githubFork/roundcubemail.git

Fix rcube_utils::normalize_string() to support unicode characters + add arg...

Thomas Bruederli

2014-12-28 e8b82c2e7b0ae2e5d45ecb600813b8990568feb9

commit \| author \| age
8fa58e	1	<?php
T	2
	3	/*
	4	+-----------------------------------------------------------------------+
e019f2	5	\| This file is part of the Roundcube Webmail client \|
48ba44	6	\| Copyright (C) 2008-2014, 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. \|
8fa58e	11	\| \|
T	12	\| PURPOSE: \|
	13	\| Logical representation of a mail message with all its data \|
	14	\| and related functions \|
	15	+-----------------------------------------------------------------------+
	16	\| Author: Thomas Bruederli <roundcube@gmail.com> \|
	17	+-----------------------------------------------------------------------+
	18	*/
	19
	20	/**
45f56c	21	* Logical representation of a mail message with all its data
T	22	* and related functions
8fa58e	23	*
9ab346	24	* @package Framework
AM	25	* @subpackage Storage
8fa58e	26	* @author Thomas Bruederli <roundcube@gmail.com>
T	27	*/
	28	class rcube_message
	29	{
5c461b	30	/**
be98df	31	* Instace of framework class.
5c461b	32	*
be98df	33	* @var rcube
5c461b	34	*/
d311d8	35	private $app;
5c461b	36
A	37	/**
8b92d2	38	* Instance of storage class
5c461b	39	*
8b92d2	40	* @var rcube_storage
5c461b	41	*/
8b92d2	42	private $storage;
1c4f23	43
A	44	/**
	45	* Instance of mime class
	46	*
	47	* @var rcube_mime
	48	*/
	49	private $mime;
d311d8	50	private $opt = array();
A	51	private $parse_alternative = false;
2ae58f	52
10562d	53	public $uid;
AM	54	public $folder;
d311d8	55	public $headers;
A	56	public $parts = array();
	57	public $mime_parts = array();
287eff	58	public $inline_parts = array();
d311d8	59	public $attachments = array();
A	60	public $subject = '';
	61	public $sender = null;
	62	public $is_safe = false;
48ba44	63
AM	64	const BODY_MAX_SIZE = 1048576; // 1MB
2ae58f	65
193fb4	66
d311d8	67	/**
A	68	* __construct
	69	*
	70	* Provide a uid, and parse message structure.
	71	*
10562d	72	* @param string $uid The message UID.
AM	73	* @param string $folder Folder name
d311d8	74	*
8b92d2	75	* @see self::$app, self::$storage, self::$opt, self::$parts
d311d8	76	*/
10562d	77	function __construct($uid, $folder = null)
8fa58e	78	{
e8cb51	79	// decode combined UID-folder identifier
188247	80	if (preg_match('/^\d+-.+/', $uid)) {
1d1fdc	81	list($uid, $folder) = explode('-', $uid, 2);
e8cb51	82	}
TB	83
1c4f23	84	$this->uid = $uid;
0c2596	85	$this->app = rcube::get_instance();
c321a9	86	$this->storage = $this->app->get_storage();
10562d	87	$this->folder = strlen($folder) ? $folder : $this->storage->get_folder();
c321a9	88	$this->storage->set_options(array('all_headers' => true));
10562d	89
AM	90	// Set current folder
	91	$this->storage->set_folder($this->folder);
64e3e8	92
c321a9	93	$this->headers = $this->storage->get_message($uid);
64e3e8	94
4fdaa0	95	if (!$this->headers) {
64e3e8	96	return;
4fdaa0	97	}
8fa58e	98
1c4f23	99	$this->mime = new rcube_mime($this->headers->charset);
A	100
4fdaa0	101	$this->subject = $this->headers->get('subject');
1c4f23	102	list(, $this->sender) = each($this->mime->decode_address_list($this->headers->from, 1));
64e3e8	103
f11142	104	$this->set_safe((intval($_GET['_safe']) \|\| $_SESSION['safe_messages'][$this->folder.':'.$uid]));
d311d8	105	$this->opt = array(
A	106	'safe' => $this->is_safe,
	107	'prefer_html' => $this->app->config->get('prefer_html'),
0c2596	108	'get_url' => $this->app->url(array(
A	109	'action' => 'get',
	110	'mbox' => $this->storage->get_folder(),
	111	'uid' => $uid))
d311d8	112	);
8fa58e	113
80152b	114	if (!empty($this->headers->structure)) {
A	115	$this->get_mime_numbers($this->headers->structure);
	116	$this->parse_structure($this->headers->structure);
aa16b4	117	}
d311d8	118	else {
c321a9	119	$this->body = $this->storage->get_body($uid);
d311d8	120	}
A	121
	122	// notify plugins and let them analyze this structured message object
	123	$this->app->plugins->exec_hook('message_load', array('object' => $this));
	124	}
2ae58f	125
A	126
d311d8	127	/**
A	128	* Return a (decoded) message header
	129	*
5c461b	130	* @param string $name Header name
A	131	* @param bool $row Don't mime-decode the value
d311d8	132	* @return string Header value
A	133	*/
	134	public function get_header($name, $raw = false)
	135	{
4fdaa0	136	if (empty($this->headers)) {
1c4f23	137	return null;
4fdaa0	138	}
1c4f23	139
4fdaa0	140	return $this->headers->get($name, !$raw);
d311d8	141	}
A	142
2ae58f	143
d311d8	144	/**
A	145	* Set is_safe var and session data
	146	*
5c461b	147	* @param bool $safe enable/disable
d311d8	148	*/
A	149	public function set_safe($safe = true)
	150	{
f11142	151	$_SESSION['safe_messages'][$this->folder.':'.$this->uid] = $this->is_safe = $safe;
d311d8	152	}
A	153
	154
	155	/**
