|
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\RendererGenerators\PHP; |
|
9
|
|
|
|
|
10
|
|
|
use Closure; |
|
11
|
|
|
use RuntimeException; |
|
12
|
|
|
use s9e\TextFormatter\Configurator\Helpers\RegexpBuilder; |
|
13
|
|
|
|
|
14
|
|
|
class Quick |
|
15
|
|
|
{ |
|
16
|
|
|
/** |
|
17
|
|
|
* Generate the Quick renderer's source |
|
18
|
|
|
* |
|
19
|
|
|
* @param array $compiledTemplates Array of tagName => compiled template |
|
20
|
|
|
* @return string |
|
21
|
|
|
*/ |
|
22
|
|
|
public static function getSource(array $compiledTemplates) |
|
23
|
|
|
{ |
|
24
|
|
|
$map = ['dynamic' => [], 'php' => [], 'static' => []]; |
|
25
|
|
|
$tagNames = []; |
|
26
|
|
|
$unsupported = []; |
|
27
|
|
|
|
|
28
|
|
|
// Ignore system tags |
|
29
|
|
|
unset($compiledTemplates['br']); |
|
30
|
|
|
unset($compiledTemplates['e']); |
|
31
|
|
|
unset($compiledTemplates['i']); |
|
32
|
|
|
unset($compiledTemplates['p']); |
|
33
|
|
|
unset($compiledTemplates['s']); |
|
34
|
|
|
|
|
35
|
|
|
foreach ($compiledTemplates as $tagName => $php) |
|
36
|
|
|
{ |
|
37
|
|
|
$renderings = self::getRenderingStrategy($php); |
|
38
|
|
|
if (empty($renderings)) |
|
39
|
|
|
{ |
|
40
|
|
|
$unsupported[] = $tagName; |
|
41
|
|
|
continue; |
|
42
|
|
|
} |
|
43
|
|
|
|
|
44
|
|
|
foreach ($renderings as $i => list($strategy, $replacement)) |
|
45
|
|
|
{ |
|
46
|
|
|
$match = (($i) ? '/' : '') . $tagName; |
|
47
|
|
|
$map[$strategy][$match] = $replacement; |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
// Record the names of tags whose template does not contain a passthrough |
|
51
|
|
|
if (!isset($renderings[1])) |
|
52
|
|
|
{ |
|
53
|
|
|
$tagNames[] = $tagName; |
|
54
|
|
|
} |
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
|
|
$php = []; |
|
58
|
|
|
$php[] = ' /** {@inheritdoc} */'; |
|
59
|
|
|
$php[] = ' public $enableQuickRenderer=true;'; |
|
60
|
|
|
$php[] = ' /** {@inheritdoc} */'; |
|
61
|
|
|
$php[] = ' protected $static=' . self::export($map['static']) . ';'; |
|
62
|
|
|
$php[] = ' /** {@inheritdoc} */'; |
|
63
|
|
|
$php[] = ' protected $dynamic=' . self::export($map['dynamic']) . ';'; |
|
64
|
|
|
|
|
65
|
|
|
$quickSource = ''; |
|
66
|
|
|
if (!empty($map['php'])) |
|
67
|
|
|
{ |
|
68
|
|
|
$quickSource = SwitchStatement::generate('$id', $map['php']); |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
// Build a regexp that matches all the tags |
|
72
|
|
|
$regexp = '(<(?:(?!/)('; |
|
73
|
|
|
$regexp .= ($tagNames) ? RegexpBuilder::fromList($tagNames) : '(?!)'; |
|
74
|
|
|
$regexp .= ')(?: [^>]*)?>.*?</\\1|(/?(?!br/|p>)[^ />]+)[^>]*?(/)?)>)s'; |
|
75
|
|
|
$php[] = ' /** {@inheritdoc} */'; |
|
76
|
|
|
$php[] = ' protected $quickRegexp=' . var_export($regexp, true) . ';'; |
|
77
|
|
|
|
|
78
|
|
|
// Build a regexp that matches tags that cannot be rendered with the Quick renderer |
|
79
|
|
|
if (!empty($unsupported)) |
|
80
|
|
|
{ |
|
81
|
|
|
$regexp = '(<(?:[!?]|' . RegexpBuilder::fromList($unsupported) . '[ />]))'; |
|
82
|
|
|
$php[] = ' /** {@inheritdoc} */'; |
|
83
|
|
|
$php[] = ' protected $quickRenderingTest=' . var_export($regexp, true) . ';'; |
|
84
|
|
|
} |
|
85
|
|
|
|
|
86
|
|
|
$php[] = ' /** {@inheritdoc} */'; |
|
87
|
|
|
$php[] = ' protected function renderQuickTemplate($id, $xml)'; |
|
88
|
|
|
$php[] = ' {'; |
|
89
|
|
|
$php[] = ' $attributes=$this->matchAttributes($xml);'; |
|
90
|
|
|
$php[] = " \$html='';" . $quickSource; |
|
91
|
|
|
$php[] = ''; |
|
92
|
|
|
$php[] = ' return $html;'; |
|
93
|
|
|
$php[] = ' }'; |
|
94
|
|
|
|
|
95
|
|
|
return implode("\n", $php); |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
/** |
|
99
|
|
|
* Export an array as PHP |
|
100
