@@ -17,392 +17,392 @@ |
||
17 | 17 | */ |
18 | 18 | class CleanHtmlService implements SingletonInterface |
19 | 19 | { |
20 | - /** |
|
21 | - * Enable Debug comment in footer. |
|
22 | - * |
|
23 | - * @var bool |
|
24 | - */ |
|
25 | - protected $debugComment = false; |
|
26 | - |
|
27 | - /** |
|
28 | - * Format Type. |
|
29 | - * |
|
30 | - * @var int |
|
31 | - */ |
|
32 | - protected $formatType = 0; |
|
33 | - |
|
34 | - /** |
|
35 | - * Tab character. |
|
36 | - * |
|
37 | - * @var string |
|
38 | - */ |
|
39 | - protected $tab = "\t"; |
|
40 | - |
|
41 | - /** |
|
42 | - * Newline character. |
|
43 | - * |
|
44 | - * @var string |
|
45 | - */ |
|
46 | - protected $newline = "\n"; |
|
47 | - |
|
48 | - /** |
|
49 | - * Configured extra header comment. |
|
50 | - * |
|
51 | - * @var string |
|
52 | - */ |
|
53 | - protected $headerComment = ''; |
|
54 | - |
|
55 | - /** |
|
56 | - * Empty space char. |
|
57 | - * |
|
58 | - * @var string |
|
59 | - */ |
|
60 | - protected $emptySpaceChar = ' '; |
|
61 | - |
|
62 | - /** |
|
63 | - * Set variables based on given config. |
|
64 | - */ |
|
65 | - public function setVariables(array $config): void |
|
66 | - { |
|
67 | - if (isset($config['headerComment']) && !empty($config['headerComment'])) { |
|
68 | - $this->headerComment = $config['headerComment']; |
|
69 | - } |
|
70 | - |
|
71 | - if (isset($config['formatHtml']) && is_numeric($config['formatHtml'])) { |
|
72 | - $this->formatType = (int) $config['formatHtml']; |
|
73 | - } |
|
74 | - |
|
75 | - if (isset($config['formatHtml.']['tabSize']) && is_numeric($config['formatHtml.']['tabSize'])) { |
|
76 | - $this->tab = str_pad('', (int) $config['formatHtml.']['tabSize'], ' '); |
|
77 | - } |
|
78 | - |
|
79 | - if (isset($config['formatHtml.']['debugComment'])) { |
|
80 | - $this->debugComment = (bool) $config['formatHtml.']['debugComment']; |
|
81 | - } |
|
82 | - |
|
83 | - if (isset($config['dropEmptySpaceChar']) && (bool) $config['dropEmptySpaceChar']) { |
|
84 | - $this->emptySpaceChar = ''; |
|
85 | - } |
|
86 | - } |
|
87 | - |
|
88 | - /** |
|
89 | - * Clean given HTML with formatter. |
|
90 | - * |
|
91 | - * @return string |
|
92 | - */ |
|
93 | - public function clean(string $html, array $config = []) |
|
94 | - { |
|
95 | - if (!empty($config)) { |
|
96 | - $this->setVariables($config); |
|
97 | - } |
|
98 | - // convert line-breaks to UNIX |
|
99 | - $this->convNlOs($html); |
|
100 | - |
|
101 | - $manipulations = []; |
|
102 | - |
|
103 | - if (isset($config['removeGenerator']) && (bool) $config['removeGenerator']) { |
|
104 | - $manipulations['removeGenerator'] = GeneralUtility::makeInstance(RemoveGenerator::class); |
|
105 | - } |
|
106 | - |
|
107 | - if (isset($config['removeComments']) && (bool) $config['removeComments']) { |
|
108 | - $manipulations['removeComments'] = GeneralUtility::makeInstance(RemoveComments::class); |
|
109 | - } |
|
110 | - |
|
111 | - if (!empty($this->headerComment)) { |
|
112 | - $this->includeHeaderComment($html); |
|
113 | - } |
|
114 | - |
|
115 | - foreach ($manipulations as $key => $manipulation) { |
|
116 | - /** @var ManipulationInterface $manipulation */ |
|
117 | - $configuration = isset($config[$key.'.']) && \is_array($config[$key.'.']) ? $config[$key.'.'] : []; |
|
118 | - $html = $manipulation->manipulate($html, $configuration); |
|
119 | - } |
|
120 | - |
|
121 | - // cleanup HTML5 self-closing elements |
|
122 | - if (!isset($GLOBALS['TSFE']->config['config']['doctype']) |
|
123 | - || 'x' !== substr($GLOBALS['TSFE']->config['config']['doctype'], 0, 1)) { |
|
124 | - $html = preg_replace( |
|
125 | - '/<((?:area|base|br|col|embed|hr|img|input|link|meta|param|source|track|wbr)\s[^>\\\\]+?)\s?\/>/', |
|
126 | - '<$1>', |
|
127 | - $html |
|
128 | - ); |
|
129 | - } |
|
130 | - |
|
131 | - if ($this->formatType > 0 && is_string($html)) { |
|
132 | - $html = $this->formatHtml($html); |
|
133 | - } |
|