	156	* Compose a valid URL for getting a message part
	157	*
5c461b	158	* @param string $mime_id Part MIME-ID
a021d6	159	* @param mixed $embed Mimetype class for parts to be embedded
d311d8	160	* @return string URL or false if part does not exist
A	161	*/
57486f	162	public function get_part_url($mime_id, $embed = false)
d311d8	163	{
A	164	if ($this->mime_parts[$mime_id])
a021d6	165	return $this->opt['get_url'] . '&_part=' . $mime_id . ($embed ? '&_embed=1&_mimeclass=' . $embed : '');
aa16b4	166	else
d311d8	167	return false;
8fa58e	168	}
T	169
d311d8	170
A	171	/**
	172	* Get content of a specific part of this message
	173	*
71950d	174	* @param string $mime_id Part MIME-ID
A	175	* @param resource $fp File pointer to save the message part
	176	* @param boolean $skip_charset_conv Disables charset conversion
dff2c7	177	* @param int $max_bytes Only read this number of bytes
ae8533	178	* @param boolean $formatted Enables formatting of text/* parts bodies
71950d	179	*
d311d8	180	* @return string Part content
48ba44	181	* @deprecated
d311d8	182	*/
ae8533	183	public function get_part_content($mime_id, $fp = null, $skip_charset_conv = false, $max_bytes = 0, $formatted = true)
d311d8	184	{
A	185	if ($part = $this->mime_parts[$mime_id]) {
	186	// stored in message structure (winmail/inline-uuencode)
8b92d2	187	if (!empty($part->body) \|\| $part->encoding == 'stream') {
d311d8	188	if ($fp) {
A	189	fwrite($fp, $part->body);
	190	}
	191	return $fp ? true : $part->body;
	192	}
10562d	193
d311d8	194	// get from IMAP
10562d	195	$this->storage->set_folder($this->folder);
AM	196
ae8533	197	return $this->storage->get_message_part($this->uid, $mime_id, $part,
AM	198	NULL, $fp, $skip_charset_conv, $max_bytes, $formatted);
10562d	199	}
48ba44	200	}
AM	201
	202
	203	/**
	204	* Get content of a specific part of this message
	205	*
	206	* @param string $mime_id Part ID
	207	* @param boolean $formatted Enables formatting of text/* parts bodies
	208	* @param int $max_bytes Only return/read this number of bytes
	209	* @param mixed $mode NULL to return a string, -1 to print body
	210	* or file pointer to save the body into
	211	*
	212	* @return string\|bool Part content or operation status
	213	*/
	214	public function get_part_body($mime_id, $formatted = false, $max_bytes = 0, $mode = null)
	215	{
	216	if (!($part = $this->mime_parts[$mime_id])) {
	217	return;
	218	}
	219
	220	// only text parts can be formatted
	221	$formatted = $formatted && $part->ctype_primary == 'text';
	222
	223	// part body not fetched yet... save in memory if it's small enough
	224	if ($part->body === null && is_numeric($mime_id) && $part->size < self::BODY_MAX_SIZE) {
d93019	225	$this->storage->set_folder($this->folder);
48ba44	226	// Warning: body here should be always unformatted
AM	227	$part->body = $this->storage->get_message_part($this->uid, $mime_id, $part,
	228	null, null, true, 0, false);
	229	}
	230
	231	// body stored in message structure (winmail/inline-uuencode)
	232	if ($part->body !== null \|\| $part->encoding == 'stream') {
	233	$body = $part->body;
	234
	235	if ($formatted && $body) {
	236	$body = self::format_part_body($body, $part, $this->headers->charset);
	237	}
	238
	239	if ($max_bytes && strlen($body) > $max_bytes) {
	240	$body = substr($body, 0, $max_bytes);
	241	}
	242
	243	if (is_resource($mode)) {
	244	if ($body !== false) {
	245	fwrite($mode, $body);
	246	rewind($mode);
	247	}
	248
	249	return $body !== false;
	250	}
	251
	252	if ($mode === -1) {
	253	if ($body !== false) {
	254	print($body);
	255	}
	256
	257	return $body !== false;
	258	}
	259
	260	return $body;
	261	}
	262
	263	// get the body from IMAP
	264	$this->storage->set_folder($this->folder);
	265
	266	$body = $this->storage->get_message_part($this->uid, $mime_id, $part,
68c41f	267	$mode === -1, is_resource($mode) ? $mode : null,
AM	268	!($mode && $formatted), $max_bytes, $mode && $formatted);
48ba44	269
AM	270	if (is_resource($mode)) {
	271	rewind($mode);
	272	return $body !== false;
	273	}
	274
68c41f	275	if (!$mode && $body && $formatted) {
AM	276	$body = self::format_part_body($body, $part, $this->headers->charset);
	277	}
	278
48ba44	279	return $body;
AM	280	}
	281
	282
	283	/**
	284	* Format text message part for display
	285	*
	286	* @param string $body Part body
	287	* @param rcube_message_part $part Part object
	288	* @param string $default_charset Fallback charset if part charset is not specified
	289	*
	290	* @return string Formatted body
	291	*/
	292	public static function format_part_body($body, $part, $default_charset = null)
	293	{
	294	// remove useless characters
	295	$body = preg_replace('/[\t\r\0\x0B]+\n/', "\n", $body);
	296
	297	// remove NULL characters if any (#1486189)
	298	if (strpos($body, "\x00") !== false) {
	299	$body = str_replace("\x00", '', $body);
	300	}
	301
	302	// detect charset...
