|
1
|
|
|
<?php |
|
2
|
|
|
namespace Netdudes\DataSourceryBundle\UQL; |
|
3
|
|
|
|
|
4
|
|
|
class Tokens |
|
5
|
|
|
{ |
|
6
|
|
|
public static $terminalTokens; |
|
7
|
|
|
|
|
8
|
|
|
public static $tokenCategories = [ |
|
9
|
|
|
'GROUP_START' => ['T_BRACKET_OPEN'], |
|
10
|
|
|
'GROUP_END' => ['T_BRACKET_CLOSE'], |
|
11
|
|
|
'ARRAY_START' => ['T_ARRAY_OPEN'], |
|
12
|
|
|
'ARRAY_END' => ['T_ARRAY_CLOSE'], |
|
13
|
|
|
'ARRAY_SEPARATOR' => ['T_ARRAY_SEPARATOR'], |
|
14
|
|
|
'OPERATOR' => [ |
|
15
|
|
|
'T_OP_EQ', |
|
16
|
|
|
'T_OP_NEQ', |
|
17
|
|
|
'T_OP_LT', |
|
18
|
|
|
'T_OP_GT', |
|
19
|
|
|
'T_OP_LTE', |
|
20
|
|
|
'T_OP_GTE', |
|
21
|
|
|
'T_OP_LIKE', |
|
22
|
|
|
'T_OP_IN', |
|
23
|
|
|
'T_OP_NIN', |
|
24
|
|
|
], |
|
25
|
|
|
'LOGIC' => [ |
|
26
|
|
|
'T_OP_AND', |
|
27
|
|
|
'T_OP_OR', |
|
28
|
|
|
'T_OP_XOR', |
|
29
|
|
|
], |
|
30
|
|
|
'LITERAL' => [ |
|
31
|
|
|
'T_LITERAL_TRUE', |
|
32
|
|
|
'T_LITERAL_FALSE', |
|
33
|
|
|
'T_LITERAL_EMPTY', |
|
34
|
|
|
'T_FUNCTION_CALL', |
|
35
|
|
|
'T_LITERAL' |
|
36
|
|
|
], |
|
37
|
|
|
'IDENTIFIER' => ['T_IDENTIFIER'], |
|
38
|
|
|
]; |
|
39
|
|
|
|
|
40
|
|
|
public static $tokenCanonicalValues = [ |
|
41
|
|
|
'T_BRACKET_OPEN' => ['('], |
|
42
|
|
|
'T_BRACKET_CLOSE' => [')'], |
|
43
|
|
|
'T_ARRAY_OPEN' => ['['], |
|
44
|
|
|
'T_ARRAY_CLOSE' => [']'], |
|
45
|
|
|
'T_ARRAY_SEPARATOR' => [','], |
|
46
|
|
|
'T_OP_NEQ' => ['!='], |
|
47
|
|
|
'T_OP_LTE' => ['<='], |
|
48
|
|
|
'T_OP_GTE' => ['>='], |
|
49
|
|
|
'T_OP_LT' => ['<'], |
|
50
|
|
|
'T_OP_GT' => ['>'], |
|
51
|
|
|
'T_OP_LIKE' => ['~'], |
|
52
|
|
|
'T_OP_EQ' => ['='], |
|
53
|
|
|
'T_OP_IN' => ['in'], |
|
54
|
|
|
'T_OP_AND' => ['and'], |
|
55
|
|
|
'T_OP_OR' => ['or'], |
|
56
|
|
|
'T_OP_XOR' => ['xor'], |
|
57
|
|
|
'T_LITERAL_FALSE' => ['false'], |
|
58
|
|
|
'T_LITERAL_TRUE' => ['true'], |
|
59
|
|
|
'T_LITERAL_EMPTY' => ['empty'], |
|
60
|
|
|
]; |
|
61
|
|
|
|
|
62
|
|
|
private static $baseTokens = [ |
|
63
|
|
|
"/^(\\()/" => "T_BRACKET_OPEN", |
|
64
|
|
|
"/^(\\))/" => "T_BRACKET_CLOSE", |
|
65
|
|
|
"/^(\\[)/" => "T_ARRAY_OPEN", |
|
66
|
|
|
"/^(\\])/" => "T_ARRAY_CLOSE", |
|
67
|
|
|
"/^(,)/" => "T_ARRAY_SEPARATOR", |
|
68
|
|
|
"/^(AND)(\s|$)/i" => "T_LOGIC_AND", |
|
69
|
|
|
"/^(XOR)(\s|$)/i" => "T_LOGIC_XOR", |
|
70
|
|
|
"/^(OR)(\s|$)/i" => "T_LOGIC_OR", |
|
71
|
|
|
"/^(false)(\s|$)/i" => "T_LITERAL_FALSE", |
|
72
|
|
|
"/^(true)(\s|$)/i" => "T_LITERAL_TRUE", |
|
73
|
|
|
"/^(empty)(\s|$)/i" => "T_LITERAL_EMPTY", |
|
74
|
|
|
"/^([a-zA-Z0-9]+\([^\(\)]*\))/" => "T_FUNCTION_CALL", |
|
75
|
|
|
"/^([0-9]+|'[^']*'|\"[^\"]*\")/" => "T_LITERAL", |
|
76
|
|
|
"/^([a-zA-Z\\_]+)/" => "T_IDENTIFIER", |
|
77
|
|
|