134 | - if ( is_string($html)) { |
|
135 | - // remove white space after line ending |
|
136 | - $this->rTrimLines($html); |
|
137 | - } |
|
20 | + /** |
|
21 | + * Enable Debug comment in footer. |
|
22 | + * |
|
23 | + * @var bool |
|
24 | + */ |
|
25 | + protected $debugComment = false; |
|
26 | + |
|
27 | + /** |
|
28 | + * Format Type. |
|
29 | + * |
|
30 | + * @var int |
|
31 | + */ |
|
32 | + protected $formatType = 0; |
|
33 | + |
|
34 | + /** |
|
35 | + * Tab character. |
|
36 | + * |
|
37 | + * @var string |
|
38 | + */ |
|
39 | + protected $tab = "\t"; |
|
40 | + |
|
41 | + /** |
|
42 | + * Newline character. |
|
43 | + * |
|
44 | + * @var string |
|
45 | + */ |
|
46 | + protected $newline = "\n"; |
|
47 | + |
|
48 | + /** |
|
49 | + * Configured extra header comment. |
|
50 | + * |
|
51 | + * @var string |
|
52 | + */ |
|
53 | + protected $headerComment = ''; |
|
54 | + |
|
55 | + /** |
|
56 | + * Empty space char. |
|
57 | + * |
|
58 | + * @var string |
|
59 | + */ |
|
60 | + protected $emptySpaceChar = ' '; |
|
61 | + |
|
62 | + /** |
|
63 | + * Set variables based on given config. |
|
64 | + */ |
|
65 | + public function setVariables(array $config): void |
|
66 | + { |
|
67 | + if (isset($config['headerComment']) && !empty($config['headerComment'])) { |
|
68 | + $this->headerComment = $config['headerComment']; |
|
69 | + } |
|
70 | + |
|
71 | + if (isset($config['formatHtml']) && is_numeric($config['formatHtml'])) { |
|
72 | + $this->formatType = (int) $config['formatHtml']; |
|
73 | + } |
|
74 | + |
|
75 | + if (isset($config['formatHtml.']['tabSize']) && is_numeric($config['formatHtml.']['tabSize'])) { |
|
76 | + $this->tab = str_pad('', (int) $config['formatHtml.']['tabSize'], ' '); |
|
77 | + } |
|
78 | + |
|
79 | + if (isset($config['formatHtml.']['debugComment'])) { |
|
80 | + $this->debugComment = (bool) $config['formatHtml.']['debugComment']; |
|
81 | + } |
|
82 | + |
|
83 | + if (isset($config['dropEmptySpaceChar']) && (bool) $config['dropEmptySpaceChar']) { |
|
84 | + $this->emptySpaceChar = ''; |
|
85 | + } |
|
86 | + } |
|
87 | + |
|
88 | + /** |
|
89 | + * Clean given HTML with formatter. |
|
90 | + * |
|
91 | + * @return string |
|
92 | + */ |
|
93 | + public function clean(string $html, array $config = []) |
|
94 | + { |
|
95 | + if (!empty($config)) { |
|
96 | + $this->setVariables($config); |
|
97 | + } |
|
98 | + // convert line-breaks to UNIX |
|
99 | + $this->convNlOs($html); |
|
100 | + |
|
101 | + $manipulations = []; |
|
102 | + |
|
103 | + if (isset($config['removeGenerator']) && (bool) $config['removeGenerator']) { |
|
104 | + $manipulations['removeGenerator'] = GeneralUtility::makeInstance(RemoveGenerator::class); |
|
105 | + } |
|
106 | + |
|
107 | + if (isset($config['removeComments']) && (bool) $config['removeComments']) { |
|
108 | + $manipulations['removeComments'] = GeneralUtility::makeInstance(RemoveComments::class); |
|
109 | + } |
|
110 | + |
|
111 | + if (!empty($this->headerComment)) { |
|
112 | + $this->includeHeaderComment($html); |
|
113 | + } |
|
114 | + |
|
115 | + foreach ($manipulations as $key => $manipulation) { |
|
116 | + /** @var ManipulationInterface $manipulation */ |
|
117 | + $configuration = isset($config[$key.'.']) && \is_array($config[$key.'.']) ? $config[$key.'.'] : []; |
|
118 | + $html = $manipulation->manipulate($html, $configuration); |
|
119 | + } |
|
120 | + |
|
121 | + // cleanup HTML5 self-closing elements |
|
122 | + if (!isset($GLOBALS['TSFE']->config['config']['doctype']) |
|
123 | + || 'x' !== substr($GLOBALS['TSFE']->config['config']['doctype'], 0, 1)) { |
|
124 | + $html = preg_replace( |
|
125 | + '/<((?:area|base|br|col|embed|hr|img|input|link|meta|param|source|track|wbr)\s[^>\\\\]+?)\s?\/>/', |
|
126 | + '<$1>', |
|
127 | + $html |
|
128 | + ); |
|
129 | + } |
|
130 | + |
|
131 | + if ($this->formatType > 0 && is_string($html)) { |
|
132 | + $html = $this->formatHtml($html); |
|
