githubFork/roundcubemail.git

Add required parameters to raise_error() calls.

svncommit

2009-01-05 6710a636170c73162560c94629656ddde85c7fa8

commit \| author \| age
4e17e6	1	<?php
T	2
	3	/*************************************************************************
8ac6fd	4	* *
A	5	* class.html2text.inc *
	6	* *
	7	*************************************************************************
	8	* *
	9	* Converts HTML to formatted plain text *
	10	* *
	11	* Copyright (c) 2005-2007 Jon Abernathy <jon@chuggnutt.com> *
	12	* All rights reserved. *
	13	* *
	14	* This script is free software; you can redistribute it and/or modify *
	15	* it under the terms of the GNU General Public License as published by *
	16	* the Free Software Foundation; either version 2 of the License, or *
	17	* (at your option) any later version. *
	18	* *
	19	* The GNU General Public License can be found at *
	20	* http://www.gnu.org/copyleft/gpl.html. *
	21	* *
	22	* This script is distributed in the hope that it will be useful, *
	23	* but WITHOUT ANY WARRANTY; without even the implied warranty of *
	24	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the *
	25	* GNU General Public License for more details. *
	26	* *
	27	* Author(s): Jon Abernathy <jon@chuggnutt.com> *
	28	* *
	29	* Last modified: 08/08/07 *
	30	* *
	31	*************************************************************************/
4e17e6	32
T	33
	34	/**
8ac6fd	35	* Takes HTML and converts it to formatted, plain text.
A	36	*
	37	* Thanks to Alexander Krug (http://www.krugar.de/) to pointing out and
	38	* correcting an error in the regexp search array. Fixed 7/30/03.
	39	*
	40	* Updated set_html() function's file reading mechanism, 9/25/03.
	41	*
	42	* Thanks to Joss Sanglier (http://www.dancingbear.co.uk/) for adding
	43	* several more HTML entity codes to the $search and $replace arrays.
	44	* Updated 11/7/03.
	45	*
	46	* Thanks to Darius Kasperavicius (http://www.dar.dar.lt/) for
	47	* suggesting the addition of $allowed_tags and its supporting function
	48	* (which I slightly modified). Updated 3/12/04.
	49	*
	50	* Thanks to Justin Dearing for pointing out that a replacement for the
	51	* <TH> tag was missing, and suggesting an appropriate fix.
	52	* Updated 8/25/04.
	53	*
	54	* Thanks to Mathieu Collas (http://www.myefarm.com/) for finding a
	55	* display/formatting bug in the _build_link_list() function: email
	56	* readers would show the left bracket and number ("[1") as part of the
	57	* rendered email address.
	58	* Updated 12/16/04.
	59	*
	60	* Thanks to Wojciech Bajon (http://histeria.pl/) for submitting code
	61	* to handle relative links, which I hadn't considered. I modified his
	62	* code a bit to handle normal HTTP links and MAILTO links. Also for
	63	* suggesting three additional HTML entity codes to search for.
	64	* Updated 03/02/05.
	65	*
	66	* Thanks to Jacob Chandler for pointing out another link condition
	67	* for the _build_link_list() function: "https".
	68	* Updated 04/06/05.
	69	*
	70	* Thanks to Marc Bertrand (http://www.dresdensky.com/) for
	71	* suggesting a revision to the word wrapping functionality; if you
	72	* specify a $width of 0 or less, word wrapping will be ignored.
	73	* Updated 11/02/06.
	74	*
	75	* *** Big housecleaning updates below:
	76	*
	77	* Thanks to Colin Brown (http://www.sparkdriver.co.uk/) for
	78	* suggesting the fix to handle </li> and blank lines (whitespace).
	79	* Christian Basedau (http://www.movetheweb.de/) also suggested the
	80	* blank lines fix.
	81	*
	82	* Special thanks to Marcus Bointon (http://www.synchromedia.co.uk/),
	83	* Christian Basedau, Norbert Laposa (http://ln5.co.uk/),
	84	* Bas van de Weijer, and Marijn van Butselaar
	85	* for pointing out my glaring error in the <th> handling. Marcus also
	86	* supplied a host of fixes.
	87	*
	88	* Thanks to Jeffrey Silverman (http://www.newtnotes.com/) for pointing
	89	* out that extra spaces should be compressed--a problem addressed with
	90	* Marcus Bointon's fixes but that I had not yet incorporated.
	91	*
	92	* Thanks to Daniel Schledermann (http://www.typoconsult.dk/) for
	93	* suggesting a valuable fix with <a> tag handling.
