|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* @package s9e\TextFormatter |
|
5
|
|
|
* @copyright Copyright (c) 2010-2018 The s9e Authors |
|
6
|
|
|
* @license http://www.opensource.org/licenses/mit-license.php The MIT License |
|
7
|
|
|
*/ |
|
8
|
|
|
namespace s9e\TextFormatter\Configurator\Helpers\TemplateParser; |
|
9
|
|
|
|
|
10
|
|
|
use DOMDocument; |
|
11
|
|
|
use DOMElement; |
|
12
|
|
|
use DOMNode; |
|
13
|
|
|
use s9e\TextFormatter\Configurator\Helpers\XPathHelper; |
|
14
|
|
|
|
|
15
|
|
|
class Normalizer extends IRProcessor |
|
16
|
|
|
{ |
|
17
|
|
|
/** |
|
18
|
|
|
* @var Optimizer |
|
19
|
|
|
*/ |
|
20
|
|
|
protected $optimizer; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* @var string Regexp that matches the names of all void elements |
|
24
|
|
|
* @link http://www.w3.org/TR/html-markup/syntax.html#void-elements |
|
25
|
|
|
*/ |
|
26
|
|
|
public $voidRegexp = '/^(?:area|base|br|col|command|embed|hr|img|input|keygen|link|meta|param|source|track|wbr)$/Di'; |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* @param Optimizer $optimizer |
|
30
|
|
|
* @return void |
|
31
|
|
|
*/ |
|
32
|
|
|
public function __construct(Optimizer $optimizer) |
|
33
|
|
|
{ |
|
34
|
|
|
$this->optimizer = $optimizer; |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* Normalize an IR |
|
39
|
|
|
* |
|
40
|
|
|
* @param DOMDocument $ir |
|
41
|
|
|
* @return void |
|
42
|
|
|
*/ |
|
43
|
|
|
public function normalize(DOMDocument $ir) |
|
44
|
|
|
{ |
|
45
|
|
|
$this->createXPath($ir); |
|
46
|
|
|
$this->addDefaultCase($ir); |
|
47
|
|
|
$this->addElementIds($ir); |
|
48
|
|
|
$this->addCloseTagElements($ir); |
|
49
|
|
|
$this->markVoidElements($ir); |
|
50
|
|
|
$this->optimizer->optimize($ir); |
|
51
|
|
|
$this->markConditionalCloseTagElements($ir); |
|
52
|
|
|
$this->setOutputContext($ir); |
|
53
|
|
|
$this->markBranchTables($ir); |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
/** |
|
57
|
|
|
* Add <closeTag/> elements everywhere an open start tag should be closed |
|
58
|
|
|
* |
|
59
|
|
|
* @param DOMDocument $ir |
|
60
|
|
|
* @return void |
|
61
|
|
|
*/ |
|
62
|
|
|
protected function addCloseTagElements(DOMDocument $ir) |
|
63
|
|
|
{ |
|
64
|
|
|
$exprs = [ |
|
65
|
|
|
'//applyTemplates[not(ancestor::attribute)]', |
|
66
|
|
|
'//comment', |
|
67
|
|
|
'//element', |
|
68
|
|
|
'//output[not(ancestor::attribute)]' |
|
69
|
|
|
]; |
|
70
|
|
|
foreach ($this->query(implode('|', $exprs)) as $node) |
|
71
|
|
|
{ |
|
72
|
|
|
$parentElementId = $this->getParentElementId($node); |
|
73
|
|
|
if (isset($parentElementId)) |
|
74
|
|
|
{ |
|
75
|
|
|
$node->parentNode |
|
76
|
|
|
->insertBefore($ir->createElement('closeTag'), $node) |
|
77
|
|
|
->setAttribute('id', $parentElementId); |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
// Append a <closeTag/> to <element/> nodes to ensure that empty elements get closed |
|
81
|
|
|
if ($node->nodeName === 'element') |
|
82
|
|
|
{ |
|
83
|
|
|
$id = $node->getAttribute('id'); |
|
84
|
|
|
$this->appendElement($node, 'closeTag')->setAttribute('id', $id); |
|
85
|
|
|
} |
|
86
|
|
|
} |
|
87
|
|
|
} |
|
88
|
|
|
|
|
89
|
|
|
/** |
|
90
|
|
|
* Add an empty default <case/> to <switch/> nodes that don't have one |
|
91
|
|
|
* |
|
92
|
|
|
* @param DOMDocument $ir |
|
93
|
|
|
* @return void |
|
94
|
|
|
*/ |
|
95
|
|
|
