1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* PHPTAL templating engine |
4
|
|
|
* |
5
|
|
|
* PHP Version 5 |
6
|
|
|
* |
7
|
|
|
* @category HTML |
8
|
|
|
* @package PHPTAL |
9
|
|
|
* @author Kornel Lesiński <[email protected]> |
10
|
|
|
* @license http://www.gnu.org/licenses/lgpl.html GNU Lesser General Public License |
11
|
|
|
* @version SVN: $Id: $ |
12
|
|
|
* @link http://phptal.org/ |
13
|
|
|
*/ |
14
|
|
|
|
15
|
|
|
/** |
16
|
|
|
* Removes all unnecessary whitespace from XHTML documents. |
17
|
|
|
* |
18
|
|
|
* extends Normalize only to re-use helper methods |
19
|
|
|
*/ |
20
|
|
|
class PHPTAL_PreFilter_Compress extends PHPTAL_PreFilter_Normalize |
21
|
|
|
{ |
22
|
|
|
/** |
23
|
|
|
* keeps track whether last element had trailing whitespace (or didn't need it). |
24
|
|
|
* If had_space==false, next element must keep leading space. |
25
|
|
|
*/ |
26
|
|
|
private $had_space=false; |
27
|
|
|
|
28
|
|
|
/** |
29
|
|
|
* last text node before closing tag that may need trailing whitespace trimmed. |
30
|
|
|
* It's often last-child, but comments, multiple end tags make that trickier. |
31
|
|
|
*/ |
32
|
|
|
private $most_recent_text_node=null; |
33
|
|
|
|
34
|
|
|
function filterDOM(PHPTAL_Dom_Element $root) |
|
|
|
|
35
|
|
|
{ |
36
|
|
|
// let xml:space=preserve preserve everything |
37
|
|
|
if ($root->getAttributeNS("http://www.w3.org/XML/1998/namespace", 'space') == 'preserve') { |
38
|
|
|
$this->most_recent_text_node = null; |
39
|
|
|
$this->findElementToFilter($root); |
40
|
|
|
return; |
41
|
|
|
} |
42
|
|
|
|
43
|
|
|
// tal:replace makes element behave like text |
44
|
|
|
if ($root->getAttributeNS('http://xml.zope.org/namespaces/tal','replace')) { |
45
|
|
|
$this->most_recent_text_node = null; |
46
|
|
|
$this->had_space = false; |
47
|
|
|
return; |
48
|
|
|
} |
49
|
|
|
|
50
|
|
|
$this->normalizeAttributes($root); |
51
|
|
|
$this->elementSpecificOptimizations($root); |
52
|
|
|
|
53
|
|
|
// <head>, <tr> don't have any significant whitespace |
54
|
|
|
$no_spaces = $this->hasNoInterelementSpace($root); |
55
|
|
|
|
56
|
|
|
// mostly block-level elements |
57
|
|
|
// if element is conditional, it may not always break the line |
58
|
|
|
$breaks_line = $no_spaces || ($this->breaksLine($root) && !$root->getAttributeNS('http://xml.zope.org/namespaces/tal','condition')); |
59
|
|
|
|
60
|
|
|
// start tag newline |
61
|
|
|
if ($breaks_line) { |
62
|
|
View Code Duplication |
if ($this->most_recent_text_node) { |
|
|
|
|
63
|
|
|
$this->most_recent_text_node->setValueEscaped(rtrim($this->most_recent_text_node->getValueEscaped())); |
64
|
|
|
$this->most_recent_text_node = null; |
65
|
|
|
} |
66
|
|
|
$this->had_space = true; |
67
|
|
|
} else if ($this->isInlineBlock($root)) { |
68
|
|
|
// spaces around <img> must be kept |
69
|
|
|
$this->most_recent_text_node = null; |
70
|
|
|
$this->had_space = false; |
71
|
|
|
} |
72
|
|
|
|
73
|
|
|
// <pre>, <textarea> are handled separately from xml:space, because they may have attributes normalized |
74
|
|
|