133 | + } |
|
134 | + if ( is_string($html)) { |
|
135 | + // remove white space after line ending |
|
136 | + $this->rTrimLines($html); |
|
137 | + } |
|
138 | 138 | |
139 | - // recover line-breaks |
|
140 | - if (Environment::isWindows() && is_string($html) ) { |
|
141 | - $html = str_replace($this->newline, "\r\n", $html); |
|
142 | - } |
|
143 | - |
|
144 | - return (string)$html; |
|
145 | - } |
|
146 | - |
|
147 | - /** |
|
148 | - * Formats the (X)HTML code: |
|
149 | - * - taps according to the hirarchy of the tags |
|
150 | - * - removes empty spaces between tags |
|
151 | - * - removes linebreaks within tags (spares where necessary: pre, textarea, comments, ..) |
|
152 | - * choose from five options: |
|
153 | - * 0 => off |
|
154 | - * 1 => no line break at all (code in one line) |
|
155 | - * 2 => minimalistic line breaks (structure defining box-elements) |
|
156 | - * 3 => aesthetic line breaks (important box-elements) |
|
157 | - * 4 => logic line breaks (all box-elements) |
|
158 | - * 5 => max line breaks (all elements). |
|
159 | - */ |
|
160 | - protected function formatHtml(string $html):string |
|
161 | - { |
|
162 | - // Save original formated pre, textarea, comments, styles and scripts & replace them with markers |
|
163 | - preg_match_all( |
|
164 | - '/(?s)((<!--.*?-->)|(<[ \n\r]*pre[^>]*>.*?<[ \n\r]*\/pre[^>]*>)|(<[ \n\r]*textarea[^>]*>.*?<[ \n\r]*\/textarea[^>]*>)|(<[ \n\r]*style[^>]*>.*?<[ \n\r]*\/style[^>]*>)|(<[ \n\r]*script[^>]*>.*?<[ \n\r]*\/script[^>]*>))/im', |
|
165 | - $html, |
|
166 | - $matches |
|
167 | - ); |
|
168 | - $noFormat = $matches[0]; // do not format these block elements |
|
169 | - for ($i = 0; $i < \count($noFormat); ++$i) { |
|
170 | - $html = str_replace($noFormat[$i], "\n<!-- ELEMENT {$i} -->", $html); |
|
171 | - } |
|
172 | - |
|
173 | - // define box elements for formatting |
|
174 | - $trueBoxElements = 'address|blockquote|center|dir|div|dl|fieldset|form|h1|h2|h3|h4|h5|h6|hr|isindex|menu|noframes|noscript|ol|p|pre|table|ul|article|aside|details|figcaption|figure|footer|header|hgroup|menu|nav|section'; |
|
175 | - $functionalBoxElements = 'dd|dt|frameset|li|tbody|td|tfoot|th|thead|tr|colgroup'; |
|
176 | - $usableBoxElements = 'applet|button|del|iframe|ins|map|object|script'; |
|
177 | - $imagineBoxElements = 'html|body|head|meta|title|link|script|base|!--'; |
|
178 | - $allBoxLikeElements = '(?>'.$trueBoxElements.'|'.$functionalBoxElements.'|'.$usableBoxElements.'|'.$imagineBoxElements.')'; |
|
179 | - $esteticBoxLikeElements = '(?>html|head|body|meta name|title|div|table|h1|h2|h3|h4|h5|h6|p|form|pre|center|!--)'; |
|
180 | - $structureBoxLikeElements = '(?>html|head|body|div|!--)'; |
|
181 | - |
|
182 | - // split html into it's elements |
|
183 | - $htmlArrayTemp = preg_split( |
|
184 | - '/(<(?:[^<>]+(?:"[^"]*"|\'[^\']*\')?)+>)/', |
|
185 | - $html, |
|
186 | - -1, |
|
187 | - \PREG_SPLIT_DELIM_CAPTURE | \PREG_SPLIT_NO_EMPTY |
|
188 | - ); |
|
189 | - |
|
190 | - if (false === $htmlArrayTemp) { |
|
191 | - // Restore saved comments, styles and scripts |
|
192 | - for ($i = 0; $i < \count($noFormat); ++$i) { |
|
193 | - $html = str_replace("<!-- ELEMENT {$i} -->", $noFormat[$i], $html); |
|
194 | - } |
|
195 | - |
|
196 | - return $html; |
|
197 | - } |
|
198 | - // remove empty lines |
|
199 | - $htmlArray = ['']; |
|
200 | - $index = 1; |
|
201 | - for ($x = 0; $x < \count($htmlArrayTemp); ++$x) { |
|
202 | - $text = trim($htmlArrayTemp[$x]); |
|
203 | - $htmlArray[$index] = '' !== $text ? $htmlArrayTemp[$x] : $this->emptySpaceChar; |
|
204 | - ++$index; |
|
205 | - } |
|
206 | - |
|
207 | - // rebuild html |
|
208 | - $html = ''; |
|
209 | - $tabs = 0; |
|
210 | - for ($x = 0; $x < \count($htmlArray); ++$x) { |
|
211 | - $htmlArrayBefore = $htmlArray[$x - 1] ?? ''; |
|
212 | - $htmlArrayCurrent = $htmlArray[$x] ?? ''; |
|
213 | - |
|
214 | - // check if the element should stand in a new line |
|
