1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Netdudes\DataSourceryBundle\UQL; |
4
|
|
|
|
5
|
|
|
use Netdudes\DataSourceryBundle\UQL\AST\ASTArray; |
6
|
|
|
use Netdudes\DataSourceryBundle\UQL\AST\ASTAssertion; |
7
|
|
|
use Netdudes\DataSourceryBundle\UQL\AST\ASTFunctionCall; |
8
|
|
|
use Netdudes\DataSourceryBundle\UQL\AST\ASTGroup; |
9
|
|
|
use Netdudes\DataSourceryBundle\UQL\Exception\Semantic\UqlUnexpectedEndOfExpressionException; |
10
|
|
|
use Netdudes\DataSourceryBundle\UQL\Exception\Semantic\UqlUnexpectedTokenException; |
11
|
|
|
use Netdudes\DataSourceryBundle\UQL\Exception\UQLSyntaxError; |
12
|
|
|
|
13
|
|
|
/** |
14
|
|
|
* Class Parser |
15
|
|
|
* |
16
|
|
|
* The parser translates a linear stream of tokens into a logical Abstract |
17
|
|
|
* Syntax Tree (AST) that represents the logical structure of the language |
18
|
|
|
* with independence of the actual end-objects (filters). |
19
|
|
|
* |
20
|
|
|
* @package Netdudes\NetdudesDataSourceryBundle\UQL |
21
|
|
|
*/ |
22
|
|
|
class Parser |
23
|
|
|
{ |
24
|
|
|
/** |
25
|
|
|
* |
26
|
|
|
* Grammar (<identifier>s and <literal>s are simple scalars defined by regular expressions on the lexer): |
27
|
|
|
* |
28
|
|
|
* <operator> ::= "<" | ">" | "<=" | ">=" | "!=" | "<>" | "=" |
29
|
|
|
* <logic> ::= "AND" | "OR" |
30
|
|
|
* <assertion> ::= <identifier> <operator> <literal> |
31
|
|
|
* <concatenation> ::= <statement> { <logic> <statement> } |
32
|
|
|
* <group> ::= "(" <concatenation> ")" |
33
|
|
|
* <statement> ::= <assertion> | <group> |
34
|
|
|
* <query> ::= <concatenation> |
35
|
|
|
* |
36
|
|
|
*/ |
37
|
|
|
|
38
|
|
|
private $tokenIndex; |
39
|
|
|
|
40
|
|
|
private $tokenStream = []; |
41
|
|
|
|
42
|
|
|
public function __construct() |
43
|
|
|
{ |
44
|
|
|
$this->tokenIndex = -1; |
45
|
|
|
} |
46
|
|
|
|
47
|
|
|
/** |
48
|
|
|
* Lex, initialise and return the AST. |
49
|
|
|
* |
50
|
|
|
* @param $string |
51
|
|
|
* |
52
|
|
|
* @return bool|ASTAssertion|ASTGroup |
53
|
|
|
*/ |
54
|
|
|
public function parse($string) |
55
|
|
|
{ |
56
|
|
|
$this->tokenStream = Lexer::lex($string); |
57
|
|
|
$this->tokenIndex = -1; |
58
|
|
|
|
59
|
|
|
return $this->getAST(); |
60
|
|
|
} |
61
|
|
|
|
62
|
|
|
/** |
63
|
|
|
* Entry point of the grammar parsing. |
64
|
|
|
* |
65
|
|
|
* @return bool|ASTAssertion|ASTGroup |
66
|
|
|
*/ |
67
|
|
|
public function getAST() |
68
|
|
|
{ |
69
|
|
|
// The top-level syntax is, in general, a concatenation of statements with logic connectors. |
70
|
|
|
$concatenation = $this->matchConcatenation(); |
71
|
|
|
|
72
|
|
|
// Make sure we are at the end of the UQL |
73
|
|
|
$token = $this->nextToken(); |
74
|
|
|
if ($token === false) { |
75
|
|
|
return $concatenation; |
76
|
|
|
} |
77
|
|
|
|
78
|
|
|
$this->throwUnexpectedTokenSyntaxError(['LOGIC'], "Logic operator or end of UQL expected after statement in first-level concatenation"); |
