1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace ntentan\nibii; |
4
|
|
|
|
5
|
|
|
/** |
6
|
|
|
* Safely compiles SQL conditions to ensure that a portable interface is provided |
7
|
|
|
* through which conditions can be specified accross database platforms. Also |
8
|
|
|
* the FilterCompiler ensures that raw data is never passed through queries. |
9
|
|
|
* This is done in order to minimize injection errors. |
10
|
|
|
*/ |
11
|
|
|
class FilterCompiler { |
12
|
|
|
|
13
|
|
|
private $lookahead; |
14
|
|
|
private $token; |
15
|
|
|
private $filter; |
16
|
|
|
private $tokens = array( |
17
|
|
|
'equals' => '\=', |
18
|
|
|
'number' => '[0-9]+', |
19
|
|
|
'cast' => 'cast\b', |
20
|
|
|
'as' => 'as\b', |
21
|
|
|
'between' => 'between\b', |
22
|
|
|
'in' => 'in\b', |
23
|
|
|
'like' => 'like\b', |
24
|
|
|
'is' => 'is\b', |
25
|
|
|
'and' => 'and\b', |
26
|
|
|
'not' => 'not\b', |
27
|
|
|
'or' => 'or\b', |
28
|
|
|
'greater_or_equal' => '\>\=', |
29
|
|
|
'less_or_equal' => '\<\=', |
30
|
|
|
'not_equal' => '\<\>', |
31
|
|
|
'greater' => '\>', |
32
|
|
|
'less' => '\<', |
33
|
|
|
'add' => '\+', |
34
|
|
|
'subtract' => '\-', |
35
|
|
|
'multiply' => '\*', |
36
|
|
|
'function' => '[a-zA-Z][a-zA-Z0-9\_]*\s*\(', |
37
|
|
|
'identifier' => '[a-zA-Z][a-zA-Z0-9\.\_\:]*\b', |
38
|
|
|
'named_bind_param' => '\:[a-z_][a-z0-9\_]+', |
39
|
|
|
'position_bind_param' => '\\?', |
40
|
|
|
'obracket' => '\(', |
41
|
|
|
'cbracket' => '\)', |
42
|
|
|
'comma' => ',' |
43
|
|
|
); |
44
|
|
|
private $operators = array( |
45
|
|
|
array('between', 'or' /* , 'like' */), |
46
|
|
|
array('and'), |
47
|
|
|
array('not'), |
48
|
|
|
array('equals', 'greater', 'less', 'greater_or_equal', 'less_or_equal', 'not_equal', 'is'), |
49
|
|
|
array('add', 'subtract'), |
50
|
|
|
array('in'), |
51
|
|
|
array('multiply') |
52
|
|
|
); |
53
|
|
|
private $numPositions = 0; |
54
|
|
|
|
55
|
14 |
|
public function compile($filter) { |
|
|
|
|
56
|
14 |
|
$this->filter = $filter; |
57
|
14 |
|
$this->getToken(); |
58
|
14 |
|
$expression = $this->parseExpression(); |
59
|
12 |
|
if ($this->token !== false) { |
60
|
4 |
|
throw new FilterCompilerException("Unexpected '" . $this->token . "' in filter [$filter]"); |
61
|
|
|
} |
62
|
8 |
|
$parsed = $this->renderExpression($expression); |
63
|
8 |
|
return $parsed; |
64
|
|
|
} |
65
|
|
|
|
66
|
8 |
|
private function renderExpression($expression) { |
|
|
|
|
67
|
8 |
|
if (is_array($expression)) { |
68
|
8 |
|
$expression = $this->renderExpression($expression['left']) . " {$expression['opr']} " . $this->renderExpression($expression['right']); |
69
|
|
|
} |
70
|
8 |
|
return $expression; |
71
|
|
|
} |
72
|
|
|
|
73
|
4 |
|
private function match($tokens) { |
74
|
4 |
|
if (is_string($tokens)) { |
75
|
4 |
|
$tokens = [$tokens]; |
76
|
|
|
} |
77
|
4 |
|
if (array_search($this->lookahead, $tokens) === false) { |
78
|
|
|
throw new FilterCompilerException("Expected " . implode(' or ', $tokens) . " but found " . $this->lookahead); |
79
|
|
|
} |
80
|
4 |
|
} |
81
|
|
|
|
82
|
2 |
|
private function parseBetween() { |
|
|
|
|
83
|
2 |
|
$this->match(['named_bind_param', 'number', 'position_bind_param']); |