215 | - $newline = false; |
|
216 | - if ('<?xml' == substr($htmlArrayBefore, 0, 5)) { |
|
217 | - $newline = true; |
|
218 | - } elseif (2 == $this->formatType && ( // minimalistic line break |
|
219 | - // this element has a line break before itself |
|
220 | - preg_match( |
|
221 | - '/<'.$structureBoxLikeElements.'(.*)>/Usi', |
|
222 | - $htmlArrayCurrent |
|
223 | - ) || preg_match( |
|
224 | - '/<'.$structureBoxLikeElements.'(.*) \/>/Usi', |
|
225 | - $htmlArrayCurrent |
|
226 | - ) // one element before is a element that has a line break after |
|
227 | - || preg_match( |
|
228 | - '/<\/'.$structureBoxLikeElements.'(.*)>/Usi', |
|
229 | - $htmlArrayBefore |
|
230 | - ) || '<!--' == substr( |
|
231 | - $htmlArrayBefore, |
|
232 | - 0, |
|
233 | - 4 |
|
234 | - ) || preg_match('/<'.$structureBoxLikeElements.'(.*) \/>/Usi', $htmlArrayBefore)) |
|
235 | - ) { |
|
236 | - $newline = true; |
|
237 | - } elseif (3 == $this->formatType && ( // aestetic line break |
|
238 | - // this element has a line break before itself |
|
239 | - preg_match( |
|
240 | - '/<'.$esteticBoxLikeElements.'(.*)>/Usi', |
|
241 | - $htmlArrayCurrent |
|
242 | - ) || preg_match( |
|
243 | - '/<'.$esteticBoxLikeElements.'(.*) \/>/Usi', |
|
244 | - $htmlArrayCurrent |
|
245 | - ) // one element before is a element that has a line break after |
|
246 | - || preg_match('/<\/'.$esteticBoxLikeElements.'(.*)>/Usi', $htmlArrayBefore) || '<!--' == substr( |
|
247 | - $htmlArrayBefore, |
|
248 | - 0, |
|
249 | - 4 |
|
250 | - ) || preg_match('/<'.$esteticBoxLikeElements.'(.*) \/>/Usi', $htmlArrayBefore)) |
|
251 | - ) { |
|
252 | - $newline = true; |
|
253 | - } elseif ($this->formatType >= 4 && ( // logical line break |
|
254 | - // this element has a line break before itself |
|
255 | - preg_match( |
|
256 | - '/<'.$allBoxLikeElements.'(.*)>/Usi', |
|
257 | - $htmlArrayCurrent |
|
258 | - ) || preg_match( |
|
259 | - '/<'.$allBoxLikeElements.'(.*) \/>/Usi', |
|
260 | - $htmlArrayCurrent |
|
261 | - ) // one element before is a element that has a line break after |
|
262 | - || preg_match('/<\/'.$allBoxLikeElements.'(.*)>/Usi', $htmlArrayBefore) || '<!--' == substr( |
|
263 | - $htmlArrayBefore, |
|
264 | - 0, |
|
265 | - 4 |
|
266 | - ) || preg_match('/<'.$allBoxLikeElements.'(.*) \/>/Usi', $htmlArrayBefore)) |
|
267 | - ) { |
|
268 | - $newline = true; |
|
269 | - } |
|
270 | - |
|
271 | - // count down a tab |
|
272 | - if ('</' == substr($htmlArrayCurrent, 0, 2)) { |
|
273 | - --$tabs; |
|
274 | - } |
|
275 | - |
|
276 | - // add tabs and line breaks in front of the current tag |
|
277 | - if ($newline) { |
|
278 | - $html .= $this->newline; |
|
279 | - for ($y = 0; $y < $tabs; ++$y) { |
|
280 | - $html .= $this->tab; |
|
281 | - } |
|
282 | - } |
|
283 | - |
|
284 | - // remove white spaces and line breaks and add current tag to the html-string |
|
285 | - if ('<![CDATA[' == substr($htmlArrayCurrent, 0, 9) // remove multiple white space in CDATA / XML |
|
286 | - || '<?xml' == substr($htmlArrayCurrent, 0, 5) |
|
287 | - ) { |
|
288 | - $html .= $this->killWhiteSpace($htmlArrayCurrent); |
|
289 | - } else { // remove all line breaks |
|
290 | - $html .= $this->killLineBreaks($htmlArrayCurrent); |
|
291 | - } |
|
292 | - |
|
293 | - // count up a tab |
|
294 | - if ('<' == substr($htmlArrayCurrent, 0, 1) && '/' != substr($htmlArrayCurrent, 1, 1)) { |
|
295 | - if (' ' !== substr($htmlArrayCurrent, 1, 1) |
|
296 | - && 'img' !== substr($htmlArrayCurrent, 1, 3) |
|
297 | - && 'source' !== substr($htmlArrayCurrent, 1, 6) |
|
298 | - && 'br' !== substr($htmlArrayCurrent, 1, 2) |
|
299 | - && 'hr' !== substr($htmlArrayCurrent, 1, 2) |
|
300 | - && 'input' !== substr($htmlArrayCurrent, 1, 5) |
|
301 | - && 'link' !== substr($htmlArrayCurrent, 1, 4) |
|
302 | - && 'meta' !== substr($htmlArrayCurrent, 1, 4) |
|
303 | - && 'col ' !== substr($htmlArrayCurrent, 1, 4) |
|
304 | - && 'frame' !== substr($htmlArrayCurrent, 1, 5) |