	94	*
	95	* Thanks to Wojciech Bajon (again!) for suggesting fixes and additions,
	96	* including the <a> tag handling that Daniel Schledermann pointed
	97	* out but that I had not yet incorporated. I haven't (yet)
	98	* incorporated all of Wojciech's changes, though I may at some
	99	* future time.
	100	*
	101	* *** End of the housecleaning updates. Updated 08/08/07.
	102	*
	103	* @author Jon Abernathy <jon@chuggnutt.com>
	104	* @version 1.0.0
	105	* @since PHP 4.0.2
	106	*/
4e17e6	107	class html2text
T	108	{
	109
	110	/**
	111	* Contains the HTML content to convert.
	112	*
	113	* @var string $html
	114	* @access public
	115	*/
	116	var $html;
	117
	118	/**
	119	* Contains the converted, formatted text.
	120	*
	121	* @var string $text
	122	* @access public
	123	*/
	124	var $text;
	125
	126	/**
	127	* Maximum width of the formatted text, in columns.
	128	*
8ac6fd	129	* Set this value to 0 (or less) to ignore word wrapping
A	130	* and not constrain text to a fixed-width column.
	131	*
4e17e6	132	* @var integer $width
T	133	* @access public
	134	*/
	135	var $width = 70;
	136
	137	/**
	138	* List of preg* regular expression patterns to search for,
	139	* used in conjunction with $replace.
	140	*
	141	* @var array $search
	142	* @access public
	143	* @see $replace
	144	*/
	145	var $search = array(
	146	"/\r/", // Non-legal carriage return
	147	"/[\n\t]+/", // Newlines and tabs
8ac6fd	148	'/[ ]{2,}/', // Runs of spaces, pre-handling
4e17e6	149	'/<script[^>]>.?<\/script>/i', // <script>s -- which strip_tags supposedly has problems with
8ac6fd	150	'/<style[^>]>.?<\/style>/i', // <style>s -- which strip_tags supposedly has problems with
4e17e6	151	//'/<!-- .* -->/', // Comments -- which strip_tags might have problem a with
T	152	'/<p[^>]*>/i', // <P>
	153	'/<br[^>]*>/i', // <br>
8ac6fd	154	'/<i[^>]>(.?)<\/i>/i', // <i>
A	155	'/<em[^>]>(.?)<\/em>/i', // <em>
4e17e6	156	'/(<ul[^>]*>\|<\/ul>)/i', // <ul> and </ul>
T	157	'/(<ol[^>]*>\|<\/ol>)/i', // <ol> and </ol>
8ac6fd	158	'/<li[^>]>(.?)<\/li>/i', // <li> and </li>
4e17e6	159	'/<li[^>]*>/i', // <li>
T	160	'/<hr[^>]*>/i', // <hr>
	161	'/(<table[^>]*>\|<\/table>)/i', // <table> and </table>
	162	'/(<tr[^>]*>\|<\/tr>)/i', // <tr> and </tr>
8ac6fd	163	'/<td[^>]>(.?)<\/td>/i', // <td> and </td>
A	164	'/&(nbsp\|#160);/i', // Non-breaking space
	165	'/&(quot\|rdquo\|ldquo\|#8220\|#8221\|#147\|#148);/i',
	166	// Double quotes
	167	'/&(apos\|rsquo\|lsquo\|#8216\|#8217);/i', // Single quotes
	168	'/>/i', // Greater-than
	169	'/</i', // Less-than
	170	'/&(amp\|#38);/i', // Ampersand
	171	'/&(copy\|#169);/i', // Copyright
	172	'/&(trade\|#8482\|#153);/i', // Trademark
	173	'/&(reg\|#174);/i', // Registered
	174	'/&(mdash\|#151\|#8212);/i', // mdash
	175	'/&(ndash\|minus\|#8211\|#8722);/i', // ndash
	176	'/&(bull\|#149\|#8226);/i', // Bullet
	177	'/&(pound\|#163);/i', // Pound sign
	178	'/&(euro\|#8364);/i', // Euro sign
	179	'/&[^&;]+;/i', // Unknown/unhandled entities
	180	'/[ ]{2,}/' // Runs of spaces, post-handling
4e17e6	181	);
T	182
	183	/**
	184	* List of pattern replacements corresponding to patterns searched.