	303	if (!$part->charset \|\| strtoupper($part->charset) == 'US-ASCII') {
	304	// try to extract charset information from HTML meta tag (#1488125)
	305	if ($part->ctype_secondary == 'html' && preg_match('/<meta[^>]+charset=([a-z0-9-_]+)/i', $body, $m)) {
	306	$part->charset = strtoupper($m[1]);
	307	}
	308	else if ($default_charset) {
	309	$part->charset = $default_charset;
	310	}
	311	else {
	312	$rcube = rcube::get_instance();
	313	$part->charset = $rcube->config->get('default_charset', RCUBE_CHARSET);
	314	}
	315	}
	316
	317	// ..convert charset encoding
	318	$body = rcube_charset::convert($body, $part->charset);
	319
	320	return $body;
8fa58e	321	}
T	322
	323
d311d8	324	/**
5c26bd	325	* Determine if the message contains a HTML part. This must to be
AM	326	* a real part not an attachment (or its part)
d311d8	327	*
5c26bd	328	* @param bool $enriched Enables checking for text/enriched parts too
33423a	329	*
d311d8	330	* @return bool True if a HTML is available, False if not
A	331	*/
5c26bd	332	function has_html_part($enriched = false)
d311d8	333	{
A	334	// check all message parts
574928	335	foreach ($this->mime_parts as $part) {
52d0d9	336	if ($part->mimetype == 'text/html' \|\| ($enriched && $part->mimetype == 'text/enriched')) {
0ef894	337	// Skip if part is an attachment, don't use is_attachment() here
AM	338	if ($part->filename) {
5c26bd	339	continue;
AM	340	}
33423a	341
5c26bd	342	$level = explode('.', $part->mime_id);
f8101f	343	$depth = count($level);
5c26bd	344
5a2d2a	345	// Check if the part belongs to higher-level's multipart part
f8101f	346	// this can be alternative/related/signed/encrypted or mixed
5c26bd	347	while (array_pop($level) !== null) {
f8101f	348	$parent_depth = count($level);
TB	349	if (!$parent_depth) {
5c26bd	350	return true;
33423a	351	}
A	352
5c26bd	353	$parent = $this->mime_parts[join('.', $level)];
f8101f	354	if (!preg_match('/^multipart\/(alternative\|related\|signed\|encrypted\|mixed)$/', $parent->mimetype)
TB	355	\|\| ($parent->mimetype == 'multipart/mixed' && $parent_depth < $depth - 1)) {
5c26bd	356	continue 2;
33423a	357	}
A	358	}
	359
5c26bd	360	if ($part->size) {
AM	361	return true;
	362	}
	363	}
	364	}
	365
	366	return false;
	367	}
	368
	369
	370	/**
	371	* Determine if the message contains a text/plain part. This must to be
	372	* a real part not an attachment (or its part)
	373	*
	374	* @return bool True if a plain text part is available, False if not
	375	*/
	376	function has_text_part()
	377	{
	378	// check all message parts
574928	379	foreach ($this->mime_parts as $part) {
5c26bd	380	if ($part->mimetype == 'text/plain') {
0ef894	381	// Skip if part is an attachment, don't use is_attachment() here
AM	382	if ($part->filename) {
5c26bd	383	continue;
AM	384	}
	385
	386	$level = explode('.', $part->mime_id);
	387
	388	// Check if the part belongs to higher-level's alternative/related
	389	while (array_pop($level) !== null) {
	390	if (!count($level)) {
	391	return true;
	392	}
	393
	394	$parent = $this->mime_parts[join('.', $level)];
	395	if ($parent->mimetype != 'multipart/alternative' && $parent->mimetype != 'multipart/related') {
	396	continue 2;
	397	}
	398	}
	399
	400	if ($part->size) {
	401	return true;
	402	}
33423a	403	}
d311d8	404	}
A	405
	406	return false;
	407	}
	408
	409
	410	/**
	411	* Return the first HTML part of this message
	412	*
	413	* @return string HTML message part content
	414	*/
	415	function first_html_part()
	416	{
	417	// check all message parts
33423a	418	foreach ($this->mime_parts as $pid => $part) {
A	419	if ($part->mimetype == 'text/html') {
48ba44	420	return $this->get_part_body($pid, true);
d311d8	421	}
A	422	}
	423	}
	424
	425
	426	/**
	427	* Return the first text part of this message
	428	*
868deb	429	* @param rcube_message_part $part Reference to the part if found
d311d8	430	* @return string Plain text message/part content
A	431	*/
868deb	432	function first_text_part(&$part=null)
d311d8	433	{
A	434	// no message structure, return complete body
	435	if (empty($this->parts))
	436	return $this->body;
	437
	438	// check all message parts
	439	foreach ($this->mime_parts as $mime_id => $part) {
33423a	440	if ($part->mimetype == 'text/plain') {
48ba44	441	return $this->get_part_body($mime_id, true);
