├── README
└── format.php


/README:
--------------------------------------------------------------------------------
 1 | A simple HTML formatter.
 2 | 
 3 | Example:
 4 | 
 5 | <?php
 6 | 
 7 | include_once('format.php');
 8 | 
 9 | $html = 'BIG STRING FULL OF HTML';
10 | 
11 | $format = new Format;
12 | 
13 | $formatted_html = $format->HTML($html);
14 | 
15 | echo $formatted_html;
16 | 


--------------------------------------------------------------------------------
/format.php:
--------------------------------------------------------------------------------
  1 | <?php
  2 | 
  3 | class Format
  4 | {
  5 | 	private $input = '';
  6 | 	private $output = '';
  7 | 	private $tabs = 0;
  8 | 	private $in_tag = FALSE;
  9 | 	private $in_comment = FALSE;
 10 | 	private $in_content = FALSE;
 11 | 	private $inline_tag = FALSE;
 12 | 	private $input_index = 0;
 13 | 	
 14 | 	public function HTML($input)
 15 | 	{
 16 | 		$this->input = $input;
 17 | 		$this->output = '';
 18 | 		
 19 | 		$starting_index = 0;
 20 | 		
 21 | 		if (preg_match('/<\!doctype/i', $this->input)) {
 22 | 			$starting_index = strpos($this->input, '>') + 1;
 23 | 			$this->output .= substr($this->input, 0, $starting_index);
 24 | 		}
 25 | 		
 26 | 		for ($this->input_index = $starting_index; $this->input_index < strlen($this->input); $this->input_index++) {
 27 | 			if ($this->in_comment) {
 28 | 				$this->parse_comment();
 29 | 			} elseif ($this->in_tag) {
 30 | 				$this->parse_inner_tag();
 31 | 			} elseif ($this->inline_tag) {
 32 | 				$this->parse_inner_inline_tag();
 33 | 			} else {
 34 | 				if (preg_match('/[\r\n\t]/', $this->input[$this->input_index])) {
 35 | 					continue;
 36 | 				} elseif ($this->input[$this->input_index] == '<') {
 37 | 					if ( ! $this->is_inline_tag()) {
 38 | 						$this->in_content = FALSE;
 39 | 					}
 40 | 					$this->parse_tag();
 41 | 				} elseif ( ! $this->in_content) {
 42 | 					if ( ! $this->inline_tag) {
 43 | 						$this->output .= "\n" . str_repeat("\t", $this->tabs);
 44 | 					}
 45 | 					$this->in_content = TRUE;
 46 | 				}
 47 | 				$this->output .= $this->input[$this->input_index];
 48 | 			}
 49 | 		}
 50 | 		
 51 | 		return $this->output;
 52 | 	}
 53 | 	
 54 | 	private function parse_comment()
 55 | 	{
 56 | 		if ($this->is_end_comment()) {
 57 | 			$this->in_comment = FALSE;
 58 | 			$this->output .= '-->';
 59 | 			$this->input_index += 3;
 60 | 		} else {
 61 | 			$this->output .= $this->input[$this->input_index];
 62 | 		}
 63 | 	}
 64 | 	
 65 | 	private function parse_inner_tag()
 66 | 	{
 67 | 		if ($this->input[$this->input_index] == '>') {
 68 | 			$this->in_tag = FALSE;
 69 | 			$this->output .= '>';
 70 | 		} else {
 71 | 			$this->output .= $this->input[$this->input_index];
 72 | 		}
 73 | 	}
 74 | 	
 75 | 	private function parse_inner_inline_tag()
 76 | 	{
 77 | 		if ($this->input[$this->input_index] == '>') {
 78 | 			$this->inline_tag = FALSE;
 79 | 			$this->decrement_tabs();
 80 | 			$this->output .= '>';
 81 | 		} else {
 82 | 			$this->output .= $this->input[$this->input_index];
 83 | 		}
 84 | 	}
 85 | 	
 86 | 	private function parse_tag()
 87 | 	{
 88 | 		if ($this->is_comment()) {
 89 | 			$this->output .= "\n" . str_repeat("\t", $this->tabs);
 90 | 			$this->in_comment = TRUE;
 91 | 		} elseif ($this->is_end_tag()) {
 92 | 			$this->in_tag = TRUE;
 93 | 			$this->inline_tag = FALSE;
 94 | 			$this->decrement_tabs();
 95 | 			if ( ! $this->is_inline_tag() AND ! $this->is_tag_empty()) {
 96 | 				$this->output .= "\n" . str_repeat("\t", $this->tabs);
 97 | 			}
 98 | 		} else {
 99 | 			$this->in_tag = TRUE;
100 | 			if ( ! $this->in_content AND ! $this->inline_tag) {
101 | 				$this->output .= "\n" . str_repeat("\t", $this->tabs);
102 | 			}
103 | 			if ( ! $this->is_closed_tag()) {
104 | 				$this->tabs++;
105 | 			}
106 | 			if ($this->is_inline_tag()) {
107 | 				$this->inline_tag = TRUE;
108 | 			}
109 | 		}
110 | 	}
111 | 	
112 | 	private function is_end_tag()
113 | 	{
114 | 		for ($input_index = $this->input_index; $input_index < strlen($this->input); $input_index++) {
115 | 			if ($this->input[$input_index] == '<' AND $this->input[$input_index + 1] == '/') {
