|
1
|
|
|
<?php |
|
2
|
|
|
namespace Thunder\Shortcode\Parser; |
|
3
|
|
|
|
|
4
|
|
|
use Thunder\Shortcode\Shortcode\ParsedShortcode; |
|
5
|
|
|
use Thunder\Shortcode\Shortcode\Shortcode; |
|
6
|
|
|
use Thunder\Shortcode\Syntax\CommonSyntax; |
|
7
|
|
|
use Thunder\Shortcode\Syntax\SyntaxInterface; |
|
8
|
|
|
|
|
9
|
|
|
/** |
|
10
|
|
|
* @author Tomasz Kowalczyk <[email protected]> |
|
11
|
|
|
*/ |
|
12
|
|
|
final class RegularParser implements ParserInterface |
|
13
|
|
|
{ |
|
14
|
|
|
private $lexerRegex; |
|
15
|
|
|
private $tokens; |
|
16
|
|
|
private $tokensCount; |
|
17
|
|
|
private $position; |
|
18
|
|
|
private $backtracks; |
|
19
|
|
|
|
|
20
|
|
|
const TOKEN_OPEN = 1; |
|
21
|
|
|
const TOKEN_CLOSE = 2; |
|
22
|
|
|
const TOKEN_MARKER = 3; |
|
23
|
|
|
const TOKEN_SEPARATOR = 4; |
|
24
|
|
|
const TOKEN_DELIMITER = 5; |
|
25
|
|
|
const TOKEN_STRING = 6; |
|
26
|
|
|
const TOKEN_WS = 7; |
|
27
|
|
|
|
|
28
|
4 |
|
public function __construct(SyntaxInterface $syntax = null) |
|
29
|
|
|
{ |
|
30
|
4 |
|
$this->lexerRegex = $this->getTokenizerRegex($syntax ?: new CommonSyntax()); |
|
31
|
4 |
|
} |
|
32
|
|
|
|
|
33
|
|
|
/** |
|
34
|
|
|
* @param string $text |
|
35
|
|
|
* |
|
36
|
|
|
* @return ParsedShortcode[] |
|
37
|
|
|
*/ |
|
38
|
39 |
|
public function parse($text) |
|
39
|
|
|
{ |
|
40
|
39 |
|
$this->tokens = $this->tokenize($text); |
|
41
|
39 |
|
$this->backtracks = array(); |
|
42
|
39 |
|
$this->position = 0; |
|
43
|
39 |
|
$this->tokensCount = count($this->tokens); |
|
44
|
|
|
|
|
45
|
39 |
|
$shortcodes = array(); |
|
46
|
39 |
|
while($this->position < $this->tokensCount) { |
|
47
|
38 |
|
while($this->position < $this->tokensCount && !$this->lookahead(self::TOKEN_OPEN)) { |
|
48
|
15 |
|
$this->position++; |
|
49
|
15 |
|
} |
|
50
|
38 |
|
$names = array(); |
|
51
|
38 |
|
$this->beginBacktrack(); |
|
52
|
38 |
|
foreach($this->shortcode($names) ?: array() as $shortcode) { |
|
|
|
|
|
|
53
|
31 |
|
$shortcodes[] = $shortcode; |
|
54
|
38 |
|
} |
|
55
|
38 |
|
} |
|
56
|
|
|
|
|
57
|
39 |
|
return $shortcodes; |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
31 |
|
private function getObject($name, $parameters, $bbCode, $offset, $content, $text) |
|
61
|
|
|
{ |
|
62
|
31 |
|
return new ParsedShortcode(new Shortcode($name, $parameters, $content, $bbCode), $text, $offset); |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
/* --- RULES ----------------------------------------------------------- */ |
|
66
|
|
|
|
|
67
|
38 |
|
private function shortcode(array &$names) |
|
68
|
|
|
{ |
|
69
|
38 |
|
$name = null; |
|
70
|
38 |
|
$offset = null; |
|
71
|
|
|
|
|
72
|
|
|
$setName = function(array $token) use(&$name) { $name = $token[1]; }; |
|
73
|
|
|
$setOffset = function(array $token) use(&$offset) { $offset = $token[2]; }; |
|
74
|
|
|
|
|
75
|
38 |
|
if(!$this->match(self::TOKEN_OPEN, $setOffset, true)) { return false; } |
|
76
|
38 |
|
if(!$this->match(self::TOKEN_STRING, $setName, false)) { return false; } |
|
77
|
35 |
|
