1 | <?php |
||
2 | declare(strict_types=1); |
||
3 | |||
4 | namespace JsonDecodeStream; |
||
5 | |||
6 | use Generator; |
||
7 | use IteratorAggregate; |
||
8 | use JsonDecodeStream\Exception\TokenizerException; |
||
9 | use JsonDecodeStream\Internal\SourceBuffer; |
||
10 | |||
11 | class Tokenizer implements IteratorAggregate |
||
12 | { |
||
13 | protected const WHITESPACE_CHARS = " \t\r\n"; |
||
14 | |||
15 | protected const LITERALS = [ |
||
16 | Token::TRUE => true, |
||
17 | Token::FALSE => false, |
||
18 | Token::NULL => null, |
||
19 | ]; |
||
20 | |||
21 | /** @var SourceBuffer|string[] */ |
||
22 | protected $buffer; |
||
23 | |||
24 | /** @var int|null */ |
||
25 | protected $lineNumber; |
||
26 | /** @var int|null */ |
||
27 | protected $charNumber; |
||
28 | |||
29 | 107 | public function __construct(SourceBuffer $buffer) |
|
30 | { |
||
31 | 107 | $this->buffer = $buffer; |
|
32 | 107 | } |
|
33 | |||
34 | /** |
||
35 | * @return \Traversable|Token[] |
||
36 | * @psalm-return \Traversable<Token> |
||
37 | * @throws TokenizerException |
||
38 | */ |
||
39 | 52 | public function getIterator() |
|
40 | { |
||
41 | 52 | return $this->tokens(); |
|
42 | } |
||
43 | |||
44 | /** |
||
45 | * @return Generator|Token[] |
||
46 | * @psalm-return \Traversable<Token> |
||
47 | * @throws TokenizerException |
||
48 | */ |
||
49 | 107 | public function tokens() |
|
50 | { |
||
51 | /** @var string|null $number collecting chars of numeric token */ |
||
52 | 107 | $number = null; |
|
53 | /** @var string|null $string collecting chars of string token */ |
||
54 | 107 | $string = null; |
|
55 | /** @var int $stringSlashes how many sequentative slashes are found in string */ |
||
56 | 107 | $stringSlashes = 0; |
|
57 | /** @var string|null $whitespace collecting chars of whitespace token */ |
||
58 | 107 | $whitespace = null; |
|
59 | /** @var string|null $literalAwaited holding the string for awaited literal */ |
||
60 | 107 | $literalAwaited = null; |
|
61 | /** @var string|null $literal collecting chars of literal token */ |
||
62 | 107 | $literal = null; |
|
63 | |||
64 | 107 | $this->lineNumber = 1; |
|
65 | 107 | $this->charNumber = 1; |
|
66 | 107 | foreach ($this->buffer as $char) { |
|
67 | 107 | if ($char == "\n") { |
|
68 | 63 | $this->lineNumber++; |
|
69 | 63 | $this->charNumber = 1; |
|
70 | } else { |
||
71 | 106 | $this->charNumber++; |
|
72 | } |
||
73 | |||
74 | 107 | if ($number !== null) { |
|
75 | 72 | if (strpos('0123456789+-Ee\.', $char) !== false) { |
|
76 | 41 | $number .= $char; |
|
77 | 41 | continue; |
|
78 | } |
||
79 | 52 | $parsedNumber = $this->parseNumber($number); |
|
80 | 52 | $number = null; |
|
81 | 52 | yield $this->createToken(Token::NUMBER, $parsedNumber); |
|
82 | } |
||
83 | |||
84 | 107 | if ($string !== null) { |
|
85 | 62 | if ($char == '\\') { |
|
86 | 7 | $string .= $char; |
|
87 | 7 | $stringSlashes++; |
|
88 | 7 | continue; |
|
89 | 62 | } elseif ($char == '"' && ($stringSlashes % 2) == 0) { |
|
90 | 62 | $parsedString = $this->parseString($string); |
|
91 | 62 | yield $this->createToken(Token::STRING, $parsedString); |
|
92 | 62 | $string = null; |
|
93 | 62 | $stringSlashes = 0; |
|
94 | 62 | continue; |
|
95 | } else { |
||
96 | 61 | $string .= $char; |
|
97 | 61 | $stringSlashes = 0; |
|
98 | 61 | continue; |
|
99 | } |
||
100 | } |
||
101 | |||
102 | 107 | if ($whitespace !== null) { |
|
103 | 69 | if (strpos(static::WHITESPACE_CHARS, $char)) { |
|
104 | 1 | $whitespace .= $char; |
|
105 | 1 | continue; |
|
106 | } else { |
||
107 | 68 | yield $this->createToken(Token::WHITESPACE, $whitespace); |
|
108 | 68 | $whitespace = null; |
|
109 | } |
||
110 | } |
||
111 | |||
112 | 107 | if ($literalAwaited !== null) { |
|
113 | 26 | if (strpos($literalAwaited, $char) !== false) { |
|
114 | 26 | $literal .= $char; |
|
115 | } else { |
||
116 | throw TokenizerException::unexpectedToken($literal, $this->lineNumber, $this->charNumber); |
||
117 | } |
||
118 | |||
