| 1 |
<?php |
| 2 |
/** |
| 3 |
* @author Niels A.D. |
| 4 |
* @author Todd Burry <todd@vanillaforums.com> |
| 5 |
* @copyright 2010 Niels A.D., 2014 Todd Burry |
| 6 |
* @license http://opensource.org/licenses/LGPL-2.1 LGPL-2.1 |
| 7 |
* @package pQuery |
| 8 |
*/ |
| 9 |
|
| 10 |
namespace pQuery; |
| 11 |
|
| 12 |
/** |
| 13 |
* Indents text |
| 14 |
* @param string $text |
| 15 |
* @param int $indent |
| 16 |
* @param string $indent_string |
| 17 |
* @return string |
| 18 |
*/ |
| 19 |
function indent_text($text, $indent, $indent_string = ' ') { |
| 20 |
if ($indent && $indent_string) { |
| 21 |
return str_replace("\n", "\n".str_repeat($indent_string, $indent), $text); |
| 22 |
} else { |
| 23 |
return $text; |
| 24 |
} |
| 25 |
} |
| 26 |
|
| 27 |
/** |
| 28 |
* Class used to format/minify HTML nodes |
| 29 |
* |
| 30 |
* Used like: |
| 31 |
* <code> |
| 32 |
* <?php |
| 33 |
* $formatter = new HtmlFormatter(); |
| 34 |
* $formatter->format($root); |
| 35 |
* ?> |
| 36 |
* </code> |
| 37 |
*/ |
| 38 |
class HtmlFormatter { |
| 39 |
|
| 40 |
/** |
| 41 |
* Determines which elements start on a new line and which function as block |
| 42 |
* @var array('element' => array('new_line' => true, 'as_block' => true, 'format_inside' => true)) |
| 43 |
*/ |
| 44 |
var $block_elements = array( |
| 45 |
'p' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 46 |
'h1' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 47 |
'h2' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 48 |
'h3' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 49 |
'h4' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 50 |
'h5' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 51 |
'h6' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 52 |
|
| 53 |
'form' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 54 |
'fieldset' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 55 |
'legend' => array('new_line' => true, 'as_block' => false, 'format_inside' => true), |
| 56 |
'dl' => array('new_line' => true, 'as_block' => false, 'format_inside' => true), |
| 57 |
'dt' => array('new_line' => true, 'as_block' => false, 'format_inside' => true), |
| 58 |
'dd' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 59 |
'ol' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 60 |
'ul' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 61 |
'li' => array('new_line' => true, 'as_block' => false, 'format_inside' => true), |
| 62 |
|
| 63 |
'table' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 64 |
'tr' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 65 |
|
| 66 |
'dir' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 67 |
'menu' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 68 |
'address' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 69 |
'blockquote' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 70 |
'center' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 71 |
'del' => array('new_line' => true, 'as_block' => false, 'format_inside' => true), |
| 72 |
//'div' => array('new_line' => false, 'as_block' => true, 'format_inside' => true), |
| 73 |
'hr' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 74 |
'ins' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 75 |
'noscript' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 76 |
'pre' => array('new_line' => true, 'as_block' => true, 'format_inside' => false), |
| 77 |
'script' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 78 |
'style' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 79 |
|
| 80 |
'html' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 81 |
'head' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 82 |
'body' => array('new_line' => true, 'as_block' => true, 'format_inside' => true), |