if($this->lookahead(self::TOKEN_STRING, null)) { return false; } |
|
78
|
35 |
|
if(!preg_match_all('/^[a-zA-Z0-9-]+$/', $name, $matches)) { return false; } |
|
79
|
34 |
|
$this->match(self::TOKEN_WS); |
|
80
|
34 |
|
if(false === ($bbCode = $this->bbCode())) { return false; } |
|
81
|
34 |
|
if(false === ($parameters = $this->parameters())) { return false; } |
|
82
|
|
|
|
|
83
|
|
|
// self-closing |
|
84
|
32 |
|
if($this->match(self::TOKEN_MARKER, null, true)) { |
|
85
|
12 |
|
if(!$this->match(self::TOKEN_CLOSE)) { return false; } |
|
86
|
|
|
|
|
87
|
11 |
|
return array($this->getObject($name, $parameters, $bbCode, $offset, null, $this->getBacktrack())); |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
// just-closed or with-content |
|
91
|
22 |
|
if(!$this->match(self::TOKEN_CLOSE)) { return false; } |
|
92
|
22 |
|
$this->beginBacktrack(); |
|
93
|
22 |
|
$names[] = $name; |
|
94
|
22 |
|
list($content, $shortcodes, $closingName) = $this->content($names); |
|
95
|
22 |
|
if(null !== $closingName && $closingName !== $name) { |
|
96
|
1 |
|
array_pop($names); |
|
97
|
1 |
|
array_pop($this->backtracks); |
|
98
|
1 |
|
array_pop($this->backtracks); |
|
99
|
|
|
|
|
100
|
1 |
|
return $closingName; |
|
101
|
|
|
} |
|
102
|
22 |
|
if(false === $content) { |
|
103
|
13 |
|
$this->backtrack(false); |
|
104
|
13 |
|
$text = $this->backtrack(false); |
|
105
|
|
|
|
|
106
|
13 |
|
return array_merge(array($this->getObject($name, $parameters, $bbCode, $offset, null, $text)), $shortcodes); |
|
107
|
|
|
} |
|
108
|
10 |
|
$content = $this->getBacktrack(); |
|
109
|
10 |
|
if(!$this->close($names)) { return false; } |
|
110
|
|
|
|
|
111
|
10 |
|
return array($this->getObject($name, $parameters, $bbCode, $offset, $content, $this->getBacktrack())); |
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
22 |
|
private function content(array &$names) |
|
115
|
|
|
{ |
|
116
|
22 |
|
$content = null; |
|
117
|
22 |
|
$shortcodes = array(); |
|
118
|
22 |
|
$closingName = null; |
|
119
|
|
|
$appendContent = function(array $token) use(&$content) { $content .= $token[1]; }; |
|
120
|
|
|
|
|
121
|
22 |
|
while($this->position < $this->tokensCount) { |
|
122
|
14 |
|
while($this->match(array(self::TOKEN_STRING, self::TOKEN_WS), $appendContent)) { |
|
123
|
13 |
|
continue; |
|
124
|
|
|
} |
|
125
|
|
|
|
|
126
|
14 |
|
$this->beginBacktrack(); |
|
127
|
14 |
|
$matchedShortcodes = $this->shortcode($names); |
|
128
|
14 |
|
if(is_string($matchedShortcodes)) { |
|
129
|
1 |
|
$closingName = $matchedShortcodes; |
|
130
|
1 |
|
break; |
|
131
|
|
|
} |
|
132
|
14 |
|
if(false !== $matchedShortcodes) { |
|
133
|
6 |
|
$shortcodes = array_merge($shortcodes, $matchedShortcodes); |
|
134
|
6 |
|
continue; |
|
135
|
|
|
} |
|
136
|
12 |
|
$this->backtrack(); |
|
137
|
|
|
|
|
138
|
12 |
|
$this->beginBacktrack(); |
|
139
|
12 |
|
if(false !== ($closingName = $this->close($names))) { |
|
140
|
10 |
|
if(null === $content) { $content = ''; } |
|
141
|
10 |
|
$this->backtrack(); |
|
142
|
10 |
|
$shortcodes = array(); |