84
|
2 |
|
$left = $this->token; |
85
|
2 |
|
$this->getToken(); |
86
|
2 |
|
$this->match('and'); |
87
|
2 |
|
$this->getToken(); |
88
|
2 |
|
$this->match(['named_bind_param', 'number', 'position_bind_param']); |
89
|
2 |
|
$right = $this->token; |
90
|
2 |
|
$this->getToken(); |
91
|
2 |
|
return "$left AND $right"; |
92
|
|
|
} |
93
|
|
|
|
94
|
2 |
|
private function parseIn() { |
95
|
2 |
|
$expression = "("; |
96
|
2 |
|
$this->match('obracket'); |
97
|
2 |
|
$this->getToken(); |
98
|
|
|
|
99
|
|
|
do { |
100
|
2 |
|
$expression .= $this->parseExpression(); |
101
|
2 |
|
if ($this->lookahead === 'comma') { |
102
|
2 |
|
$expression .= ','; |
103
|
2 |
|
$this->getToken(); |
104
|
2 |
|
continue; |
105
|
|
|
} else { |
106
|
2 |
|
break; |
107
|
|
|
} |
108
|
2 |
|
} while (true); |
109
|
|
|
|
110
|
2 |
|
$this->match('cbracket'); |
111
|
|
|
|
112
|
2 |
|
$this->getToken(); |
113
|
|
|
|
114
|
2 |
|
$expression .= ')'; |
115
|
2 |
|
return $expression; |
116
|
|
|
} |
117
|
|
|
|
118
|
4 |
|
private function parseFunctionParams() { |
119
|
4 |
|
$parameters = ''; |
120
|
4 |
|
$size = 0; |
121
|
|
|
do { |
122
|
4 |
|
$size++; |
123
|
4 |
|
$parameters .= $this->renderExpression($this->parseExpression()); |
124
|
4 |
|
if ($this->lookahead == 'comma') { |
125
|
2 |
|
$this->getToken(); |
126
|
2 |
|
$parameters .= ", "; |
127
|
4 |
|
} else if ($this->lookahead == 'cbracket') { |
128
|
4 |
|
break; |
129
|
|
|
} |
130
|
2 |
|
} while ($size < 100); |
131
|
4 |
|
return $parameters; |
132
|
|
|
} |
133
|
|
|
|
134
|
2 |
|
private function parseCast() { |
135
|
2 |
|
$return = 'cast('; |
136
|
2 |
|
$this->getToken(); |
137
|
2 |
|
$this->match('obracket'); |
138
|
2 |
|
$this->getToken(); |
139
|
2 |
|
$return .= $this->renderExpression($this->parseExpression()); |
140
|
2 |
|
$this->match('as'); |
141
|
2 |
|
$return .= ' as '; |
142
|
2 |
|
$this->getToken(); |
143
|
2 |
|
$this->match('identifier'); |
144
|
2 |
|
$return .= $this->token; |
145
|
2 |
|
$this->getToken(); |
146
|
2 |
|
$this->match('cbracket'); |
147
|
2 |
|
$return .= ')'; |
148
|
2 |
|
return $return; |
149
|
|
|
} |
150
|
|
|
|
151
|
4 |
|
private function parseFunction() { |
|
|
|
|
152
|
4 |
|
$name = $this->token; |
153
|
4 |
|
$this->getToken(); |
154
|
4 |
|
$parameters = $this->parseFunctionParams(); |
155
|
4 |
|
return "$name$parameters)"; |
156
|
|
|
} |
157
|
|
|
|
158
|
12 |
|
private function returnToken() { |
159
|
12 |
|
return $this->token; |
160
|
|
|
} |
161
|
|
|
|
162
|
10 |
|
private function returnPositionTag() { |
163
|
10 |
|
return ":filter_bind_" . ( ++$this->numPositions); |
164
|
|
|
} |
165
|
|
|
|
166
|
2 |
|
private function parseObracket() { |
|
|
|
|
167
|
2 |
|
$this->getToken(); |
168
|
2 |
|
$expression = $this->parseExpression(); |
169
|
2 |
|
return $this->renderExpression($expression); |
170
|
|
|
} |
171
|
|
|
|
172
|
14 |
|
private function parseFactor() { |
|
|
|
|
173
|
14 |
|
$return = null; |
174
|
|
|
$methods = [ |
175
|
14 |
|
'cast' => 'parseCast', |
176
|
|
|
'function' => 'parseFunction', |
177
|
|
|
'identifier' => 'returnToken', |
178
|
|
|
'named_bind_param' => 'returnToken', |
179
|
|
|
'number' => 'returnToken', |
