1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
namespace Cloudstek\SCIM\FilterParser; |
6
|
|
|
|
7
|
|
|
use Cloudstek\SCIM\FilterParser\Exception\UnexpectedValueException; |
8
|
|
|
use Nette\Tokenizer; |
9
|
|
|
|
10
|
|
|
/** |
11
|
|
|
* Abstract parser. |
12
|
|
|
*/ |
13
|
|
|
abstract class AbstractParser |
14
|
|
|
{ |
15
|
|
|
protected const T_NUMBER = 10; |
16
|
|
|
protected const T_STRING = 11; |
17
|
|
|
protected const T_BOOL = 12; |
18
|
|
|
protected const T_NULL = 13; |
19
|
|
|
|
20
|
|
|
protected const T_PAREN_OPEN = 21; |
21
|
|
|
protected const T_PAREN_CLOSE = 22; |
22
|
|
|
protected const T_BRACKET_OPEN = 23; |
23
|
|
|
protected const T_BRACKET_CLOSE = 24; |
24
|
|
|
|
25
|
|
|
protected const T_NEGATION = 30; |
26
|
|
|
protected const T_LOG_OP = 31; |
27
|
|
|
protected const T_COMP_OP = 32; |
28
|
|
|
|
29
|
|
|
protected const T_NAME = 40; |
30
|
|
|
protected const T_SUBATTR = 41; |
31
|
|
|
|
32
|
|
|
protected Tokenizer\Tokenizer $tokenizer; |
33
|
|
|
|
34
|
|
|
protected ParserMode $mode; |
35
|
|
|
|
36
|
|
|
/** |
37
|
|
|
* Abstract parser. |
38
|
|
|
* |
39
|
|
|
* @param ParserMode $mode Parser mode. |
40
|
|
|
*/ |
41
|
|
|
public function __construct(ParserMode $mode) |
42
|
|
|
{ |
43
|
|
|
$patterns = [ |
44
|
|
|
self::T_NUMBER => '\d+(?:\.\d+)?', |
45
|
|
|
self::T_STRING => '\"[^\"]*\"', |
46
|
|
|
self::T_BOOL => 'true|false', |
47
|
|
|
self::T_NULL => 'null', |
48
|
|
|
self::T_PAREN_OPEN => '\(', |
49
|
|
|
self::T_PAREN_CLOSE => '\)', |
50
|
|
|
self::T_BRACKET_OPEN => '\[', |
51
|
|
|
self::T_BRACKET_CLOSE => '\]', |
52
|
|
|
self::T_NEGATION => 'not\s+', |
53
|
|
|
self::T_LOG_OP => '\s+(?:and|or)\s+', |
54
|
|
|
self::T_COMP_OP => '\s(?:(?:eq|ne|co|sw|ew|gt|lt|ge|le)\s+|pr)', |
55
|
|
|
self::T_NAME => '(?:(?:[^\"]+\:)+)?[\-\_a-z0-9]+(?:\.[\-\_a-z0-9]+)?', |
56
|
|
|
]; |
57
|
|
|
|
58
|
|
|
if ($mode === ParserMode::PATH()) { |
59
|
|
|
$patterns[self::T_SUBATTR] = '\.[\-\_a-z0-9]+$'; |
60
|
|
|
} |
61
|
|
|
|
62
|
|
|
// Sort patterns by key as constant value determines order. |
63
|
|
|
ksort($patterns); |
64
|
|
|
|
65
|
|
|
$this->tokenizer = new Tokenizer\Tokenizer( |
66
|
|
|
$patterns, |
67
|
|
|
'i' |
68
|
|
|
); |
69
|
|
|
|
70
|
|
|
$this->mode = $mode; |
71
|
|
|
} |
72
|
|
|
|
73
|
|
|
/** |
74
|
|
|
* Parse. |
75
|
|
|
* |
76
|
|
|
* @param string $input |
77
|
|
|
* |
78
|
|
|
* @return AST\Node|AST\Path|null |
79
|
|
|
*/ |
80
|
|
|
abstract public function parse(string $input); |
81
|
|
|
|
82
|
|
|
/** |
83
|
|
|
* Parse inner. |
84
|
|
|
* |
85
|
|
|
* @param Tokenizer\Stream $stream |
86
|
|
|
* @param bool $inValuePath |
87
|
|
|
* |
88
|
|
|
* @throws Tokenizer\Exception |
89
|
|
|
* |
90
|
|
|
* @return AST\Node|null |
91
|
|
|
*/ |
92
|
|
|
protected function parseInner(Tokenizer\Stream $stream, bool $inValuePath = false): ?AST\Node |