|
305 | - && 'isindex' !== substr($htmlArrayCurrent, 1, 7) |
|
306 | - && 'param' !== substr($htmlArrayCurrent, 1, 5) |
|
307 | - && 'area' !== substr($htmlArrayCurrent, 1, 4) |
|
308 | - && 'base' !== substr($htmlArrayCurrent, 1, 4) |
|
309 | - && '<!' !== substr($htmlArrayCurrent, 0, 2) |
|
310 | - && '<?xml' !== substr($htmlArrayCurrent, 0, 5) |
|
311 | - ) { |
|
312 | - ++$tabs; |
|
313 | - } |
|
314 | - } |
|
315 | - } |
|
316 | - |
|
317 | - // Remove empty lines |
|
318 | - if ($this->formatType > 1) { |
|
319 | - $this->removeEmptyLines($html); |
|
320 | - } |
|
321 | - |
|
322 | - // Restore saved comments, styles and scripts |
|
323 | - for ($i = 0; $i < \count($noFormat); ++$i) { |
|
324 | - $html = str_replace("<!-- ELEMENT {$i} -->", $noFormat[$i], $html); |
|
325 | - } |
|
326 | - |
|
327 | - // include debug comment at the end |
|
328 | - if (0 != $tabs && true === $this->debugComment) { |
|
329 | - $html .= "<!-- {$tabs} open elements found -->"; |
|
330 | - } |
|
331 | - |
|
332 | - return $html; |
|
333 | - } |
|
334 | - |
|
335 | - /** |
|
336 | - * Remove ALL line breaks and multiple white space. |
|
337 | - * |
|
338 | - * @param string $html |
|
339 | - * |
|
340 | - * @return string |
|
341 | - */ |
|
342 | - protected function killLineBreaks($html) |
|
343 | - { |
|
344 | - $html = str_replace($this->newline, '', $html); |
|
345 | - |
|
346 | - return preg_replace('/\s\s+/u', ' ', $html); |
|
347 | - // ? return preg_replace('/\n|\s+(\s)/u', '$1', $html); |
|
348 | - } |
|
349 | - |
|
350 | - /** |
|
351 | - * Remove multiple white space, keeps line breaks. |
|
352 | - */ |
|
353 | - protected function killWhiteSpace(string $html): string |
|
354 | - { |
|
355 | - $temp = explode($this->newline, $html); |
|
356 | - for ($i = 0; $i < \count($temp); ++$i) { |
|
357 | - if (!trim($temp[$i])) { |
|
358 | - unset($temp[$i]); |
|
359 | - continue; |
|
360 | - } |
|
361 | - |
|
362 | - $temp[$i] = trim($temp[$i]); |
|
363 | - $temp[$i] = preg_replace('/\s\s+/', ' ', $temp[$i]); |
|
364 | - } |
|
365 | - |
|
366 | - return implode($this->newline, $temp); |
|
367 | - } |
|
368 | - |
|
369 | - /** |
|
370 | - * Remove white space at the end of lines, keeps other white space and line breaks. |
|
371 | - */ |
|
372 | - protected function rTrimLines(string &$html): void |
|
373 | - { |
|
374 | - $html = preg_replace('/\s+$/m', '', $html); |
|
375 | - } |
|
376 | - |
|
377 | - /** |
|
378 | - * Convert newlines according to the current OS. |
|
379 | - */ |
|
380 | - protected function convNlOs(string &$html): void |
|
381 | - { |
|
382 | - $html = preg_replace("(\r\n|\r)", $this->newline, $html); |
|
383 | - } |
|
384 | - |
|
385 | - /** |
|
386 | - * Remove empty lines. |
|
387 | - */ |
|
388 | - protected function removeEmptyLines(string &$html): void |
|
389 | - { |
|
390 | - $temp = explode($this->newline, $html); |
|
391 | - $result = []; |
|
392 | - for ($i = 0; $i < \count($temp); ++$i) { |
|
393 | - if ('' == trim($temp[$i])) { |
|
394 | - continue; |
|
395 | - } |
|
396 | - $result[] = $temp[$i]; |
|
397 | - } |
|
398 | - $html = implode($this->newline, $result); |
|
399 | - } |
|
400 | - |
|
401 | - /** |
|
402 | - * Include configured header comment in HTML content block. |
|
403 | - */ |
|
404 | - public function includeHeaderComment(string &$html): void |
|
405 | - { |
|
406 | - $html = preg_replace('/^(-->)$/m', "\n\t".$this->headerComment."\n$1", $html); |
|
407 | - } |
|
139 | + // recover line-breaks |
|
140 | + if (Environment::isWindows() && is_string($html) ) { |
|
141 | + $html = str_replace($this->newline, "\r\n", $html); |
|
142 | + } |
|
143 | + |
|
144 | + return (string)$html; |
|
145 | + } |
|
146 | + |
|
147 | + /** |
|
148 | + * Formats the (X)HTML code: |
|
149 | + * - taps according to the hirarchy of the tags |
|
150 | + * - removes empty spaces between tags |
|