180
|
|
|
'position_bind_param' => 'returnPositionTag', |
181
|
|
|
'obracket' => 'parseObracket' |
182
|
|
|
]; |
183
|
|
|
|
184
|
14 |
|
if (isset($methods[$this->lookahead])) { |
185
|
12 |
|
$method = $methods[$this->lookahead]; |
186
|
12 |
|
$return = $this->$method(); |
187
|
|
|
} |
188
|
|
|
|
189
|
14 |
|
$this->getToken(); |
190
|
14 |
|
return $return; |
191
|
|
|
} |
192
|
|
|
|
193
|
10 |
|
private function parseRightExpression($level, $opr) { |
|
|
|
|
194
|
|
|
switch ($opr) { |
195
|
10 |
|
case 'between': return $this->parseBetween(); |
|
|
|
|
196
|
10 |
|
case 'in': return $this->parseIn(); |
|
|
|
|
197
|
10 |
|
default: return $this->parseExpression($level); |
|
|
|
|
198
|
|
|
} |
199
|
|
|
} |
200
|
|
|
|
201
|
14 |
|
private function parseExpression($level = 0) { |
|
|
|
|
202
|
14 |
|
if ($level === count($this->operators)) { |
203
|
14 |
|
return $this->parseFactor(); |
204
|
|
|
} else { |
205
|
14 |
|
$expression = $this->parseExpression($level + 1); |
206
|
|
|
} |
207
|
|
|
|
208
|
14 |
|
while ($this->token != false) { |
209
|
14 |
|
if (array_search($this->lookahead, $this->operators[$level]) !== false) { |
210
|
12 |
|
$left = $expression; |
211
|
12 |
|
$opr = $this->token; |
212
|
12 |
|
$this->getToken(); |
213
|
10 |
|
$right = $this->parseRightExpression($level + 1, strtolower($opr)); |
214
|
|
|
$expression = array( |
215
|
10 |
|
'left' => $left, |
216
|
10 |
|
'opr' => $opr, |
217
|
10 |
|
'right' => $right |
218
|
|
|
); |
219
|
|
|
} else { |
220
|
14 |
|
break; |
221
|
|
|
} |
222
|
|
|
} |
223
|
|
|
|
224
|
14 |
|
return $expression; |
225
|
|
|
} |
226
|
|
|
|
227
|
14 |
|
private function getToken() { |
228
|
14 |
|
$this->eatWhite(); |
229
|
14 |
|
$this->token = false; |
230
|
14 |
|
foreach ($this->tokens as $token => $regex) { |
231
|
14 |
|
if (preg_match("/^$regex/i", $this->filter, $matches)) { |
232
|
14 |
|
$this->filter = substr($this->filter, strlen($matches[0])); |
233
|
14 |
|
$this->lookahead = $token; |
234
|
14 |
|
$this->token = $matches[0]; |
235
|
14 |
|
break; |
236
|
|
|
} |
237
|
|
|
} |
238
|
|
|
|
239
|
14 |
|
if ($this->token === false && strlen($this->filter) > 0) { |
240
|
2 |
|
throw new FilterCompilerException("Unexpected character [" . $this->filter[0] . "] begining " . $this->filter . "."); |
241
|
|
|
} |
242
|
14 |
|
} |
243
|
|
|
|
244
|
14 |
|
private function eatWhite() { |
245
|
14 |
|
if (preg_match("/^\s*/", $this->filter, $matches)) { |
246
|
14 |
|
$this->filter = substr($this->filter, strlen($matches[0])); |
247
|
|
|
} |
248
|
14 |
|
} |
249
|
|
|
|
250
|
6 |
|
public function rewriteBoundData($data) { |
251
|
6 |
|
$rewritten = []; |
252
|
6 |
|
foreach ($data as $key => $value) { |
253
|
6 |
|
if (is_numeric($key)) { |
254
|
6 |
|
$rewritten["filter_bind_" . ($key + 1)] = $value; |
255
|
|
|
} else { |
256
|
6 |
|
$rewritten[$key] = $value; |
257
|
|
|
} |
258
|
|
|
} |
259
|
6 |
|
return $rewritten; |
260
|
|
|
} |
261
|
|
|
|
262
|
|
|
} |
263
|
|
|
|
Our type inference engine in quite powerful, but sometimes the code does not provide enough clues to go by. In these cases we request you to add a
@return
annotation as described here.