|
|
|
* |
|
101
|
|
|
* @param array $arr |
|
102
|
|
|
* @return string |
|
103
|
|
|
*/ |
|
104
|
|
|
protected static function export(array $arr) |
|
105
|
|
|
{ |
|
106
|
|
|
$exportKeys = (array_keys($arr) !== range(0, count($arr) - 1)); |
|
107
|
|
|
ksort($arr); |
|
108
|
|
|
|
|
109
|
|
|
$entries = []; |
|
110
|
|
|
foreach ($arr as $k => $v) |
|
111
|
|
|
{ |
|
112
|
|
|
$entries[] = (($exportKeys) ? var_export($k, true) . '=>' : '') |
|
113
|
|
|
. ((is_array($v)) ? self::export($v) : var_export($v, true)); |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
return '[' . implode(',', $entries) . ']'; |
|
117
|
|
|
} |
|
118
|
|
|
|
|
119
|
|
|
/** |
|
120
|
|
|
* Compute the rendering strategy for a compiled template |
|
121
|
|
|
* |
|
122
|
|
|
* @param string $php Template compiled for the PHP renderer |
|
123
|
|
|
* @return array[] An array containing 0 to 2 pairs of [<rendering type>, <replacement>] |
|
124
|
|
|
*/ |
|
125
|
|
|
public static function getRenderingStrategy($php) |
|
126
|
|
|
{ |
|
127
|
|
|
$renderings = self::getStringRenderings($php); |
|
128
|
|
|
|
|
129
|
|
|
// Keep string rendering where possible, use PHP rendering wherever else |
|
130
|
|
|
foreach (self::getQuickRendering($php) as $i => $phpRendering) |
|
131
|
|
|
{ |
|
132
|
|
|
if (!isset($renderings[$i]) || strpos($phpRendering, '$this->attributes[]') !== false) |
|
133
|
|
|
{ |
|
134
|
|
|
$renderings[$i] = ['php', $phpRendering]; |
|
135
|
|
|
} |
|
136
|
|
|
} |
|
137
|
|
|
|
|
138
|
|
|
return $renderings; |
|
139
|
|
|
} |
|
140
|
|
|
|
|
141
|
|
|
/** |
|
142
|
|
|
* Generate the code for rendering a compiled template with the Quick renderer |
|
143
|
|
|
* |
|
144
|
|
|
* Parse and record every code path that contains a passthrough. Parse every if-else structure. |
|
145
|
|
|
* When the whole structure is parsed, there are 2 possible situations: |
|
146
|
|
|
* - no code path contains a passthrough, in which case we discard the data |
|
147
|
|
|
* - all the code paths including the mandatory "else" branch contain a passthrough, in which |
|
148
|
|
|
* case we keep the data |
|
149
|
|
|
* |
|
150
|
|
|
* @param string $php Template compiled for the PHP renderer |
|
151
|
|
|
* @return string[] An array containing one or two strings of PHP, or an empty array |
|
152
|
|
|
* if the PHP cannot be converted |
|
153
|
|
|
*/ |
|
154
|
|
|
protected static function getQuickRendering($php) |
|
155
|
|
|
{ |
|
156
|
|
|
// xsl:apply-templates elements with a select expression are not supported |
|
157
|
|
|
if (preg_match('(\\$this->at\\((?!\\$node\\);))', $php)) |
|
158
|
|
|
{ |
|
159
|
|
|
return []; |
|
160
|
|
|
} |
|
161
|
|
|
|
|
162
|
|
|
// Tokenize the PHP and add an empty token as terminator |
|
163
|
|
|
$tokens = token_get_all('<?php ' . $php); |
|
164
|
|
|
$tokens[] = [0, '']; |
|
165
|
|
|
|
|
166
|
|
|
// Remove the first token, which is a T_OPEN_TAG |
|
167
|
|
|
array_shift($tokens); |
|
168
|
|
|
$cnt = count($tokens); |
|
169
|
|
|
|
|
170
|
|
|
// Prepare the main branch |
|
171
|
|
|
$branch = [ |
|
172
|
|
|
// We purposefully use a value that can never match |
|
173
|
|
|
'braces' => -1, |
|
174
|
|
|
'branches' => [], |
|
175
|
|
|
'head' => '', |
|
176
|
|
|
'passthrough' => 0, |
|
177
|
|
|
'statement' => '', |
|
178
|
|
|
'tail' => '' |
|
179
|
|
|
]; |
|
180
|
|
|
|
|
181
|
|
|
$braces = 0; |
|
182
|
|
|
$i = 0; |
|
183
|
|
|
do |
|
184
|
|
|
{ |
|
185
|
|
|
// Test whether we've reached a passthrough |
|
186
|
|
|
if ($tokens[$i ][0] === T_VARIABLE |
|
187
|
|
|
&& $tokens[$i ][1] === '$this' |
|
188
|
|
|
&& $tokens[$i + 1][0] === T_OBJECT_OPERATOR |
|
189
|
|
|
&& $tokens[$i + 2][0] === T_STRING |