| 83 |
'title' => array('new_line' => true, 'as_block' => false, 'format_inside' => false) |
| 84 |
); |
| 85 |
|
| 86 |
/** |
| 87 |
* Determines which characters are considered whitespace |
| 88 |
* @var array("\t" => true) True to recognize as new line |
| 89 |
*/ |
| 90 |
var $whitespace = array( |
| 91 |
' ' => false, |
| 92 |
"\t" => false, |
| 93 |
"\x0B" => false, |
| 94 |
"\0" => false, |
| 95 |
"\n" => true, |
| 96 |
"\r" => true |
| 97 |
); |
| 98 |
|
| 99 |
/** |
| 100 |
* String that is used to generate correct indenting |
| 101 |
* @var string |
| 102 |
*/ |
| 103 |
var $indent_string = ' '; |
| 104 |
|
| 105 |
/** |
| 106 |
* String that is used to break lines |
| 107 |
* @var string |
| 108 |
*/ |
| 109 |
var $linebreak_string = "\n"; |
| 110 |
|
| 111 |
/** |
| 112 |
* Other formatting options |
| 113 |
* @var array |
| 114 |
*/ |
| 115 |
public $options = array( |
| 116 |
'img_alt' => '', |
| 117 |
'self_close_str' => null, |
| 118 |
'attribute_shorttag' => false, |
| 119 |
'sort_attributes' => false, |
| 120 |
'attributes_case' => CASE_LOWER, |
| 121 |
'minify_script' => true |
| 122 |
); |
| 123 |
|
| 124 |
/** |
| 125 |
* Errors found during formatting |
| 126 |
* @var array |
| 127 |
*/ |
| 128 |
var $errors = array(); |
| 129 |
|
| 130 |
|
| 131 |
/** |
| 132 |
* Class constructor |
| 133 |
* @param array $options {@link $options} |
| 134 |
*/ |
| 135 |
function __construct($options = array()) { |
| 136 |
$this->options = array_merge($this->options, $options); |
| 137 |
|
| 138 |
if (isset($options['indent_str'])) |
| 139 |
$this->indent_string = $options['indent_str']; |
| 140 |
|
| 141 |
if (isset($options['linebreak_str'])) |
| 142 |
$this->linebreak_string = $options['linebreak_str']; |
| 143 |
} |
| 144 |
|
| 145 |
#php4 PHP4 class constructor compatibility |
| 146 |
#function HtmlFormatter($options = array()) {return $this->__construct($options);} |
| 147 |
#php4e |
| 148 |
|
| 149 |
/** |
| 150 |
* Class magic invoke method, performs {@link format()} |
| 151 |
* @access private |
| 152 |
*/ |
| 153 |
function __invoke(&$node) { |
| 154 |
return $this->format($node); |
| 155 |
} |
| 156 |
|
| 157 |
/** |
| 158 |
* Minifies HTML / removes unneeded whitespace |
| 159 |
* @param DomNode $root |
| 160 |
* @param bool $strip_comments |
| 161 |
* @param bool $recursive |
| 162 |
*/ |
| 163 |
static function minify_html(&$root, $strip_comments = true, $recursive = true) { |
| 164 |
if ($strip_comments) { |
| 165 |
foreach($root->select(':comment', false, $recursive, true) as $c) { |
| 166 |
$prev = $c->getSibling(-1); |
| 167 |
$next = $c->getSibling(1); |
| 168 |
$c->delete(); |
| 169 |
if ($prev && $next && ($prev->isText()) && ($next->isText())) { |
| 170 |
$prev->text .= $next->text; |
| 171 |
$next->delete(); |
| 172 |
} |
| 173 |
} |
| 174 |
} |
| 175 |
foreach($root->select('(!pre + !xmp + !style + !script + !"?php" + !"~text~" + !"~comment~"):not-empty > "~text~"', false, $recursive, true) as $c) { |
| 176 |
$c->text = preg_replace('`\s+`', ' ', $c->text); |
| 177 |
} |
| 178 |
} |
| 179 |
|
| 180 |
/** |
| 181 |
* Minifies javascript using JSMin+ |
| 182 |
* @param DomNode $root |
| 183 |
* @param string $indent_string |
| 184 |
* @param bool $wrap_comment Wrap javascript in HTML comments (<!-- ~text~ //-->) |
| 185 |
* @param bool $recursive |
| 186 |
* @return bool|array Array of errors on failure, true on succes |
| 187 |
*/ |
| 188 |
static function minify_javascript(&$root, $indent_string = ' ', $wrap_comment = true, $recursive = true) { |
| 189 |
#php4 JSMin+ doesn't support PHP4 |
| 190 |
#return true; |
| 191 |
#php4e |
| 192 |
#php5 |
| 193 |
include_once('third party/jsminplus.php'); |
| 194 |
|
| 195 |
$errors = array(); |
