1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* Buffered query utilities. |
5
|
|
|
* |
6
|
|
|
* @package SqlParser |
7
|
|
|
* @subpackage Utils |
8
|
|
|
*/ |
9
|
|
|
namespace SqlParser\Utils; |
10
|
|
|
|
11
|
|
|
use SqlParser\Context; |
12
|
|
|
|
13
|
|
|
/** |
14
|
|
|
* Buffer query utilities. |
15
|
|
|
* |
16
|
|
|
* Implements a specialized lexer used to extract statements from large inputs |
17
|
|
|
* that are being buffered. After each statement has been extracted, a lexer or |
18
|
|
|
* a parser may be used. |
19
|
|
|
* |
20
|
|
|
* All comments are skipped, with one exception: MySQL commands inside `/*!`. |
21
|
|
|
* |
22
|
|
|
* @category Lexer |
23
|
|
|
* @package SqlParser |
24
|
|
|
* @subpackage Utils |
25
|
|
|
* @author Dan Ungureanu <[email protected]> |
26
|
|
|
* @license http://opensource.org/licenses/GPL-2.0 GNU Public License |
27
|
|
|
*/ |
28
|
|
|
class BufferedQuery |
29
|
|
|
{ |
30
|
|
|
|
31
|
|
|
// Constants that describe the current status of the parser. |
32
|
|
|
|
33
|
|
|
// A string is being parsed. |
34
|
|
|
const STATUS_STRING = 16; // 0001 0000 |
35
|
|
|
const STATUS_STRING_SINGLE_QUOTES = 17; // 0001 0001 |
36
|
|
|
const STATUS_STRING_DOUBLE_QUOTES = 18; // 0001 0010 |
37
|
|
|
const STATUS_STRING_BACKTICK = 20; // 0001 0100 |
38
|
|
|
|
39
|
|
|
// A comment is being parsed. |
40
|
|
|
const STATUS_COMMENT = 32; // 0010 0000 |
41
|
|
|
const STATUS_COMMENT_BASH = 33; // 0010 0001 |
42
|
|
|
const STATUS_COMMENT_C = 34; // 0010 0010 |
43
|
|
|
const STATUS_COMMENT_SQL = 36; // 0010 0100 |
44
|
|
|
|
45
|
|
|
/** |
46
|
|
|
* The query that is being processed. |
47
|
|
|
* |
48
|
|
|
* This field can be modified just by appending to it! |
49
|
|
|
* |
50
|
|
|
* @var string |
51
|
|
|
*/ |
52
|
|
|
public $query = ''; |
53
|
|
|
|
54
|
|
|
/** |
55
|
|
|
* The options of this parser. |
56
|
|
|
* |
57
|
|
|
* @var array |
58
|
|
|
*/ |
59
|
|
|
public $options = array(); |
60
|
|
|
|
61
|
|
|
/** |
62
|
|
|
* The last delimiter used. |
63
|
|
|
* |
64
|
|
|
* @var string |
65
|
|
|
*/ |
66
|
|
|
public $delimiter; |
67
|
|
|
|
68
|
|
|
/** |
69
|
|
|
* The length of the delimiter. |
70
|
|
|
* |
71
|
|
|
* @var int |
72
|
|
|
*/ |
73
|
|
|
public $delimiterLen; |
74
|
|
|
|
75
|
|
|
/** |
76
|
|
|
* The current status of the parser. |
77
|
|
|
* |
78
|
|
|
* @var int |
79
|
|
|
*/ |
80
|
|
|
public $status; |
81
|
|
|
|
82
|
|
|
/** |
83
|
|
|
* The last incomplete query that was extracted. |
84
|
|
|
* |
85
|
|
|
* @var string |
86
|
|
|
*/ |
87
|
|
|
public $current = ''; |
88
|
|
|
|
89
|
|
|
/** |
90
|
|
|
* Constructor. |
91
|
|
|
* |
92
|
|
|
* @param string $query The query to be parsed. |
93
|
|
|
* @param array $options The options of this parser. |
94
|
|
|
*/ |
95
|
7 |
|
public function __construct($query = '', array $options = array()) |
96
|
|
|
{ |
97
|
|
|
// Merges specified options with defaults. |
98
|
7 |
|
$this->options = array_merge( |
99
|
|
|
array( |
100
|
|
|
|