|
190
|
|
|
&& $tokens[$i + 2][1] === 'at' |
|
191
|
|
|
&& $tokens[$i + 3] === '(' |
|
192
|
|
|
&& $tokens[$i + 4][0] === T_VARIABLE |
|
193
|
|
|
&& $tokens[$i + 4][1] === '$node' |
|
194
|
|
|
&& $tokens[$i + 5] === ')' |
|
195
|
|
|
&& $tokens[$i + 6] === ';') |
|
196
|
|
|
{ |
|
197
|
|
|
if (++$branch['passthrough'] > 1) |
|
198
|
|
|
{ |
|
199
|
|
|
// Multiple passthroughs are not supported |
|
200
|
|
|
return []; |
|
201
|
|
|
} |
|
202
|
|
|
|
|
203
|
|
|
// Skip to the semi-colon |
|
204
|
|
|
$i += 6; |
|
205
|
|
|
|
|
206
|
|
|
continue; |
|
207
|
|
|
} |
|
208
|
|
|
|
|
209
|
|
|
$key = ($branch['passthrough']) ? 'tail' : 'head'; |
|
210
|
|
|
$branch[$key] .= (is_array($tokens[$i])) ? $tokens[$i][1] : $tokens[$i]; |
|
211
|
|
|
|
|
212
|
|
|
if ($tokens[$i] === '{') |
|
213
|
|
|
{ |
|
214
|
|
|
++$braces; |
|
215
|
|
|
continue; |
|
216
|
|
|
} |
|
217
|
|
|
|
|
218
|
|
|
if ($tokens[$i] === '}') |
|
219
|
|
|
{ |
|
220
|
|
|
--$braces; |
|
221
|
|
|
|
|
222
|
|
|
if ($branch['braces'] === $braces) |
|
223
|
|
|
{ |
|
224
|
|
|
// Remove the last brace from the branch's content |
|
225
|
|
|
$branch[$key] = substr($branch[$key], 0, -1); |
|
226
|
|
|
|
|
227
|
|
|
// Jump back to the parent branch |
|
228
|
|
|
$branch =& $branch['parent']; |
|
229
|
|
|
|
|
230
|
|
|
// Copy the current index to look ahead |
|
231
|
|
|
$j = $i; |
|
232
|
|
|
|
|
233
|
|
|
// Skip whitespace |
|
234
|
|
|
while ($tokens[++$j][0] === T_WHITESPACE); |
|
235
|
|
|
|
|
236
|
|
|
// Test whether this is the last brace of an if-else structure by looking for |
|
237
|
|
|
// an additional elseif/else case |
|
238
|
|
|
if ($tokens[$j][0] !== T_ELSEIF && $tokens[$j][0] !== T_ELSE) |
|
239
|
|
|
{ |
|
240
|
|
|
$passthroughs = self::getBranchesPassthrough($branch['branches']); |
|
241
|
|
|
if ($passthroughs === [0]) |
|
242
|
|
|
{ |
|
243
|
|
|
// No branch was passthrough, move their PHP source back to this branch |
|
244
|
|
|
// then discard the data |
|
245
|
|
|
foreach ($branch['branches'] as $child) |
|
246
|
|
|
{ |
|
247
|
|
|
$branch['head'] .= $child['statement'] . '{' . $child['head'] . '}'; |
|
248
|
|
|
} |
|
249
|
|
|
|
|
250
|
|
|
$branch['branches'] = []; |
|
251
|
|
|
continue; |
|
252
|
|
|
} |
|
253
|
|
|
|
|
254
|
|
|
if ($passthroughs === [1]) |
|
255
|
|
|
{ |
|
256
|
|
|
// All branches were passthrough, so their parent is passthrough |
|
257
|
|
|
++$branch['passthrough']; |
|
258
|
|
|
|
|
259
|
|
|
continue; |
|
260
|
|
|
} |
|
261
|
|
|
|
|
262
|
|
|
// Mixed branches (with/out passthrough) are not supported |
|
263
|
|
|
return []; |
|
264
|
|
|
} |
|
265
|
|
|
} |
|
266
|
|
|
|
|
267
|
|
|
continue; |
|
268
|
|
|
} |
|
269
|
|
|
|
|
270
|
|
|
// We don't have to record child branches if we know that current branch is passthrough. |
|
271
|
|
|
// If a child branch contains a passthrough, it will be treated as a multiple |
|
272
|
|
|
// passthrough and we will abort |
|
273
|
|
|
if ($branch['passthrough']) |
|
274
|
|
|
{ |
|
275
|
|
|
continue; |
|
276
|
|
|
} |
|
277
|
|
|
|
|
278
|
|
|
if ($tokens[$i][0] === T_IF |
|
279
|
|
|
|| $tokens[$i][0] === T_ELSEIF |
|
280
|
|
|
|| $tokens[$i][0] === T_ELSE) |
|
281
|
|
|
{ |
|
282
|
|
|
// Remove the statement from the branch's content |
|
283
|
|
|
$branch[$key] = substr($branch[$key], 0, -strlen($tokens[$i][1])); |
|
284
|
|
|
|
|
285
|
|
|
// Create a new branch |
|
286
|
|
|
$branch['branches'][] = [ |
|
287
|
|
|
'braces' => $braces, |
|
288
|
|
|
'branches' => [], |
|
289
|
|
|
'head' => '', |
|
290
|
|
|
'parent' => &$branch, |
|
291
|
|
|
'passthrough' => 0, |
|
292
|
|
|
'statement' => '', |
|
293
|
|
|
'tail' => '' |
|
294
|
|
|
]; |
|