116 | 				return true;
117 | 			} elseif ($this->input[$input_index] == '<' AND $this->input[$input_index + 1] == '!') {
118 | 				return true;
119 | 			} elseif ($this->input[$input_index] == '>') {
120 | 				return false;
121 | 			}
122 | 		}
123 | 		return false;
124 | 	}
125 | 	
126 | 	private function decrement_tabs()
127 | 	{
128 | 		$this->tabs--;
129 | 		if ($this->tabs < 0) {
130 | 			$this->tabs = 0;
131 | 		}
132 | 	}
133 | 	
134 | 	private function is_comment()
135 | 	{
136 | 		if ($this->input[$this->input_index] == '<'
137 | 		AND $this->input[$this->input_index + 1] == '!'
138 | 		AND $this->input[$this->input_index + 2] == '-'
139 | 		AND $this->input[$this->input_index + 3] == '-') {
140 | 			return true;
141 | 		} else {
142 | 			return false;
143 | 		}
144 | 	}
145 | 	
146 | 	private function is_end_comment()
147 | 	{
148 | 		if ($this->input[$this->input_index] == '-'
149 | 		AND $this->input[$this->input_index + 1] == '-'
150 | 		AND $this->input[$this->input_index + 2] == '>') {
151 | 			return TRUE;
152 | 		} else {
153 | 			return FALSE;
154 | 		}
155 | 	}
156 | 	
157 | 	private function is_tag_empty()
158 | 	{
159 | 		$current_tag = $this->get_current_tag($this->input_index + 2);
160 | 		$in_tag = FALSE;
161 | 		
162 | 		for ($input_index = $this->input_index - 1; $input_index >= 0; $input_index--) {
163 | 			if ( ! $in_tag) {
164 | 				if ($this->input[$input_index] == '>') {
165 | 					$in_tag = TRUE;
166 | 				} elseif ( ! preg_match('/\s/', $this->input[$input_index])) {
167 | 					return FALSE;
168 | 				}
169 | 			} else {
170 | 				if ($this->input[$input_index] == '<') {
171 | 					if ($current_tag == $this->get_current_tag($input_index + 1)) {
172 | 						return TRUE;
173 | 					} else {
174 | 						return FALSE;
175 | 					}
176 | 				}
177 | 			}
178 | 		}
179 | 		return TRUE;
180 | 	}
181 | 	
182 | 	private function get_current_tag($input_index)
183 | 	{
184 | 		$current_tag = '';
185 | 		
186 | 		for ($input_index; $input_index < strlen($this->input); $input_index++) {
187 | 			if ($this->input[$input_index] == '<') {
188 | 				continue;
189 | 			} elseif ($this->input[$input_index] == '>' OR preg_match('/\s/', $this->input[$input_index])) {
190 | 				return $current_tag;
191 | 			} else {
192 | 				$current_tag .= $this->input[$input_index];
193 | 			}
194 | 		}
195 | 		
196 | 		return $current_tag;
197 | 	}
198 | 	
199 | 	private function is_closed_tag()
200 | 	{
201 | 		$closed_tags = array(
202 | 			'meta', 'link', 'img', 'hr', 'br', 'input',
203 | 		);
204 | 		
205 | 		$current_tag = '';
206 | 		
207 | 		for ($input_index = $this->input_index; $input_index < strlen($this->input); $input_index++) {
208 | 			if ($this->input[$input_index] == '<') {
209 | 				continue;
210 | 			} elseif (preg_match('/\s/', $this->input[$input_index])) {
211 | 				break;
212 | 			} else {
213 | 				$current_tag .= $this->input[$input_index];
214 | 			}
215 | 		}
216 | 		
217 | 		if (in_array($current_tag, $closed_tags)) {
218 | 			return true;
219 | 		} else {
220 | 			return false;
221 | 		}
222 | 	}
223 | 	
224 | 	private function is_inline_tag()
225 | 	{
226 | 		$inline_tags = array(
227 | 			'title', 'a', 'span', 'abbr', 'acronym', 'b', 'basefont', 'bdo', 'big', 'cite', 'code', 'dfn', 'em', 'font', 'i', 'kbd', 'q', 's', 'samp', 'small', 'strike', 'strong', 'sub', 'sup', 'textarea', 'tt', 'u', 'var', 'del', 'pre',
228 | 		);
229 | 		
230 | 		$current_tag = '';
231 | 		
232 | 		for ($input_index = $this->input_index; $input_index < strlen($this->input); $input_index++) {
233 | 			if ($this->input[$input_index] == '<' OR $this->input[$input_index] == '/') {
234 | 				continue;
235 | 			} elseif (preg_match('/\s/', $this->input[$input_index]) OR $this->input[$input_index] == '>') {
236 | 				break;
237 | 			} else {
238 | 				$current_tag .= $this->input[$input_index];
239 | 			}
240 | 		}
241 | 		
242 | 		if (in_array($current_tag, $inline_tags)) {
243 | 			return true;
244 | 		} else {
245 | 			return false;
246 | 		}
247 | 	}
248 | }
249 | 


--------------------------------------------------------------------------------