d311d8	442	}
33423a	443	else if ($part->mimetype == 'text/html') {
48ba44	444	$out = $this->get_part_body($mime_id, true);
d311d8	445
A	446	// create instance of html2text class
66afd7	447	$txt = new rcube_html2text($out);
868deb	448	return $txt->get_text();
d311d8	449	}
A	450	}
	451
868deb	452	$part = null;
A	453	return null;
d311d8	454	}
A	455
	456
	457	/**
3efc74	458	* Checks if part of the message is an attachment (or part of it)
AM	459	*
	460	* @param rcube_message_part $part Message part
	461	*
	462	* @return bool True if the part is an attachment part
	463	*/
	464	public function is_attachment($part)
	465	{
	466	foreach ($this->attachments as $att_part) {
	467	if ($att_part->mime_id == $part->mime_id) {
	468	return true;
	469	}
	470
	471	// check if the part is a subpart of another attachment part (message/rfc822)
	472	if ($att_part->mimetype == 'message/rfc822') {
	473	if (in_array($part, (array)$att_part->parts)) {
	474	return true;
	475	}
	476	}
	477	}
	478
	479	return false;
	480	}
	481
	482
	483	/**
8b92d2	484	* Read the message structure returend by the IMAP server
d311d8	485	* and build flat lists of content parts and attachments
A	486	*
5c461b	487	* @param rcube_message_part $structure Message structure node
A	488	* @param bool $recursive True when called recursively
d311d8	489	*/
A	490	private function parse_structure($structure, $recursive = false)
	491	{
	492	// real content-type of message/rfc822 part
a8a72e	493	if ($structure->mimetype == 'message/rfc822' && $structure->real_mimetype) {
5ced9c	494	$mimetype = $structure->real_mimetype;
a8a72e	495
TB	496	// parse headers from message/rfc822 part
7ae7cd	497	if (!isset($structure->headers['subject']) && !isset($structure->headers['from'])) {
48ba44	498	list($headers, ) = explode("\r\n\r\n", $this->get_part_body($structure->mime_id, false, 32768));
a8a72e	499	$structure->headers = rcube_mime::parse_headers($headers);
TB	500	}
	501	}
5ced9c	502	else
A	503	$mimetype = $structure->mimetype;
d311d8	504
A	505	// show message headers
ddfdd8	506	if ($recursive && is_array($structure->headers) &&
TB	507	(isset($structure->headers['subject']) \|\| $structure->headers['from'] \|\| $structure->headers['to'])) {
d311d8	508	$c = new stdClass;
A	509	$c->type = 'headers';
c5d7c9	510	$c->headers = $structure->headers;
d311d8	511	$this->parts[] = $c;
A	512	}
5ced9c	513
A	514	// Allow plugins to handle message parts
	515	$plugin = $this->app->plugins->exec_hook('message_part_structure',
	516	array('object' => $this, 'structure' => $structure,
	517	'mimetype' => $mimetype, 'recursive' => $recursive));
	518
	519	if ($plugin['abort'])
	520	return;
	521
	522	$structure = $plugin['structure'];
	523	list($message_ctype_primary, $message_ctype_secondary) = explode('/', $plugin['mimetype']);
d311d8	524
A	525	// print body if message doesn't have multiple parts
	526	if ($message_ctype_primary == 'text' && !$recursive) {
c23dc8	527	// parts with unsupported type add to attachments list
AM	528	if (!in_array($message_ctype_secondary, array('plain', 'html', 'enriched'))) {
	529	$this->attachments[] = $structure;
	530	return;
	531	}
	532
d311d8	533	$structure->type = 'content';
c5d7c9	534	$this->parts[] = $structure;
8757f5	535
d311d8	536	// Parse simple (plain text) message body
c23dc8	537	if ($message_ctype_secondary == 'plain') {
d311d8	538	foreach ((array)$this->uu_decode($structure) as $uupart) {
A	539	$this->mime_parts[$uupart->mime_id] = $uupart;
	540	$this->attachments[] = $uupart;
	541	}
c23dc8	542	}
d311d8	543	}
A	544	// the same for pgp signed messages
	545	else if ($mimetype == 'application/pgp' && !$recursive) {
	546	$structure->type = 'content';
c5d7c9	547	$this->parts[] = $structure;
d311d8	548	}
e730cd	549	// message contains (more than one!) alternative parts
A	550	else if ($mimetype == 'multipart/alternative'
	551	&& is_array($structure->parts) && count($structure->parts) > 1
	552	) {
c5d7c9	553	// get html/plaintext parts, other add to attachments list
d311d8	554	foreach ($structure->parts as $p => $sub_part) {
A	555	$sub_mimetype = $sub_part->mimetype;
cb0f03	556	$is_multipart = preg_match('/^multipart\/(related\|relative\|mixed\|alternative)/', $sub_mimetype);
8757f5	557