295
|
|
|
|
|
296
|
|
|
// Jump to the new branch |
|
297
|
|
|
$branch =& $branch['branches'][count($branch['branches']) - 1]; |
|
298
|
|
|
|
|
299
|
|
|
// Record the PHP statement |
|
300
|
|
|
do |
|
301
|
|
|
{ |
|
302
|
|
|
$branch['statement'] .= (is_array($tokens[$i])) ? $tokens[$i][1] : $tokens[$i]; |
|
303
|
|
|
} |
|
304
|
|
|
while ($tokens[++$i] !== '{'); |
|
305
|
|
|
|
|
306
|
|
|
// Account for the brace in the statement |
|
307
|
|
|
++$braces; |
|
308
|
|
|
} |
|
309
|
|
|
} |
|
310
|
|
|
while (++$i < $cnt); |
|
311
|
|
|
|
|
312
|
|
|
list($head, $tail) = self::buildPHP($branch['branches']); |
|
313
|
|
|
$head = $branch['head'] . $head; |
|
314
|
|
|
$tail .= $branch['tail']; |
|
315
|
|
|
|
|
316
|
|
|
// Convert the PHP renderer source to the format used in the Quick renderer |
|
317
|
|
|
self::convertPHP($head, $tail, (bool) $branch['passthrough']); |
|
318
|
|
|
|
|
319
|
|
|
// Test whether any method call was left unconverted. If so, we cannot render this template |
|
320
|
|
|
if (preg_match('((?<!-|\\$this)->)', $head . $tail)) |
|
321
|
|
|
{ |
|
322
|
|
|
return []; |
|
323
|
|
|
} |
|
324
|
|
|
|
|
325
|
|
|
return ($branch['passthrough']) ? [$head, $tail] : [$head]; |
|
326
|
|
|
} |
|
327
|
|
|
|
|
328
|
|
|
/** |
|
329
|
|
|
* Convert the two sides of a compiled template to quick rendering |
|
330
|
|
|
* |
|
331
|
|
|
* @param string &$head |
|
332
|
|
|
* @param string &$tail |
|
333
|
|
|
* @param bool $passthrough |
|
334
|
|
|
* @return void |
|
335
|
|
|
*/ |
|
336
|
|
|
protected static function convertPHP(&$head, &$tail, $passthrough) |
|
337
|
|
|
{ |
|
338
|
|
|
// Test whether the attributes must be saved when rendering the head because they're needed |
|
339
|
|
|
// when rendering the tail |
|
340
|
|
|
$saveAttributes = (bool) preg_match('(\\$node->(?:get|has)Attribute)', $tail); |
|
341
|
|
|
|
|
342
|
|
|
// Collect the names of all the attributes so that we can initialize them with a null value |
|
343
|
|
|
// to avoid undefined variable notices. We exclude attributes that seem to be in an if block |
|
344
|
|
|
// that tests its existence beforehand. This last part is not an accurate process as it |
|
345
|
|
|
// would be much more expensive to do it accurately but where it fails the only consequence |
|
346
|
|
|
// is we needlessly add the attribute to the list. There is no difference in functionality |
|
347
|
|
|
preg_match_all( |
|
348
|
|
|
"(\\\$node->getAttribute\\('([^']+)'\\))", |
|
349
|
|
|
preg_replace_callback( |
|
350
|
|
|
'(if\\(\\$node->hasAttribute\\(([^\\)]+)[^}]+)', |
|
351
|
|
|
function ($m) |
|
352
|
|
|
{ |
|
353
|
|
|
return str_replace('$node->getAttribute(' . $m[1] . ')', '', $m[0]); |
|
354
|
|
|
}, |
|
355
|
|
|
$head . $tail |
|
356
|
|
|
), |
|
357
|
|
|
$matches |
|
358
|
|
|
); |
|
359
|
|
|
$attrNames = array_unique($matches[1]); |
|
360
|
|
|
|
|
361
|
|
|
// Replace the source in $head and $tail |
|
362
|
|
|
self::replacePHP($head); |
|
363
|
|
|
self::replacePHP($tail); |
|
364
|
|
|
|
|
365
|
|
|
if (!$passthrough && strpos($head, '$node->textContent') !== false) |
|
366
|
|
|
{ |
|
367
|
|
|
$head = '$textContent=$this->getQuickTextContent($xml);' . str_replace('$node->textContent', '$textContent', $head); |
|
368
|
|
|
} |
|
369
|
|
|
|
|
370
|
|
|
if (!empty($attrNames)) |
|
371
|
|
|
{ |
|
372
|
|
|
ksort($attrNames); |
|
373
|
|
|
$head = "\$attributes+=['" . implode("'=>null,'", $attrNames) . "'=>null];" . $head; |
|
374
|
|
|
} |
|
375
|
|
|
|
|
376
|
|
|
if ($saveAttributes) |
|
377
|
|
|
{ |
|
378
|
|
|
$head .= '$this->attributes[]=$attributes;'; |
|
379
|
|
|
$tail = '$attributes=array_pop($this->attributes);' . $tail; |
|