|
143
|
10 |
|
break; |
|
144
|
|
|
} |
|
145
|
3 |
|
$closingName = null; |
|
146
|
3 |
|
$this->backtrack(); |
|
147
|
3 |
|
if($this->position < $this->tokensCount) { |
|
148
|
|
|
$shortcodes = array(); |
|
149
|
|
|
break; |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
3 |
|
$this->match(null, $appendContent); |
|
153
|
3 |
|
} |
|
154
|
|
|
|
|
155
|
22 |
|
return array($this->position < $this->tokensCount ? $content : false, $shortcodes, $closingName); |
|
156
|
|
|
} |
|
157
|
|
|
|
|
158
|
12 |
|
private function close(array &$names) |
|
159
|
|
|
{ |
|
160
|
12 |
|
$closingName = null; |
|
161
|
|
|
$setName = function(array $token) use(&$closingName) { $closingName = $token[1]; }; |
|
162
|
|
|
|
|
163
|
12 |
|
if(!$this->match(self::TOKEN_OPEN, null, true)) { return false; } |
|
164
|
10 |
|
if(!$this->match(self::TOKEN_MARKER, null, true)) { return false; } |
|
165
|
10 |
|
if(!$this->match(self::TOKEN_STRING, $setName, true)) { return false; } |
|
166
|
10 |
|
if(!$this->match(self::TOKEN_CLOSE)) { return false; } |
|
167
|
|
|
|
|
168
|
10 |
|
return in_array($closingName, $names) ? $closingName : false; |
|
169
|
|
|
} |
|
170
|
|
|
|
|
171
|
34 |
|
private function bbCode() |
|
172
|
|
|
{ |
|
173
|
34 |
|
return $this->match(self::TOKEN_SEPARATOR, null, true) ? $this->value() : null; |
|
174
|
|
|
} |
|
175
|
|
|
|
|
176
|
34 |
|
private function parameters() |
|
177
|
|
|
{ |
|
178
|
34 |
|
$parameters = array(); |
|
179
|
|
|
$setName = function(array $token) use(&$name) { $name = $token[1]; }; |
|
180
|
|
|
|
|
181
|
34 |
|
while(true) { |
|
182
|
34 |
|
$name = null; |
|
|
|
|
|
|
183
|
|
|
|
|
184
|
34 |
|
$this->match(self::TOKEN_WS); |
|
185
|
34 |
|
if($this->lookahead(array(self::TOKEN_MARKER, self::TOKEN_CLOSE))) { break; } |
|
186
|
22 |
|
if(!$this->match(self::TOKEN_STRING, $setName, true)) { return false; } |
|
187
|
21 |
|
if(!$this->match(self::TOKEN_SEPARATOR, null, true)) { $parameters[$name] = null; continue; } |
|
188
|
20 |
|
if(false === ($value = $this->value())) { return false; } |
|
189
|
19 |
|
$this->match(self::TOKEN_WS); |
|
190
|
|
|
|
|
191
|
19 |
|
$parameters[$name] = $value; |
|
192
|
19 |
|
} |
|
193
|
|
|
|
|
194
|
32 |
|
return $parameters; |
|
195
|
|
|
} |
|
196
|
|
|
|
|
197
|
21 |
|
private function value() |
|
198
|
|
|
{ |
|
199
|
21 |
|
$value = ''; |
|
200
|
|
|
$appendValue = function(array $token) use(&$value) { $value .= $token[1]; }; |
|
201
|
|
|
|
|
202
|
21 |
|
if($this->match(self::TOKEN_DELIMITER)) { |
|
203
|
18 |
|
while($this->position < $this->tokensCount && !$this->lookahead(self::TOKEN_DELIMITER)) { |
|
204
|
18 |
|
$this->match(null, $appendValue); |
|
205
|
18 |
|
} |
|
206
|
|
|
|
|
207
|
18 |
|
return $this->match(self::TOKEN_DELIMITER) ? $value : false; |
|
208
|
|
|
} |
|
209
|
|
|
|
|
210
|
8 |
|
return $this->match(self::TOKEN_STRING, $appendValue) ? $value : false; |
|
211
|
|
|
} |
|
212
|
|
|
|
|
213
|
|
|
/* --- PARSER ---------------------------------------------------------- */ |
|