101
|
|
|
/** |
102
|
|
|
* The starting delimiter. |
103
|
|
|
* |
104
|
|
|
* @var string |
105
|
|
|
*/ |
106
|
7 |
|
'delimiter' => ';', |
107
|
|
|
|
108
|
|
|
/** |
109
|
|
|
* Whether `DELIMITER` statements should be parsed. |
110
|
|
|
* |
111
|
|
|
* @var bool |
112
|
|
|
*/ |
113
|
7 |
|
'parse_delimiter' => false, |
114
|
|
|
|
115
|
|
|
/** |
116
|
|
|
* Whether a delimiter should be added at the end of the |
117
|
|
|
* statement. |
118
|
|
|
* |
119
|
|
|
* @var bool |
120
|
|
|
*/ |
121
|
7 |
|
'add_delimiter' => false, |
122
|
7 |
|
), |
123
|
|
|
$options |
124
|
7 |
|
); |
125
|
|
|
|
126
|
7 |
|
$this->query = $query; |
127
|
7 |
|
$this->setDelimiter($this->options['delimiter']); |
128
|
7 |
|
} |
129
|
|
|
|
130
|
|
|
/** |
131
|
|
|
* Sets the delimiter. |
132
|
|
|
* |
133
|
|
|
* Used to update the length of it too. |
134
|
|
|
* |
135
|
|
|
* @param string $delimiter |
136
|
|
|
*/ |
137
|
7 |
|
public function setDelimiter($delimiter) |
138
|
|
|
{ |
139
|
7 |
|
$this->delimiter = $delimiter; |
140
|
7 |
|
$this->delimiterLen = strlen($delimiter); |
141
|
7 |
|
} |
142
|
|
|
|
143
|
|
|
/** |
144
|
|
|
* Extracts a statement from the buffer. |
145
|
|
|
* |
146
|
|
|
* @param bool $end Whether the end of the buffer was reached. |
147
|
|
|
* |
148
|
|
|
* @return string |
|
|
|
|
149
|
|
|
*/ |
150
|
7 |
|
public function extract($end = false) |
151
|
|
|
{ |
152
|
|
|
/** |
153
|
|
|
* The last parsed position. |
154
|
|
|
* |
155
|
|
|
* This is statically defined because it is not used outside anywhere |
156
|
|
|
* outside this method and there is probably a (minor) performance |
157
|
|
|
* improvement to it. |
158
|
|
|
* |
159
|
|
|
* @var int |
160
|
|
|
*/ |
161
|
7 |
|
static $i = 0; |
162
|
|
|
|
163
|
7 |
|
if (empty($this->query)) { |
164
|
7 |
|
return false; |
165
|
|
|
} |
166
|
|
|
|
167
|
|
|
/** |
168
|
|
|
* The length of the buffer. |
169
|
|
|
* |
170
|
|
|
* @var int $len |
171
|
|
|
*/ |
172
|
7 |
|
$len = strlen($this->query); |
173
|
|
|
|
174
|
|
|
/** |
175
|
|
|
* The last index of the string that is going to be parsed. |
176
|
|
|
* |
177
|
|
|
* There must be a few characters left in the buffer so the parser can |
178
|
|
|
* avoid confusing some symbols that may have multiple meanings. |
179
|
|
|
* |
180
|
|
|
* For example, if the buffer ends in `-` that may be an operator or the |
181
|
|
|
* beginning of a comment. |
182
|
|
|
* |
183
|
|
|
* Another example if the buffer ends in `DELIMITE`. The parser is going |
184
|
|
|
* to require a few more characters because that may be a part of the |
185
|
|
|
* `DELIMITER` keyword or just a column named `DELIMITE`. |
186
|
|
|
* |
187
|
|
|
* Those extra characters are required only if there is more data |
188
|
|
|
* expected (the end of the buffer was not reached). |
189
|
|
|
* |
190
|
|
|
* @var int $loopLen |
191
|
|
|
*/ |
192
|
7 |
|
$loopLen = $end ? $len : $len - 16; |
193
|
|
|
|
194
|
7 |
|
for (; $i < $loopLen; ++$i) { |
195
|
|
|
/** |