93
|
|
|
{ |
94
|
|
|
$node = null; |
95
|
|
|
|
96
|
|
|
if ($stream->isNext(self::T_PAREN_OPEN)) { |
97
|
|
|
// Parentheses |
98
|
|
|
$node = $this->parseParentheses($stream); |
99
|
|
|
|
100
|
|
|
if ($node === null) { |
101
|
|
|
return null; |
102
|
|
|
} |
103
|
|
|
} elseif ($stream->isNext(self::T_NEGATION)) { |
104
|
|
|
// Negation |
105
|
|
|
$node = $this->parseNegation($stream, $inValuePath); |
106
|
|
|
|
107
|
|
|
if ($node === null) { |
108
|
|
|
return null; |
109
|
|
|
} |
110
|
|
|
} elseif ($stream->isNext(self::T_NAME)) { |
111
|
|
|
// Comparison or value path |
112
|
|
|
$node = $this->parseAttributePath($stream, $inValuePath); |
113
|
|
|
|
114
|
|
|
if ($node instanceof AST\AttributePath) { |
115
|
|
|
$node = $this->parseComparison($stream, $node); |
116
|
|
|
} |
117
|
|
|
} |
118
|
|
|
|
119
|
|
|
// Make sure we only return nodes, not paths. |
120
|
|
|
if ($node !== null && $node instanceof AST\Node) { |
121
|
|
|
// Logical connective |
122
|
|
|
if ($stream->isNext(self::T_LOG_OP)) { |
123
|
|
|
return $this->parseConnective($stream, $node, $inValuePath); |
124
|
|
|
} |
125
|
|
|
|
126
|
|
|
return $node; |
127
|
|
|
} |
128
|
|
|
|
129
|
|
|
throw new UnexpectedValueException($stream); |
130
|
|
|
} |
131
|
|
|
|
132
|
|
|
/** |
133
|
|
|
* Parse filter in parentheses. |
134
|
|
|
* |
135
|
|
|
* @param Tokenizer\Stream $stream |
136
|
|
|
* |
137
|
|
|
* @throws Tokenizer\Exception |
138
|
|
|
* |
139
|
|
|
* @return AST\Node|AST\Path|null |
140
|
|
|
*/ |
141
|
|
|
protected function parseParentheses(Tokenizer\Stream $stream) |
142
|
|
|
{ |
143
|
|
|
$stream->consumeToken(self::T_PAREN_OPEN); |
144
|
|
|
$filter = $this->joinUntilMatchingParenthesis($stream); |
145
|
|
|
$stream->consumeToken(self::T_PAREN_CLOSE); |
146
|
|
|
|
147
|
|
|
if (empty($filter)) { |
148
|
|
|
return null; |
149
|
|
|
} |
150
|
|
|
|
151
|
|
|
return $this->parse($filter); |
152
|
|
|
} |
153
|
|
|
|
154
|
|
|
/** |
155
|
|
|
* Parse negation. |
156
|
|
|
* |
157
|
|
|
* @param Tokenizer\Stream $stream |
158
|
|
|
* @param bool $inValuePath |
159
|
|
|
* |
160
|
|
|
* @throws Tokenizer\Exception |
161
|
|
|
* |
162
|
|
|
* @return AST\Node|null |
163
|
|
|
*/ |
164
|
|
|
protected function parseNegation(Tokenizer\Stream $stream, bool $inValuePath = false): ?AST\Node |
165
|
|
|
{ |
166
|
|
|
$stream->consumeToken(self::T_NEGATION); |
167
|
|
|
$stream->consumeToken(self::T_PAREN_OPEN); |
168
|
|
|
$filter = $this->joinUntilMatchingParenthesis($stream); |
169
|
|
|
$stream->consumeToken(self::T_PAREN_CLOSE); |
170
|
|
|
|
171
|
|
|
if (empty($filter)) { |
172
|
|
|
return null; |
173
|
|
|
} |
174
|
|
|
|
175
|
|
|
$node = $this->parseInner($this->tokenizer->tokenize($filter), $inValuePath); |
176
|
|
|
|
177
|
|
|
if ($node === null) { |
178
|
|
|