380
|
|
|
} |
|
381
|
|
|
} |
|
382
|
|
|
|
|
383
|
|
|
/** |
|
384
|
|
|
* Replace the PHP code used in a compiled template to be used by the Quick renderer |
|
385
|
|
|
* |
|
386
|
|
|
* @param string &$php |
|
387
|
|
|
* @return void |
|
388
|
|
|
*/ |
|
389
|
|
|
protected static function replacePHP(&$php) |
|
390
|
|
|
{ |
|
391
|
|
|
// Expression that matches a $node->getAttribute() call and captures its string argument |
|
392
|
|
|
$getAttribute = "\\\$node->getAttribute\\(('[^']+')\\)"; |
|
393
|
|
|
|
|
394
|
|
|
$replacements = [ |
|
395
|
|
|
'$this->out' => '$html', |
|
396
|
|
|
|
|
397
|
|
|
// An attribute value escaped as ENT_NOQUOTES. We only need to unescape quotes |
|
398
|
|
|
'(htmlspecialchars\\(' . $getAttribute . ',' . ENT_NOQUOTES . '\\))' |
|
399
|
|
|
=> "str_replace('"','\"',\$attributes[\$1])", |
|
400
|
|
|
|
|
401
|
|
|
// One or several attribute values escaped as ENT_COMPAT can be used as-is |
|
402
|
|
|
'(htmlspecialchars\\((' . $getAttribute . '(?:\\.' . $getAttribute . ')*),' . ENT_COMPAT . '\\))' |
|
403
|
|
|
=> function ($m) use ($getAttribute) |
|
404
|
|
|
{ |
|
405
|
|
|
return preg_replace('(' . $getAttribute . ')', '$attributes[$1]', $m[1]); |
|
406
|
|
|
}, |
|
407
|
|
|
|
|
408
|
|
|
// Character replacement can be performed directly on the escaped value provided that it |
|
409
|
|
|
// is then escaped as ENT_COMPAT and that replacements do not interfere with the escaping |
|
410
|
|
|
// of the characters &<>" or their representation &<>" |
|
411
|
|
|
'(htmlspecialchars\\(strtr\\(' . $getAttribute . ",('[^\"&\\\\';<>aglmopqtu]+'),('[^\"&\\\\'<>]+')\\)," . ENT_COMPAT . '\\))' |
|
412
|
|
|
=> 'strtr($attributes[$1],$2,$3)', |
|
413
|
|
|
|
|
414
|
|
|
// A comparison between two attributes. No need to unescape |
|
415
|
|
|
'(' . $getAttribute . '(!?=+)' . $getAttribute . ')' |
|
416
|
|
|
=> '$attributes[$1]$2$attributes[$3]', |
|
417
|
|
|
|
|
418
|
|
|
// A comparison between an attribute and a literal string. Rather than unescape the |
|
419
|
|
|
// attribute value, we escape the literal. This applies to comparisons using XPath's |
|
420
|
|
|
// contains() as well (translated to PHP's strpos()) |
|
421
|
|
|
'(' . $getAttribute . "===('.*?(?<!\\\\)(?:\\\\\\\\)*'))s" |
|
422
|
|
|
=> function ($m) |
|
423
|
|
|
{ |
|
424
|
|
|
return '$attributes[' . $m[1] . ']===' . htmlspecialchars($m[2], ENT_COMPAT); |
|
425
|
|
|
}, |
|
426
|
|
|
|
|
427
|
|
|
"(('.*?(?<!\\\\)(?:\\\\\\\\)*')===" . $getAttribute . ')s' |
|
428
|
|
|
=> function ($m) |
|
429
|
|
|
{ |
|
430
|
|
|
return htmlspecialchars($m[1], ENT_COMPAT) . '===$attributes[' . $m[2] . ']'; |
|
431
|
|
|
}, |
|
432
|
|
|
|
|
433
|
|
|
'(strpos\\(' . $getAttribute . ",('.*?(?<!\\\\)(?:\\\\\\\\)*')\\)([!=]==(?:0|false)))s" |
|
434
|
|
|
=> function ($m) |
|
435
|
|
|
{ |
|
436
|
|
|
return 'strpos($attributes[' . $m[1] . "]," . htmlspecialchars($m[2], ENT_COMPAT) . ')' . $m[3]; |
|
437
|
|
|
}, |
|
438
|
|
|
|
|
439
|
|
|
"(strpos\\(('.*?(?<!\\\\)(?:\\\\\\\\)*')," . $getAttribute . '\\)([!=]==(?:0|false)))s' |
|
440
|
|
|
=> function ($m) |
|
441
|
|
|
{ |
|
442
|
|
|
return 'strpos(' . htmlspecialchars($m[1], ENT_COMPAT) . ',$attributes[' . $m[2] . '])' . $m[3]; |
|
443
|
|
|
}, |
|
444
|
|
|
|
|
445
|
|
|
// An attribute value used in an arithmetic comparison or operation does not need to be |
|
446
|
|
|
// unescaped. The same applies to empty(), isset() and conditionals |
|
447
|
|
|
'(' . $getAttribute . '(?=(?:==|[-+*])\\d+))' => '$attributes[$1]', |
|
448
|
|
|
'((?<!\\w)(\\d+(?:==|[-+*]))' . $getAttribute . ')' => '$1$attributes[$2]', |
|
449
|
|
|
"(empty\\(\\\$node->getAttribute\\(('[^']+')\\)\\))" => 'empty($attributes[$1])', |