151 | + * - removes linebreaks within tags (spares where necessary: pre, textarea, comments, ..) |
|
152 | + * choose from five options: |
|
153 | + * 0 => off |
|
154 | + * 1 => no line break at all (code in one line) |
|
155 | + * 2 => minimalistic line breaks (structure defining box-elements) |
|
156 | + * 3 => aesthetic line breaks (important box-elements) |
|
157 | + * 4 => logic line breaks (all box-elements) |
|
158 | + * 5 => max line breaks (all elements). |
|
159 | + */ |
|
160 | + protected function formatHtml(string $html):string |
|
161 | + { |
|
162 | + // Save original formated pre, textarea, comments, styles and scripts & replace them with markers |
|
163 | + preg_match_all( |
|
164 | + '/(?s)((<!--.*?-->)|(<[ \n\r]*pre[^>]*>.*?<[ \n\r]*\/pre[^>]*>)|(<[ \n\r]*textarea[^>]*>.*?<[ \n\r]*\/textarea[^>]*>)|(<[ \n\r]*style[^>]*>.*?<[ \n\r]*\/style[^>]*>)|(<[ \n\r]*script[^>]*>.*?<[ \n\r]*\/script[^>]*>))/im', |
|
165 | + $html, |
|
166 | + $matches |
|
167 | + ); |
|
168 | + $noFormat = $matches[0]; // do not format these block elements |
|
169 | + for ($i = 0; $i < \count($noFormat); ++$i) { |
|
170 | + $html = str_replace($noFormat[$i], "\n<!-- ELEMENT {$i} -->", $html); |
|
171 | + } |
|
172 | + |
|
173 | + // define box elements for formatting |
|
174 | + $trueBoxElements = 'address|blockquote|center|dir|div|dl|fieldset|form|h1|h2|h3|h4|h5|h6|hr|isindex|menu|noframes|noscript|ol|p|pre|table|ul|article|aside|details|figcaption|figure|footer|header|hgroup|menu|nav|section'; |
|
175 | + $functionalBoxElements = 'dd|dt|frameset|li|tbody|td|tfoot|th|thead|tr|colgroup'; |
|
176 | + $usableBoxElements = 'applet|button|del|iframe|ins|map|object|script'; |
|
177 | + $imagineBoxElements = 'html|body|head|meta|title|link|script|base|!--'; |
|
178 | + $allBoxLikeElements = '(?>'.$trueBoxElements.'|'.$functionalBoxElements.'|'.$usableBoxElements.'|'.$imagineBoxElements.')'; |
|
179 | + $esteticBoxLikeElements = '(?>html|head|body|meta name|title|div|table|h1|h2|h3|h4|h5|h6|p|form|pre|center|!--)'; |
|
180 | + $structureBoxLikeElements = '(?>html|head|body|div|!--)'; |
|
181 | + |
|
182 | + // split html into it's elements |
|
183 | + $htmlArrayTemp = preg_split( |
|
184 | + '/(<(?:[^<>]+(?:"[^"]*"|\'[^\']*\')?)+>)/', |
|
185 | + $html, |
|
186 | + -1, |
|
187 | + \PREG_SPLIT_DELIM_CAPTURE | \PREG_SPLIT_NO_EMPTY |
|
188 | + ); |
|
189 | + |
|
190 | + if (false === $htmlArrayTemp) { |
|
191 | + // Restore saved comments, styles and scripts |
|
192 | + for ($i = 0; $i < \count($noFormat); ++$i) { |
|
193 | + $html = str_replace("<!-- ELEMENT {$i} -->", $noFormat[$i], $html); |
|
194 | + } |
|
195 | + |
|
196 | + return $html; |
|
197 | + } |
|
198 | + // remove empty lines |
|
199 | + $htmlArray = ['']; |
|
200 | + $index = 1; |
|
201 | + for ($x = 0; $x < \count($htmlArrayTemp); ++$x) { |
|
202 | + $text = trim($htmlArrayTemp[$x]); |
|
203 | + $htmlArray[$index] = '' !== $text ? $htmlArrayTemp[$x] : $this->emptySpaceChar; |
|
204 | + ++$index; |
|
205 | + } |
|
206 | + |
|
207 | + // rebuild html |
|
208 | + $html = ''; |
|
209 | + $tabs = 0; |
|
210 | + for ($x = 0; $x < \count($htmlArray); ++$x) { |
|
211 | + $htmlArrayBefore = $htmlArray[$x - 1] ?? ''; |
|
212 | + $htmlArrayCurrent = $htmlArray[$x] ?? ''; |
|
213 | + |
|
214 | + // check if the element should stand in a new line |
|
215 | + $newline = false; |
|
216 | + if ('<?xml' == substr($htmlArrayBefore, 0, 5)) { |
|
217 | + $newline = true; |
|
218 | + } elseif (2 == $this->formatType && ( // minimalistic line break |
|
219 | + // this element has a line break before itself |
|
220 | + preg_match( |
|
221 | + '/<'.$structureBoxLikeElements.'(.*)>/Usi', |
|
222 | + $htmlArrayCurrent |
|
223 | + ) || preg_match( |
|
224 | + '/<'.$structureBoxLikeElements.'(.*) \/>/Usi', |
|
225 | + $htmlArrayCurrent |
|
226 | + ) // one element before is a element that has a line break after |
|
227 | + || preg_match( |
|
228 | + '/<\/'.$structureBoxLikeElements.'(.*)>/Usi', |