protected function addDefaultCase(DOMDocument $ir) |
|
96
|
|
|
{ |
|
97
|
|
|
foreach ($this->query('//switch[not(case[not(@test)])]') as $switch) |
|
98
|
|
|
{ |
|
99
|
|
|
$this->appendElement($switch, 'case'); |
|
100
|
|
|
} |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
|
|
/** |
|
104
|
|
|
* Add an id attribute to <element/> nodes |
|
105
|
|
|
* |
|
106
|
|
|
* @param DOMDocument $ir |
|
107
|
|
|
* @return void |
|
108
|
|
|
*/ |
|
109
|
|
|
protected function addElementIds(DOMDocument $ir) |
|
110
|
|
|
{ |
|
111
|
|
|
$id = 0; |
|
112
|
|
|
foreach ($ir->getElementsByTagName('element') as $element) |
|
113
|
|
|
{ |
|
114
|
|
|
$element->setAttribute('id', ++$id); |
|
115
|
|
|
} |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
/** |
|
119
|
|
|
* Get the context type for given output element |
|
120
|
|
|
* |
|
121
|
|
|
* @param DOMNode $output |
|
122
|
|
|
* @return string |
|
123
|
|
|
*/ |
|
124
|
|
|
protected function getOutputContext(DOMNode $output) |
|
125
|
|
|
{ |
|
126
|
|
|
$contexts = [ |
|
127
|
|
|
'boolean(ancestor::attribute)' => 'attribute', |
|
128
|
|
|
'@disable-output-escaping="yes"' => 'raw', |
|
129
|
|
|
'count(ancestor::element[@name="script"])' => 'raw' |
|
130
|
|
|
]; |
|
131
|
|
|
foreach ($contexts as $expr => $context) |
|
132
|
|
|
{ |
|
133
|
|
|
if ($this->evaluate($expr, $output)) |
|
134
|
|
|
{ |
|
135
|
|
|
return $context; |
|
136
|
|
|
} |
|
137
|
|
|
} |
|
138
|
|
|
|
|
139
|
|
|
return 'text'; |
|
140
|
|
|
} |
|
141
|
|
|
|
|
142
|
|
|
/** |
|
143
|
|
|
* Get the ID of the closest "element" ancestor |
|
144
|
|
|
* |
|
145
|
|
|
* @param DOMNode $node Context node |
|
146
|
|
|
* @return string|null |
|
147
|
|
|
*/ |
|
148
|
|
|
protected function getParentElementId(DOMNode $node) |
|
149
|
|
|
{ |
|
150
|
|
|
$parentNode = $node->parentNode; |
|
151
|
|
|
while (isset($parentNode)) |
|
152
|
|
|
{ |
|
153
|
|
|
if ($parentNode->nodeName === 'element') |
|
154
|
|
|
{ |
|
155
|
|
|
return $parentNode->getAttribute('id'); |
|
156
|
|
|
} |
|
157
|
|
|
$parentNode = $parentNode->parentNode; |
|
158
|
|
|
} |
|
159
|
|
|
} |
|
160
|
|
|
|
|
161
|
|
|
/** |
|
162
|
|
|
* Mark switch elements that are used as branch tables |
|
163
|
|
|
* |
|
164
|
|
|
* If a switch is used for a series of equality tests against the same attribute or variable, the |
|
165
|
|
|
* attribute/variable is stored within the switch as "branch-key" and the values it is compared |
|
166
|
|
|
* against are stored JSON-encoded in the case as "branch-values". It can be used to create |
|
167
|
|
|
* optimized branch tables |
|
168
|
|
|
* |
|
169
|
|
|
* @param DOMDocument $ir |
|
170
|
|
|
* @return void |
|
171
|
|
|
*/ |
|
172
|
|
|
protected function markBranchTables(DOMDocument $ir) |
|
173
|
|
|
{ |
|
174
|
|
|
// Iterate over switch elements that have at least two case children with a test attribute |
|
175
|
|
|
foreach ($this->query('//switch[case[2][@test]]') as $switch) |
|
176
|
|
|
{ |
|
177
|
|
|
$this->markSwitchTable($switch); |
|
178
|
|
|
} |
|
179
|
|
|
} |
|
180
|
|
|
|
|
181
|
|
|
/** |
|
182
|
|
|
* Mark given switch element if it's used as a branch table |
|
183
|
|
|
* |
|
184
|
|
|
* @param DOMElement $switch |
|
185
|
|
|
* @return void |
|
186
|
|
|
*/ |
|
187
|
|
|
protected function markSwitchTable(DOMElement $switch) |
|
188
|
|
|
{ |
|
189
|
|
|
$branches = []; |
|
190
|
|
|
$values = []; |
|
191