5fbfde	558	// skip empty text parts
c5d7c9	559	if (!$sub_part->size && !$is_multipart) {
5f4095	560	continue;
5fbfde	561	}
5f4095	562
170702	563	// We've encountered (malformed) messages with more than
AM	564	// one text/plain or text/html part here. There's no way to choose
	565	// which one is better, so we'll display first of them and add
	566	// others as attachments (#1489358)
	567
d311d8	568	// check if sub part is
c5d7c9	569	if ($is_multipart)
AM	570	$related_part = $p;
170702	571	else if ($sub_mimetype == 'text/plain' && !$plain_part)
d311d8	572	$plain_part = $p;
170702	573	else if ($sub_mimetype == 'text/html' && !$html_part)
d311d8	574	$html_part = $p;
170702	575	else if ($sub_mimetype == 'text/enriched' && !$enriched_part)
d311d8	576	$enriched_part = $p;
170702	577	else {
AM	578	// add unsupported/unrecognized parts to attachments list
	579	$this->attachments[] = $sub_part;
	580	}
d311d8	581	}
A	582
	583	// parse related part (alternative part could be in here)
	584	if ($related_part !== null && !$this->parse_alternative) {
	585	$this->parse_alternative = true;
	586	$this->parse_structure($structure->parts[$related_part], true);
	587	$this->parse_alternative = false;
8757f5	588
d311d8	589	// if plain part was found, we should unset it if html is preferred
A	590	if ($this->opt['prefer_html'] && count($this->parts))
	591	$plain_part = null;
	592	}
	593
	594	// choose html/plain part to print
	595	if ($html_part !== null && $this->opt['prefer_html']) {
c5d7c9	596	$print_part = $structure->parts[$html_part];
d311d8	597	}
A	598	else if ($enriched_part !== null) {
c5d7c9	599	$print_part = $structure->parts[$enriched_part];
d311d8	600	}
A	601	else if ($plain_part !== null) {
c5d7c9	602	$print_part = $structure->parts[$plain_part];
d311d8	603	}
A	604
	605	// add the right message body
	606	if (is_object($print_part)) {
	607	$print_part->type = 'content';
	608	$this->parts[] = $print_part;
	609	}
	610	// show plaintext warning
	611	else if ($html_part !== null && empty($this->parts)) {
	612	$c = new stdClass;
	613	$c->type = 'content';
	614	$c->ctype_primary = 'text';
	615	$c->ctype_secondary = 'plain';
0c2596	616	$c->mimetype = 'text/plain';
A	617	$c->realtype = 'text/html';
d311d8	618
A	619	$this->parts[] = $c;
	620	}
	621	}
	622	// this is an ecrypted message -> create a plaintext body with the according message
	623	else if ($mimetype == 'multipart/encrypted') {
	624	$p = new stdClass;
	625	$p->type = 'content';
	626	$p->ctype_primary = 'text';
	627	$p->ctype_secondary = 'plain';
0c2596	628	$p->mimetype = 'text/plain';
A	629	$p->realtype = 'multipart/encrypted';
c054ec	630
A	631	$this->parts[] = $p;
d311d8	632	}
ee89c6	633	// this is an S/MIME ecrypted message -> create a plaintext body with the according message
AM	634	else if ($mimetype == 'application/pkcs7-mime') {
	635	$p = new stdClass;
	636	$p->type = 'content';
	637	$p->ctype_primary = 'text';
	638	$p->ctype_secondary = 'plain';
	639	$p->mimetype = 'text/plain';
	640	$p->realtype = 'application/pkcs7-mime';
	641
	642	$this->parts[] = $p;
	643	}
d311d8	644	// message contains multiple parts
A	645	else if (is_array($structure->parts) && !empty($structure->parts)) {
	646	// iterate over parts
	647	for ($i=0; $i < count($structure->parts); $i++) {
	648	$mail_part = &$structure->parts[$i];
	649	$primary_type = $mail_part->ctype_primary;
	650	$secondary_type = $mail_part->ctype_secondary;
8fa58e	651
d311d8	652	// real content-type of message/rfc822
A	653	if ($mail_part->real_mimetype) {
	654	$part_orig_mimetype = $mail_part->mimetype;
	655	$part_mimetype = $mail_part->real_mimetype;
	656	list($primary_type, $secondary_type) = explode('/', $part_mimetype);
	657	}
596301	658	else {
TB	659	$part_mimetype = $part_orig_mimetype = $mail_part->mimetype;
	660	}
6b6f2e	661
d311d8	662	// multipart/alternative
A	663	if ($primary_type == 'multipart') {
	664	$this->parse_structure($mail_part, true);
	665
	666	// list message/rfc822 as attachment as well (mostly .eml)
	667	if ($part_orig_mimetype == 'message/rfc822' && !empty($mail_part->filename))
	668	$this->attachments[] = $mail_part;
	669	}
f22ea7	670	// part text/[plain\|html] or delivery status