"/^(\\s+)/" => "T_WHITESPACE", |
|
78
|
|
|
]; |
|
79
|
|
|
|
|
80
|
|
|
private static $operators = [ |
|
81
|
|
|
"STRING_NEQ" => ["!=", "not", "is not"], |
|
82
|
|
|
"NUMERIC_NEQ" => ["!=", "not", "is not"], |
|
83
|
|
|
"DATETIME_NEQ" => ["!=", "<>", "not", "is not"], |
|
84
|
|
|
"NUMERIC_LTE" => ["<="], |
|
85
|
|
|
"DATETIME_LTE" => ["<="], |
|
86
|
|
|
"NUMERIC_GTE" => [">="], |
|
87
|
|
|
"DATETIME_GTE" => [">="], |
|
88
|
|
|
"NUMERIC_LT" => ["<", "less than"], |
|
89
|
|
|
"DATETIME_LT" => ["<", "before"], |
|
90
|
|
|
"NUMERIC_GT" => [">", "more than"], |
|
91
|
|
|
"DATETIME_GT" => [">", "after"], |
|
92
|
|
|
"STRING_LIKE" => ["~", "like"], |
|
93
|
|
|
"STRING_EQ" => ["=", "equals", "is"], |
|
94
|
|
|
"NUMERIC_EQ" => ["=", "equals", "is"], |
|
95
|
|
|
"DATETIME_EQ" => ["=", "on", "is", "at"], |
|
96
|
|
|
"IN" => ["in"], |
|
97
|
|
|
"NIN" => ["not in"], |
|
98
|
|
|
]; |
|
99
|
|
|
|
|
100
|
|
|
private static $operatorTokenAssociations = [ |
|
101
|
|
|
"T_OP_IN" => ["IN"], |
|
102
|
|
|
"T_OP_NIN" => ["NIN"], |
|
103
|
|
|
"T_OP_NEQ" => ["STRING_NEQ", "NUMERIC_NEQ", "DATETIME_NEQ"], |
|
104
|
|
|
"T_OP_EQ" => ["STRING_EQ", "NUMERIC_EQ", "DATETIME_EQ"], |
|
105
|
|
|
"T_OP_GTE" => ["NUMERIC_GTE", "DATETIME_GTE"], |
|
106
|
|
|
"T_OP_GT" => ["NUMERIC_GT", "DATETIME_GT"], |
|
107
|
|
|
"T_OP_LTE" => ["NUMERIC_LTE", "DATETIME_LTE"], |
|
108
|
|
|
"T_OP_LT" => ["NUMERIC_LT", "DATETIME_LT"], |
|
109
|
|
|
"T_OP_LIKE" => ["STRING_LIKE"], |
|
110
|
|
|
]; |
|
111
|
|
|
|
|
112
|
|
|
/** |
|
113
|
|
|
* @var bool |
|
114
|
|
|
*/ |
|
115
|
|
|
private static $initialised = false; |
|
116
|
|
|
|
|
117
|
|
|
public static function initialize() |
|
118
|
|
|
{ |
|
119
|
|
|
if (self::$initialised) { |
|
120
|
|
|
return; |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
$tokens = []; |
|
124
|
|
|
foreach (self::$operatorTokenAssociations as $token => $operators) { |
|
125
|
|
|
$strings = []; |
|
126
|
|
|
foreach ($operators as $operator) { |
|
127
|
|
|
$strings = array_merge($strings, self::$operators[$operator]); |
|
128
|
|
|
} |
|
129
|
|
|
$strings = array_unique($strings); |
|
130
|
|
|
usort( |
|
131
|
|
|
$strings, |
|
132
|
|
|
function ($a, $b) { |
|
133
|
|
|
return strlen($a) - strlen($b); |
|
134
|
|
|
} |
|
135
|
|
|
); |
|
136
|
|
|
$regex = "/^(" . implode("|", $strings) . ")(\s|$)/i"; |
|
137
|
|
|
$tokens[$regex] = $token; |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
|
|
self::$terminalTokens = array_merge($tokens, self::$baseTokens); |
|
141
|
|
|
self::$initialised = true; |
|
142
|
|
|
} |
|
143
|
|
|
} |
|
144
|
|
|
|
|
145
|
|
|
// Initialise the static class |
|
146
|
|
|
Tokens::initialize(); |
|
147
|
|
|
|