|
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
|
|
|
* Baseclass for Parsers. |
|
15
|
|
|
*/ |
|
16
|
|
|
abstract class Parser { |
|
17
|
|
|
/** |
|
18
|
|
|
* @var SymbolTable |
|
19
|
|
|
*/ |
|
20
|
|
|
private $symbol_table; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* @var Tokenizer|null |
|
24
|
|
|
*/ |
|
25
|
|
|
protected $tokenizer = null; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* @var array|null (Symbol, array $matches) |
|
29
|
|
|
*/ |
|
30
|
|
|
protected $token = null; |
|
31
|
|
|
|
|
32
|
47 |
|
public function __construct() { |
|
33
|
47 |
|
$this->symbol_table = $this->create_symbol_table(); |
|
34
|
47 |
|
$this->add_symbols_to($this->symbol_table); |
|
35
|
47 |
|
} |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* Parse the string according to this parser. |
|
39
|
|
|
* |
|
40
|
|
|
* @return mixed |
|
41
|
|
|
*/ |
|
42
|
47 |
|
public function parse($source) { |
|
43
|
|
|
try { |
|
44
|
47 |
|
$this->tokenizer = $this->create_tokenizer($source); |
|
45
|
47 |
|
$this->token = $this->tokenizer->current(); |
|
46
|
47 |
|
return $this->root(); |
|
47
|
|
|
} |
|
48
|
1 |
|
catch (ParserException $e) { |
|
49
|
1 |
|
list($l, $c) = $this->tokenizer->source_position(); |
|
50
|
1 |
|
$e->setPosition($l, $c); |
|
51
|
1 |
|
throw $e; |
|
52
|
|
|
} |
|
53
|
|
|
finally { |
|
54
|
47 |
|
$this->tokenizer = null; |
|
55
|
47 |
|
$this->token = null; |
|
56
|
47 |
|
} |
|
57
|
|
|
} |
|
58
|
|
|
|
|
59
|
|
|
/** |
|
60
|
|
|
* The root for the parse tree. |
|
61
|
|
|
* |
|
62
|
|
|
* @return mixed |
|
63
|
|
|
*/ |
|
64
|
|
|
abstract protected function root(); |
|
65
|
|
|
|
|
66
|
|
|
// Factory Methods |
|
67
|
|
|
|
|
68
|
|
|
/** |
|
69
|
|
|
* Build the Tokenizer. |
|
70
|
|
|
* |
|
71
|
|
|
* @return Tokenizer |
|
72
|
|
|
*/ |
|
73
|
47 |
|
public function create_tokenizer($source) { |
|
74
|
47 |
|
assert('is_string($source)'); |
|
75
|
47 |
|
return new Tokenizer($this->symbol_table, $source); |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
|
|
/** |
|
79
|
|
|
* Build the SymbolTable |
|
80
|
|
|
* |
|
81
|
|
|
* @return SymbolTable |
|
82
|
|
|
*/ |
|
83
|
47 |
|
protected function create_symbol_table() { |
|
84
|
47 |
|
return new SymbolTable(); |
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
|
|
/** |
|
88
|
|
|
* @param SymbolTable |
|
89
|
|
|
* @return null |
|
90
|
|
|
*/ |
|
91
|
|
|
abstract protected function add_symbols_to(SymbolTable $table); |
|
92
|
|
|
|
|
93
|
|
|
// Helpers for defining the grammar. |
|
94
|
|
|
|
|
95
|
|
|
|
|
96
|
|
|
// Helpers for actual parsing. |
|
97
|
|
|
|
|
98
|
|
|
/** |
|
99
|
|
|
* Set the current token to the next token from the tokenizer. |
|
100
|
|
|
* |
|
101
|
|
|
* @return null |
|
102
|
|
|
*/ |
|
103
|
45 |
|
protected function fetch_next_token() { |
|
104
|
45 |
|
assert('is_array($this->token)'); |
|
105
|
45 |
|
assert('$this->tokenizer !== null'); |
|
106
|
45 |
|
$this->tokenizer->next(); |