d311d8	671	else if ((($part_mimetype == 'text/plain' \|\| $part_mimetype == 'text/html') && $mail_part->disposition != 'attachment') \|\|
f22ea7	672	in_array($part_mimetype, array('message/delivery-status', 'text/rfc822-headers', 'message/disposition-notification'))
d311d8	673	) {
1a2f83	674	// Allow plugins to handle also this part
A	675	$plugin = $this->app->plugins->exec_hook('message_part_structure',
	676	array('object' => $this, 'structure' => $mail_part,
	677	'mimetype' => $part_mimetype, 'recursive' => true));
	678
	679	if ($plugin['abort'])
	680	continue;
	681
9c299e	682	if ($part_mimetype == 'text/html' && $mail_part->size) {
63f9de	683	$got_html_part = true;
A	684	}
	685
1a2f83	686	$mail_part = $plugin['structure'];
A	687	list($primary_type, $secondary_type) = explode('/', $plugin['mimetype']);
	688
d311d8	689	// add text part if it matches the prefs
A	690	if (!$this->parse_alternative \|\|
	691	($secondary_type == 'html' && $this->opt['prefer_html']) \|\|
	692	($secondary_type == 'plain' && !$this->opt['prefer_html'])
	693	) {
	694	$mail_part->type = 'content';
	695	$this->parts[] = $mail_part;
	696	}
fd371a	697
d311d8	698	// list as attachment as well
f7c11e	699	if (!empty($mail_part->filename)) {
AM	700	$this->attachments[] = $mail_part;
	701	}
d311d8	702	}
A	703	// part message/*
8757f5	704	else if ($primary_type == 'message') {
d311d8	705	$this->parse_structure($mail_part, true);
A	706
	707	// list as attachment as well (mostly .eml)
	708	if (!empty($mail_part->filename))
	709	$this->attachments[] = $mail_part;
	710	}
	711	// ignore "virtual" protocol parts
	712	else if ($primary_type == 'protocol') {
	713	continue;
	714	}
	715	// part is Microsoft Outlook TNEF (winmail.dat)
	716	else if ($part_mimetype == 'application/ms-tnef') {
	717	foreach ((array)$this->tnef_decode($mail_part) as $tpart) {
	718	$this->mime_parts[$tpart->mime_id] = $tpart;
	719	$this->attachments[] = $tpart;
	720	}
	721	}
	722	// part is a file/attachment
	723	else if (preg_match('/^(inline\|attach)/', $mail_part->disposition) \|\|
8794f1	724	$mail_part->headers['content-id'] \|\|
A	725	($mail_part->filename &&
	726	(empty($mail_part->disposition) \|\| preg_match('/^[a-z0-9!#$&.+^_-]+$/i', $mail_part->disposition)))
d311d8	727	) {
A	728	// skip apple resource forks
	729	if ($message_ctype_secondary == 'appledouble' && $secondary_type == 'applefile')
	730	continue;
	731
	732	// part belongs to a related message and is linked
cb0f03	733	if (preg_match('/^multipart\/(related\|relative)/', $mimetype)
d311d8	734	&& ($mail_part->headers['content-id'] \|\| $mail_part->headers['content-location'])) {
A	735	if ($mail_part->headers['content-id'])
	736	$mail_part->content_id = preg_replace(array('/^</', '/>$/'), '', $mail_part->headers['content-id']);
	737	if ($mail_part->headers['content-location'])
	738	$mail_part->content_location = $mail_part->headers['content-base'] . $mail_part->headers['content-location'];
	739
	740	$this->inline_parts[] = $mail_part;
	741	}
	742	// attachment encapsulated within message/rfc822 part needs further decoding (#1486743)
	743	else if ($part_orig_mimetype == 'message/rfc822') {
	744	$this->parse_structure($mail_part, true);
fd371a	745
A	746	// list as attachment as well (mostly .eml)
	747	if (!empty($mail_part->filename))
	748	$this->attachments[] = $mail_part;
d311d8	749	}
b38925	750	// regular attachment with valid content type
A	751	// (content-type name regexp according to RFC4288.4.2)
2ae58f	752	else if (preg_match('/^[a-z0-9!#$&.+^_-]+\/[a-z0-9!#$&.+^_-]+$/i', $part_mimetype)) {
b38925	753	$this->attachments[] = $mail_part;
A	754	}
	755	// attachment with invalid content type
	756	// replace malformed content type with application/octet-stream (#1487767)
	757	else if ($mail_part->filename) {
	758	$mail_part->ctype_primary = 'application';
	759	$mail_part->ctype_secondary = 'octet-stream';
	760	$mail_part->mimetype = 'application/octet-stream';
	761
d311d8	762	$this->attachments[] = $mail_part;
A	763	}
	764	}
8757f5	765	// attachment part as message/rfc822 (#1488026)
A	766	else if ($mail_part->mimetype == 'message/rfc822') {
	767	$this->parse_structure($mail_part);
	768	}
d311d8	769	}
A	770
	771	// if this was a related part try to resolve references