	185	*
	186	* @var array $replace
	187	* @access public
	188	* @see $search
	189	*/
	190	var $replace = array(
	191	'', // Non-legal carriage return
	192	' ', // Newlines and tabs
8ac6fd	193	' ', // Runs of spaces, pre-handling
4e17e6	194	'', // <script>s -- which strip_tags supposedly has problems with
8ac6fd	195	'', // <style>s -- which strip_tags supposedly has problems with
A	196	//'', // Comments -- which strip_tags might have problem a with
	197	"\n\n", // <P>
4e17e6	198	"\n", // <br>
T	199	'_\\1_', // <i>
8ac6fd	200	'_\\1_', // <em>
4e17e6	201	"\n\n", // <ul> and </ul>
T	202	"\n\n", // <ol> and </ol>
8ac6fd	203	"\t* \\1\n", // <li> and </li>
A	204	"\n\t* ", // <li>
	205	"\n-------------------------\n", // <hr>
	206	"\n\n", // <table> and </table>
4e17e6	207	"\n", // <tr> and </tr>
T	208	"\t\t\\1\n", // <td> and </td>
8ac6fd	209	' ', // Non-breaking space
A	210	'"', // Double quotes
	211	"'", // Single quotes
4e17e6	212	'>',
T	213	'<',
	214	'&',
	215	'(c)',
	216	'(tm)',
	217	'(R)',
8ac6fd	218	'--',
A	219	'-',
4e17e6	220	'*',
300fc6	221	'£',
8ac6fd	222	'EUR', // Euro sign. ?
A	223	'', // Unknown/unhandled entities
	224	' ' // Runs of spaces, post-handling
f50cc7	225	);
A	226
	227	/**
	228	* List of preg* regular expression patterns to search for
	229	* and replace using callback function.
	230	*
	231	* @var array $callback_search
	232	* @access public
	233	*/
	234	var $callback_search = array(
	235	'/<(h)[123456][^>]>(.?)<\/h[123456]>/i', // H1 - H3
	236	'/<(b)[^>]>(.?)<\/b>/i', // <b>
	237	'/<(strong)[^>]>(.?)<\/strong>/i', // <strong>
	238	'/<(a) [^>]href=("\|\')([^"\']+)\2[^>]>(.*?)<\/a>/i',
	239	// <a href="">
	240	'/<(th)[^>]>(.?)<\/th>/i', // <th> and </th>
7353fa	241	);
A	242
8ac6fd	243	/**
A	244	* List of preg* regular expression patterns to search for in PRE body,
	245	* used in conjunction with $pre_replace.
	246	*
	247	* @var array $pre_search
	248	* @access public
	249	* @see $pre_replace
	250	*/
7789fd	251	var $pre_search = array(
7353fa	252	"/\n/",
A	253	"/\t/",
	254	'/ /',
	255	'/<pre[^>]*>/',
	256	'/<\/pre>/'
	257	);
	258
	259	/**
	260	* List of pattern replacements corresponding to patterns searched for PRE body.
	261	*
	262	* @var array $pre_replace
	263	* @access public
	264	* @see $pre_search
	265	*/
7789fd	266	var $pre_replace = array(
7353fa	267	'<br>',
A	268	'    ',
	269	' ',
	270	'',
	271	''
4e17e6	272	);
T	273
	274	/**
	275	* Contains a list of HTML tags to allow in the resulting text.
	276	*
	277	* @var string $allowed_tags
	278	* @access public
	279	* @see set_allowed_tags()
	280	*/
	281	var $allowed_tags = '';
	282
	283	/**
	284	* Contains the base URL that relative links should resolve to.
	285	*
	286	* @var string $url
	287	* @access public
	288	*/
	289	var $url;
	290
	291	/**
	292	* Indicates whether content in the $html variable has been converted yet.
	293	*
8ac6fd	294	* @var boolean $_converted
4e17e6	295	* @access private
T	296	* @see $html, $text
	297	*/
	298	var $_converted = false;
	299
	300	/**
	301	* Contains URL addresses from links to be rendered in plain text.
	302	*
8ac6fd	303	* @var string $_link_list
4e17e6	304	* @access private
T	305	* @see _build_link_list()
	306	*/
8ac6fd	307	var $_link_list = '';
dd792e	308
S	309	/**
8ac6fd	310	* Number of valid links detected in the text, used for plain text
A	311	* display (rendered similar to footnotes).
	312	*
	313	* @var integer $_link_count
	314	* @access private
	315	* @see _build_link_list()
dd792e	316	*/
8ac6fd	317	var $_link_count = 0;
4e17e6	318
e7f85b	319	/**
A	320	* Boolean flag, true if a table of link URLs should be listed after the text.
	321	*
	322	* @var boolean $_do_links
	323	* @access private
	324	* @see html2text()
	325	*/
	326	var $_do_links = true;
	327
4e17e6	328	/**
T	329	* Constructor.