196
|
|
|
* Handling backslash. |
197
|
|
|
* |
198
|
|
|
* Even if the next character is a special character that should be |
199
|
|
|
* treated differently, because of the preceding backslash, it will |
200
|
|
|
* be ignored. |
201
|
|
|
*/ |
202
|
7 |
|
if ((($this->status & static::STATUS_COMMENT) == 0) && ($this->query[$i] === '\\')) { |
203
|
3 |
|
$this->current .= $this->query[$i] . $this->query[++$i]; |
204
|
3 |
|
continue; |
205
|
|
|
} |
206
|
|
|
|
207
|
|
|
/* |
208
|
|
|
* Handling special parses statuses. |
209
|
|
|
*/ |
210
|
7 |
|
if ($this->status === static::STATUS_STRING_SINGLE_QUOTES) { |
211
|
|
|
// Single-quoted strings like 'foo'. |
212
|
5 |
|
if ($this->query[$i] === '\'') { |
213
|
5 |
|
$this->status = 0; |
214
|
5 |
|
} |
215
|
5 |
|
$this->current .= $this->query[$i]; |
216
|
5 |
|
continue; |
217
|
7 |
View Code Duplication |
} elseif ($this->status === static::STATUS_STRING_DOUBLE_QUOTES) { |
218
|
|
|
// Double-quoted strings like "bar". |
219
|
4 |
|
if ($this->query[$i] === '"') { |
220
|
4 |
|
$this->status = 0; |
221
|
4 |
|
} |
222
|
4 |
|
$this->current .= $this->query[$i]; |
223
|
4 |
|
continue; |
224
|
7 |
|
} elseif ($this->status === static::STATUS_STRING_BACKTICK) { |
225
|
4 |
|
if ($this->query[$i] === '`') { |
226
|
4 |
|
$this->status = 0; |
227
|
4 |
|
} |
228
|
4 |
|
$this->current .= $this->query[$i]; |
229
|
4 |
|
continue; |
230
|
7 |
|
} elseif (($this->status === static::STATUS_COMMENT_BASH) |
231
|
7 |
|
|| ($this->status === static::STATUS_COMMENT_SQL) |
232
|
7 |
|
) { |
233
|
|
|
// Bash-like (#) or SQL-like (-- ) comments end in new line. |
234
|
3 |
|
if ($this->query[$i] === "\n") { |
235
|
3 |
|
$this->status = 0; |
236
|
3 |
|
} |
237
|
3 |
|
continue; |
238
|
7 |
|
} elseif ($this->status === static::STATUS_COMMENT_C) { |
239
|
|
|
// C-like comments end in */. |
240
|
3 |
|
if (($this->query[$i - 1] === '*') && ($this->query[$i] === '/')) { |
241
|
3 |
|
$this->status = 0; |
242
|
3 |
|
} |
243
|
3 |
|
continue; |
244
|
|
|
} |
245
|
|
|
|
246
|
|
|
/* |
247
|
|
|
* Checking if a string started. |
248
|
|
|
*/ |
249
|
7 |
|
if ($this->query[$i] === '\'') { |
250
|
5 |
|
$this->status = static::STATUS_STRING_SINGLE_QUOTES; |
251
|
5 |
|
$this->current .= $this->query[$i]; |
252
|
5 |
|
continue; |
253
|
7 |
View Code Duplication |
} elseif ($this->query[$i] === '"') { |
254
|
4 |
|
$this->status = static::STATUS_STRING_DOUBLE_QUOTES; |
255
|
4 |
|
$this->current .= $this->query[$i]; |
256
|
4 |
|
continue; |
257
|
7 |
|
} elseif ($this->query[$i] === '`') { |
258
|
4 |
|
$this->status = static::STATUS_STRING_BACKTICK; |
259
|
4 |
|
$this->current .= $this->query[$i]; |
260
|
4 |
|
continue; |
261
|
|
|
} |
262
|
|
|
|
263
|
|
|
/* |
264
|
|
|
* Checking if a comment started. |
265
|
|
|
*/ |
266
|
7 |
|
if ($this->query[$i] === '#') { |
267
|
3 |
|
$this->status = static::STATUS_COMMENT_BASH; |
268
|
3 |
|
continue; |
269
|
7 |
|
} elseif (($i + 2 < $len) |
270
|
7 |
|