cb0f03	772	if (preg_match('/^multipart\/(related\|relative)/', $mimetype) && sizeof($this->inline_parts)) {
d311d8	773	$a_replaces = array();
89d19c	774	$img_regexp = '/^image\/(gif\|jpe?g\|png\|tiff\|bmp\|svg)/';
d311d8	775
A	776	foreach ($this->inline_parts as $inline_object) {
a021d6	777	$part_url = $this->get_part_url($inline_object->mime_id, $inline_object->ctype_primary);
8cfba1	778	if (isset($inline_object->content_id))
d311d8	779	$a_replaces['cid:'.$inline_object->content_id] = $part_url;
63f9de	780	if ($inline_object->content_location) {
d311d8	781	$a_replaces[$inline_object->content_location] = $part_url;
63f9de	782	}
89d19c	783
A	784	if (!empty($inline_object->filename)) {
	785	// MS Outlook sends sometimes non-related attachments as related
	786	// In this case multipart/related message has only one text part
	787	// We'll add all such attachments to the attachments list
	788	if (!isset($got_html_part) && empty($inline_object->content_id)) {
	789	$this->attachments[] = $inline_object;
	790	}
	791	// MS Outlook sometimes also adds non-image attachments as related
	792	// We'll add all such attachments to the attachments list
	793	// Warning: some browsers support pdf in <img/>
	794	else if (!preg_match($img_regexp, $inline_object->mimetype)) {
	795	$this->attachments[] = $inline_object;
	796	}
	797	// @TODO: we should fetch HTML body and find attachment's content-id
	798	// to handle also image attachments without reference in the body
	799	// @TODO: should we list all image attachments in text mode?
02b6e6	800	}
d311d8	801	}
A	802
	803	// add replace array to each content part
	804	// (will be applied later when part body is available)
	805	foreach ($this->parts as $i => $part) {
	806	if ($part->type == 'content')
	807	$this->parts[$i]->replaces = $a_replaces;
	808	}
	809	}
	810	}
	811	// message is a single part non-text
	812	else if ($structure->filename) {
	813	$this->attachments[] = $structure;
	814	}
3e58bf	815	// message is a single part non-text (without filename)
A	816	else if (preg_match('/application\//i', $mimetype)) {
	817	$this->attachments[] = $structure;
c5d7c9	818	}
AM	819	}
	820
	821
	822	/**
d311d8	823	* Fill aflat array with references to all parts, indexed by part numbers
A	824	*
5c461b	825	* @param rcube_message_part $part Message body structure
d311d8	826	*/
A	827	private function get_mime_numbers(&$part)
	828	{
	829	if (strlen($part->mime_id))
	830	$this->mime_parts[$part->mime_id] = &$part;
f19d86	831
d311d8	832	if (is_array($part->parts))
A	833	for ($i=0; $i<count($part->parts); $i++)
	834	$this->get_mime_numbers($part->parts[$i]);
	835	}
	836
	837
	838	/**
	839	* Decode a Microsoft Outlook TNEF part (winmail.dat)
	840	*
5c461b	841	* @param rcube_message_part $part Message part to decode
A	842	* @return array
d311d8	843	*/
A	844	function tnef_decode(&$part)
	845	{
48ba44	846	// @TODO: attachment may be huge, handle body via file
AM	847	$body = $this->get_part_body($part->mime_id);
2883fc	848	$tnef = new rcube_tnef_decoder;
48ba44	849	$tnef_arr = $tnef->decompress($body);
AM	850	$parts = array();
d311d8	851
48ba44	852	unset($body);
d311d8	853
A	854	foreach ($tnef_arr as $pid => $winatt) {
	855	$tpart = new rcube_message_part;
	856
2da830	857	$tpart->filename = $this->fix_attachment_name(trim($winatt['name']), $part);
d311d8	858	$tpart->encoding = 'stream';
f19d86	859	$tpart->ctype_primary = trim(strtolower($winatt['type']));
A	860	$tpart->ctype_secondary = trim(strtolower($winatt['subtype']));
d311d8	861	$tpart->mimetype = $tpart->ctype_primary . '/' . $tpart->ctype_secondary;
A	862	$tpart->mime_id = 'winmail.' . $part->mime_id . '.' . $pid;
	863	$tpart->size = $winatt['size'];
	864	$tpart->body = $winatt['stream'];
	865
	866	$parts[] = $tpart;
	867	unset($tnef_arr[$pid]);
	868	}
f19d86	869
d311d8	870	return $parts;
A	871	}
	872
	873
	874	/**
	875	* Parse message body for UUencoded attachments bodies
	876	*
5c461b	877	* @param rcube_message_part $part Message part to decode
A	878	* @return array
d311d8	879	*/
A	880	function uu_decode(&$part)
	881	{
48ba44	882	// @TODO: messages may be huge, handle body via file
AM	883	$part->body = $this->get_part_body($part->mime_id);
	884	$parts = array();
	885	$pid = 0;
d311d8	886
A	887	// FIXME: line length is max.65?
