1<?php
2
3/**
4 * Formats any html output (must be valid xml where every tag opened is closed)
5 * using a single tab for indenting. 'pre' and other whitespace sensitive
6 * tags should not be affected.
7 *
8 * It is not recommended to use this on every template if you render multiple
9 * templates per page, you should only use it once on the main page template so that
10 * everything is formatted in one pass.
11 *
12 * This software is provided 'as-is', without any express or implied warranty.
13 * In no event will the authors be held liable for any damages arising from the use of this software.
14 *
15 * @author     Jordi Boggiano <j.boggiano@seld.be>
16 * @copyright  Copyright (c) 2008, Jordi Boggiano
17 * @license    http://dwoo.org/LICENSE   Modified BSD License
18 * @link       http://dwoo.org/
19 * @version    1.0.0
20 * @date       2008-10-23
21 * @package    Dwoo
22 */
23class Dwoo_Filter_html_format extends Dwoo_Filter
24{
25	/**
26	 * tab count to auto-indent the source
27	 *
28	 * @var int
29	 */
30	protected static $tabCount = -1;
31
32	/**
33	 * stores the additional data (following a tag) of the last call to open/close/singleTag
34	 *
35	 * @var string
36	 */
37	protected static $lastCallAdd = '';
38
39	/**
40	 * formats the input using the singleTag/closeTag/openTag functions
41	 *
42	 * It is auto indenting the whole code, excluding <textarea>, <code> and <pre> tags that must be kept intact.
43	 * Those tags must however contain only htmlentities-escaped text for everything to work properly.
44	 * Inline tags are presented on a single line with their content
45	 *
46	 * @param Dwoo_Core $dwoo the dwoo instance rendering this
47	 * @param string $input the xhtml to format
48	 * @return string formatted xhtml
49	 */
50	public function process($input)
51	{
52		self::$tabCount = -1;
53
54		// auto indent all but textareas & pre (or we have weird tabs inside)
55		$input = preg_replace_callback("#(<[^>]+>)(\s*)([^<]*)#", array('self', 'tagDispatcher'), $input);
56
57		return $input;
58	}
59
60	/**
61	 * helper function for format()'s preg_replace call
62	 *
63	 * @param array	$input	array of matches (1=>tag, 2=>whitespace(optional), 3=>additional non-html content)
64	 * @return string the indented tag
65	 */
66	protected static function tagDispatcher($input)
67	{
68		// textarea, pre, code tags and comments are to be left alone to avoid any non-wanted whitespace inside them so it just outputs them as they were
69		if (substr($input[1],0,9) == "<textarea" || substr($input[1],0,4) == "<pre" || substr($input[1],0,5) == "<code" || substr($input[1],0,4) == "<!--" || substr($input[1],0,9) == "<![CDATA[") {
70			return $input[1] . $input[3];
71		}
72		// closing textarea, code and pre tags and self-closed tags (i.e. <br />) are printed as singleTags because we didn't use openTag for the formers and the latter is a single tag
73		if (substr($input[1],0,10) == "</textarea" || substr($input[1],0,5) == "</pre" || substr($input[1],0,6) == "</code" || substr($input[1],-2) == "/>") {
74			return self::singleTag($input[1],$input[3],$input[2]);
75		}
76		// it's the closing tag
77		if ($input[0][1]=="/"){
78			return self::closeTag($input[1],$input[3],$input[2]);
79		}
80		// opening tag
81		return self::openTag($input[1],$input[3],$input[2]);
82	}
83
84	/**
85	 * returns an open tag and adds a tab into the auto indenting
86	 *
87	 * @param string $tag content of the tag
88	 * @param string $add additional data (anything before the following tag)
89	 * @param string $whitespace white space between the tag and the additional data
90	 * @return string
91	 */
92	protected static function openTag($tag,$add,$whitespace)
93	{
94		$tabs = str_pad('',self::$tabCount++,"\t");
95
96		if (preg_match('#^<(a|label|option|textarea|h1|h2|h3|h4|h5|h6|strong|b|em|i|abbr|acronym|cite|span|sub|sup|u|s|title)(?: [^>]*|)>#', $tag)) {
97			// if it's one of those tag it's inline so it does not require a leading line break
98			$result = $tag . $whitespace . str_replace("\n","\n".$tabs,$add);
99		} elseif (substr($tag,0,9) == '<!DOCTYPE') {
100			// it's the doctype declaration so no line break here either
101			$result = $tabs . $tag;
102		} else {
103			// normal block tag
104			$result = "\n".$tabs . $tag;
105
106			if (!empty($add)) {
107				$result .= "\n".$tabs."\t".str_replace("\n","\n\t".$tabs,$add);
108			}
109		}
110
111		self::$lastCallAdd = $add;
112
113		return $result;
114	}
115
116	/**
117	 * returns a closing tag and removes a tab from the auto indenting
118	 *
119	 * @param string $tag content of the tag
120	 * @param string $add additional data (anything before the following tag)
121	 * @param string $whitespace white space between the tag and the additional data
122	 * @return string
123	 */
124	protected static function closeTag($tag,$add,$whitespace)
125	{
126		$tabs = str_pad('',--self::$tabCount,"\t");
127
128		// if it's one of those tag it's inline so it does not require a leading line break
129		if (preg_match('#^</(a|label|option|textarea|h1|h2|h3|h4|h5|h6|strong|b|em|i|abbr|acronym|cite|span|sub|sup|u|s|title)>#', $tag)) {
130			$result = $tag . $whitespace . str_replace("\n","\n".$tabs,$add);
131		} else {
132			$result = "\n".$tabs.$tag;
133
134			if (!empty($add)) {
135				$result .= "\n".$tabs."\t".str_replace("\n","\n\t".$tabs,$add);
136			}
137		}
138
139		self::$lastCallAdd = $add;
140
141		return $result;
142	}
143
144	/**
145	 * returns a single tag with auto indenting
146	 *
147	 * @param string $tag content of the tag
148	 * @param string $add additional data (anything before the following tag)
149	 * @return string
150	 */
151	protected static function singleTag($tag,$add,$whitespace)
152	{
153		$tabs = str_pad('',self::$tabCount,"\t");
154
155		// if it's img, br it's inline so it does not require a leading line break
156		// if it's a closing textarea, code or pre tag, it does not require a leading line break either or it creates whitespace at the end of those blocks
157		if (preg_match('#^<(img|br|/textarea|/pre|/code)(?: [^>]*|)>#', $tag)) {
158			$result = $tag.$whitespace;
159
160			if (!empty($add)) {
161				$result .= str_replace("\n","\n".$tabs,$add);
162			}
163		} else {
164			$result = "\n".$tabs.$tag;
165
166			if (!empty($add)) {
167				$result .= "\n".$tabs.str_replace("\n","\n".$tabs,$add);
168			}
169		}
170
171		self::$lastCallAdd = $add;
172
173		return $result;
174	}
175}
176