . ':' . self::vcard_quote($value) . "\n";
	401	}
	402	}
	403
0dbac3	404	return "BEGIN:VCARD\nVERSION:3.0\n{$vcard}END:VCARD";
ed132e	405	}
T	406
	407
	408	/**
	409	* Join indexed data array to a vcard quoted string
	410	*
	411	* @param array Field data
	412	* @param string Separator
	413	* @return string Joined and quoted string
	414	*/
	415	private static function vcard_quote($s, $sep = ';')
	416	{
	417	if (is_array($s)) {
	418	foreach($s as $part) {
	419	$r[] = self::vcard_quote($part, $sep);
	420	}
	421	return(implode($sep, (array)$r));
	422	}
	423	else {
	424	return strtr($s, array('\\' => '\\\\', "\r" => '', "\n" => '\n', ';' => '\;', ':' => '\:'));
	425	}
	426	}
	427
	428
	429	/**
	430	* Returns UNICODE type based on BOM (Byte Order Mark)
	431	*
	432	* @param string Input string to test
	433	* @return string Detected encoding
	434	*/
	435	private static function detect_encoding($string)
	436	{
	437	if (substr($string, 0, 4) == "\0\0\xFE\xFF") return 'UTF-32BE'; // Big Endian
	438	if (substr($string, 0, 4) == "\xFF\xFE\0\0") return 'UTF-32LE'; // Little Endian
	439	if (substr($string, 0, 2) == "\xFE\xFF") return 'UTF-16BE'; // Big Endian
	440	if (substr($string, 0, 2) == "\xFF\xFE") return 'UTF-16LE'; // Little Endian
	441	if (substr($string, 0, 3) == "\xEF\xBB\xBF") return 'UTF-8';
	442
6fa87f	443	// use mb_detect_encoding()
T	444	$encodings = array('UTF-8', 'ISO-8859-1', 'ISO-8859-2', 'ISO-8859-3',
	445	'ISO-8859-4', 'ISO-8859-5', 'ISO-8859-6', 'ISO-8859-7', 'ISO-8859-8', 'ISO-8859-9',
	446	'ISO-8859-10', 'ISO-8859-13', 'ISO-8859-14', 'ISO-8859-15', 'ISO-8859-16',
	447	'WINDOWS-1252', 'WINDOWS-1251', 'BIG5', 'GB2312');
	448
ae9124	449	if (function_exists('mb_detect_encoding') && ($enc = mb_detect_encoding($string, $encodings)))
abdc58	450	return $enc;
A	451
ed132e	452	// No match, check for UTF-8
T	453	// from http://w3.org/International/questions/qa-forms-utf-8.html
	454	if (preg_match('/\A(
	455	[\x09\x0A\x0D\x20-\x7E]
	456	\| [\xC2-\xDF][\x80-\xBF]
	457	\| \xE0[\xA0-\xBF][\x80-\xBF]
	458	\| [\xE1-\xEC\xEE\xEF][\x80-\xBF]{2}
	459	\| \xED[\x80-\x9F][\x80-\xBF]
	460	\| \xF0[\x90-\xBF][\x80-\xBF]{2}
	461	\| [\xF1-\xF3][\x80-\xBF]{3}
	462	\| \xF4[\x80-\x8F][\x80-\xBF]{2}
	463	)*\z/xs', substr($string, 0, 2048)))
	464	return 'UTF-8';
	465
ae9124	466	return rcmail::get_instance()->config->get('default_charset', 'ISO-8859-1'); # fallback to Latin-1
ed132e	467	}
T	468
	469	}
	470
	471