	330	*
	331	* If the HTML source string (or file) is supplied, the class
	332	* will instantiate with that source propagated, all that has
	333	* to be done it to call get_text().
	334	*
	335	* @param string $source HTML content
	336	* @param boolean $from_file Indicates $source is a file to pull content from
e7f85b	337	* @param boolean $do_links Indicate whether a table of link URLs is desired
A	338	* @param integer $width Maximum width of the formatted text, 0 for no limit
4e17e6	339	* @access public
T	340	* @return void
	341	*/
e7f85b	342	function html2text( $source = '', $from_file = false, $do_links = true, $width = 75 )
4e17e6	343	{
T	344	if ( !empty($source) ) {
	345	$this->set_html($source, $from_file);
	346	}
e7f85b	347
4e17e6	348	$this->set_base_url();
e7f85b	349	$this->_do_links = $do_links;
A	350	$this->width = $width;
4e17e6	351	}
T	352
	353	/**
	354	* Loads source HTML into memory, either from $source string or a file.
	355	*
	356	* @param string $source HTML content
	357	* @param boolean $from_file Indicates $source is a file to pull content from
	358	* @access public
	359	* @return void
	360	*/
	361	function set_html( $source, $from_file = false )
	362	{
	363	if ( $from_file && file_exists($source) ) {
8ac6fd	364	$this->html = file_get_contents($source);
4e17e6	365	}
8ac6fd	366	else
A	367	$this->html = $source;
4e17e6	368
T	369	$this->_converted = false;
	370	}
	371
	372	/**
	373	* Returns the text, converted from HTML.
	374	*
	375	* @access public
	376	* @return string
	377	*/
	378	function get_text()
	379	{
	380	if ( !$this->_converted ) {
	381	$this->_convert();
	382	}
	383
	384	return $this->text;
	385	}
	386
	387	/**
	388	* Prints the text, converted from HTML.
	389	*
	390	* @access public
	391	* @return void
	392	*/
	393	function print_text()
	394	{
	395	print $this->get_text();
	396	}
	397
	398	/**
	399	* Alias to print_text(), operates identically.
	400	*
	401	* @access public
	402	* @return void
	403	* @see print_text()
	404	*/
	405	function p()
	406	{
	407	print $this->get_text();
	408	}
	409
	410	/**
	411	* Sets the allowed HTML tags to pass through to the resulting text.
	412	*
	413	* Tags should be in the form "<p>", with no corresponding closing tag.
	414	*
	415	* @access public
	416	* @return void
	417	*/
	418	function set_allowed_tags( $allowed_tags = '' )
	419	{
	420	if ( !empty($allowed_tags) ) {
	421	$this->allowed_tags = $allowed_tags;
	422	}
	423	}
	424
	425	/**
	426	* Sets a base URL to handle relative links.
	427	*
	428	* @access public
	429	* @return void
	430	*/
	431	function set_base_url( $url = '' )
	432	{
	433	if ( empty($url) ) {
8ac6fd	434	if ( !empty($_SERVER['HTTP_HOST']) ) {
A	435	$this->url = 'http://' . $_SERVER['HTTP_HOST'];
	436	} else {
	437	$this->url = '';
	438	}
4e17e6	439	} else {
T	440	// Strip any trailing slashes for consistency (relative
	441	// URLs may already start with a slash like "/file.html")
	442	if ( substr($url, -1) == '/' ) {
	443	$url = substr($url, 0, -1);
	444	}
	445	$this->url = $url;
	446	}
	447	}
	448
	449	/**
	450	* Workhorse function that does actual conversion.
	451	*
	452	* First performs custom tag replacement specified by $search and
	453	* $replace arrays. Then strips any remaining HTML tags, reduces whitespace
	454	* and newlines to a readable format, and word wraps the text to
	455	* $width characters.