| 196 |
foreach($root->select('script:not-empty > "~text~"', false, $recursive, true) as $c) { |
| 197 |
try { |
| 198 |
$text = $c->text; |
| 199 |
while ($text) { |
| 200 |
$text = trim($text); |
| 201 |
//Remove comment/CDATA tags at begin and end |
| 202 |
if (substr($text, 0, 4) === '<!--') { |
| 203 |
$text = substr($text, 5); |
| 204 |
continue; |
| 205 |
} elseif (strtolower(substr($text, 0, 9)) === '<![cdata[') { |
| 206 |
$text = substr($text, 10); |
| 207 |
continue; |
| 208 |
} |
| 209 |
|
| 210 |
if (($end = substr($text, -3)) && (($end === '-->') || ($end === ']]>'))) { |
| 211 |
$text = substr($text, 0, -3); |
| 212 |
continue; |
| 213 |
} |
| 214 |
|
| 215 |
break; |
| 216 |
} |
| 217 |
|
| 218 |
if (trim($text)) { |
| 219 |
$text = \JSMinPlus::minify($text); |
| 220 |
if ($wrap_comment) { |
| 221 |
$text = "<!--\n".$text."\n//-->"; |
| 222 |
} |
| 223 |
if ($indent_string && ($wrap_comment || (strpos($text, "\n") !== false))) { |
| 224 |
$text = indent_text("\n".$text, $c->indent(), $indent_string); |
| 225 |
} |
| 226 |
} |
| 227 |
$c->text = $text; |
| 228 |
} catch (\Exception $e) { |
| 229 |
$errors[] = array($e, $c->parent->dumpLocation()); |
| 230 |
} |
| 231 |
} |
| 232 |
|
| 233 |
return (($errors) ? $errors : true); |
| 234 |
#php5e |
| 235 |
} |
| 236 |
|
| 237 |
/** |
| 238 |
* Formats HTML |
| 239 |
* @param DomNode $root |
| 240 |
* @param bool $recursive |
| 241 |
* @access private |
| 242 |
*/ |
| 243 |
function format_html(&$root, $recursive = null) { |
| 244 |
if ($recursive === null) { |
| 245 |
$recursive = true; |
| 246 |
self::minify_html($root); |
| 247 |
} elseif (is_int($recursive)) { |
| 248 |
$recursive = (($recursive > 1) ? $recursive - 1 : false); |
| 249 |
} |
| 250 |
|
| 251 |
$root_tag = strtolower($root->tag); |
| 252 |
$in_block = isset($this->block_elements[$root_tag]) && $this->block_elements[$root_tag]['as_block']; |
| 253 |
$child_count = count($root->children); |
| 254 |
|
| 255 |
if (isset($this->options['attributes_case']) && $this->options['attributes_case']) { |
| 256 |
$root->attributes = array_change_key_case($root->attributes, $this->options['attributes_case']); |
| 257 |
$root->attributes_ns = null; |
| 258 |
} |
| 259 |
|
| 260 |
if (isset($this->options['sort_attributes']) && $this->options['sort_attributes']) { |
| 261 |
if ($this->options['sort_attributes'] === 'reverse') { |
| 262 |
krsort($root->attributes); |
| 263 |
} else { |
| 264 |
ksort($root->attributes); |
| 265 |
} |
| 266 |
} |
| 267 |
|
| 268 |
if ($root->select(':element', true, false, true)) { |
| 269 |
$root->setTag(strtolower($root->tag), true); |
| 270 |
if (($this->options['img_alt'] !== null) && ($root_tag === 'img') && (!isset($root->alt))) { |
| 271 |
$root->setAttribute('alt', $this->options['img_alt']); |
| 272 |
} |
| 273 |
} |
| 274 |
if ($this->options['self_close_str'] !== null) { |
| 275 |
$root->self_close_str = $this->options['self_close_str']; |
| 276 |
} |
| 277 |
if ($this->options['attribute_shorttag'] !== null) { |
| 278 |
$root->attribute_shorttag = $this->options['attribute_shorttag']; |
| 279 |
} |
| 280 |
|
| 281 |
$prev = null; |
| 282 |
$n_tag = ''; |
| 283 |
// $prev_tag = ''; |
| 284 |
$as_block = false; |
| 285 |
$prev_asblock = false; |
| 286 |
for($i = 0; $i < $child_count; $i++) { |
| 287 |
$n =& $root->children[$i]; |
| 288 |
$indent = $n->indent(); |
| 289 |
|
| 290 |
if (!$n->isText()) { |
| 291 |
$n_tag = strtolower($n->tag); |
| 292 |
$new_line = isset($this->block_elements[$n_tag]) && $this->block_elements[$n_tag]['new_line']; |
| 293 |
$as_block = isset($this->block_elements[$n_tag]) && $this->block_elements[$n_tag]['as_block']; |
| 294 |
$format_inside = ((!isset($this->block_elements[$n_tag])) || $this->block_elements[$n_tag]['format_inside']); |