|
229 | + $htmlArrayBefore |
|
230 | + ) || '<!--' == substr( |
|
231 | + $htmlArrayBefore, |
|
232 | + 0, |
|
233 | + 4 |
|
234 | + ) || preg_match('/<'.$structureBoxLikeElements.'(.*) \/>/Usi', $htmlArrayBefore)) |
|
235 | + ) { |
|
236 | + $newline = true; |
|
237 | + } elseif (3 == $this->formatType && ( // aestetic line break |
|
238 | + // this element has a line break before itself |
|
239 | + preg_match( |
|
240 | + '/<'.$esteticBoxLikeElements.'(.*)>/Usi', |
|
241 | + $htmlArrayCurrent |
|
242 | + ) || preg_match( |
|
243 | + '/<'.$esteticBoxLikeElements.'(.*) \/>/Usi', |
|
244 | + $htmlArrayCurrent |
|
245 | + ) // one element before is a element that has a line break after |
|
246 | + || preg_match('/<\/'.$esteticBoxLikeElements.'(.*)>/Usi', $htmlArrayBefore) || '<!--' == substr( |
|
247 | + $htmlArrayBefore, |
|
248 | + 0, |
|
249 | + 4 |
|
250 | + ) || preg_match('/<'.$esteticBoxLikeElements.'(.*) \/>/Usi', $htmlArrayBefore)) |
|
251 | + ) { |
|
252 | + $newline = true; |
|
253 | + } elseif ($this->formatType >= 4 && ( // logical line break |
|
254 | + // this element has a line break before itself |
|
255 | + preg_match( |
|
256 | + '/<'.$allBoxLikeElements.'(.*)>/Usi', |
|
257 | + $htmlArrayCurrent |
|
258 | + ) || preg_match( |
|
259 | + '/<'.$allBoxLikeElements.'(.*) \/>/Usi', |
|
260 | + $htmlArrayCurrent |
|
261 | + ) // one element before is a element that has a line break after |
|
262 | + || preg_match('/<\/'.$allBoxLikeElements.'(.*)>/Usi', $htmlArrayBefore) || '<!--' == substr( |
|
263 | + $htmlArrayBefore, |
|
264 | + 0, |
|
265 | + 4 |
|
266 | + ) || preg_match('/<'.$allBoxLikeElements.'(.*) \/>/Usi', $htmlArrayBefore)) |
|
267 | + ) { |
|
268 | + $newline = true; |
|
269 | + } |
|
270 | + |
|
271 | + // count down a tab |
|
272 | + if ('</' == substr($htmlArrayCurrent, 0, 2)) { |
|
273 | + --$tabs; |
|
274 | + } |
|
275 | + |
|
276 | + // add tabs and line breaks in front of the current tag |
|
277 | + if ($newline) { |
|
278 | + $html .= $this->newline; |
|
279 | + for ($y = 0; $y < $tabs; ++$y) { |
|
280 | + $html .= $this->tab; |
|
281 | + } |
|
282 | + } |
|
283 | + |
|
284 | + // remove white spaces and line breaks and add current tag to the html-string |
|
285 | + if ('<![CDATA[' == substr($htmlArrayCurrent, 0, 9) // remove multiple white space in CDATA / XML |
|
286 | + || '<?xml' == substr($htmlArrayCurrent, 0, 5) |
|
287 | + ) { |
|
288 | + $html .= $this->killWhiteSpace($htmlArrayCurrent); |
|
289 | + } else { // remove all line breaks |
|
290 | + $html .= $this->killLineBreaks($htmlArrayCurrent); |
|
291 | + } |
|
292 | + |
|
293 | + // count up a tab |
|
294 | + if ('<' == substr($htmlArrayCurrent, 0, 1) && '/' != substr($htmlArrayCurrent, 1, 1)) { |
|
295 | + if (' ' !== substr($htmlArrayCurrent, 1, 1) |
|
296 | + && 'img' !== substr($htmlArrayCurrent, 1, 3) |
|
297 | + && 'source' !== substr($htmlArrayCurrent, 1, 6) |
|
298 | + && 'br' !== substr($htmlArrayCurrent, 1, 2) |
|
299 | + && 'hr' !== substr($htmlArrayCurrent, 1, 2) |
|
300 | + && 'input' !== substr($htmlArrayCurrent, 1, 5) |
|
301 | + && 'link' !== substr($htmlArrayCurrent, 1, 4) |
|
302 | + && 'meta' !== substr($htmlArrayCurrent, 1, 4) |
|
303 | + && 'col ' !== substr($htmlArrayCurrent, 1, 4) |
|
304 | + && 'frame' !== substr($htmlArrayCurrent, 1, 5) |
|
305 | + && 'isindex' !== substr($htmlArrayCurrent, 1, 7) |
|
306 | + && 'param' !== substr($htmlArrayCurrent, 1, 5) |
|
307 | + && 'area' !== substr($htmlArrayCurrent, 1, 4) |
|
308 | + && 'base' !== substr($htmlArrayCurrent, 1, 4) |
|
309 | + && '<!' !== substr($htmlArrayCurrent, 0, 2) |
|
310 | + && '<?xml' !== substr($htmlArrayCurrent, 0, 5) |
|
311 | + ) { |
|
312 | + ++$tabs; |
|
313 | + } |
|
314 | + } |
|
315 | + } |
|
316 | + |
|
317 | + // Remove empty lines |
|
318 | + if ($this->formatType > 1) { |
|
319 | + $this->removeEmptyLines($html); |
|
320 | + } |
|
321 | + |
|
