1
|
|
|
<?php declare(strict_types=1); |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* @package s9e\Bencode |
5
|
|
|
* @copyright Copyright (c) 2014-2020 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 InvalidArgumentException; |
12
|
|
|
use RuntimeException; |
13
|
|
|
use s9e\Bencode\Exceptions\ComplianceError; |
14
|
|
|
use s9e\Bencode\Exceptions\IllegalCharacter; |
15
|
|
|
use s9e\Bencode\Exceptions\PrematureEndOfData; |
16
|
|
|
|
17
|
|
|
class Decoder |
18
|
|
|
{ |
19
|
|
|
/** |
20
|
|
|
* @var string Bencoded string being decoded |
21
|
|
|
*/ |
22
|
|
|
protected string $bencoded; |
23
|
|
|
|
24
|
|
|
/** |
25
|
|
|
* @var int Length of the bencoded string |
26
|
|
|
*/ |
27
|
|
|
protected int $len; |
28
|
|
|
|
29
|
|
|
/** |
30
|
|
|
* @var int Safe rightmost boundary |
31
|
|
|
*/ |
32
|
|
|
protected int $max; |
33
|
|
|
|
34
|
|
|
/** |
35
|
|
|
* @var int Position of the cursor while decoding |
36
|
|
|
*/ |
37
|
|
|
protected int $offset; |
38
|
|
|
|
39
|
61 |
|
public static function decode(string $bencoded) |
40
|
|
|
{ |
41
|
61 |
|
$decoder = new static($bencoded); |
42
|
50 |
|
$value = $decoder->decodeAnything(); |
43
|
|
|
|
44
|
26 |
|
$decoder->checkCursorPosition(); |
45
|
|
|
|
46
|
18 |
|
return $value; |
47
|
|
|
} |
48
|
|
|
|
49
|
61 |
|
protected function __construct(string $bencoded) |
50
|
|
|
{ |
51
|
61 |
|
$this->bencoded = $bencoded; |
52
|
61 |
|
$this->len = strlen($bencoded); |
53
|
61 |
|
$this->offset = 0; |
54
|
|
|
|
55
|
61 |
|
if ($bencoded === '') |
56
|
|
|
{ |
57
|
1 |
|
$this->prematureEndOfDataError(); |
58
|
|
|
} |
59
|
60 |
|
$this->computeSafeBoundary(); |
60
|
|
|
} |
61
|
|
|
|
62
|
60 |
|
protected function checkBoundary(): void |
63
|
|
|
{ |
64
|
60 |
|
if ($this->max < 1) |
65
|
|
|
{ |
66
|
10 |
|
if (strpos('-e', $this->bencoded[0]) !== false) |
67
|
|
|
{ |
68
|
2 |
|
throw new IllegalCharacter(0); |
69
|
|
|
} |
70
|
|
|
|
71
|
8 |
|
$this->prematureEndOfDataError(); |
72
|
|
|
} |
73
|
|
|
} |
74
|
|
|
|
75
|
|
|
/** |
76
|
|
|
* Check the cursor's position after decoding is done |
77
|
|
|
*/ |
78
|
26 |
|
protected function checkCursorPosition(): void |
79
|
|
|
{ |
80
|
26 |
|
if ($this->offset !== $this->len) |
81
|
|
|
{ |
82
|
8 |
|
if ($this->offset > $this->len) |
83
|
|
|
{ |
84
|
3 |
|
$this->prematureEndOfDataError(); |
85
|
|
|
} |
86
|
|
|
|
87
|
5 |
|
$this->complianceError($this->offset, 'Superfluous content'); |
88
|
|
|
} |
89
|
|
|
} |
90
|
|
|
|
91
|
12 |
|
protected function complianceError(int $pos, string $message = 'Illegal character'): void |
92
|
|
|
{ |
93
|
12 |
|
throw new ComplianceError($pos, $message); |
94
|
|
|
} |
95
|
|
|
|
96
|
|
|
/** |
97
|