&& ($this->query[$i] === '-') |
271
|
7 |
|
&& ($this->query[$i + 1] === '-') |
272
|
7 |
|
&& (Context::isWhitespace($this->query[$i + 2])) |
273
|
7 |
|
) { |
274
|
3 |
|
$this->status = static::STATUS_COMMENT_SQL; |
275
|
3 |
|
continue; |
276
|
7 |
|
} elseif (($i + 2 < $len) |
277
|
7 |
|
&& ($this->query[$i] === '/') |
278
|
7 |
|
&& ($this->query[$i + 1] === '*') |
279
|
7 |
|
&& ($this->query[$i + 2] !== '!') |
280
|
7 |
|
) { |
281
|
3 |
|
$this->status = static::STATUS_COMMENT_C; |
282
|
3 |
|
continue; |
283
|
|
|
} |
284
|
|
|
|
285
|
|
|
/* |
286
|
|
|
* Handling `DELIMITER` statement. |
287
|
|
|
* |
288
|
|
|
* The code below basically checks for |
289
|
|
|
* `strtoupper(substr($this->query, $i, 9)) === 'DELIMITER'` |
290
|
|
|
* |
291
|
|
|
* This optimization makes the code about 3 times faster. |
292
|
|
|
* |
293
|
|
|
* `DELIMITER` is not being considered a keyword. The only context |
294
|
|
|
* it has a special meaning is when it is the beginning of a |
295
|
|
|
* statement. This is the reason for the last condition. |
296
|
|
|
*/ |
297
|
7 |
|
if (($i + 9 < $len) |
298
|
7 |
|
&& (($this->query[$i ] === 'D') || ($this->query[$i ] === 'd')) |
299
|
7 |
|
&& (($this->query[$i + 1] === 'E') || ($this->query[$i + 1] === 'e')) |
300
|
7 |
|
&& (($this->query[$i + 2] === 'L') || ($this->query[$i + 2] === 'l')) |
301
|
7 |
|
&& (($this->query[$i + 3] === 'I') || ($this->query[$i + 3] === 'i')) |
302
|
7 |
|
&& (($this->query[$i + 4] === 'M') || ($this->query[$i + 4] === 'm')) |
303
|
7 |
|
&& (($this->query[$i + 5] === 'I') || ($this->query[$i + 5] === 'i')) |
304
|
7 |
|
&& (($this->query[$i + 6] === 'T') || ($this->query[$i + 6] === 't')) |
305
|
7 |
|
&& (($this->query[$i + 7] === 'E') || ($this->query[$i + 7] === 'e')) |
306
|
7 |
|
&& (($this->query[$i + 8] === 'R') || ($this->query[$i + 8] === 'r')) |
307
|
7 |
|
&& (Context::isWhitespace($this->query[$i + 9])) |
308
|
7 |
|
&& (trim($this->current) === '') |
309
|
7 |
|
) { |
310
|
|
|
// Saving the current index to be able to revert any parsing |
311
|
|
|
// done in this block. |
312
|
4 |
|
$iBak = $i; |
313
|
4 |
|
$i += 9; // Skipping `DELIMITER`. |
314
|
|
|
|
315
|
|
|
// Skipping whitespaces. |
316
|
4 |
|
while (($i < $len) && (Context::isWhitespace($this->query[$i]))) { |
317
|
4 |
|
++$i; |
318
|
4 |
|
} |
319
|
|
|
|
320
|
|
|
// Parsing the delimiter. |
321
|
4 |
|
$delimiter = ''; |
322
|
4 |
|
while (($i < $len) && (!Context::isWhitespace($this->query[$i]))) { |
323
|
4 |
|
$delimiter .= $this->query[$i++]; |
324
|
4 |
|
} |
325
|
|
|
|
326
|
|
|
// Checking if the delimiter definition ended. |
327
|
4 |
|
if (($delimiter != '') |
328
|
4 |
|
&& ((($i < $len) && (Context::isWhitespace($this->query[$i]))) |
329
|
1 |
|
|| (($i === $len) && ($end))) |
330
|
4 |
|
) { |
331
|
|
|
// Saving the delimiter. |
332
|
4 |
|
$this->setDelimiter($delimiter); |
333
|
|
|
|
334
|
|
|
// Whether this statement should be returned or not. |
335
|
4 |
|
$ret = ''; |
336
|
4 |
|