|
|
|
foreach ($this->query('./case[@test]', $switch) as $i => $case) |
|
192
|
|
|
{ |
|
193
|
|
|
$map = XPathHelper::parseEqualityExpr($case->getAttribute('test')); |
|
194
|
|
|
if ($map === false) |
|
195
|
|
|
{ |
|
196
|
|
|
return; |
|
197
|
|
|
} |
|
198
|
|
|
$values += $map; |
|
199
|
|
|
$branches[$i] = end($map); |
|
200
|
|
|
} |
|
201
|
|
|
if (count($values) !== 1) |
|
202
|
|
|
{ |
|
203
|
|
|
return; |
|
204
|
|
|
} |
|
205
|
|
|
|
|
206
|
|
|
$switch->setAttribute('branch-key', key($values)); |
|
207
|
|
|
foreach ($branches as $i => $values) |
|
208
|
|
|
{ |
|
209
|
|
|
sort($values); |
|
210
|
|
|
$switch->childNodes->item($i)->setAttribute('branch-values', serialize($values)); |
|
211
|
|
|
} |
|
212
|
|
|
} |
|
213
|
|
|
|
|
214
|
|
|
/** |
|
215
|
|
|
* Mark conditional <closeTag/> nodes |
|
216
|
|
|
* |
|
217
|
|
|
* @param DOMDocument $ir |
|
218
|
|
|
* @return void |
|
219
|
|
|
*/ |
|
220
|
|
|
protected function markConditionalCloseTagElements(DOMDocument $ir) |
|
221
|
|
|
{ |
|
222
|
|
|
foreach ($ir->getElementsByTagName('closeTag') as $closeTag) |
|
223
|
|
|
{ |
|
224
|
|
|
$id = $closeTag->getAttribute('id'); |
|
225
|
|
|
|
|
226
|
|
|
// For each <switch/> ancestor, look for a <closeTag/> and that is either a sibling or |
|
227
|
|
|
// the descendant of a sibling, and that matches the id |
|
228
|
|
|
$query = 'ancestor::switch/' |
|
229
|
|
|
. 'following-sibling::*/' |
|
230
|
|
|
. 'descendant-or-self::closeTag[@id = "' . $id . '"]'; |
|
231
|
|
|
foreach ($this->query($query, $closeTag) as $following) |
|
232
|
|
|
{ |
|
233
|
|
|
// Mark following <closeTag/> nodes to indicate that the status of this tag must |
|
234
|
|
|
// be checked before it is closed |
|
235
|
|
|
$following->setAttribute('check', ''); |
|
236
|
|
|
|
|
237
|
|
|
// Mark the current <closeTag/> to indicate that it must set a flag to indicate |
|
238
|
|
|
// that its tag has been closed |
|
239
|
|
|
$closeTag->setAttribute('set', ''); |
|
240
|
|
|
} |
|
241
|
|
|
} |
|
242
|
|
|
} |
|
243
|
|
|
|
|
244
|
|
|
/** |
|
245
|
|
|
* Mark void elements |
|
246
|
|
|
* |
|
247
|
|
|
* @param DOMDocument $ir |
|
248
|
|
|
* @return void |
|
249
|
|
|
*/ |
|
250
|
|
|
protected function markVoidElements(DOMDocument $ir) |
|
251
|
|
|
{ |
|
252
|
|
|
foreach ($ir->getElementsByTagName('element') as $element) |
|
253
|
|
|
{ |
|
254
|
|
|
// Test whether this element is (maybe) void |
|
255
|
|
|
$elName = $element->getAttribute('name'); |
|
256
|
|
|
if (strpos($elName, '{') !== false) |
|
257
|
|
|
{ |
|
258
|
|
|
// Dynamic element names must be checked at runtime |
|
259
|
|
|
$element->setAttribute('void', 'maybe'); |
|
260
|
|
|
} |
|
261
|
|
|
elseif (preg_match($this->voidRegexp, $elName)) |
|
262
|
|
|
{ |
|
263
|
|
|
// Static element names can be checked right now |
|
264
|
|
|
$element->setAttribute('void', 'yes'); |
|
265
|
|
|
} |
|
266
|
|
|
} |
|
267
|
|
|
} |
|
268
|
|
|
|
|
269
|
|
|
/** |
|
270
|
|
|
* Fill in output context |
|
271
|
|
|
* |
|
272
|
|
|
* @param DOMDocument $ir |
|
273
|
|
|
* @return void |
|
274
|
|
|
*/ |
|
275
|
|
|
protected function setOutputContext(DOMDocument $ir) |
|
276
|
|
|
{ |
|
277
|
|
|
foreach ($ir->getElementsByTagName('output') as $output) |
|
278
|
|
|
{ |
|
279
|
|
|
$output->setAttribute('escape', $this->getOutputContext($output)); |
|
280
|
|
|
} |
|
281
|
|
|
} |
|
282
|
|
|
} |