|
|
* Adjust the rightmost boundary to the last safe character that can start a value |
98
|
|
|
* |
99
|
|
|
* Will rewind the boundary to skip the rightmost digits, optionally preceded by "i" or "i-" |
100
|
|
|
*/ |
101
|
60 |
|
protected function computeSafeBoundary(): void |
102
|
|
|
{ |
103
|
60 |
|
$boundary = $this->len - 1; |
104
|
60 |
|
$c = $this->bencoded[$boundary]; |
105
|
60 |
|
while (is_numeric($c) && --$boundary >= 0) |
106
|
|
|
{ |
107
|
7 |
|
$c = $this->bencoded[$boundary]; |
108
|
|
|
} |
109
|
60 |
|
if ($c === '-') |
110
|
|
|
{ |
111
|
6 |
|
$boundary -= 2; |
112
|
|
|
} |
113
|
54 |
|
elseif ($c === 'i') |
114
|
|
|
{ |
115
|
6 |
|
--$boundary; |
116
|
|
|
} |
117
|
|
|
|
118
|
60 |
|
$this->max = $boundary; |
119
|
60 |
|
$this->checkBoundary(); |
120
|
|
|
} |
121
|
|
|
|
122
|
50 |
|
protected function decodeAnything() |
123
|
|
|
{ |
124
|
50 |
|
$c = $this->bencoded[$this->offset]; |
125
|
50 |
|
if ($c === 'i') |
126
|
|
|
{ |
127
|
17 |
|
return $this->decodeInteger(); |
128
|
|
|
} |
129
|
41 |
|
if ($c === 'd') |
130
|
|
|
{ |
131
|
15 |
|
return $this->decodeDictionary(); |
132
|
|
|
} |
133
|
32 |
|
if ($c === 'l') |
134
|
|
|
{ |
135
|
12 |
|
return $this->decodeList(); |
136
|
|
|
} |
137
|
|
|
|
138
|
25 |
|
return $this->decodeString(); |
139
|
|
|
} |
140
|
|
|
|
141
|
15 |
|
protected function decodeDictionary(): ArrayObject |
142
|
|
|
{ |
143
|
15 |
|
$values = []; |
144
|
15 |
|
$lastKey = null; |
145
|
|
|
|
146
|
15 |
|
++$this->offset; |
147
|
15 |
|
while ($this->offset <= $this->max) |
148
|
|
|
{ |
149
|
15 |
|
if ($this->bencoded[$this->offset] === 'e') |
150
|
|
|
{ |
151
|
7 |
|
++$this->offset; |
152
|
|
|
|
153
|
7 |
|
return new ArrayObject($values, ArrayObject::ARRAY_AS_PROPS); |
154
|
|
|
} |
155
|
|
|
|
156
|
14 |
|
$offset = $this->offset; |
157
|
14 |
|
$key = $this->decodeString(); |
158
|
13 |
|
if ($key <= $lastKey) |
159
|
|
|
{ |
160
|
4 |
|
$this->dictionaryComplianceError($offset, $key, $lastKey); |
161
|
|
|
} |
162
|
13 |
|
if ($this->offset > $this->max) |
163
|
|
|
{ |
164
|
1 |
|
break; |
165
|
|
|
} |
166
|
12 |
|
$values[$key] = $this->decodeAnything(); |
167
|
9 |
|
$lastKey = $key; |
168
|
|
|
} |
169
|
|
|
|
170
|
1 |
|
$this->prematureEndOfDataError(); |
|
|
|
|
171
|
|
|
} |
172
|
|
|
|
173
|
43 |
|
protected function decodeDigits(string $terminator): int |
174
|
|
|
{ |
175
|
|
|
// Digits sorted by decreasing frequency as observed on a random sample of torrent files |
176
|
43 |
|
$spn = strspn($this->bencoded, '4615302879', $this->offset); |
177
|
43 |
|
if (!$spn) |
178
|
|
|
{ |
179
|
7 |
|
throw new IllegalCharacter($this->offset); |
180
|
|
|
} |
181
|
38 |
|