if ($this->isSpaceSensitiveInXHTML($root)) { |
75
|
|
|
$this->most_recent_text_node = null; |
76
|
|
|
|
77
|
|
|
// HTML 5 (9.1.2.5) specifies quirk that a first *single* newline in <pre> can be removed |
78
|
|
|
if (count($root->childNodes) && $root->childNodes[0] instanceof PHPTAL_Dom_Text) { |
79
|
|
|
if (preg_match('/^\n[^\n]/', $root->childNodes[0]->getValueEscaped())) { |
80
|
|
|
$root->childNodes[0]->setValueEscaped(substr($root->childNodes[0]->getValueEscaped(),1)); |
81
|
|
|
} |
82
|
|
|
} |
83
|
|
|
$this->findElementToFilter($root); |
84
|
|
|
return; |
85
|
|
|
} |
86
|
|
|
|
87
|
|
|
foreach ($root->childNodes as $node) { |
88
|
|
|
|
89
|
|
|
if ($node instanceof PHPTAL_Dom_Text) { |
90
|
|
|
// replaces runs of whitespace with ' ' |
91
|
|
|
$norm = $this->normalizeSpace($node->getValueEscaped(), $node->getEncoding()); |
92
|
|
|
|
93
|
|
|
if ($no_spaces) { |
94
|
|
|
$norm = trim($norm); |
95
|
|
|
} elseif ($this->had_space) { |
96
|
|
|
$norm = ltrim($norm); |
97
|
|
|
} |
98
|
|
|
|
99
|
|
|
$node->setValueEscaped($norm); |
100
|
|
|
|
101
|
|
|
// collapsed whitespace-only nodes are ignored (otherwise trimming of most_recent_text_node would be useless) |
102
|
|
|
if ($norm !== '') { |
103
|
|
|
$this->most_recent_text_node = $node; |
104
|
|
|
$this->had_space = (substr($norm,-1) == ' '); |
105
|
|
|
} |
106
|
|
|
} else if ($node instanceof PHPTAL_Dom_Element) { |
107
|
|
|
$this->filterDOM($node); |
108
|
|
|
} else if ($node instanceof PHPTAL_Dom_DocumentType || $node instanceof PHPTAL_Dom_XMLDeclaration) { |
109
|
|
|
$this->had_space = true; |
110
|
|
|
} else if ($node instanceof PHPTAL_Dom_ProcessingInstruction) { |
111
|
|
|
// PI may output something requiring spaces |
112
|
|
|
$this->most_recent_text_node = null; |
113
|
|
|
$this->had_space = false; |
114
|
|
|
} |
115
|
|
|
} |
116
|
|
|
|
117
|
|
|
// repeated element may need trailing space. |
118
|
|
|
if (!$breaks_line && $root->getAttributeNS('http://xml.zope.org/namespaces/tal','repeat')) { |
119
|
|
|
$this->most_recent_text_node = null; |
120
|
|
|
} |
121
|
|
|
|
122
|
|
|
// tal:content may replace element with something without space |
123
|
|
|
if (!$breaks_line && $root->getAttributeNS('http://xml.zope.org/namespaces/tal','content')) { |
124
|
|
|
$this->had_space = false; |
125
|
|
|
$this->most_recent_text_node = null; |
126
|
|
|
} |
127
|
|
|
|
128
|
|
|
// line break caused by end tag |
129
|
|
View Code Duplication |
if ($breaks_line) { |
|
|
|
|
130
|
|
|
if ($this->most_recent_text_node) { |
131
|
|
|
$this->most_recent_text_node->setValueEscaped(rtrim($this->most_recent_text_node->getValueEscaped())); |
132
|
|
|
$this->most_recent_text_node = null; |
133
|
|
|
} |
134
|
|
|
$this->had_space = true; |
135
|
|
|
} |
136
|
|
|
} |
137
|
|
|
|
138
|
|
|
private static $no_interelement_space = array( |
139
|
|
|
'html','head','table','thead','tfoot','select','optgroup','dl','ol','ul','tr','datalist', |
140
|
|
|
); |
141
|
|
|
|
142
|
|
|
private function hasNoInterelementSpace(PHPTAL_Dom_Element $element) |
143
|
|
|
{ |
144
|
|
|