return null; |
179
|
|
|
} |
180
|
|
|
|
181
|
|
|
return new AST\Negation($node); |
182
|
|
|
} |
183
|
|
|
|
184
|
|
|
/** |
185
|
|
|
* Parse attribute path. |
186
|
|
|
* |
187
|
|
|
* @param Tokenizer\Stream $stream |
188
|
|
|
* @param bool $inValuePath |
189
|
|
|
* |
190
|
|
|
* @throws Tokenizer\Exception |
191
|
|
|
* |
192
|
|
|
* @return AST\Path |
193
|
|
|
*/ |
194
|
|
|
protected function parseAttributePath(Tokenizer\Stream $stream, bool $inValuePath = false): AST\Path |
195
|
|
|
{ |
196
|
|
|
$name = $stream->consumeToken(self::T_NAME)->value; |
197
|
|
|
|
198
|
|
|
// Attribute scheme |
199
|
|
|
$scheme = null; |
200
|
|
|
|
201
|
|
|
if (strpos($name, ':') !== false) { |
202
|
|
|
/** @var int $lastColonPos */ |
203
|
|
|
$lastColonPos = strrpos($name, ':'); |
204
|
|
|
$scheme = substr($name, 0, $lastColonPos); |
205
|
|
|
$name = substr($name, $lastColonPos + 1); |
206
|
|
|
} |
207
|
|
|
|
208
|
|
|
// Attribute path |
209
|
|
|
$attributePath = new AST\AttributePath($scheme, explode('.', $name)); |
210
|
|
|
|
211
|
|
|
// Value path (only if not in value path already |
212
|
|
|
if ($stream->isNext(self::T_BRACKET_OPEN)) { |
213
|
|
|
if ($inValuePath === true || count($attributePath) !== 1) { |
214
|
|
|
throw new UnexpectedValueException($stream); |
215
|
|
|
} |
216
|
|
|
|
217
|
|
|
return $this->parseValuePath($stream, $attributePath); |
218
|
|
|
} |
219
|
|
|
|
220
|
|
|
return $attributePath; |
221
|
|
|
} |
222
|
|
|
|
223
|
|
|
/** |
224
|
|
|
* Parse value path. |
225
|
|
|
* |
226
|
|
|
* @param Tokenizer\Stream $stream |
227
|
|
|
* @param AST\AttributePath $attributePath |
228
|
|
|
* |
229
|
|
|
* @throws Tokenizer\Exception |
230
|
|
|
* |
231
|
|
|
* @return AST\ValuePath |
232
|
|
|
*/ |
233
|
|
|
protected function parseValuePath(Tokenizer\Stream $stream, AST\AttributePath $attributePath): AST\ValuePath |
234
|
|
|
{ |
235
|
|
|
// Save position to report exception coordinates later |
236
|
|
|
$startPos = $stream->position + 1; |
237
|
|
|
|
238
|
|
|
// Parse |
239
|
|
|
$stream->consumeToken(self::T_BRACKET_OPEN); |
240
|
|
|
$node = $this->parseInner($stream, true); |
241
|
|
|
$stream->consumeToken(self::T_BRACKET_CLOSE); |
242
|
|
|
|
243
|
|
|
// Correct attribute path for node. |
244
|
|
|
try { |
245
|
|
|
$node = $this->updateValuePathAttributePath($attributePath, $node); |
246
|
|
|
} catch (Tokenizer\Exception $ex) { |
247
|
|
|
// Reset stream position to value path start to correct exception coordinates. |
248
|
|
|
$stream->position = $startPos; |
249
|
|
|
|
250
|
|
|
throw new UnexpectedValueException($stream); |
251
|
|
|
} |
252
|
|
|
|
253
|
|
|
return new AST\ValuePath($attributePath, $node); |
254
|
|
|
} |
255
|
|
|
|
256
|
|
|
/** |
257
|
|
|
* Parse comparison. |
258
|
|
|
* |
259
|
|
|
* @param Tokenizer\Stream $stream |
260
|
|