79
|
|
|
} |
80
|
|
|
|
81
|
|
|
/** |
82
|
|
|
* Tries to match the following tokens to a <concatenation> grammar. |
83
|
|
|
* |
84
|
|
|
* @return bool|ASTAssertion|ASTGroup |
85
|
|
|
*/ |
86
|
|
|
public function matchConcatenation() |
87
|
|
|
{ |
88
|
|
|
$elements = []; |
89
|
|
|
|
90
|
|
|
$firstStatement = $this->matchStatement(); |
91
|
|
|
|
92
|
|
|
if ($firstStatement === false) { |
93
|
|
|
$this->throwUnexpectedTokenSyntaxError(['IDENTIFIER', 'GROUP_START'], 'Expected statement at beginning of concatenation.'); |
94
|
|
|
} |
95
|
|
|
|
96
|
|
|
$elements[] = $firstStatement; |
97
|
|
|
|
98
|
|
|
$firstLogic = $this->matchLogic(); |
99
|
|
|
if ($firstLogic === false) { |
100
|
|
|
// There is no actual concatenation. This is a single statement. Return as such. |
101
|
|
|
return $firstStatement; |
102
|
|
|
} |
103
|
|
|
$logic = $firstLogic; |
104
|
|
|
|
105
|
|
|
// While there are concatenating logic operators, keep adding elements. |
106
|
|
|
while ($logic !== false) { |
107
|
|
|
if ($logic['token'] != $firstLogic['token']) { |
108
|
|
|
$this->throwSyntaxError('Can\'t mix ORs and ANDs in same-level expression, ambiguous statement.'); |
109
|
|
|
} |
110
|
|
|
$statement = $this->matchStatement(); |
111
|
|
|
if ($statement === false) { |
112
|
|
|
$this->throwUnexpectedTokenSyntaxError(['IDENTIFIER', 'GROUP_START'], 'Expected statement after logic operator'); |
113
|
|
|
} |
114
|
|
|
$elements[] = $statement; |
115
|
|
|
$logic = $this->matchLogic(); |
116
|
|
|
} |
117
|
|
|
|
118
|
|
|
return new ASTGroup($firstLogic['token'], $elements); |
119
|
|
|
} |
120
|
|
|
|
121
|
|
|
/** |
122
|
|
|
* Tries to match a general <statement>, that is a <group> or <assertion> |
123
|
|
|
* |
124
|
|
|
* @return bool|ASTAssertion|ASTGroup |
125
|
|
|
*/ |
126
|
|
|
public function matchStatement() |
127
|
|
|
{ |
128
|
|
|
// Try <group> |
129
|
|
|
$matchGroup = $this->matchGroup(); |
130
|
|
|
|
131
|
|
|
if ($matchGroup !== false) { |
132
|
|
|
return $matchGroup; |
133
|
|
|
} |
134
|
|
|
|
135
|
|
|
// Try <assertion> |
136
|
|
|
$matchAssertion = $this->matchAssertion(); |
137
|
|
|
|
138
|
|
|
if ($matchAssertion !== false) { |
139
|
|
|
return $matchAssertion; |
140
|
|
|
} |
141
|
|
|
|
142
|
|
|
// None found |
143
|
|
|
$this->rewindToken(); |
144
|
|
|
|
145
|
|
|
return false; |
146
|
|
|
} |
147
|
|
|
|
148
|
|
|
/** |
149
|
|
|
* Tries to match a <group> grammar to the following tokens |
150
|
|
|
* |
151
|
|
|
* @return bool|ASTAssertion|ASTGroup |
152
|
|
|
*/ |
153
|
|
|
public function matchGroup() |
154
|
|
|
{ |
155
|
|
|
$token = $this->nextToken(); |
156
|
|
|
|
157
|
|
|
// Check for the open parenthesis |
158
|
|
|
if ($token['token'] != "T_BRACKET_OPEN") { |
159
|
|
|
$this->rewindToken(); |
160
|
|
|
|
161
|
|
|
return false; |
162
|
|
|
} |
163
|