if ($element->getLocalName() === 'block' |
145
|
|
|
&& $element->parentNode |
146
|
|
|
&& $element->getNamespaceURI() === 'http://xml.zope.org/namespaces/tal') { |
147
|
|
|
return $this->hasNoInterelementSpace($element->parentNode); |
148
|
|
|
} |
149
|
|
|
|
150
|
|
|
return in_array($element->getLocalName(), self::$no_interelement_space) |
151
|
|
|
&& ($element->getNamespaceURI() === 'http://www.w3.org/1999/xhtml' || $element->getNamespaceURI() === ''); |
152
|
|
|
} |
153
|
|
|
|
154
|
|
|
/** |
155
|
|
|
* li is deliberately omitted, as it's commonly used with display:inline in menus. |
156
|
|
|
*/ |
157
|
|
|
private static $breaks_line = array( |
158
|
|
|
'address','article','aside','base','blockquote','body','br','dd','div','dl','dt','fieldset','figure', |
159
|
|
|
'footer','form','h1','h2','h3','h4','h5','h6','head','header','hgroup','hr','html','legend','link', |
160
|
|
|
'meta','nav','ol','option','p','param','pre','section','style','table','tbody','td','th','thead', |
161
|
|
|
'title','tr','ul','details', |
162
|
|
|
); |
163
|
|
|
|
164
|
|
|
private function breaksLine(PHPTAL_Dom_Element $element) |
165
|
|
|
{ |
166
|
|
|
if ($element->getAttributeNS('http://xml.zope.org/namespaces/metal','define-macro')) { |
167
|
|
|
return true; |
168
|
|
|
} |
169
|
|
|
|
170
|
|
|
if (!$element->parentNode) { |
171
|
|
|
return true; |
172
|
|
|
} |
173
|
|
|
|
174
|
|
|
if ($element->getNamespaceURI() !== 'http://www.w3.org/1999/xhtml' |
175
|
|
|
&& $element->getNamespaceURI() !== '') { |
176
|
|
|
return false; |
177
|
|
|
} |
178
|
|
|
|
179
|
|
|
return in_array($element->getLocalName(), self::$breaks_line); |
180
|
|
|
} |
181
|
|
|
|
182
|
|
|
/** |
183
|
|
|
* replaced elements need to preserve spaces before and after |
184
|
|
|
*/ |
185
|
|
|
private static $inline_blocks = array( |
186
|
|
|
'select','input','button','img','textarea','output','progress','meter', |
187
|
|
|
); |
188
|
|
|
|
189
|
|
|
private function isInlineBlock(PHPTAL_Dom_Element $element) |
190
|
|
|
{ |
191
|
|
|
if ($element->getNamespaceURI() !== 'http://www.w3.org/1999/xhtml' |
192
|
|
|
&& $element->getNamespaceURI() !== '') { |
193
|
|
|
return false; |
194
|
|
|
} |
195
|
|
|
|
196
|
|
|
return in_array($element->getLocalName(), self::$inline_blocks); |
197
|
|
|
} |
198
|
|
|
|
199
|
|
|
/** |
200
|
|
|
* Consistent sorting of attributes might give slightly better gzip performance |
201
|
|
|
*/ |
202
|
|
|
protected function normalizeAttributes(PHPTAL_Dom_Element $element) |
203
|
|
|
{ |
204
|
|
|
parent::normalizeAttributes($element); |
205
|
|
|
|
206
|
|
|
$attrs_by_qname = array(); |
207
|
|
|
foreach ($element->getAttributeNodes() as $attrnode) { |
208
|
|
|
// safe, as there can't be two attrs with same qname |
209
|
|
|
$attrs_by_qname[$attrnode->getQualifiedName()] = $attrnode; |
210
|
|
|
} |
211
|
|
|
|
212
|
|
|
if (count($attrs_by_qname) > 1) { |
213
|
|
|
uksort($attrs_by_qname, array($this, 'compareQNames')); |
214
|
|
|
$element->setAttributeNodes(array_values($attrs_by_qname)); |
215
|
|
|
} |
216
|
|
