1
|
|
|
<?php declare(strict_types=1); |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* @copyright (c) 2017-present brian ridley |
5
|
|
|
* @author brian ridley <[email protected]> |
6
|
|
|
* @license http://opensource.org/licenses/MIT MIT |
7
|
|
|
*/ |
8
|
|
|
|
9
|
|
|
namespace ptlis\SerializedDataEditor\Tokenizer; |
10
|
|
|
|
11
|
|
|
/** |
12
|
|
|
* Tokenizes serialized PHP data. |
13
|
|
|
* |
14
|
|
|
* See http://www.phpinternalsbook.com/classes_objects/serialization.html |
15
|
|
|
*/ |
16
|
|
|
final class Tokenizer |
17
|
|
|
{ |
18
|
|
|
/** |
19
|
|
|
* Processes serialized data into an array of tokens. |
20
|
|
|
* |
21
|
|
|
* @param string $serializedData |
22
|
|
|
* @return Token[] |
23
|
|
|
*/ |
24
|
22 |
|
public function tokenize(string $serializedData): array |
25
|
|
|
{ |
26
|
22 |
|
$tokenList = []; |
27
|
22 |
|
for ($i = 0; $i < strlen($serializedData); $i++) { |
28
|
22 |
|
$character = substr($serializedData, $i, 1); |
29
|
|
|
|
30
|
|
|
switch ($character) { |
31
|
|
|
// Null |
32
|
22 |
|
case Token::PREFIX_NULL: |
33
|
1 |
|
$tokenList[] = new Token(Token::NULL); |
34
|
1 |
|
$i++; // Skip terminator character ';' |
35
|
1 |
|
break; |
36
|
|
|
|
37
|
|
|
// Boolean |
38
|
21 |
|
case Token::PREFIX_BOOL: |
39
|
4 |
|
$tokenList[] = new Token(Token::BOOL, substr($serializedData, $i + 2, 1)); |
40
|
4 |
|
$i += 3; |
41
|
4 |
|
break; |
42
|
|
|
|
43
|
|
|
// Integer |
44
|
19 |
|
case Token::PREFIX_INTEGER: |
45
|
12 |
|
$tokenList[] = $this->getNumberToken(Token::INTEGER, $serializedData, $i); |
46
|
12 |
|
break; |
47
|
|
|
|
48
|
|
|
// Float |
49
|
17 |
|
case Token::PREFIX_FLOAT: |
50
|
2 |
|
$tokenList[] = $this->getNumberToken(Token::FLOAT, $serializedData, $i); |
51
|
2 |
|
break; |
52
|
|
|
|
53
|
|
|
// String |
54
|
16 |
|
case Token::PREFIX_STRING: |
55
|
13 |
|
$tokenList[] = $this->getStringToken($serializedData, $i); |
56
|
13 |
|
break; |
57
|
|
|
|
58
|
|
|
// Reference |
59
|
14 |
|
case Token::PREFIX_REFERENCE: |
60
|
1 |
|
$tokenList[] = $this->getNumberToken(Token::REFERENCE, $serializedData, $i); |
61
|
1 |
|
break; |
62
|
|
|
|
63
|
|
|
// Array |
64
|
14 |
|
case Token::PREFIX_ARRAY_START: |
65
|
10 |
|
$tokenList[] = $this->getArrayToken($serializedData, $i); |
66
|
10 |
|
break; |
67
|
|
|
|
68
|
|
|
// Object, default serialization |
69
|
14 |
|
case Token::PREFIX_OBJECT_DEFAULT_NAME: |
70
|
7 |
|
$tokenList = array_merge($tokenList, $this->getObjectDefaultToken($serializedData, $i)); |
71
|
7 |
|
break; |
72
|
|
|
|
73
|
|
|
// Object, custom serialization |
74
|
14 |
|
case Token::PREFIX_OBJECT_CUSTOM_NAME: |
75
|
1 |
|
$tokenList = array_merge($tokenList, $this->getObjectCustomToken($serializedData, $i)); |
76
|
1 |
|
break; |
77
|
|
|
|
78
|
|
|
// Array or object end |
79
|
13 |
|
case Token::PREFIX_COMPOUND_END: |
80
|
12 |
|
$tokenList[] = new Token(Token::COMPOUND_END); |
