|
1
|
|
|
<?php |
|
2
|
|
|
/****************************************************************************** |
|
3
|
|
|
* An implementation of dicto (scg.unibe.ch/dicto) in and for PHP. |
|
4
|
|
|
* |
|
5
|
|
|
* Copyright (c) 2016 Richard Klees <[email protected]> |
|
6
|
|
|
* |
|
7
|
|
|
* This software is licensed under The MIT License. You should have received |
|
8
|
|
|
* a copy of the license along with the code. |
|
9
|
|
|
*/ |
|
10
|
|
|
|
|
11
|
|
|
namespace Lechimp\Dicto\Definition; |
|
12
|
|
|
|
|
13
|
|
|
/** |
|
14
|
|
|
* Tokenizes a rules file. |
|
15
|
|
|
*/ |
|
16
|
|
|
class Tokenizer implements \Iterator { |
|
17
|
|
|
/** |
|
18
|
|
|
* @var SymbolTable |
|
19
|
|
|
*/ |
|
20
|
|
|
protected $symbol_table; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* @var mixed[] |
|
24
|
|
|
*/ |
|
25
|
|
|
protected $tokens; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* @var int |
|
29
|
|
|
*/ |
|
30
|
|
|
protected $position; |
|
31
|
|
|
|
|
32
|
|
|
/** |
|
33
|
|
|
* @var string |
|
34
|
|
|
*/ |
|
35
|
|
|
protected $source; |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* @var string |
|
39
|
|
|
*/ |
|
40
|
|
|
protected $unparsed; |
|
41
|
|
|
|
|
42
|
|
|
/** |
|
43
|
|
|
* @var int |
|
44
|
|
|
*/ |
|
45
|
|
|
protected $parsing_position; |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* @var bool |
|
49
|
|
|
*/ |
|
50
|
|
|
protected $is_end_token_added; |
|
51
|
|
|
|
|
52
|
|
|
static $UNPARSED_PREVIEW_FOR_ERROR = 10; |
|
|
|
|
|
|
53
|
|
|
|
|
54
|
46 |
|
public function __construct(SymbolTable $symbol_table, $source) { |
|
55
|
46 |
|
assert('is_string($source)'); |
|
56
|
46 |
|
$this->symbol_table = $symbol_table; |
|
57
|
46 |
|
$this->tokens = array(); |
|
58
|
46 |
|
$this->position = 0; |
|
59
|
46 |
|
$this->source = $source; |
|
60
|
46 |
|
$this->unparsed = $source; |
|
61
|
46 |
|
$this->is_end_token_added = false; |
|
62
|
46 |
|
} |
|
63
|
|
|
|
|
64
|
|
|
// Methods from Iterator-interface |
|
65
|
|
|
|
|
66
|
|
|
/** |
|
67
|
|
|
* @return array (Symbol,$matches) |
|
68
|
|
|
*/ |
|
69
|
44 |
|
public function current() { |
|
70
|
44 |
|
$this->maybe_parse_next_token(); |
|
71
|
41 |
|
return $this->tokens[$this->position]; |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
/** |
|
75
|
|
|
* @inheritdocs |
|
76
|
|
|
*/ |
|
77
|
1 |
|
public function key() { |
|
78
|
1 |
|
return $this->position; |
|
79
|
|
|
} |
|
80
|
|
|
|
|
81
|
|
|
/** |
|
82
|
|
|
* @inheritdocs |
|
83
|
|
|
*/ |
|
84
|
39 |
|
public function next() { |
|
85
|
39 |
|
$this->position++; |
|
86
|
39 |
|
} |
|
87
|
|
|
|
|
88
|
|
|
/** |
|
89
|
|
|
* @inheritdocs |
|
90
|
|
|
*/ |
|
91
|
1 |
|
public function rewind() { |
|
92
|
1 |
|
$this->position = 0; |
|
93
|
1 |
|
} |
|
94
|
|
|
|
|
95
|
|
|
/** |
|
96
|
|
|
* @inheritdocs |
|
97
|
|
|
*/ |
|
98
|
1 |
|
public function valid() { |
|
99
|
1 |
|