|
* @param AST\AttributePath $attributePath |
261
|
|
|
* |
262
|
|
|
* @throws Tokenizer\Exception |
263
|
|
|
* |
264
|
|
|
* @return AST\Connective|AST\Comparison |
265
|
|
|
*/ |
266
|
|
|
protected function parseComparison( |
267
|
|
|
Tokenizer\Stream $stream, |
268
|
|
|
AST\AttributePath $attributePath |
269
|
|
|
) { |
270
|
|
|
$operator = trim($stream->consumeValue(self::T_COMP_OP)); |
271
|
|
|
$value = null; |
272
|
|
|
|
273
|
|
|
if (strcasecmp($operator, 'pr') <> 0) { |
274
|
|
|
$value = $stream->consumeToken(self::T_STRING, self::T_NUMBER, self::T_BOOL, self::T_NULL); |
275
|
|
|
|
276
|
|
|
switch ($value->type) { |
277
|
|
|
case self::T_STRING: |
278
|
|
|
$value = trim($value->value, '"'); |
279
|
|
|
break; |
280
|
|
|
case self::T_NUMBER: |
281
|
|
|
if (strpos($value->value, '.') !== false) { |
282
|
|
|
$value = floatval($value->value); |
283
|
|
|
break; |
284
|
|
|
} |
285
|
|
|
|
286
|
|
|
$value = intval($value->value); |
287
|
|
|
break; |
288
|
|
|
case self::T_BOOL: |
289
|
|
|
$value = strcasecmp($value->value, 'true') === 0; |
290
|
|
|
break; |
291
|
|
|
default: |
292
|
|
|
$value = null; |
293
|
|
|
break; |
294
|
|
|
} |
295
|
|
|
} |
296
|
|
|
|
297
|
|
|
return new AST\Comparison($attributePath, $operator, $value); |
298
|
|
|
} |
299
|
|
|
|
300
|
|
|
/** |
301
|
|
|
* Parse logical connective (and/or). |
302
|
|
|
* |
303
|
|
|
* @param Tokenizer\Stream $stream |
304
|
|
|
* @param AST\Node $leftNode |
305
|
|
|
* @param bool $inValuePath |
306
|
|
|
* |
307
|
|
|
* @throws Tokenizer\Exception |
308
|
|
|
* |
309
|
|
|
* @return AST\Connective |
310
|
|
|
*/ |
311
|
|
|
protected function parseConnective( |
312
|
|
|
Tokenizer\Stream $stream, |
313
|
|
|
AST\Node $leftNode, |
314
|
|
|
bool $inValuePath = false |
315
|
|
|
): AST\Connective { |
316
|
|
|
// Logical operator |
317
|
|
|
$logOp = trim($stream->consumeToken(self::T_LOG_OP)->value); |
318
|
|
|
|
319
|
|
|
$isConjunction = strcasecmp($logOp, 'and') === 0; |
320
|
|
|
|
321
|
|
|
// Save position to report exception coordinates later |
322
|
|
|
$startPos = $stream->position + 1; |
323
|
|
|
|
324
|
|
|
// Parse right hand node |
325
|
|
|
$rightNode = $this->parseInner($stream, $inValuePath); |
326
|
|
|
|
327
|
|
|
if ($rightNode === null) { |
328
|
|
|
// Reset stream position to value path start to correct exception coordinates. |
329
|
|
|
$stream->position = $startPos; |
330
|
|
|
|
331
|
|
|
throw new UnexpectedValueException($stream); |
332
|
|
|
} |
333
|
|
|
|
334
|
|
|
// Connective nodes |
335
|
|
|
$nodes = [$leftNode, $rightNode]; |
336
|
|
|
|
337
|
|
|
// Merge consecutive connectives of the same type into one connective. |
338
|
|
|
if ( |
339
|
|
|
($rightNode instanceof AST\Conjunction && $isConjunction === true) |
340
|
|
|
|| ($rightNode instanceof AST\Disjunction && $isConjunction === false) |