322 | + // Restore saved comments, styles and scripts |
|
323 | + for ($i = 0; $i < \count($noFormat); ++$i) { |
|
324 | + $html = str_replace("<!-- ELEMENT {$i} -->", $noFormat[$i], $html); |
|
325 | + } |
|
326 | + |
|
327 | + // include debug comment at the end |
|
328 | + if (0 != $tabs && true === $this->debugComment) { |
|
329 | + $html .= "<!-- {$tabs} open elements found -->"; |
|
330 | + } |
|
331 | + |
|
332 | + return $html; |
|
333 | + } |
|
334 | + |
|
335 | + /** |
|
336 | + * Remove ALL line breaks and multiple white space. |
|
337 | + * |
|
338 | + * @param string $html |
|
339 | + * |
|
340 | + * @return string |
|
341 | + */ |
|
342 | + protected function killLineBreaks($html) |
|
343 | + { |
|
344 | + $html = str_replace($this->newline, '', $html); |
|
345 | + |
|
346 | + return preg_replace('/\s\s+/u', ' ', $html); |
|
347 | + // ? return preg_replace('/\n|\s+(\s)/u', '$1', $html); |
|
348 | + } |
|
349 | + |
|
350 | + /** |
|
351 | + * Remove multiple white space, keeps line breaks. |
|
352 | + */ |
|
353 | + protected function killWhiteSpace(string $html): string |
|
354 | + { |
|
355 | + $temp = explode($this->newline, $html); |
|
356 | + for ($i = 0; $i < \count($temp); ++$i) { |
|
357 | + if (!trim($temp[$i])) { |
|
358 | + unset($temp[$i]); |
|
359 | + continue; |
|
360 | + } |
|
361 | + |
|
362 | + $temp[$i] = trim($temp[$i]); |
|
363 | + $temp[$i] = preg_replace('/\s\s+/', ' ', $temp[$i]); |
|
364 | + } |
|
365 | + |
|
366 | + return implode($this->newline, $temp); |
|
367 | + } |
|
368 | + |
|
369 | + /** |
|
370 | + * Remove white space at the end of lines, keeps other white space and line breaks. |
|
371 | + */ |
|
372 | + protected function rTrimLines(string &$html): void |
|
373 | + { |
|
374 | + $html = preg_replace('/\s+$/m', '', $html); |
|
375 | + } |
|
376 | + |
|
377 | + /** |
|
378 | + * Convert newlines according to the current OS. |
|
379 | + */ |
|
380 | + protected function convNlOs(string &$html): void |
|
381 | + { |
|
382 | + $html = preg_replace("(\r\n|\r)", $this->newline, $html); |
|
383 | + } |
|
384 | + |
|
385 | + /** |
|
386 | + * Remove empty lines. |
|
387 | + */ |
|
388 | + protected function removeEmptyLines(string &$html): void |
|
389 | + { |
|
390 | + $temp = explode($this->newline, $html); |
|
391 | + $result = []; |
|
392 | + for ($i = 0; $i < \count($temp); ++$i) { |
|
393 | + if ('' == trim($temp[$i])) { |
|
394 | + continue; |
|
395 | + } |
|
396 | + $result[] = $temp[$i]; |
|
397 | + } |
|
398 | + $html = implode($this->newline, $result); |
|
399 | + } |
|
400 | + |
|
401 | + /** |
|
402 | + * Include configured header comment in HTML content block. |
|
403 | + */ |
|
404 | + public function includeHeaderComment(string &$html): void |
|
405 | + { |
|
406 | + $html = preg_replace('/^(-->)$/m', "\n\t".$this->headerComment."\n$1", $html); |
|
407 | + } |
|
408 | 408 | } |
@@ -131,17 +131,17 @@ discard block |
||
131 | 131 | if ($this->formatType > 0 && is_string($html)) { |
132 | 132 | $html = $this->formatHtml($html); |
133 | 133 | } |
134 | - if ( is_string($html)) { |
|
134 | + if (is_string($html)) { |
|
135 | 135 | // remove white space after line ending |
136 | 136 | $this->rTrimLines($html); |
137 | 137 | } |
138 | 138 | |
139 | 139 | // recover line-breaks |
140 | - if (Environment::isWindows() && is_string($html) ) { |
|
140 | + if (Environment::isWindows() && is_string($html)) { |
|
141 | 141 | $html = str_replace($this->newline, "\r\n", $html); |
142 | 142 | } |
143 | 143 | |
144 | - return (string)$html; |
|
144 | + return (string) $html; |
|
145 | 145 | } |
146 | 146 | |
147 | 147 | /** |
@@ -184,7 +184,7 @@ discard block |
||
184 | 184 | '/(<(?:[^<>]+(?:"[^"]*"|\'[^\']*\')?)+>)/', |
185 | 185 | $html, |
186 | 186 | -1, |
187 | - \PREG_SPLIT_DELIM_CAPTURE | \PREG_SPLIT_NO_EMPTY |
|
187 | + \PREG_SPLIT_DELIM_CAPTURE|\PREG_SPLIT_NO_EMPTY |
|
188 | 188 | ); |
189 | 189 | |
190 | 190 | if (false === $htmlArrayTemp) { |