$this->maybe_parse_next_token(); |
|
100
|
1 |
|
return count($this->tokens) > $this->position; |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
|
|
/** |
|
104
|
|
|
* Try to parse the next token if there are currently not enough tokens |
|
105
|
|
|
* in the tokens to get a token for the current position. |
|
106
|
|
|
* |
|
107
|
|
|
* @throws ParserException if next token can not be parsed. |
|
108
|
|
|
*/ |
|
109
|
45 |
|
public function maybe_parse_next_token() { |
|
110
|
45 |
|
if (count($this->tokens) <= $this->position) { |
|
111
|
45 |
|
$this->parse_next_token(); |
|
112
|
42 |
|
} |
|
113
|
42 |
|
} |
|
114
|
|
|
|
|
115
|
|
|
|
|
116
|
|
|
/** |
|
117
|
|
|
* Try to parse the next token from the source. |
|
118
|
|
|
* |
|
119
|
|
|
* @throws ParserException if next token can not be parsed. |
|
120
|
|
|
*/ |
|
121
|
45 |
|
protected function parse_next_token() { |
|
122
|
45 |
|
if ($this->is_everything_parsed()) { |
|
123
|
39 |
|
if (!$this->is_end_token_added) { |
|
124
|
39 |
|
$this->tokens[] = array(new Symbol("", 0), array("")); |
|
125
|
39 |
|
$this->is_end_token_added = true; |
|
126
|
39 |
|
} |
|
127
|
39 |
|
return; |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
42 |
|
foreach ($this->symbol_table->symbols() as $symbol) { |
|
131
|
41 |
|
$re = $symbol->regexp(); |
|
132
|
41 |
|
$matches = array(); |
|
133
|
41 |
|
if (preg_match("%^($re)%s", $this->unparsed, $matches) == 1) { |
|
134
|
39 |
|
unset($matches[1]); |
|
135
|
39 |
|
$this->advance($matches[0]); |
|
136
|
39 |
|
$this->tokens[] = array($symbol, array_values($matches)); |
|
137
|
39 |
|
return; |
|
138
|
|
|
} |
|
139
|
37 |
|
} |
|
140
|
|
|
|
|
141
|
3 |
|
$next = substr($this->unparsed, 0, static::$UNPARSED_PREVIEW_FOR_ERROR); |
|
|
|
|
|
|
142
|
3 |
|
throw new ParserException("Could not match \"$next\"."); |
|
143
|
|
|
} |
|
144
|
|
|
|
|
145
|
|
|
/** |
|
146
|
|
|
* Go forward in the string we have parsed so far. |
|
147
|
|
|
* |
|
148
|
|
|
* @param string $match |
|
149
|
|
|
* @return null |
|
150
|
|
|
*/ |
|
151
|
39 |
|
public function advance($match) { |
|
152
|
39 |
|
assert('is_string($match)'); |
|
153
|
39 |
|
$this->unparsed = ltrim |
|
154
|
39 |
|
( substr($this->unparsed, strlen($match)) |
|
155
|
39 |
|
, "\t \0\x0B" // don't trim linebreaks |
|
156
|
39 |
|
); |
|
157
|
39 |
|
} |
|
158
|
|
|
|
|
159
|
|
|
/** |
|
160
|
|
|
* Checkout if everything is parsed. |
|
161
|
|
|
* |
|
162
|
|
|
* @return bool |
|
163
|
|
|
*/ |
|
164
|
45 |
|
protected function is_everything_parsed() { |
|
165
|
45 |
|
return empty($this->unparsed); |
|
166
|
|
|
} |
|
167
|
|
|
} |
|
168
|
|
|
|
The PSR-2 coding standard requires that all properties in a class have their visibility explicitly declared. If you declare a property using
the property is implicitly global.
To learn more about the PSR-2, please see the PHP-FIG site on the PSR-2.