|
} |
217
|
|
|
|
218
|
|
|
/** |
219
|
|
|
* pre-defined order of attributes roughly by popularity |
220
|
|
|
*/ |
221
|
|
|
private static $attributes_order = array( |
222
|
|
|
'href','src','class','rel','type','title','width','height','alt','content','name','style','lang','id', |
223
|
|
|
); |
224
|
|
|
|
225
|
|
|
/** |
226
|
|
|
* compare names according to $attributes_order array. |
227
|
|
|
* Elements that are not in array, are considered greater than all elements in array, |
228
|
|
|
* and are sorted alphabetically. |
229
|
|
|
*/ |
230
|
|
|
private static function compareQNames($a, $b) { |
231
|
|
|
$a_index = array_search($a, self::$attributes_order); |
232
|
|
|
$b_index = array_search($b, self::$attributes_order); |
233
|
|
|
|
234
|
|
|
if ($a_index !== false && $b_index !== false) { |
235
|
|
|
return $a_index - $b_index; |
236
|
|
|
} |
237
|
|
|
if ($a_index === false && $b_index === false) { |
238
|
|
|
return strcmp($a, $b); |
239
|
|
|
} |
240
|
|
|
return ($a_index === false) ? 1 : -1; |
241
|
|
|
} |
242
|
|
|
|
243
|
|
|
/** |
244
|
|
|
* HTML5 doesn't care about boilerplate |
245
|
|
|
*/ |
246
|
|
|
private function elementSpecificOptimizations(PHPTAL_Dom_Element $element) |
247
|
|
|
{ |
248
|
|
|
if ($element->getNamespaceURI() !== 'http://www.w3.org/1999/xhtml' |
249
|
|
|
&& $element->getNamespaceURI() !== '') { |
250
|
|
|
return; |
251
|
|
|
} |
252
|
|
|
|
253
|
|
|
if ($this->getPHPTAL()->getOutputMode() !== PHPTAL::HTML5) { |
254
|
|
|
return; |
255
|
|
|
} |
256
|
|
|
|
257
|
|
|
// <meta charset> |
258
|
|
|
if ('meta' === $element->getLocalName() && |
259
|
|
|
$element->getAttributeNS('','http-equiv') === 'Content-Type') { |
260
|
|
|
$element->removeAttributeNS('','http-equiv'); |
261
|
|
|
$element->removeAttributeNS('','content'); |
262
|
|
|
$element->setAttributeNS('','charset',strtolower($this->getPHPTAL()->getEncoding())); |
263
|
|
|
} |
264
|
|
|
elseif (('link' === $element->getLocalName() && $element->getAttributeNS('','rel') === 'stylesheet') || |
265
|
|
|
('style' === $element->getLocalName())) { |
266
|
|
|
// There's only one type of stylesheets that works. |
267
|
|
|
$element->removeAttributeNS('','type'); |
268
|
|
|
|
269
|
|
|
} elseif ('script' === $element->getLocalName()) { |
270
|
|
|
$element->removeAttributeNS('','language'); |
271
|
|
|
|
272
|
|
|
// Only remove type that matches default. E4X, vbscript, coffeescript, etc. must be preserved |
273
|
|
|
$type = $element->getAttributeNS('','type'); |
274
|
|
|
$is_std = preg_match('/^(?:text|application)\/(?:ecma|java)script(\s*;\s*charset\s*=\s*[^;]*)?$/', $type); |
275
|
|
|
|
276
|
|
|
// Remote scripts should have type specified in HTTP headers. |
277
|
|
|
if ($is_std || $element->getAttributeNS('','src')) { |
278
|
|
|
$element->removeAttributeNS('','type'); |
279
|
|
|
} |
280
|
|
|
} |
281
|
|
|
} |
282
|
|
|
} |
283
|
|
|
|
Adding explicit visibility (
private
,protected
, orpublic
) is generally recommend to communicate to other developers how, and from where this method is intended to be used.