81
|
12 |
|
break; |
82
|
|
|
|
83
|
|
|
default: |
84
|
1 |
|
throw new \RuntimeException('Invalid serialized data - unexpected character "' . $character . '" encountered'); |
85
|
|
|
} |
86
|
|
|
} |
87
|
|
|
|
88
|
21 |
|
return $tokenList; |
89
|
|
|
} |
90
|
|
|
|
91
|
|
|
/** |
92
|
|
|
* Create token for an integer or float. |
93
|
|
|
*/ |
94
|
13 |
|
private function getNumberToken(string $type, string $serializedData, int &$currentIndex): Token |
95
|
|
|
{ |
96
|
|
|
// Skip first characters 'i:' |
97
|
13 |
|
$currentIndex += 2; |
98
|
|
|
|
99
|
13 |
|
return new Token( |
100
|
13 |
|
$type, |
101
|
13 |
|
$this->readUntil($serializedData, $currentIndex, ';') |
102
|
|
|
); |
103
|
|
|
} |
104
|
|
|
|
105
|
|
|
/** |
106
|
|
|
* Create token for a string. |
107
|
|
|
*/ |
108
|
13 |
|
private function getStringToken(string $serializedData, int &$currentIndex): Token |
109
|
|
|
{ |
110
|
|
|
// Skip first characters 's:' |
111
|
13 |
|
$currentIndex += 2; |
112
|
|
|
|
113
|
|
|
// Get string length |
114
|
13 |
|
$stringLength = $this->readUntil($serializedData, $currentIndex, ':'); |
115
|
|
|
|
116
|
|
|
// Skip separator and open quote ':"' |
117
|
13 |
|
$currentIndex += 2; |
118
|
|
|
|
119
|
|
|
// Read string |
120
|
13 |
|
$string = substr($serializedData, $currentIndex, intval($stringLength)); |
121
|
13 |
|
$currentIndex += intval($stringLength); |
122
|
|
|
|
123
|
|
|
// Skip close terminator '";' |
124
|
13 |
|
$currentIndex++; |
125
|
|
|
|
126
|
13 |
|
return new Token(Token::STRING, $string); |
127
|
|
|
} |
128
|
|
|
|
129
|
|
|
/** |
130
|
|
|
* Create token for the start of an array. |
131
|
|
|
*/ |
132
|
10 |
|
private function getArrayToken(string $serializedData, int &$currentIndex): Token |
133
|
|
|
{ |
134
|
|
|
// Skip first characters 'a:' |
135
|
10 |
|
$currentIndex += 2; |
136
|
|
|
|
137
|
|
|
// Get array length |
138
|
10 |
|
$arrayLength = $this->readUntil($serializedData, $currentIndex, ':'); |
139
|
|
|
|
140
|
|
|
// Skip array open |
141
|
10 |
|
$currentIndex += 1; |
142
|
|
|
|
143
|
10 |
|
return new Token(Token::ARRAY_START, $arrayLength); |
144
|
|
|
} |
145
|
|
|
|
146
|
|
|
/** |
147
|
|
|
* Create tokens for the start of an object serialized with PHP's default serialization format. |
148
|
|
|
* |
149
|
|
|
* @return Token[] |
150
|
|
|
*/ |
151
|
7 |
|
private function getObjectDefaultToken(string $serializedData, int &$currentIndex): array |
152
|
|
|
{ |
153
|
|
|
// Skip first characters 'O:' |
154
|
7 |
|
$currentIndex += 2; |
155
|
|
|
|
156
|
|
|
// Get class name |
157
|
7 |
|
$classNameLength = $this->readUntil($serializedData, $currentIndex, ':'); |
158
|
|
|
|
159
|
|
|
// Skip separator and open quote ':"' |
160
|
7 |
|
$currentIndex += 2; |
161
|
|
|
|
162
|
|
|
// Read class name |
163
|
7 |
|
$className = substr($serializedData, $currentIndex, intval($classNameLength)); |