	456	*
	457	* @access private
	458	* @return void
	459	*/
	460	function _convert()
	461	{
	462	// Variables used for building the link list
8ac6fd	463	$this->_link_count = 0;
A	464	$this->_link_list = '';
4e17e6	465
T	466	$text = trim(stripslashes($this->html));
	467
7353fa	468	// Convert <PRE>
8ac6fd	469	$this->_convert_pre($text);
300fc6	470
4e17e6	471	// Run our defined search-and-replace
T	472	$text = preg_replace($this->search, $this->replace, $text);
f50cc7	473	$text = preg_replace_callback($this->callback_search, array('html2text', '_preg_callback'), $text);
4e17e6	474
4e0419	475	// Replace known html entities
A	476	$text = html_entity_decode($text, ENT_COMPAT, 'UTF-8');
	477
4e17e6	478	// Strip any other HTML tags
T	479	$text = strip_tags($text, $this->allowed_tags);
	480
	481	// Bring down number of empty lines to 2 max
8ac6fd	482	$text = preg_replace("/\n\s+\n/", "\n\n", $text);
4e17e6	483	$text = preg_replace("/[\n]{3,}/", "\n\n", $text);
T	484
	485	// Add link list
8ac6fd	486	if ( !empty($this->_link_list) ) {
A	487	$text .= "\n\nLinks:\n------\n" . $this->_link_list;
4e17e6	488	}
T	489
	490	// Wrap the text to a readable format
	491	// for PHP versions >= 4.0.2. Default width is 75
8ac6fd	492	// If width is 0 or less, don't wrap the text.
A	493	if ( $this->width > 0 ) {
	494	$text = wordwrap($text, $this->width);
	495	}
4e17e6	496
T	497	$this->text = $text;
	498
	499	$this->_converted = true;
	500	}
	501
	502	/**
	503	* Helper function called by preg_replace() on link replacement.
	504	*
	505	* Maintains an internal list of links to be displayed at the end of the
	506	* text, with numeric indices to the original point in the text they
	507	* appeared. Also makes an effort at identifying and handling absolute
	508	* and relative links.
	509	*
	510	* @param string $link URL of the link
	511	* @param string $display Part of the text to associate number with
	512	* @access private
	513	* @return string
8ac6fd	514	*/
A	515	function _build_link_list( $link, $display )
	516	{
e7f85b	517	if ( !$this->_do_links ) return $display;
A	518
	519	if ( substr($link, 0, 7) == 'http://' \|\| substr($link, 0, 8) == 'https://' \|\|
8ac6fd	520	substr($link, 0, 7) == 'mailto:' ) {
A	521	$this->_link_count++;
	522	$this->_link_list .= "[" . $this->_link_count . "] $link\n";
	523	$additional = ' [' . $this->_link_count . ']';
	524	} elseif ( substr($link, 0, 11) == 'javascript:' ) {
	525	// Don't count the link; ignore it
	526	$additional = '';
	527	// what about href="#anchor" ?
	528	} else {
	529	$this->_link_count++;
	530	$this->_link_list .= "[" . $this->_link_count . "] " . $this->url;
	531	if ( substr($link, 0, 1) != '/' ) {
	532	$this->_link_list .= '/';
4e17e6	533	}
8ac6fd	534	$this->_link_list .= "$link\n";
A	535	$additional = ' [' . $this->_link_count . ']';
4e17e6	536	}
T	537
8ac6fd	538	return $display . $additional;
A	539	}
	540
7353fa	541	/**
A	542	* Helper function for PRE body conversion.
	543	*
	544	* @param string HTML content
	545	* @access private
8ac6fd	546	*/
7353fa	547	function _convert_pre(&$text)
8ac6fd	548	{
A	549	while(preg_match('/<pre[^>]>(.)<\/pre>/ismU', $text, $matches))
	550	{
	551	$result = preg_replace($this->pre_search, $this->pre_replace, $matches[1]);
f35a22	552	$text = preg_replace('/<pre[^>]>.<\/pre>/ismU', '<div><br>' . $result . '<br></div>', $text, 1);
8ac6fd	553	}
A	554	}
f50cc7	555
A	556	/**
	557	* Callback function for preg_replace_callback use.
	558	*
	559	* @param array PREG matches
	560	* @return string
	561	* @access private
	562	*/
	563	function _preg_callback($matches)
	564	{
	565	switch($matches[1])
	566	{
	567	case 'b':
	568	case 'strong':
	569	return $this->_strtoupper($matches[2]);
	570	case 'hr':
	571	return $this->_strtoupper("\t\t". $matches[2] ."\n");
	572	case 'h':
	573	return $this->_strtoupper("\n\n". $matches[2] ."\n\n");
	574	case 'a':
	575	return $this->_build_link_list($matches[3], $matches[4]);
	576	}
	577	}
	578
	579	/**
	580	* Strtoupper multibyte wrapper function
	581	*
	582	* @param string
	583	* @return string
	584	* @access private
	585	*/
	586	function _strtoupper($str)
	587	{
	588	if (function_exists('mb_strtoupper'))
	589	return mb_strtoupper($str);
	590	else
	591	return strtoupper($str);
	592	}
4e17e6	593	}
T	594
7353fa	595	?>