119 | 26 | if (strlen($literal) == strlen($literalAwaited)) { |
|
120 | 26 | if ($literal === $literalAwaited) { |
|
121 | 26 | yield $this->createToken($literalAwaited, static::LITERALS[$literalAwaited]); |
|
122 | 26 | $literalAwaited = null; |
|
123 | 26 | $literal = null; |
|
124 | } else { |
||
125 | throw TokenizerException::unexpectedToken($literal, $this->lineNumber, $this->charNumber); |
||
126 | } |
||
127 | } |
||
128 | 26 | continue; |
|
129 | } |
||
130 | |||
131 | 107 | if (strpos(static::WHITESPACE_CHARS, $char) !== false) { |
|
132 | 73 | $whitespace = $char; |
|
133 | 102 | } elseif ($char == "{") { |
|
134 | 56 | yield $this->createToken(Token::OBJECT_START); |
|
135 | 102 | } elseif ($char == "}") { |
|
136 | 56 | yield $this->createToken(Token::OBJECT_END); |
|
137 | 100 | } elseif ($char == "[") { |
|
138 | 51 | yield $this->createToken(Token::ARRAY_START); |
|
139 | 100 | } elseif ($char == "]") { |
|
140 | 51 | yield $this->createToken(Token::ARRAY_END); |
|
141 | 98 | } elseif ($char == ":") { |
|
142 | 55 | yield $this->createToken(Token::KEY_DELIMITER); |
|
143 | 97 | } elseif ($char == ",") { |
|
144 | 63 | yield $this->createToken(Token::COMA); |
|
145 | 96 | } elseif ($char == '"') { |
|
146 | 62 | $string = ''; |
|
147 | 79 | } elseif (strpos('+-0123456789', $char) !== false) { |
|
148 | 74 | $number = $char; |
|
149 | 26 | } elseif ($char == 't') { |
|
150 | 20 | $literalAwaited = Token::TRUE; |
|
151 | 20 | $literal = $char; |
|
152 | 8 | } elseif ($char == 'f') { |
|
153 | 7 | $literalAwaited = Token::FALSE; |
|
154 | 7 | $literal = $char; |
|
155 | 7 | } elseif ($char == 'n') { |
|
156 | 7 | $literalAwaited = Token::NULL; |
|
157 | 7 | $literal = $char; |
|
158 | } else { |
||
159 | throw TokenizerException::unexpectedCharacter($char, $this->lineNumber, $this->charNumber); |
||
160 | } |
||
161 | } |
||
162 | |||
163 | 107 | if ($string !== null || $literal !== null) { |
|
164 | throw TokenizerException::malformedString($string ?? $literal, $this->lineNumber, $this->charNumber); |
||
165 | } |
||
166 | |||
167 | 107 | if ($number !== null) { |
|
168 | 22 | $parsedNumber = $this->parseNumber($number); |
|
169 | 16 | yield $this->createToken(Token::NUMBER, $parsedNumber); |
|
170 | 16 | $number = null; |
|
0 ignored issues
–
show
Unused Code
introduced
by
![]() |
|||
171 | } |
||
172 | |||
173 | 101 | if ($whitespace !== null) { |
|
174 | 17 | yield $this->createToken(Token::WHITESPACE, $whitespace); |
|
175 | } |
||
176 | 101 | } |
|
177 | |||
178 | /** |
||
179 | * @param string $number |
||
180 | * @return float|int|null |
||
181 | * @throws TokenizerException |
||
182 | */ |
||
183 | 74 | protected function parseNumber(string $number) |
|
184 | { |
||
185 | 74 | if (preg_match_all('/^[+-]?(0|[1-9][0-9]*)$/', $number)) { |
|
186 | 56 | return intval($number); |
|
187 | 22 | } elseif (preg_match('/^[+-]?(0|[1-9][0-9]*)(\.[0-9]+)?([eE][-+]?([0-9]+))?$/', $number)) { |
|
188 | 16 | return doubleval($number); |
|
189 | } else { |
||
190 | 6 | throw TokenizerException::malformedNumber($number, $this->lineNumber, $this->charNumber); |
|
191 | } |
||
192 | } |
||
193 | |||
194 | /** |
||
195 | * @param string $string |
||
196 | * @return string|null |
||
197 | * @throws TokenizerException |
||
198 | */ |
||
199 | 62 | protected function parseString(string $string) |
|
200 | { |
||
201 | 62 | $parsed = json_decode('"' . $string . '"'); |
|
202 | 62 | if (!is_string($parsed)) { |
|
203 | throw TokenizerException::malformedString($string, $this->lineNumber, $this->charNumber); |
||
204 | } |
||
205 | 62 | return $parsed; |
|
206 | } |
||
207 | |||
208 | /** |
||
209 | * @param string $tokenId |
||
210 | * @param mixed $value |
||
211 | * @return Token |
||
212 | */ |
||
213 | 101 | protected function createToken(string $tokenId, $value = null): Token |
|
214 | { |
||
215 | 101 | return new Token($tokenId, $value, $this->lineNumber, $this->charNumber); |
|
216 | } |
||
217 | } |
||
218 |