|
450
|
|
|
"(\\\$node->hasAttribute\\(('[^']+')\\))" => 'isset($attributes[$1])', |
|
451
|
|
|
'if($node->attributes->length)' => 'if($this->hasNonNullValues($attributes))', |
|
452
|
|
|
|
|
453
|
|
|
// In all other situations, unescape the attribute value before use |
|
454
|
|
|
"(\\\$node->getAttribute\\(('[^']+')\\))" => 'htmlspecialchars_decode($attributes[$1])' |
|
455
|
|
|
]; |
|
456
|
|
|
|
|
457
|
|
|
foreach ($replacements as $match => $replace) |
|
458
|
|
|
{ |
|
459
|
|
|
if ($replace instanceof Closure) |
|
460
|
|
|
{ |
|
461
|
|
|
$php = preg_replace_callback($match, $replace, $php); |
|
462
|
|
|
} |
|
463
|
|
|
elseif ($match[0] === '(') |
|
464
|
|
|
{ |
|
465
|
|
|
$php = preg_replace($match, $replace, $php); |
|
466
|
|
|
} |
|
467
|
|
|
else |
|
468
|
|
|
{ |
|
469
|
|
|
$php = str_replace($match, $replace, $php); |
|
470
|
|
|
} |
|
471
|
|
|
} |
|
472
|
|
|
} |
|
473
|
|
|
|
|
474
|
|
|
/** |
|
475
|
|
|
* Build the source for the two sides of a templates based on the structure extracted from its |
|
476
|
|
|
* original source |
|
477
|
|
|
* |
|
478
|
|
|
* @param array $branches |
|
479
|
|
|
* @return string[] |
|
480
|
|
|
*/ |
|
481
|
|
|
protected static function buildPHP(array $branches) |
|
482
|
|
|
{ |
|
483
|
|
|
$return = ['', '']; |
|
484
|
|
|
foreach ($branches as $branch) |
|
485
|
|
|
{ |
|
486
|
|
|
$return[0] .= $branch['statement'] . '{' . $branch['head']; |
|
487
|
|
|
$return[1] .= $branch['statement'] . '{'; |
|
488
|
|
|
|
|
489
|
|
|
if ($branch['branches']) |
|
490
|
|
|
{ |
|
491
|
|
|
list($head, $tail) = self::buildPHP($branch['branches']); |
|
492
|
|
|
|
|
493
|
|
|
$return[0] .= $head; |
|
494
|
|
|
$return[1] .= $tail; |
|
495
|
|
|
} |
|
496
|
|
|
|
|
497
|
|
|
$return[0] .= '}'; |
|
498
|
|
|
$return[1] .= $branch['tail'] . '}'; |
|
499
|
|
|
} |
|
500
|
|
|
|
|
501
|
|
|
return $return; |
|
502
|
|
|
} |
|
503
|
|
|
|
|
504
|
|
|
/** |
|
505
|
|
|
* Get the unique values for the "passthrough" key of given branches |
|
506
|
|
|
* |
|
507
|
|
|
* @param array $branches |
|
508
|
|
|
* @return integer[] |
|
509
|
|
|
*/ |
|
510
|
|
|
protected static function getBranchesPassthrough(array $branches) |
|
511
|
|
|
{ |
|
512
|
|
|
$values = []; |
|
513
|
|
|
foreach ($branches as $branch) |
|
514
|
|
|
{ |
|
515
|
|
|
$values[] = $branch['passthrough']; |
|
516
|
|
|
} |
|
517
|
|
|
|
|
518
|
|
|
// If the last branch isn't an "else", we act as if there was an additional branch with no |
|
519
|
|
|
// passthrough |
|
520
|
|
|
if ($branch['statement'] !== 'else') |
|
521
|
|
|
{ |
|
522
|
|
|
$values[] = 0; |
|
523
|
|
|
} |
|
524
|
|
|
|
|
525
|
|
|
return array_unique($values); |
|
526
|
|
|
} |
|
527
|
|
|
|
|
528
|
|
|
/** |
|
529
|
|
|
* Get a string suitable as a preg_replace() replacement for given PHP code |
|
530
|
|
|
* |
|
531
|
|
|
* @param string $php Original code |
|
532
|
|
|
* @return array|bool Array of [regexp, replacement] if possible, or FALSE otherwise |
|
533
|
|
|
*/ |
|
534
|
|
|
protected static function getDynamicRendering($php) |
|
535
|
|
|
{ |
|
536
|
|
|
$rendering = ''; |
|
537
|
|
|
|
|
538
|
|
|
$literal = "(?<literal>'((?>[^'\\\\]+|\\\\['\\\\])*)')"; |
|
539
|
|
|
$attribute = "(?<attribute>htmlspecialchars\\(\\\$node->getAttribute\\('([^']+)'\\),2\\))"; |
|
540
|
|
|
$value = "(?<value>$literal|$attribute)"; |
|
541
|
|
|
$output = "(?<output>\\\$this->out\\.=$value(?:\\.(?&value))*;)"; |
|
542
|
|
|
|
|
543
|
|
|
$copyOfAttribute = "(?<copyOfAttribute>if\\(\\\$node->hasAttribute\\('([^']+)'\\)\\)\\{\\\$this->out\\.=' \\g-1=\"'\\.htmlspecialchars\\(\\\$node->getAttribute\\('\\g-1'\\),2\\)\\.'\"';\\})"; |
|
544
|
|
