|
1
|
|
|
<?php |
|
2
|
|
|
/* @description Transformation Style Sheets - Revolutionising PHP templating * |
|
3
|
|
|
* @author Tom Butler [email protected] * |
|
4
|
|
|
* @copyright 2015 Tom Butler <[email protected]> | https://r.je/ * |
|
5
|
|
|
* @license http://www.opensource.org/licenses/bsd-license.php BSD License * |
|
6
|
|
|
* @version 1.0 */ |
|
7
|
|
|
namespace Transphporm\Parser; |
|
8
|
|
|
class Tokenizer { |
|
9
|
|
|
private $str; |
|
10
|
|
|
const NAME = 'LITERAL'; |
|
11
|
|
|
const STRING = 'STRING'; |
|
12
|
|
|
const OPEN_BRACKET = 'OPEN_BRACKET'; |
|
13
|
|
|
const CLOSE_BRACKET = 'CLOSE_BRACKET'; |
|
14
|
|
|
const OPEN_SQUARE_BRACKET = 'SQUARE_BRACKET'; |
|
15
|
|
|
const CLOSE_SQUARE_BRACKET = 'CLOSE_SQUARE_BRACKET'; |
|
16
|
|
|
const CONCAT = 'CONCAT'; |
|
17
|
|
|
const ARG = 'ARG'; |
|
18
|
|
|
const WHITESPACE = 'WHITESPACE'; |
|
19
|
|
|
const NEW_LINE = 'NEW_LINE'; |
|
20
|
|
|
const DOT = 'DOT'; |
|
21
|
|
|
const NUMERIC = 'NUMERIC'; |
|
22
|
|
|
const EQUALS = 'EQUALS'; |
|
23
|
|
|
const NOT = 'NOT'; |
|
24
|
|
|
const OPEN_BRACE = 'OPEN_BRACE'; |
|
25
|
|
|
const CLOSE_BRACE = 'CLOSE_BRACE'; |
|
26
|
|
|
const BOOL = 'BOOL'; |
|
27
|
|
|
const COLON = 'COLON'; |
|
28
|
|
|
const SEMI_COLON = 'SEMI_COLON'; |
|
29
|
|
|
const NUM_SIGN = 'NUM_SIGN'; |
|
30
|
|
|
const GREATER_THAN = 'GREATER_THAN'; |
|
31
|
|
|
const AT_SIGN = 'AT_SIGN'; |
|
32
|
|
|
const SUBTRACT = 'SUBTRACT'; |
|
33
|
|
|
const MULTIPLY = 'MULTIPLY'; |
|
34
|
|
|
const DIVIDE = 'DIVIDE'; |
|
35
|
|
|
|
|
36
|
|
|
private $chars = [ |
|
37
|
|
|
'"' => self::STRING, |
|
38
|
|
|
'\'' => self::STRING, |
|
39
|
|
|
'(' => self::OPEN_BRACKET, |
|
40
|
|
|
')' => self::CLOSE_BRACKET, |
|
41
|
|
|
'[' => self::OPEN_SQUARE_BRACKET, |
|
42
|
|
|
']' => self::CLOSE_SQUARE_BRACKET, |
|
43
|
|
|
'+' => self::CONCAT, |
|
44
|
|
|
',' => self::ARG, |
|
45
|
|
|
'.' => self::DOT, |
|
46
|
|
|
'!' => self::NOT, |
|
47
|
|
|
'=' => self::EQUALS, |
|
48
|
|
|
'{' => self::OPEN_BRACE, |
|
49
|
|
|
'}' => self::CLOSE_BRACE, |
|
50
|
|
|
':' => self::COLON, |
|
51
|
|
|
';' => self::SEMI_COLON, |
|
52
|
|
|
'#' => self::NUM_SIGN, |
|
53
|
|
|
'>' => self::GREATER_THAN, |
|
54
|
|
|
'@' => self::AT_SIGN, |
|
55
|
|
|
'-' => self::SUBTRACT, |
|
56
|
|
|
'*' => self::MULTIPLY, |
|
57
|
|
|
'/' => self::DIVIDE, |
|
58
|
|
|
' ' => self::WHITESPACE, |
|
59
|
|
|
"\n" => self::NEW_LINE, |
|
60
|
|
|
"\r" => self::WHITESPACE, |
|
61
|
|
|
"\t" => self::WHITESPACE |
|
62
|
|
|
]; |
|
63
|
|
|
|
|
64
|
|
|
public function __construct($str) { |
|
65
|
|
|
$this->str = $str; |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
public function getTokens($returnObj = true) { |
|
69
|
|
|
$tokens = []; |
|
70
|
|
|
|
|
71
|
|
|
for ($i = 0; $i < strlen($this->str); $i++) { |
|
72
|
|
|
$char = $this->identifyChar($this->str[$i]); |
|
73
|
|
|
|
|
74
|
|
|
$this->doSimpleTokens($tokens, $char); |
|
75
|
|
|
$this->doLiterals($tokens, $char, $i); |
|
76
|
|
|
$i += $this->doStrings($tokens, $char, $i); |
|
77
|
|
|
$i += $this->doBrackets($tokens, $char, $i); |
|
78
|
|
|
} |
|
79
|
|
|
if ($returnObj) return new Tokens($tokens); |
|
80
|
|
|
else return $tokens; |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
private function doSimpleTokens(&$tokens, $char) { |
|
84
|
|
|
if (in_array($char, [Tokenizer::ARG, Tokenizer::CONCAT, Tokenizer::DOT, Tokenizer::NOT, Tokenizer::EQUALS, |
|
85
|
|
|
Tokenizer::COLON, Tokenizer::SEMI_COLON, Tokenizer::WHITESPACE, Tokenizer::NEW_LINE, Tokenizer::NUM_SIGN, |
|
86
|
|
|
Tokenizer::GREATER_THAN, Tokenizer::AT_SIGN, Tokenizer::SUBTRACT, Tokenizer::MULTIPLY, Tokenizer::DIVIDE])) { |
|
87
|
|
|
$tokens[] = ['type' => $char]; |
|
88
|
|
|
} |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