164
|
7 |
|
$currentIndex += intval($classNameLength); |
165
|
|
|
|
166
|
|
|
// Skip closing quote and seperator '":' |
167
|
7 |
|
$currentIndex += 2; |
168
|
|
|
|
169
|
|
|
// Get object property count |
170
|
7 |
|
$objectPropertyCount = $this->readUntil($serializedData, $currentIndex, ':'); |
171
|
|
|
|
172
|
|
|
// Skip terminator |
173
|
7 |
|
$currentIndex++; |
174
|
|
|
|
175
|
|
|
return [ |
176
|
7 |
|
new Token(Token::OBJECT_DEFAULT_NAME, $className), |
177
|
7 |
|
new Token(Token::OBJECT_MEMBER_COUNT, $objectPropertyCount) |
178
|
|
|
]; |
179
|
|
|
} |
180
|
|
|
|
181
|
|
|
/** |
182
|
|
|
* Create tokens for the start of an object serialized with a custom serialization format. |
183
|
|
|
* |
184
|
|
|
* @return Token[] |
185
|
|
|
*/ |
186
|
1 |
|
private function getObjectCustomToken(string $serializedData, int &$currentIndex): array |
187
|
|
|
{ |
188
|
|
|
// Skip first characters 'O:' |
189
|
1 |
|
$currentIndex += 2; |
190
|
|
|
|
191
|
|
|
// Get class name |
192
|
1 |
|
$classNameLength = $this->readUntil($serializedData, $currentIndex, ':'); |
193
|
|
|
|
194
|
|
|
// Skip separator and open quote ':"' |
195
|
1 |
|
$currentIndex += 2; |
196
|
|
|
|
197
|
|
|
// Read class name |
198
|
1 |
|
$className = substr($serializedData, $currentIndex, intval($classNameLength)); |
199
|
1 |
|
$currentIndex += intval($classNameLength); |
200
|
|
|
|
201
|
|
|
// Skip closing quote and seperator '":' |
202
|
1 |
|
$currentIndex += 2; |
203
|
|
|
|
204
|
|
|
// Get length of serialized data |
205
|
1 |
|
$serializedDataLength = $this->readUntil($serializedData, $currentIndex, ':'); |
206
|
|
|
|
207
|
|
|
// Skip seperator and opening delimiter ':{' |
208
|
1 |
|
$currentIndex += 2; |
209
|
|
|
|
210
|
|
|
// Read serialized data |
211
|
1 |
|
$data = substr($serializedData, $currentIndex, intval($serializedDataLength)); |
212
|
1 |
|
$currentIndex += intval($serializedDataLength); |
213
|
|
|
|
214
|
|
|
// Skip closing delimiter '}' |
215
|
1 |
|
$currentIndex++; |
216
|
|
|
|
217
|
|
|
return [ |
218
|
1 |
|
new Token(Token::OBJECT_CUSTOM_NAME, $className), |
219
|
1 |
|
new Token(Token::OBJECT_CUSTOM_DATA, $data), |
220
|
1 |
|
new Token(Token::COMPOUND_END) |
221
|
|
|
]; |
222
|
|
|
} |
223
|
|
|
|
224
|
|
|
/** |
225
|
|
|
* Returns the segment of $serializedData from $currentIndex until the first instance of $char is found. |
226
|
|
|
*/ |
227
|
18 |
|
private function readUntil( |
228
|
|
|
string $serializedData, |
229
|
|
|
int &$currentIndex, |
230
|
|
|
string $char |
231
|
|
|
): string { |
232
|
18 |
|
$string = ''; |
233
|
18 |
|
while (substr($serializedData, $currentIndex, strlen($char)) !== $char) { |
234
|
18 |
|
$string .= substr($serializedData, $currentIndex, 1); |
235
|
18 |
|
$currentIndex++; |
236
|
|
|
} |
237
|
18 |
|
return $string; |
238
|
|
|
} |
239
|
|
|
} |