1
|
|
|
<?php declare(strict_types=1); |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* @package s9e\Bencode |
5
|
|
|
* @copyright Copyright (c) 2014-2021 The s9e authors |
6
|
|
|
* @license http://www.opensource.org/licenses/mit-license.php The MIT License |
7
|
|
|
*/ |
8
|
|
|
namespace s9e\Bencode; |
9
|
|
|
|
10
|
|
|
use ArrayObject; |
11
|
|
|
use const PHP_INT_MAX, PHP_INT_MIN; |
12
|
|
|
use function is_float, str_contains, strcmp, strlen, strspn, substr; |
13
|
|
|
use s9e\Bencode\Exceptions\ComplianceError; |
14
|
|
|
use s9e\Bencode\Exceptions\DecodingException; |
15
|
|
|
|
16
|
|
|
class Decoder |
17
|
|
|
{ |
18
|
|
|
/** |
19
|
|
|
* @var string Bencoded string being decoded |
20
|
|
|
*/ |
21
|
|
|
protected string $bencoded; |
22
|
|
|
|
23
|
|
|
/** |
24
|
|
|
* @var int Length of the bencoded string |
25
|
|
|
*/ |
26
|
|
|
protected int $len; |
27
|
|
|
|
28
|
|
|
/** |
29
|
|
|
* @var int Safe rightmost boundary |
30
|
|
|
*/ |
31
|
|
|
protected int $max; |
32
|
|
|
|
33
|
|
|
/** |
34
|
|
|
* @var int Position of the cursor while decoding |
35
|
|
|
*/ |
36
|
|
|
protected int $offset = 0; |
37
|
|
|
|
38
|
70 |
|
public static function decode(string $bencoded): ArrayObject|array|int|string |
39
|
|
|
{ |
40
|
70 |
|
$decoder = new static($bencoded); |
41
|
58 |
|
$value = $decoder->decodeAnything(); |
42
|
|
|
|
43
|
30 |
|
$decoder->checkCursorPosition(); |
44
|
|
|
|
45
|
22 |
|
return $value; |
46
|
|
|
} |
47
|
|
|
|
48
|
70 |
|
protected function __construct(string $bencoded) |
49
|
|
|
{ |
50
|
70 |
|
$this->bencoded = $bencoded; |
51
|
70 |
|
$this->len = strlen($bencoded); |
52
|
|
|
|
53
|
70 |
|
$this->computeSafeBoundary(); |
54
|
70 |
|
$this->checkBoundary(); |
55
|
58 |
|
} |
56
|
|
|
|
57
|
|
|
/** |
58
|
|
|
* Cast given string as an integer and check for clamping |
59
|
|
|
*/ |
60
|
16 |
|
protected function castInteger(string $string, bool $negative): int |
61
|
|
|
{ |
62
|
16 |
|
if ($negative) |
63
|
|
|
{ |
64
|
2 |
|
$value = (int) "-$string"; |
65
|
2 |
|
if ($value === PHP_INT_MIN) |
66
|
|
|
{ |
67
|
2 |
|
$this->checkIntegerOverflow("-$string"); |
68
|
|
|
} |
69
|
|
|
} |
70
|
|
|
else |
71
|
|
|
{ |
72
|
14 |
|
$value = (int) $string; |
73
|
14 |
|
if ($value === PHP_INT_MAX) |
74
|
|
|
{ |
75
|
2 |
|
$this->checkIntegerOverflow($string); |
76
|
|
|
} |
77
|
|
|
} |
78
|
|
|
|
79
|
15 |
|
return $value; |
80
|
|
|
} |
81
|
|
|
|
82
|
70 |
|
protected function checkBoundary(): void |
83
|
|
|
{ |
84
|
70 |
|
if ($this->max < 1) |
85
|
|
|
{ |
86
|
12 |
|
throw match (substr($this->bencoded, 0, 1)) |
87
|
|
|
{ |
88
|
2 |
|
'-', 'e' => new DecodingException('Illegal character', 0), |
89
|
1 |
|
'' => new DecodingException('Premature end of data', 0), |
90
|
9 |
|
default => new DecodingException('Premature end of data', $this->len - 1) |
91
|
|
|
}; |
92
|
|
|
} |
93
|
58 |
|
} |
94
|
|
|
|
95
|
|
|
/** |
96
|
|
|
* Check the cursor's position after decoding is done |