private function isLiteral($n) { |
|
92
|
|
|
//Is it a normal literal character |
|
93
|
|
|
return isset($this->str[$n]) && ($this->identifyChar($this->str[$n]) == self::NAME |
|
94
|
|
|
//but a subtract can be part of a class name or a mathematical operation |
|
95
|
|
|
|| ($this->identifyChar($this->str[$n]) == self::SUBTRACT && !is_numeric($this->str[$n-1]))); |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
private function doLiterals(&$tokens, $char, &$i) { |
|
99
|
|
|
if ($char === self::NAME) { |
|
100
|
|
|
$name = $this->str[$i]; |
|
101
|
|
|
while ($this->isLiteral($i+1)) { |
|
102
|
|
|
$name .= $this->str[$i+1]; |
|
103
|
|
|
$i++; |
|
104
|
|
|
} |
|
105
|
|
|
$this->processLiterals($tokens, $name); |
|
106
|
|
|
} |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
|
|
private function processLiterals(&$tokens, $name) { |
|
110
|
|
|
if (is_numeric($name)) $tokens[] = ['type' => self::NUMERIC, 'value' => $name]; |
|
111
|
|
|
else if ($name == 'true') $tokens[] = ['type' => self::BOOL, 'value' => true]; |
|
112
|
|
|
else if ($name == 'false') $tokens[] = ['type' => self::BOOL, 'value' => false]; |
|
113
|
|
|
else $tokens[] = ['type' => self::NAME, 'value' => $name]; |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
private function doBrackets(&$tokens, $char, $i) { |
|
117
|
|
|
$types = [ |
|
118
|
|
|
self::OPEN_BRACKET => ['(', ')'], |
|
119
|
|
|
self::OPEN_BRACE => ['{', '}'], |
|
120
|
|
|
self::OPEN_SQUARE_BRACKET => ['[', ']'] |
|
121
|
|
|
]; |
|
122
|
|
|
|
|
123
|
|
|
foreach ($types as $type => $brackets) { |
|
124
|
|
|
if ($char === $type) { |
|
125
|
|
|
$contents = $this->extractBrackets($i, $brackets[0], $brackets[1]); |
|
126
|
|
|
$tokenizer = new Tokenizer($contents); |
|
127
|
|
|
$tokens[] = ['type' => $type, 'value' => $tokenizer->getTokens(), 'string' => $contents]; |
|
128
|
|
|
return strlen($contents); |
|
129
|
|
|
} |
|
130
|
|
|
} |
|
131
|
|
|
} |
|
132
|
|
|
|
|
133
|
|
|
private function doStrings(&$tokens, $char, $i) { |
|
134
|
|
|
if ($char === self::STRING) { |
|
135
|
|
|
$string = $this->extractString($i); |
|
136
|
|
|
$length = strlen($string)+1; |
|
137
|
|
|
$char = $this->getChar($char); |
|
138
|
|
|
$string = str_replace('\\' . $char, $char, $string); |
|
139
|
|
|
$tokens[] = ['type' => self::STRING, 'value' => $string]; |
|
140
|
|
|
return $length; |
|
141
|
|
|
} |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
|
|
private function extractString($pos) { |
|
145
|
|
|
$char = $this->str[$pos]; |
|
146
|
|
|
$end = strpos($this->str, $char, $pos+1); |
|
147
|
|
|
while ($end !== false && $this->str[$end-1] == '\\') $end = strpos($this->str, $char, $end+1); |
|
148
|
|
|
|
|
149
|
|
|
return substr($this->str, $pos+1, $end-$pos-1); |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
|
|
private function extractBrackets($open, $startBracket = '(', $closeBracket = ')') { |
|
153
|
|
|
$close = strpos($this->str, $closeBracket, $open); |
|
154
|
|
|
|
|
155
|
|
|
$cPos = $open+1; |
|
156
|
|
|
while (($cPos = strpos($this->str, $startBracket, $cPos+1)) !== false && $cPos < $close) $close = strpos($this->str, $closeBracket, $close+1); |
|
157
|
|
|
return substr($this->str, $open+1, $close-$open-1); |
|
158
|
|
|
} |
|
159
|
|
|
|
|
160
|
|
|
private function identifyChar($chr) { |
|
161
|
|
|
if (isset($this->chars[$chr])) return $this->chars[$chr]; |
|
162
|
|
|
else return self::NAME; |
|
163
|
|
|
} |
|
164
|
|
|
|
|
165
|
|
|
private function getChar($num) { |
|
166
|
|
|
$chars = array_reverse($this->chars); |
|
167
|
|
|
if (isset($chars[$num])) return $chars[$num]; |
|
168
|
|
|
else return false; |
|
169
|
|
|
} |
|
170
|
|
|
} |
|
171
|
|
|
|
Level-2 /
Transphporm