48ba44	888	$uu_regexp_begin = '/begin [0-7]{3,4} ([^\r\n]+)\r?\n/s';
AM	889	$uu_regexp_end = '/`\r?\nend((\r?\n)\|($))/s';
d311d8	890
48ba44	891	while (preg_match($uu_regexp_begin, $part->body, $matches, PREG_OFFSET_CAPTURE)) {
AM	892	$startpos = $matches[0][1];
d311d8	893
48ba44	894	if (!preg_match($uu_regexp_end, $part->body, $m, PREG_OFFSET_CAPTURE, $startpos)) {
AM	895	break;
d311d8	896	}
14f22f	897
48ba44	898	$endpos = $m[0][1];
AM	899	$begin_len = strlen($matches[0][0]);
	900	$end_len = strlen($m[0][0]);
	901
	902	// extract attachment body
	903	$filebody = substr($part->body, $startpos + $begin_len, $endpos - $startpos - $begin_len - 1);
	904	$filebody = str_replace("\r\n", "\n", $filebody);
	905
	906	// remove attachment body from the message body
	907	$part->body = substr_replace($part->body, '', $startpos, $endpos + $end_len - $startpos);
2268aa	908	// mark body as modified so it will not be cached by rcube_imap_cache
AM	909	$part->body_modified = true;
48ba44	910
AM	911	// add attachments to the structure
	912	$uupart = new rcube_message_part;
	913	$uupart->filename = trim($matches[1][0]);
	914	$uupart->encoding = 'stream';
	915	$uupart->body = convert_uudecode($filebody);
	916	$uupart->size = strlen($uupart->body);
	917	$uupart->mime_id = 'uu.' . $part->mime_id . '.' . $pid;
	918
	919	$ctype = rcube_mime::file_content_type($uupart->body, $uupart->filename, 'application/octet-stream', true);
	920	$uupart->mimetype = $ctype;
	921	list($uupart->ctype_primary, $uupart->ctype_secondary) = explode('/', $ctype);
	922
	923	$parts[] = $uupart;
	924	$pid++;
d311d8	925	}
f19d86	926
d311d8	927	return $parts;
A	928	}
b91f04	929
2da830	930	/**
AM	931	* Fix attachment name encoding if needed/possible
	932	*/
	933	protected function fix_attachment_name($name, $part)
	934	{
	935	if ($name == rcube_charset::clean($name)) {
	936	return $name;
	937	}
	938
	939	// find charset from part or its parent(s)
	940	if ($part->charset) {
	941	$charsets[] = $part->charset;
	942	}
	943	else {
	944	// check first part (common case)
	945	$n = strpos($part->mime_id, '.') ? preg_replace('/\.[0-9]+$/', '', $part->mime_id) . '.1' : 1;
	946	if (($_part = $this->mime_parts[$n]) && $_part->charset) {
	947	$charsets[] = $_part->charset;
	948	}
	949
	950	// check parents' charset
	951	$items = explode('.', $part->mime_id);
	952	for ($i = count($items)-1; $i > 0; $i--) {
	953	$last = array_pop($items);
	954	$parent = $this->mime_parts[join('.', $items)];
	955
	956	if ($parent && $parent->charset) {
	957	$charsets[] = $parent->charset;
	958	}
	959	}
	960	}
	961
	962	if ($this->headers->charset) {
	963	$charsets[] = $this->headers->charset;
	964	}
	965
	966	if (empty($charsets)) {
	967	$rcube = rcube::get_instance();
	968	$charsets[] = rcube_charset::detect($name, $rcube->config->get('default_charset', RCUBE_CHARSET));
	969	}
	970
	971	foreach (array_unique($charsets) as $charset) {
	972	$_name = rcube_charset::convert($name, $charset);
	973
	974	if ($_name == rcube_charset::clean($_name)) {
	975	if (!$part->charset) {
	976	$part->charset = $charset;
	977	}
	978
	979	return $_name;
	980	}
	981	}
	982
	983	return $name;
	984	}
b91f04	985
T	986	/**
	987	* Deprecated methods (to be removed)
	988	*/
	989
	990	public static function unfold_flowed($text)
	991	{
	992	return rcube_mime::unfold_flowed($text);
	993	}
	994
	995	public static function format_flowed($text, $length = 72)
	996	{
	997	return rcube_mime::format_flowed($text, $length);
	998	}
	999
8fa58e	1000	}