97
|
|
|
*/ |
98
|
30 |
|
protected function checkCursorPosition(): void |
99
|
|
|
{ |
100
|
30 |
|
if ($this->offset !== $this->len) |
101
|
|
|
{ |
102
|
8 |
|
if ($this->offset > $this->len) |
103
|
|
|
{ |
104
|
3 |
|
throw new DecodingException('Premature end of data', $this->len - 1); |
105
|
|
|
} |
106
|
|
|
|
107
|
5 |
|
$this->complianceError('Superfluous content', $this->offset); |
108
|
|
|
} |
109
|
22 |
|
} |
110
|
|
|
|
111
|
7 |
|
protected function checkDictionaryCompliance(int $offset, string $key, string $lastKey): void |
112
|
|
|
{ |
113
|
7 |
|
$cmp = strcmp($key, $lastKey); |
114
|
7 |
|
if ($cmp <= 0) |
115
|
|
|
{ |
116
|
4 |
|
$msg = ($cmp === 0) ? 'Duplicate' : 'Out of order'; |
117
|
4 |
|
$this->complianceError($msg . " dictionary entry '" . $key . "'", $offset); |
118
|
|
|
} |
119
|
3 |
|
} |
120
|
|
|
|
121
|
3 |
|
protected function checkIntegerOverflow(string $str): void |
122
|
|
|
{ |
123
|
3 |
|
if (is_float(+$str)) |
124
|
|
|
{ |
125
|
1 |
|
throw new DecodingException('Integer overflow', $this->offset - 1 - strlen($str)); |
126
|
|
|
} |
127
|
2 |
|
} |
128
|
|
|
|
129
|
13 |
|
protected function complianceError(string $message, int $offset): void |
130
|
|
|
{ |
131
|
13 |
|
throw new ComplianceError($message, $offset); |
132
|
|
|
} |
133
|
|
|
|
134
|
|
|
/** |
135
|
|
|
* Adjust the rightmost boundary to the last safe character that can start a value |
136
|
|
|
* |
137
|
|
|
* Will rewind the boundary to skip the rightmost digits, optionally preceded by "i" or "i-" |
138
|
|
|
*/ |
139
|
70 |
|
protected function computeSafeBoundary(): void |
140
|
|
|
{ |
141
|
70 |
|
$boundary = $this->len - 1; |
142
|
|
|
do |
143
|
|
|
{ |
144
|
70 |
|
$c = substr($this->bencoded, $boundary, 1); |
145
|
|
|
} |
146
|
70 |
|
while (str_contains('0123456789', $c) && --$boundary >= 0); |
147
|
|
|
|
148
|
70 |
|
$this->max = match ($c) |
149
|
|
|
{ |
150
|
8 |
|
'-' => $boundary - 2, |
151
|
6 |
|
'i' => $boundary - 1, |
152
|
56 |
|
default => $boundary |
153
|
|
|
}; |
154
|
70 |
|
} |
155
|
|
|
|
156
|
58 |
|
protected function decodeAnything(): ArrayObject|array|int|string |
157
|
|
|
{ |
158
|
58 |
|
return match ($this->bencoded[$this->offset]) |
159
|
|
|
{ |
160
|
16 |
|
'd' => $this->decodeDictionary(), |
161
|
22 |
|
'i' => $this->decodeInteger(), |
162
|
15 |
|
'l' => $this->decodeList(), |
163
|
46 |
|
default => $this->decodeString() |
164
|
|
|
}; |
165
|
|
|
} |
166
|
|
|
|
167
|
16 |
|
protected function decodeDictionary(): ArrayObject |
168
|
|
|
{ |
169
|
16 |
|
$values = []; |
170
|
16 |
|
$lastKey = null; |
171
|
|
|
|
172
|
16 |
|
++$this->offset; |
173
|
16 |
|
while ($this->offset <= $this->max) |
174
|
|
|
{ |
175
|
16 |
|
if ($this->bencoded[$this->offset] === 'e') |
176
|
|
|
{ |
177
|
7 |
|
++$this->offset; |
178
|
|
|
|
179
|
7 |
|
return new ArrayObject($values, ArrayObject::ARRAY_AS_PROPS); |
180
|
|
|
} |
181
|
|
|
|
182
|
15 |
|
$offset = $this->offset; |
183
|