341
|
|
|
) { |
342
|
|
|
$rightNodes = $rightNode->getNodes(); |
343
|
|
|
|
344
|
|
|
$nodes = array_merge([$leftNode], $rightNodes); |
345
|
|
|
} |
346
|
|
|
|
347
|
|
|
return $isConjunction === true |
348
|
|
|
? new AST\Conjunction($nodes) |
349
|
|
|
: new AST\Disjunction($nodes); |
350
|
|
|
} |
351
|
|
|
|
352
|
|
|
/** |
353
|
|
|
* Prepend attribute path to node. |
354
|
|
|
* |
355
|
|
|
* @param AST\AttributePath $attributePath |
356
|
|
|
* @param AST\Node|null $node |
357
|
|
|
* |
358
|
|
|
* @throws Tokenizer\Exception When node is null or not a valid node. |
359
|
|
|
* |
360
|
|
|
* @return AST\Node |
361
|
|
|
*/ |
362
|
|
|
protected function updateValuePathAttributePath(AST\AttributePath $attributePath, ?AST\Node $node): AST\Node |
363
|
|
|
{ |
364
|
|
|
if ($node instanceof AST\Comparison) { |
365
|
|
|
$names = array_merge($attributePath->getNames(), $node->getAttributePath()->getNames()); |
366
|
|
|
|
367
|
|
|
$node->setAttributePath(new AST\AttributePath($attributePath->getSchema(), $names)); |
368
|
|
|
|
369
|
|
|
return $node; |
370
|
|
|
} |
371
|
|
|
|
372
|
|
|
if ($node instanceof AST\Negation) { |
373
|
|
|
$node->setNode($this->updateValuePathAttributePath($attributePath, $node->getNode())); |
374
|
|
|
|
375
|
|
|
return $node; |
376
|
|
|
} |
377
|
|
|
|
378
|
|
|
if ($node instanceof AST\Connective) { |
379
|
|
|
foreach ($node->getNodes() as $subNode) { |
380
|
|
|
$this->updateValuePathAttributePath($attributePath, $subNode); |
381
|
|
|
} |
382
|
|
|
|
383
|
|
|
return $node; |
384
|
|
|
} |
385
|
|
|
|
386
|
|
|
throw new Tokenizer\Exception('Invalid value path.'); |
387
|
|
|
} |
388
|
|
|
|
389
|
|
|
/** |
390
|
|
|
* Join values until matching |
391
|
|
|
* |
392
|
|
|
* @param Tokenizer\Stream $stream |
393
|
|
|
* |
394
|
|
|
* @throws Tokenizer\Exception |
395
|
|
|
* @return string |
396
|
|
|
*/ |
397
|
|
|
protected function joinUntilMatchingParenthesis(Tokenizer\Stream $stream): string |
398
|
|
|
{ |
399
|
|
|
$pos = $stream->position + 1; |
400
|
|
|
$numTokens = count($stream->tokens); |
401
|
|
|
$level = 1; |
402
|
|
|
$result = ''; |
403
|
|
|
|
404
|
|
|
if ($pos >= $numTokens) { |
405
|
|
|
throw new Tokenizer\Exception('Unexpected end of string'); |
406
|
|
|
} |
407
|
|
|
|
408
|
|
|
for ($i = $pos; $pos < $numTokens; $i++) { |
409
|
|
|
$token = $stream->tokens[$i]; |
410
|
|
|
|
411
|
|
|
if ($token->type === self::T_PAREN_OPEN) { |
412
|
|
|
$level++; |
413
|
|
|
} elseif ($token->type === self::T_PAREN_CLOSE) { |
414
|
|
|
$level--; |
415
|
|
|
|
416
|
|
|
if ($level === 0) { |
417
|
|
|
$stream->position = $i - 1; |
418
|
|
|
break; |
419
|
|
|
} |
420
|
|
|
} |
421
|
|
|
|
422
|
|
|
$result .= $token->value; |
423
|
|
|
} |
424
|
|
|
|
425
|
|
|
return $result; |
426
|
|
|
} |
427
|
|
|
} |
428
|
|
|
|