|
|
|
164
|
|
|
// The interior of a group is a <concatenation> |
165
|
|
|
$concatenation = $this->matchConcatenation(); |
166
|
|
|
|
167
|
|
|
$token = $this->nextToken(); |
168
|
|
|
|
169
|
|
|
// Check for closed parenthesis. Mismatch is a Syntax Error. |
170
|
|
|
if ($token['token'] != "T_BRACKET_CLOSE") { |
171
|
|
|
$this->throwUnexpectedTokenSyntaxError(['GROUP_END'], 'Expected closing bracket.'); |
172
|
|
|
} |
173
|
|
|
|
174
|
|
|
return $concatenation; |
175
|
|
|
} |
176
|
|
|
|
177
|
|
|
/** |
178
|
|
|
* Tries to match the following tokens to an <assertion>. |
179
|
|
|
* |
180
|
|
|
* @throws Exception\UQLSyntaxError |
181
|
|
|
* @return bool|ASTAssertion |
182
|
|
|
*/ |
183
|
|
|
public function matchAssertion() |
184
|
|
|
{ |
185
|
|
|
$identifier = $this->nextToken(); |
186
|
|
|
|
187
|
|
|
if ($identifier['token'] != 'T_IDENTIFIER') { |
188
|
|
|
// If a stream doesn't start with an identifier, it's not an <assertion>. |
189
|
|
|
$this->rewindToken(); |
190
|
|
|
|
191
|
|
|
return false; |
192
|
|
|
} |
193
|
|
|
|
194
|
|
|
$operator = $this->matchOperator(); |
195
|
|
|
|
196
|
|
|
if ($operator === false) { |
197
|
|
|
$this->nextToken(); // MatchOperator rewinds |
198
|
|
|
$this->throwUnexpectedTokenSyntaxError(['OPERATOR'], 'Comparison operator expected after identifier'); |
199
|
|
|
} |
200
|
|
|
|
201
|
|
|
$array = $this->matchArray(); |
202
|
|
|
if ($array) { |
203
|
|
|
if (!in_array($operator['token'], ['T_OP_IN', 'T_OP_NIN'])) { |
204
|
|
|
throw new UQLSyntaxError("Arrays are only valid after IN or NOT IN operators"); |
205
|
|
|
} |
206
|
|
|
|
207
|
|
|
return new ASTAssertion($identifier['match'], $operator['token'], $array); |
208
|
|
|
} |
209
|
|
|
|
210
|
|
|
$literal = $this->nextToken(); |
211
|
|
|
|
212
|
|
|
if ($literal['token'] == 'T_FUNCTION_CALL') { |
213
|
|
|
return new ASTAssertion($identifier['match'], $operator['token'], ASTFunctionCall::createFromExpression($literal['match'])); |
214
|
|
|
} |
215
|
|
|
|
216
|
|
|
if (strpos($literal['token'], 'T_LITERAL') !== 0) { |
217
|
|
|
$this->throwUnexpectedTokenSyntaxError(['ARRAY_START', 'LITERAL'], 'Array, value or function call expected after comparison operator'); |
218
|
|
|
} |
219
|
|
|
$literal = $this->transformLiteral($literal); |
220
|
|
|
|
221
|
|
|
return new ASTAssertion($identifier['match'], $operator['token'], $literal['match']); |
222
|
|
|
} |
223
|
|
|
|
224
|
|
|
/** |
225
|
|
|
* Tries to match the next token to an <operator>. |
226
|
|
|
* |
227
|
|
|
* @return bool |
228
|
|
|
*/ |
229
|
|
|
public function matchOperator() |
230
|
|
|
{ |
231
|
|
|
$operator = $this->nextToken(); |
232
|
|
|
|
233
|
|
|
switch ($operator['token']) { |
234
|
|
|
case 'T_OP_NEQ': |
235
|
|
|
case 'T_OP_LTE': |
236
|
|
|
case 'T_OP_LT': |
237
|
|
|
case 'T_OP_GTE': |
238
|
|
|
case 'T_OP_GT': |
239
|
|
|
case 'T_OP_EQ': |