| 295 |
|
| 296 |
if ($prev && ($prev->isText()) && $prev->text && ($char = $prev->text[strlen($prev->text) - 1]) && isset($this->whitespace[$char])) { |
| 297 |
if ($this->whitespace[$char]) { |
| 298 |
$prev->text .= str_repeat($this->indent_string, $indent); |
| 299 |
} else { |
| 300 |
$prev->text = substr_replace($prev->text, $this->linebreak_string.str_repeat($this->indent_string, $indent), -1, 1); |
| 301 |
} |
| 302 |
} elseif (($new_line || $prev_asblock || ($in_block && ($i === 0)))){ |
| 303 |
if ($prev && ($prev->isText())) { |
| 304 |
$prev->text .= $this->linebreak_string.str_repeat($this->indent_string, $indent); |
| 305 |
} else { |
| 306 |
$root->addText($this->linebreak_string.str_repeat($this->indent_string, $indent), $i); |
| 307 |
++$child_count; |
| 308 |
} |
| 309 |
} |
| 310 |
|
| 311 |
if ($format_inside && count($n->children)) { |
| 312 |
//$last = end($n->children); |
| 313 |
$last = $n->children[count($n->children) - 1]; |
| 314 |
$last_tag = ($last) ? strtolower($last->tag) : ''; |
| 315 |
$last_asblock = ($last_tag && isset($this->block_elements[$last_tag]) && $this->block_elements[$last_tag]['as_block']); |
| 316 |
|
| 317 |
if (($n->childCount(true) > 0) || (trim($n->getPlainText()))) { |
| 318 |
if ($last && ($last->isText()) && $last->text && ($char = $last->text[strlen($last->text) - 1]) && isset($this->whitespace[$char])) { |
| 319 |
if ($as_block || ($last->index() > 0) || isset($this->whitespace[$last->text[0]])) { |
| 320 |
if ($this->whitespace[$char]) { |
| 321 |
$last->text .= str_repeat($this->indent_string, $indent); |
| 322 |
} else { |
| 323 |
$last->text = substr_replace($last->text, $this->linebreak_string.str_repeat($this->indent_string, $indent), -1, 1); |
| 324 |
} |
| 325 |
} |
| 326 |
} elseif (($as_block || $last_asblock || ($in_block && ($i === 0))) && $last) { |
| 327 |
if ($last && ($last->isText())) { |
| 328 |
$last->text .= $this->linebreak_string.str_repeat($this->indent_string, $indent); |
| 329 |
} else { |
| 330 |
$n->addText($this->linebreak_string.str_repeat($this->indent_string, $indent)); |
| 331 |
} |
| 332 |
} |
| 333 |
} elseif (!trim($n->getInnerText())) { |
| 334 |
$n->clear(); |
| 335 |
} |
| 336 |
|
| 337 |
if ($recursive) { |
| 338 |
$this->format_html($n, $recursive); |
| 339 |
} |
| 340 |
} |
| 341 |
|
| 342 |
} elseif (trim($n->text) && ((($i - 1 < $child_count) && ($char = $n->text[0]) && isset($this->whitespace[$char])) || ($in_block && ($i === 0)))) { |
| 343 |
if (isset($this->whitespace[$char])) { |
| 344 |
if ($this->whitespace[$char]) { |
| 345 |
$n->text = str_repeat($this->indent_string, $indent).$n->text; |
| 346 |
} else { |
| 347 |
$n->text = substr_replace($n->text, $this->linebreak_string.str_repeat($this->indent_string, $indent), 0, 1); |
| 348 |
} |
| 349 |
} else { |
| 350 |
$n->text = $this->linebreak_string.str_repeat($this->indent_string, $indent).$n->text; |
| 351 |
} |
| 352 |
} |
| 353 |
|
| 354 |
$prev = $n; |
| 355 |
// $prev_tag = $n_tag; |
| 356 |
$prev_asblock = $as_block; |
| 357 |
} |
| 358 |
|
| 359 |
return true; |
| 360 |
} |
| 361 |
|
| 362 |
/** |
| 363 |
* Formats HTML/Javascript |
| 364 |
* @param DomNode $root |
| 365 |
* @see format_html() |
| 366 |
*/ |
| 367 |
function format(&$node) { |
| 368 |
$this->errors = array(); |
| 369 |
if ($this->options['minify_script']) { |
| 370 |
$a = self::minify_javascript($node, $this->indent_string, true, true); |
| 371 |
if (is_array($a)) { |
| 372 |
foreach($a as $error) { |
| 373 |
$this->errors[] = $error[0]->getMessage().' >>> '.$error[1]; |
| 374 |
} |
| 375 |
} |
| 376 |
} |
| 377 |
return $this->format_html($node); |
| 378 |
} |
| 379 |
} |
| 380 |
|
| 381 |
?> |