15 |
|
$key = $this->decodeString(); |
184
|
14 |
|
if (isset($lastKey)) |
185
|
|
|
{ |
186
|
7 |
|
$this->checkDictionaryCompliance($offset, $key, $lastKey); |
187
|
|
|
} |
188
|
14 |
|
if ($this->offset > $this->max) |
189
|
|
|
{ |
190
|
1 |
|
break; |
191
|
|
|
} |
192
|
13 |
|
$values[$key] = $this->decodeAnything(); |
193
|
10 |
|
$lastKey = $key; |
194
|
|
|
} |
195
|
|
|
|
196
|
1 |
|
throw new DecodingException('Premature end of data', $this->len - 1); |
197
|
|
|
} |
198
|
|
|
|
199
|
22 |
|
protected function decodeInteger(): int |
200
|
|
|
{ |
201
|
22 |
|
$negative = ($this->bencoded[++$this->offset] === '-'); |
202
|
22 |
|
if ($negative && $this->bencoded[++$this->offset] === '0') |
203
|
|
|
{ |
204
|
1 |
|
$this->complianceError('Illegal character', $this->offset); |
205
|
|
|
} |
206
|
|
|
|
207
|
21 |
|
return $this->castInteger($this->readDigits('e'), $negative); |
208
|
|
|
} |
209
|
|
|
|
210
|
15 |
|
protected function decodeList(): array |
211
|
|
|
{ |
212
|
15 |
|
++$this->offset; |
213
|
|
|
|
214
|
15 |
|
$list = []; |
215
|
15 |
|
while ($this->offset <= $this->max) |
216
|
|
|
{ |
217
|
14 |
|
if ($this->bencoded[$this->offset] === 'e') |
218
|
|
|
{ |
219
|
8 |
|
++$this->offset; |
220
|
|
|
|
221
|
8 |
|
return $list; |
222
|
|
|
} |
223
|
|
|
|
224
|
11 |
|
$list[] = $this->decodeAnything(); |
225
|
|
|
} |
226
|
|
|
|
227
|
5 |
|
throw new DecodingException('Premature end of data', $this->len - 1); |
228
|
|
|
} |
229
|
|
|
|
230
|
37 |
|
protected function decodeString(): string |
231
|
|
|
{ |
232
|
37 |
|
$len = (int) $this->readDigits(':'); |
233
|
29 |
|
if ($this->offset + $len >= PHP_INT_MAX) |
234
|
|
|
{ |
235
|
1 |
|
throw new DecodingException('String length overflow', $this->offset - 1 - strlen((string) $len)); |
236
|
|
|
} |
237
|
|
|
|
238
|
28 |
|
$string = substr($this->bencoded, $this->offset, $len); |
239
|
28 |
|
$this->offset += $len; |
240
|
|
|
|
241
|
28 |
|
return $string; |
242
|
|
|
} |
243
|
|
|
|
244
|
50 |
|
protected function readDigits(string $terminator): string |
245
|
|
|
{ |
246
|
|
|
// Digits sorted by decreasing frequency as observed on a random sample of torrent files |
247
|
50 |
|
$spn = strspn($this->bencoded, '4615302879', $this->offset); |
248
|
50 |
|
if (!$spn) |
249
|
|
|
{ |
250
|
7 |
|
throw new DecodingException('Illegal character', $this->offset); |
251
|
|
|
} |
252
|
45 |
|
if ($this->bencoded[$this->offset] === '0' && $spn !== 1) |
253
|
|
|
{ |
254
|
3 |
|
$this->complianceError('Illegal character', 1 + $this->offset); |
255
|
|
|
} |
256
|
|
|
|
257
|
42 |
|
$string = substr($this->bencoded, $this->offset, $spn); |
258
|
|
|
|
259
|
42 |
|
$this->offset += $spn; |
260
|
42 |
|
if ($this->bencoded[$this->offset] !== $terminator) |
261
|
|
|
{ |
262
|
5 |
|
throw new DecodingException('Illegal character', $this->offset); |
263
|
|
|
} |
264
|
37 |
|
++$this->offset; |
265
|
|
|
|
266
|
37 |
|
return $string; |
267
|
|
|
} |
268
|
|
|
} |