240
|
|
|
case 'T_OP_LIKE': |
241
|
|
|
case 'T_OP_IN': |
242
|
|
|
case 'T_OP_NIN': |
243
|
|
|
return $operator; |
244
|
|
|
break; |
|
|
|
|
245
|
|
|
default: |
246
|
|
|
$this->rewindToken(); |
247
|
|
|
|
248
|
|
|
return false; |
249
|
|
|
} |
250
|
|
|
} |
251
|
|
|
|
252
|
|
|
public function matchArray() |
253
|
|
|
{ |
254
|
|
|
$token = $this->nextToken(); |
255
|
|
|
if ($token['token'] != "T_ARRAY_OPEN") { |
256
|
|
|
$this->rewindToken(); |
257
|
|
|
|
258
|
|
|
return false; |
259
|
|
|
} |
260
|
|
|
|
261
|
|
|
$element = $this->nextToken(); |
262
|
|
|
if ($element['token'] == "T_ARRAY_CLOSE") { |
263
|
|
|
// Empty array |
264
|
|
|
return new ASTArray(); |
265
|
|
|
} |
266
|
|
|
|
267
|
|
|
$elements = [$element['match']]; |
268
|
|
|
$comma = $this->nextToken(); |
269
|
|
|
while ($comma['token'] == "T_ARRAY_SEPARATOR") { |
270
|
|
|
$element = $this->nextToken(); |
271
|
|
|
if ($element['token'] !== 'T_LITERAL') { |
272
|
|
|
$this->throwUnexpectedTokenSyntaxError(['LITERAL'], "An array must consist of literals"); |
273
|
|
|
} |
274
|
|
|
$elements[] = $element['match']; |
275
|
|
|
$comma = $this->nextToken(); |
276
|
|
|
} |
277
|
|
|
if ($comma['token'] != 'T_ARRAY_CLOSE') { |
278
|
|
|
// Unterminated array |
279
|
|
|
$this->throwUnexpectedTokenSyntaxError(['ARRAY_END'], "An array must end with ']'."); |
280
|
|
|
} |
281
|
|
|
|
282
|
|
|
return new ASTArray($elements); |
283
|
|
|
} |
284
|
|
|
|
285
|
|
|
/** |
286
|
|
|
* Tries to match the next token to a <logic> operator |
287
|
|
|
* |
288
|
|
|
* @return bool |
289
|
|
|
*/ |
290
|
|
|
public function matchLogic() |
291
|
|
|
{ |
292
|
|
|
$token = $this->nextToken(); |
293
|
|
|
|
294
|
|
|
if ($token['token'] == 'T_LOGIC_AND' || $token['token'] == 'T_LOGIC_OR' || $token['token'] == 'T_LOGIC_XOR') { |
295
|
|
|
return $token; |
296
|
|
|
} |
297
|
|
|
|
298
|
|
|
// None found |
299
|
|
|
$this->rewindToken(); |
300
|
|
|
|
301
|
|
|
return false; |
302
|
|
|
} |
303
|
|
|
|
304
|
|
|
/** |
305
|
|
|
* @return mixed |
306
|
|
|
*/ |
307
|
|
|
public function getTokenStream() |
308
|
|
|
{ |
309
|
|
|
return $this->tokenStream; |
310
|
|
|
} |
311
|
|
|
|
312
|
|
|
/** |
313
|
|
|
* @param mixed $tokenStream |
314
|
|
|
*/ |
315
|
|
|
public function setTokenStream($tokenStream) |
316
|
|
|
{ |
317
|
|
|
$this->tokenStream = $tokenStream; |
|
|
|
|
318
|
|
|
} |
319
|
|
|
|
320
|
|
|
/** |
321
|
|
|
* @return mixed |
322
|
|
|
*/ |
323
|
|
|
public function getTokenIndex() |
324
|
|
|
{ |
325
|
|
|
return $this->tokenIndex; |
326
|
|
|
} |
327
|
|
|
|
328
|
|
|
/** |
329
|
|
|
* @param mixed $tokenIndex |
330
|
|
|
*/ |
331
|
|
|
public function setTokenIndex($tokenIndex) |
332
|
|
|
{ |
333
|
|
|
$this->tokenIndex = $tokenIndex; |
334
|
|
|
} |
335
|
|
|
|
336
|
|
|
/** |
337
|
|
|
* Advance the token index and return. |
338
|
|
|
* |
339
|
|
|
* @return bool |
340
|
|
|
*/ |
341
|
|
|
private function nextToken() |
342
|
|
|
{ |
343
|
|
|
$this->tokenIndex++; |
344
|
|
|
|
345
|
|
|
return $this->currentToken(); |
346
|
|
|
} |
347
|
|
|
|
348
|
|
|
/** |
349
|
|
|
* Return the current token, without advancing the index. |
350
|
|
|
* |
351
|
|
|
* @return bool |
352
|
|
|
*/ |
353
|
|
|
private function currentToken() |
354
|
|
|
{ |
355
|
|
|
return isset($this->tokenStream[$this->tokenIndex]) ? $this->tokenStream[$this->tokenIndex] : false; |
356
|
|
|
} |
357
|
|
|
|
358
|
|
|
/** |
359
|
|
|
* Move back the token index once. |
360
|
|
|
*/ |
361
|
|
|
private function rewindToken() |
362
|
|
|
{ |
363
|
|
|
$this->tokenIndex--; |
364
|
|
|
} |
365
|
|
|
|
366
|
|
|
/** |
367
|
|
|
* Helper method. Throws an Exception representing a Syntax Error. |
368
|
|
|
* |
369
|
|
|
* @param $message |
370
|
|
|
* |
371
|
|
|
* @throws \Exception |
372
|
|
|
*/ |
373
|
|
|
private function throwUnexpectedTokenSyntaxError(array $expectedTokenCategories, $message = null) |
374
|
|
|
{ |
375
|
|
|
$parsedTokenStream = array_slice($this->tokenStream, 0, $this->tokenIndex + 1); |
376
|
|
|
if ($this->currentToken() === false) { |
377
|
|
|
throw new UqlUnexpectedEndOfExpressionException( |
378
|
|
|
$expectedTokenCategories, |
379
|
|
|
$parsedTokenStream, |
380
|
|
|
$message); |
381
|
|
|
} |
382
|
|
|
throw new UqlUnexpectedTokenException( |
383
|
|
|
$this->currentToken()['token'], |
384
|
|
|
$this->currentToken()['match'], |
385
|
|
|
$expectedTokenCategories, |
386
|
|
|
$parsedTokenStream, |
387
|
|
|
$message |
388
|
|
|
); |
389
|
|
|
} |
390
|
|
|
|
391
|
|
|
private function throwSyntaxError($message) |
392
|
|
|
{ |
393
|
|
|
throw new UQLSyntaxError('Syntax error: ' . $message); |
394
|
|
|
} |
395
|
|
|
|
396
|
|
|
/** |
397
|
|
|
* Transforms a literal subtype (e.g. T_LITERAL_FALSE) into a plain |
398
|
|
|
* literal match. Plain literals are unchanged. |
399
|
|
|
* |
400
|
|
|
* @param $literal |
401
|
|
|
* |
402
|
|
|
* @return array |
403
|
|
|
*/ |
404
|
|
|
private function transformLiteral($literal) |
405
|
|
|
{ |
406
|
|
|
switch ($literal['token']) { |
407
|
|
|
case 'T_LITERAL_FALSE': |
408
|
|
|
$match = false; |
409
|
|
|
break; |
410
|
|
|
case 'T_LITERAL_TRUE': |
411
|
|
|
$match = true; |
412
|
|
|
break; |
413
|
|
|
case 'T_LITERAL_EMPTY': |
414
|
|
|
$match = null; |
415
|
|
|
break; |
416
|
|
|
default: |
417
|
|
|
$match = $literal['match']; |
418
|
|
|
} |
419
|
|
|
|
420
|
|
|
return [ |
421
|
|
|
'token' => 'T_LITERAL', |
422
|
|
|
'match' => $match |
423
|
|
|
]; |
424
|
|
|
} |
425
|
|
|
} |
426
|
|
|
|
The break statement is not necessary if it is preceded for example by a return statement:
If you would like to keep this construct to be consistent with other case statements, you can safely mark this issue as a false-positive.