if (!empty($this->options['parse_delimiter'])) { |
337
|
|
|
// Appending the `DELIMITER` statement that was just |
338
|
|
|
// found to the current statement. |
339
|
2 |
|
$ret = trim( |
340
|
2 |
|
$this->current . ' ' . substr($this->query, $iBak, $i - $iBak) |
341
|
2 |
|
); |
342
|
2 |
|
} |
343
|
|
|
|
344
|
|
|
// Removing the statement that was just extracted from the |
345
|
|
|
// query. |
346
|
4 |
|
$this->query = substr($this->query, $i); |
347
|
4 |
|
$i = 0; |
348
|
|
|
|
349
|
|
|
// Resetting the current statement. |
350
|
4 |
|
$this->current = ''; |
351
|
|
|
|
352
|
4 |
|
return $ret; |
353
|
|
|
} |
354
|
|
|
|
355
|
|
|
// Incomplete statement. Reverting |
356
|
1 |
|
$i = $iBak; |
357
|
1 |
|
return false; |
358
|
|
|
} |
359
|
|
|
|
360
|
|
|
/* |
361
|
|
|
* Checking if the current statement finished. |
362
|
|
|
* |
363
|
|
|
* The first letter of the delimiter is being checked as an |
364
|
|
|
* optimization. This code is almost as fast as the one above. |
365
|
|
|
* |
366
|
|
|
* There is no point in checking if two strings match if not even |
367
|
|
|
* the first letter matches. |
368
|
|
|
*/ |
369
|
7 |
|
if (($this->query[$i] === $this->delimiter[0]) |
370
|
7 |
|
&& (($this->delimiterLen === 1) |
371
|
4 |
|
|| (substr($this->query, $i, $this->delimiterLen) === $this->delimiter)) |
372
|
7 |
|
) { |
373
|
|
|
// Saving the statement that just ended. |
374
|
7 |
|
$ret = $this->current; |
375
|
|
|
|
376
|
|
|
// If needed, adds a delimiter at the end of the statement. |
377
|
7 |
|
if (!empty($this->options['add_delimiter'])) { |
378
|
5 |
|
$ret .= $this->delimiter; |
379
|
5 |
|
} |
380
|
|
|
|
381
|
|
|
// Removing the statement that was just extracted from the |
382
|
|
|
// query. |
383
|
7 |
|
$this->query = substr($this->query, $i + $this->delimiterLen); |
384
|
7 |
|
$i = 0; |
385
|
|
|
|
386
|
|
|
// Resetting the current statement. |
387
|
7 |
|
$this->current = ''; |
388
|
|
|
|
389
|
|
|
// Returning the statement. |
390
|
7 |
|
return trim($ret); |
391
|
|
|
} |
392
|
|
|
|
393
|
|
|
/* |
394
|
|
|
* Appending current character to current statement. |
395
|
|
|
*/ |
396
|
7 |
|
$this->current .= $this->query[$i]; |
397
|
7 |
|
} |
398
|
|
|
|
399
|
7 |
|
if (($end) && ($i === $len)) { |
400
|
|
|
// If the end of the buffer was reached, the buffer is emptied and |
401
|
|
|
// the current statement that was extracted is returned. |
402
|
5 |
|
$ret = $this->current; |
403
|
|
|
|
404
|
|
|
// Emptying the buffer. |
405
|
5 |
|
$this->query = ''; |
406
|
5 |
|
$i = 0; |
407
|
|
|
|
408
|
|
|
// Resetting the current statement. |
409
|
5 |
|
$this->current = ''; |
410
|
|
|
|
411
|
|
|
// Returning the statement. |
412
|
5 |
|
return trim($ret); |
413
|
|
|
} |
414
|
|
|
|
415
|
7 |
|
return ''; |
416
|
|
|
} |
417
|
|
|
} |
418
|
|
|
|
This check compares the return type specified in the
@return
annotation of a function or method doc comment with the types returned by the function and raises an issue if they mismatch.