214
|
|
|
|
|
215
|
|
|
/** |
|
216
|
|
|
* This method is used only for debugging purposes. DO NOT DELETE. |
|
217
|
|
|
* |
|
218
|
|
|
* @return array |
|
219
|
|
|
*/ |
|
220
|
|
|
private function listBacktracks() |
|
|
|
|
|
|
221
|
|
|
{ |
|
222
|
|
|
return array_map(function(array $backtrack) { |
|
223
|
|
|
return implode('', array_column($backtrack, 1)); |
|
224
|
|
|
}, $this->backtracks); |
|
225
|
|
|
} |
|
226
|
|
|
|
|
227
|
38 |
|
private function beginBacktrack() |
|
228
|
|
|
{ |
|
229
|
38 |
|
$this->backtracks[] = array(); |
|
230
|
38 |
|
} |
|
231
|
|
|
|
|
232
|
|
|
private function getBacktrack() |
|
233
|
|
|
{ |
|
234
|
|
|
// switch from array_map() to array_column() when dropping support for PHP <5.5 |
|
235
|
|
|
return implode('', array_map(function(array $token) { return $token[1]; }, array_pop($this->backtracks))); |
|
236
|
|
|
} |
|
237
|
|
|
|
|
238
|
22 |
|
private function backtrack($modifyPosition = true) |
|
239
|
|
|
{ |
|
240
|
22 |
|
$tokens = array_pop($this->backtracks); |
|
241
|
22 |
|
$count = count($tokens); |
|
242
|
22 |
|
if($modifyPosition) { |
|
243
|
12 |
|
$this->position -= $count; |
|
244
|
12 |
|
} |
|
245
|
|
|
|
|
246
|
22 |
|
foreach($this->backtracks as &$backtrack) { |
|
247
|
|
|
// array_pop() in loop is much faster than array_slice() because |
|
248
|
|
|
// it operates directly on the passed array |
|
249
|
22 |
|
for($i = 0; $i < $count; $i++) { |
|
250
|
14 |
|
array_pop($backtrack); |
|
251
|
14 |
|
} |
|
252
|
22 |
|
} |
|
253
|
|
|
|
|
254
|
|
|
return implode('', array_map(function(array $token) { return $token[1]; }, $tokens)); |
|
255
|
|
|
} |
|
256
|
|
|
|
|
257
|
38 |
|
private function lookahead($type, $callback = null) |
|
258
|
|
|
{ |
|
259
|
38 |
|
if($this->position >= $this->tokensCount) { |
|
260
|
1 |
|
return false; |
|
261
|
|
|
} |
|
262
|
|
|
|
|
263
|
38 |
|
$type = (array)$type; |
|
264
|
38 |
|
$token = $this->tokens[$this->position]; |
|
265
|
38 |
|
if(!empty($type) && !in_array($token[0], $type)) { |
|
266
|
37 |
|
return false; |
|
267
|
|
|
} |
|
268
|
|
|
|
|
269
|
|
|
/** @var $callback callable */ |
|
270
|
38 |
|
$callback && $callback($token); |
|
271
|
|
|
|
|
272
|
38 |
|
return true; |
|
273
|
|
|
} |
|
274
|
|
|
|
|
275
|
38 |
|
private function match($type, $callbacks = null, $ws = false) |
|
276
|
|
|
{ |
|
277
|
38 |
|
if($this->position >= $this->tokensCount) { |
|
278
|
13 |
|
return false; |
|
279
|
|
|
} |
|
280
|
|
|
|
|
281
|
38 |
|
$type = (array)$type; |
|
282
|
38 |
|
$token = $this->tokens[$this->position]; |
|
283
|
38 |
|
if(!empty($type) && !in_array($token[0], $type)) { |
|
284
|
38 |
|
return false; |
|
285
|
|
|
} |
|
286
|
38 |
|
foreach($this->backtracks as &$backtrack) { |
|
287
|
38 |
|
$backtrack[] = $token; |
|
288
|
38 |
|
} |
|
289
|
|
|
|
|
290
|
38 |
|
$this->position++; |
|
291
|
38 |
|
foreach((array)$callbacks as $callback) { |
|
292
|
38 |
|
$callback($token); |
|
293
|
38 |
|
} |
|
294
|
|
|
|
|
295
|
38 |
|
$ws && $this->match(self::TOKEN_WS); |
|