|
|
|
545
|
|
|
$regexp = '(^(' . $output . '|' . $copyOfAttribute . ')*$)'; |
|
546
|
|
|
if (!preg_match($regexp, $php, $m)) |
|
547
|
|
|
{ |
|
548
|
|
|
return false; |
|
549
|
|
|
} |
|
550
|
|
|
|
|
551
|
|
|
// Attributes that are copied in the replacement |
|
552
|
|
|
$copiedAttributes = []; |
|
553
|
|
|
|
|
554
|
|
|
// Attributes whose value is used in the replacement |
|
555
|
|
|
$usedAttributes = []; |
|
556
|
|
|
|
|
557
|
|
|
$regexp = '(' . $output . '|' . $copyOfAttribute . ')A'; |
|
558
|
|
|
$offset = 0; |
|
559
|
|
|
while (preg_match($regexp, $php, $m, 0, $offset)) |
|
560
|
|
|
{ |
|
561
|
|
|
// Test whether it's normal output or a copy of attribute |
|
562
|
|
|
if ($m['output']) |
|
563
|
|
|
{ |
|
564
|
|
|
// 12 === strlen('$this->out.=') |
|
565
|
|
|
$offset += 12; |
|
566
|
|
|
|
|
567
|
|
|
while (preg_match('(' . $value . ')A', $php, $m, 0, $offset)) |
|
568
|
|
|
{ |
|
569
|
|
|
// Test whether it's a literal or an attribute value |
|
570
|
|
|
if ($m['literal']) |
|
571
|
|
|
{ |
|
572
|
|
|
// Unescape the literal |
|
573
|
|
|
$str = stripslashes(substr($m[0], 1, -1)); |
|
574
|
|
|
|
|
575
|
|
|
// Escape special characters |
|
576
|
|
|
$rendering .= preg_replace('([\\\\$](?=\\d))', '\\\\$0', $str); |
|
577
|
|
|
} |
|
578
|
|
|
else |
|
579
|
|
|
{ |
|
580
|
|
|
$attrName = end($m); |
|
581
|
|
|
|
|
582
|
|
|
// Generate a unique ID for this attribute name, we'll use it as a |
|
583
|
|
|
// placeholder until we have the full list of captures and we can replace it |
|
584
|
|
|
// with the capture number |
|
585
|
|
|
if (!isset($usedAttributes[$attrName])) |
|
586
|
|
|
{ |
|
587
|
|
|
$usedAttributes[$attrName] = uniqid($attrName, true); |
|
588
|
|
|
} |
|
589
|
|
|
|
|
590
|
|
|
$rendering .= $usedAttributes[$attrName]; |
|
591
|
|
|
} |
|
592
|
|
|
|
|
593
|
|
|
// Skip the match plus the next . or ; |
|
594
|
|
|
$offset += 1 + strlen($m[0]); |
|
595
|
|
|
} |
|
596
|
|
|
} |
|
597
|
|
|
else |
|
598
|
|
|
{ |
|
599
|
|
|
$attrName = end($m); |
|
600
|
|
|
|
|
601
|
|
|
if (!isset($copiedAttributes[$attrName])) |
|
602
|
|
|
{ |
|
603
|
|
|
$copiedAttributes[$attrName] = uniqid($attrName, true); |
|
604
|
|
|
} |
|
605
|
|
|
|
|
606
|
|
|
$rendering .= $copiedAttributes[$attrName]; |
|
607
|
|
|
$offset += strlen($m[0]); |
|
608
|
|
|
} |
|
609
|
|
|
} |
|
610
|
|
|
|
|
611
|
|
|
// Gather the names of the attributes used in the replacement either by copy or by value |
|
612
|
|
|
$attrNames = array_keys($copiedAttributes + $usedAttributes); |
|
613
|
|
|
|
|
614
|
|
|
// Sort them alphabetically |
|
615
|
|
|
sort($attrNames); |
|
616
|
|
|
|
|
617
|
|
|
// Keep a copy of the attribute names to be used in the fillter subpattern |
|
618
|
|
|
$remainingAttributes = array_combine($attrNames, $attrNames); |
|
619
|
|
|
|
|
620
|
|
|
// Prepare the final regexp |
|
621
|
|
|
$regexp = '(^[^ ]+'; |
|
622
|
|
|
$index = 0; |
|
623
|
|
|
foreach ($attrNames as $attrName) |
|
624
|
|
|
{ |
|
625
|
|
|
// Add a subpattern that matches (and skips) any attribute definition that is not one of |
|
626
|
|
|
// the remaining attributes we're trying to match |
|
627
|
|
|
$regexp .= '(?> (?!' . RegexpBuilder::fromList($remainingAttributes) . '=)[^=]+="[^"]*")*'; |
|
628
|
|
|
unset($remainingAttributes[$attrName]); |
|
629
|
|
|
|
|
630
|
|
|
$regexp .= '('; |
|
631
|
|
|
|
|
632
|
|
|
if (isset($copiedAttributes[$attrName])) |
|
633
|
|
|
{ |
|
634
|
|
|
self::replacePlaceholder($rendering, $copiedAttributes[$attrName], ++$index); |
|
635
|
|
|
} |
|
636
|
|
|
else |
|
637
|
|
|
{ |
|
638
|
|
|
$regexp .= '?>'; |
|
639
|
|
|
} |
|
640
|
|
|
|
|
641
|
|
|
$regexp .= ' ' . $attrName . '="'; |
|