|
107
|
45 |
|
$this->token = $this->tokenizer->current(); |
|
108
|
45 |
|
} |
|
109
|
|
|
|
|
110
|
|
|
/** |
|
111
|
|
|
* Get the current symbol. |
|
112
|
|
|
* |
|
113
|
|
|
* @return Symbol |
|
114
|
|
|
*/ |
|
115
|
41 |
|
protected function current_symbol() { |
|
116
|
41 |
|
return $this->token[0]; |
|
117
|
|
|
} |
|
118
|
|
|
|
|
119
|
|
|
/** |
|
120
|
|
|
* Get the current match. |
|
121
|
|
|
* |
|
122
|
|
|
* @return string[] |
|
123
|
|
|
*/ |
|
124
|
45 |
|
protected function current_match() { |
|
125
|
45 |
|
return $this->token[1]; |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
|
|
/** |
|
129
|
|
|
* Advance the tokenizer to the next token if current token |
|
130
|
|
|
* was matched by the given regexp. |
|
131
|
|
|
* |
|
132
|
|
|
* @param string $regexp |
|
133
|
|
|
* @return null |
|
134
|
|
|
*/ |
|
135
|
21 |
|
protected function advance($regexp) { |
|
136
|
21 |
|
assert('is_string($regexp)'); |
|
137
|
21 |
|
assert('is_array($this->token)'); |
|
138
|
21 |
|
assert('$this->tokenizer !== null'); |
|
139
|
21 |
|
if (!$this->is_current_token_matched_by($regexp)) { |
|
140
|
|
|
$match = $this->current_match()[0]; |
|
141
|
|
|
throw new ParserException("Syntax Error: Expected '$regexp', found '$match'"); |
|
142
|
|
|
} |
|
143
|
21 |
|
$this->tokenizer->next(); |
|
144
|
21 |
|
$this->token = $this->tokenizer->current(); |
|
145
|
21 |
|
} |
|
146
|
|
|
|
|
147
|
|
|
/** |
|
148
|
|
|
* Advance the tokenizer to the next token if current token |
|
149
|
|
|
* was matched by the given operator. |
|
150
|
|
|
* |
|
151
|
|
|
* @param string $op |
|
152
|
|
|
* @return null |
|
153
|
|
|
*/ |
|
154
|
9 |
|
protected function advance_operator($op) { |
|
155
|
9 |
|
$this->advance($this->symbol_table->operator_regexp($op)); |
|
156
|
9 |
|
} |
|
157
|
|
|
|
|
158
|
|
|
/** |
|
159
|
|
|
* Is the end of the file reached? |
|
160
|
|
|
* |
|
161
|
|
|
* @return bool |
|
162
|
|
|
*/ |
|
163
|
37 |
|
public function is_end_of_file_reached() { |
|
164
|
37 |
|
return $this->is_current_token_matched_by(""); |
|
165
|
|
|
} |
|
166
|
|
|
|
|
167
|
|
|
/** |
|
168
|
|
|
* Check if the current token was matched by the given regexp. |
|
169
|
|
|
* |
|
170
|
|
|
* @param string $regexp |
|
171
|
|
|
* @return bool |
|
172
|
|
|
*/ |
|
173
|
44 |
|
protected function is_current_token_matched_by($regexp) { |
|
174
|
44 |
|
assert('is_string($regexp)'); |
|
175
|
44 |
|
return $this->token[0]->regexp() == $regexp; |
|
176
|
|
|
} |
|
177
|
|
|
|
|
178
|
|
|
/** |
|
179
|
|
|
* Check if the current token is the given operator. |
|
180
|
|
|
* |
|
181
|
|
|
* @param string $operator |
|
182
|
|
|
* @return bool |
|
183
|
|
|
*/ |
|
184
|
6 |
|
protected function is_current_token_operator($operator) { |
|
185
|
6 |
|
return $this->is_current_token_matched_by |
|
186
|
6 |
|
($this->symbol_table->operator_regexp($operator)); |
|
187
|
|
|
} |
|
188
|
|
|
} |
|
189
|
|
|
|
lechimp-p /
dicto.php