296
|
|
|
|
|
297
|
38 |
|
return true; |
|
298
|
|
|
} |
|
299
|
|
|
|
|
300
|
|
|
/* --- LEXER ----------------------------------------------------------- */ |
|
301
|
|
|
|
|
302
|
39 |
|
private function tokenize($text) |
|
303
|
|
|
{ |
|
304
|
39 |
|
preg_match_all($this->lexerRegex, $text, $matches, PREG_SET_ORDER | PREG_OFFSET_CAPTURE); |
|
305
|
39 |
|
$tokens = array(); |
|
306
|
39 |
|
$position = 0; |
|
307
|
39 |
|
$type = null; |
|
308
|
39 |
|
$token = null; |
|
309
|
39 |
|
foreach($matches as $match) { |
|
310
|
38 |
|
switch(true) { |
|
311
|
38 |
View Code Duplication |
case -1 !== $match['open'][1]: $token = $match['open'][0]; $type = self::TOKEN_OPEN; break; |
|
|
|
|
|
|
312
|
38 |
View Code Duplication |
case -1 !== $match['close'][1]: $token = $match['close'][0]; $type = self::TOKEN_CLOSE; break; |
|
|
|
|
|
|
313
|
37 |
View Code Duplication |
case -1 !== $match['marker'][1]: $token = $match['marker'][0]; $type = self::TOKEN_MARKER; break; |
|
|
|
|
|
|
314
|
37 |
View Code Duplication |
case -1 !== $match['separator'][1]: $token = $match['separator'][0]; $type = self::TOKEN_SEPARATOR; break; |
|
|
|
|
|
|
315
|
37 |
View Code Duplication |
case -1 !== $match['delimiter'][1]: $token = $match['delimiter'][0]; $type = self::TOKEN_DELIMITER; break; |
|
|
|
|
|
|
316
|
37 |
View Code Duplication |
case -1 !== $match['ws'][1]: $token = $match['ws'][0]; $type = self::TOKEN_WS; break; |
|
|
|
|
|
|
317
|
37 |
View Code Duplication |
case -1 !== $match['string'][1]: $token = $match['string'][0]; $type = self::TOKEN_STRING; break; |
|
|
|
|
|
|
318
|
|
|
} |
|
319
|
38 |
|
$tokens[] = array($type, $token, $position); |
|
320
|
38 |
|
$position += mb_strlen($token, 'utf-8'); |
|
321
|
39 |
|
} |
|
322
|
|
|
|
|
323
|
39 |
|
return $tokens; |
|
324
|
|
|
} |
|
325
|
|
|
|
|
326
|
|
|
private function getTokenizerRegex(SyntaxInterface $syntax) |
|
327
|
|
|
{ |
|
328
|
4 |
|
$quote = function($text, $group) { |
|
329
|
4 |
|
return '(?<'.$group.'>'.preg_replace('/(.)/us', '\\\\$0', $text).')'; |
|
330
|
4 |
|
}; |
|
331
|
|
|
|
|
332
|
|
|
$rules = array( |
|
333
|
4 |
|
$quote($syntax->getOpeningTag(), 'open'), |
|
334
|
4 |
|
$quote($syntax->getClosingTag(), 'close'), |
|
335
|
4 |
|
$quote($syntax->getClosingTagMarker(), 'marker'), |
|
336
|
4 |
|
$quote($syntax->getParameterValueSeparator(), 'separator'), |
|
337
|
4 |
|
$quote($syntax->getParameterValueDelimiter(), 'delimiter'), |
|
338
|
4 |
|
'(?<ws>\s+)', |
|
339
|
4 |
|
'(?<string>[\w-]+|\\\\.|.)', |
|
340
|
4 |
|
); |
|
341
|
|
|
|
|
342
|
4 |
|
return '~('.implode('|', $rules).')~us'; |
|
343
|
|
|
} |
|
344
|
|
|
} |
|
345
|
|
|
|
There are different options of fixing this problem.
If you want to be on the safe side, you can add an additional type-check:
If you are sure that the expression is traversable, you might want to add a doc comment cast to improve IDE auto-completion and static analysis:
Mark the issue as a false-positive: Just hover the remove button, in the top-right corner of this issue for more options.