642
|
|
|
|
|
643
|
|
|
if (isset($usedAttributes[$attrName])) |
|
644
|
|
|
{ |
|
645
|
|
|
$regexp .= '('; |
|
646
|
|
|
|
|
647
|
|
|
self::replacePlaceholder($rendering, $usedAttributes[$attrName], ++$index); |
|
648
|
|
|
} |
|
649
|
|
|
|
|
650
|
|
|
$regexp .= '[^"]*'; |
|
651
|
|
|
|
|
652
|
|
|
if (isset($usedAttributes[$attrName])) |
|
653
|
|
|
{ |
|
654
|
|
|
$regexp .= ')'; |
|
655
|
|
|
} |
|
656
|
|
|
|
|
657
|
|
|
$regexp .= '")?'; |
|
658
|
|
|
} |
|
659
|
|
|
|
|
660
|
|
|
$regexp .= '.*)s'; |
|
661
|
|
|
|
|
662
|
|
|
return [$regexp, $rendering]; |
|
663
|
|
|
} |
|
664
|
|
|
|
|
665
|
|
|
/** |
|
666
|
|
|
* Get a string suitable as a str_replace() replacement for given PHP code |
|
667
|
|
|
* |
|
668
|
|
|
* @param string $php Original code |
|
669
|
|
|
* @return bool|string Static replacement if possible, or FALSE otherwise |
|
670
|
|
|
*/ |
|
671
|
|
|
protected static function getStaticRendering($php) |
|
672
|
|
|
{ |
|
673
|
|
|
if ($php === '') |
|
674
|
|
|
{ |
|
675
|
|
|
return ''; |
|
676
|
|
|
} |
|
677
|
|
|
|
|
678
|
|
|
$regexp = "(^\\\$this->out\.='((?>[^'\\\\]|\\\\['\\\\])*+)';\$)"; |
|
679
|
|
|
if (preg_match($regexp, $php, $m)) |
|
680
|
|
|
{ |
|
681
|
|
|
return stripslashes($m[1]); |
|
682
|
|
|
} |
|
683
|
|
|
|
|
684
|
|
|
return false; |
|
685
|
|
|
} |
|
686
|
|
|
|
|
687
|
|
|
/** |
|
688
|
|
|
* Get string rendering strategies for given chunks |
|
689
|
|
|
* |
|
690
|
|
|
* @param string $php |
|
691
|
|
|
* @return array |
|
692
|
|
|
*/ |
|
693
|
|
|
protected static function getStringRenderings($php) |
|
694
|
|
|
{ |
|
695
|
|
|
$chunks = explode('$this->at($node);', $php); |
|
696
|
|
|
if (count($chunks) > 2) |
|
697
|
|
|
{ |
|
698
|
|
|
// Can't use string replacements if there are more than one xsl:apply-templates |
|
699
|
|
|
return []; |
|
700
|
|
|
} |
|
701
|
|
|
|
|
702
|
|
|
$renderings = []; |
|
703
|
|
|
foreach ($chunks as $k => $chunk) |
|
704
|
|
|
{ |
|
705
|
|
|
// Try a static replacement first |
|
706
|
|
|
$rendering = self::getStaticRendering($chunk); |
|
707
|
|
|
if ($rendering !== false) |
|
708
|
|
|
{ |
|
709
|
|
|
$renderings[$k] = ['static', $rendering]; |
|
710
|
|
|
} |
|
711
|
|
|
elseif ($k === 0) |
|
712
|
|
|
{ |
|
713
|
|
|
// If this is the first chunk, we can try a dynamic replacement. This wouldn't work |
|
714
|
|
|
// for the second chunk because we wouldn't have access to the attribute values |
|
715
|
|
|
$rendering = self::getDynamicRendering($chunk); |
|
716
|
|
|
if ($rendering !== false) |
|
717
|
|
|
{ |
|
718
|
|
|
$renderings[$k] = ['dynamic', $rendering]; |
|
719
|
|
|
} |
|
720
|
|
|
} |
|
721
|
|
|
} |
|
722
|
|
|
|
|
723
|
|
|
return $renderings; |
|
724
|
|
|
} |
|
725
|
|
|
|
|
726
|
|
|
/** |
|
727
|
|
|
* Replace all instances of a uniqid with a PCRE replacement in a string |
|
728
|
|
|
* |
|
729
|
|
|
* @param string &$str PCRE replacement |
|
730
|
|
|
* @param string $uniqid Unique ID |
|
731
|
|
|
* @param integer $index Capture index |
|
732
|
|
|
* @return void |
|
733
|
|
|
*/ |
|
734
|
|
|
protected static function replacePlaceholder(&$str, $uniqid, $index) |
|
735
|
|
|
{ |
|
736
|
|
|
$str = preg_replace_callback( |
|
737
|
|
|
'(' . preg_quote($uniqid) . '(.))', |
|
738
|
|
|
function ($m) use ($index) |
|
739
|
|
|
{ |
|
740
|
|
|
// Replace with $1 where unambiguous and ${1} otherwise |
|
741
|
|
|
if (is_numeric($m[1])) |
|
742
|
|
|
{ |
|
743
|
|
|
return '${' . $index . '}' . $m[1]; |
|
744
|
|
|
} |
|
745
|
|
|
else |
|
746
|
|
|
{ |
|
747
|
|
|
return '$' . $index . $m[1]; |
|
748
|
|
|
} |
|
749
|
|
|
}, |
|
750
|
|
|
$str |
|
751
|
|
|
); |
|
752
|
|
|
} |
|
753
|
|
|
} |
s9e /
TextFormatter