if ($this->bencoded[$this->offset] === '0' && $spn > 1) |
182
|
|
|
{ |
183
|
3 |
|
$this->complianceError(1 + $this->offset); |
184
|
|
|
} |
185
|
|
|
|
186
|
|
|
// Capture the value and cast it as an integer |
187
|
35 |
|
$value = (int) substr($this->bencoded, $this->offset, $spn); |
188
|
|
|
|
189
|
35 |
|
$this->offset += $spn; |
190
|
35 |
|
if ($this->bencoded[$this->offset] !== $terminator) |
191
|
|
|
{ |
192
|
5 |
|
throw new IllegalCharacter($this->offset); |
193
|
|
|
} |
194
|
30 |
|
++$this->offset; |
195
|
|
|
|
196
|
30 |
|
return $value; |
197
|
|
|
} |
198
|
|
|
|
199
|
17 |
|
protected function decodeInteger(): int |
200
|
|
|
{ |
201
|
17 |
|
$negative = ($this->bencoded[++$this->offset] === '-'); |
202
|
17 |
|
if ($negative && $this->bencoded[++$this->offset] === '0') |
203
|
|
|
{ |
204
|
1 |
|
$this->complianceError($this->offset); |
205
|
|
|
} |
206
|
|
|
|
207
|
16 |
|
$value = $this->decodeDigits('e'); |
208
|
|
|
|
209
|
11 |
|
return ($negative) ? -$value : $value; |
210
|
|
|
} |
211
|
|
|
|
212
|
12 |
|
protected function decodeList(): array |
213
|
|
|
{ |
214
|
12 |
|
++$this->offset; |
215
|
|
|
|
216
|
12 |
|
$list = []; |
217
|
12 |
|
while ($this->offset <= $this->max) |
218
|
|
|
{ |
219
|
11 |
|
if ($this->bencoded[$this->offset] === 'e') |
220
|
|
|
{ |
221
|
7 |
|
++$this->offset; |
222
|
|
|
|
223
|
7 |
|
return $list; |
224
|
|
|
} |
225
|
|
|
|
226
|
8 |
|
$list[] = $this->decodeAnything(); |
227
|
|
|
} |
228
|
|
|
|
229
|
4 |
|
$this->prematureEndOfDataError(); |
|
|
|
|
230
|
|
|
} |
231
|
|
|
|
232
|
34 |
|
protected function decodeString(): string |
233
|
|
|
{ |
234
|
34 |
|
$len = $this->decodeDigits(':'); |
235
|
26 |
|
$string = substr($this->bencoded, $this->offset, $len); |
236
|
26 |
|
$this->offset += $len; |
237
|
|
|
|
238
|
26 |
|
return $string; |
239
|
|
|
} |
240
|
|
|
|
241
|
4 |
|
protected function dictionaryComplianceError(int $offset, string $key, ?string $lastKey): void |
242
|
|
|
{ |
243
|
4 |
|
if ($key === $lastKey) |
244
|
|
|
{ |
245
|
2 |
|
$this->complianceError($offset, "Duplicate dictionary entry '" . $key . "'"); |
246
|
|
|
} |
247
|
2 |
|
elseif ($key < $lastKey) |
248
|
|
|
{ |
249
|
1 |
|
$this->complianceError($offset, "Out of order dictionary entry '" . $key . "'"); |
250
|
|
|
} |
251
|
|
|
} |
252
|
|
|
|
253
|
17 |
|
protected function prematureEndOfDataError() |
254
|
|
|
{ |
255
|
17 |
|
throw new PrematureEndOfData(max(0, $this->len - 1)); |
256
|
|
|
} |
257
|
|
|
} |
For hinted functions/methods where all return statements with the correct type are only reachable via conditions, ?null? gets implicitly returned which may be incompatible with the hinted type. Let?s take a look at an example: