1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace MaxMind\Db\Reader; |
4
|
|
|
|
5
|
|
|
// @codingStandardsIgnoreLine |
6
|
|
|
use RuntimeException; |
7
|
|
|
|
8
|
|
|
/** |
9
|
|
|
* @ignore |
10
|
|
|
* |
11
|
|
|
* We subtract 1 from the log to protect against precision loss. |
12
|
|
|
*/ |
13
|
|
|
\define(__NAMESPACE__ . '\_MM_MAX_INT_BYTES', (log(PHP_INT_MAX, 2) - 1) / 8); |
14
|
|
|
|
15
|
|
|
class Decoder |
16
|
|
|
{ |
17
|
|
|
private $fileStream; |
18
|
|
|
private $pointerBase; |
19
|
|
|
private $pointerBaseByteSize; |
20
|
|
|
// This is only used for unit testing |
21
|
|
|
private $pointerTestHack; |
22
|
|
|
private $switchByteOrder; |
23
|
|
|
|
24
|
|
|
/** @ignore */ |
25
|
|
|
const _EXTENDED = 0; |
26
|
|
|
/** @ignore */ |
27
|
|
|
const _POINTER = 1; |
28
|
|
|
/** @ignore */ |
29
|
|
|
const _UTF8_STRING = 2; |
30
|
|
|
/** @ignore */ |
31
|
|
|
const _DOUBLE = 3; |
32
|
|
|
/** @ignore */ |
33
|
|
|
const _BYTES = 4; |
34
|
|
|
/** @ignore */ |
35
|
|
|
const _UINT16 = 5; |
36
|
|
|
/** @ignore */ |
37
|
|
|
const _UINT32 = 6; |
38
|
|
|
/** @ignore */ |
39
|
|
|
const _MAP = 7; |
40
|
|
|
/** @ignore */ |
41
|
|
|
const _INT32 = 8; |
42
|
|
|
/** @ignore */ |
43
|
|
|
const _UINT64 = 9; |
44
|
|
|
/** @ignore */ |
45
|
|
|
const _UINT128 = 10; |
46
|
|
|
/** @ignore */ |
47
|
|
|
const _ARRAY = 11; |
48
|
|
|
/** @ignore */ |
49
|
|
|
const _CONTAINER = 12; |
50
|
|
|
/** @ignore */ |
51
|
|
|
const _END_MARKER = 13; |
52
|
|
|
/** @ignore */ |
53
|
|
|
const _BOOLEAN = 14; |
54
|
|
|
/** @ignore */ |
55
|
|
|
const _FLOAT = 15; |
56
|
|
|
|
57
|
|
|
public function __construct( |
58
|
|
|
$fileStream, |
59
|
|
|
$pointerBase = 0, |
60
|
|
|
$pointerTestHack = false |
61
|
|
|
) { |
62
|
|
|
$this->fileStream = $fileStream; |
63
|
|
|
$this->pointerBase = $pointerBase; |
64
|
|
|
|
65
|
|
|
$this->pointerBaseByteSize = $pointerBase > 0 ? log($pointerBase, 2) / 8 : 0; |
66
|
|
|
$this->pointerTestHack = $pointerTestHack; |
67
|
|
|
|
68
|
|
|
$this->switchByteOrder = $this->isPlatformLittleEndian(); |
69
|
|
|
} |
70
|
|
|
|
71
|
|
|
public function decode($offset) |
72
|
|
|
{ |
73
|
|
|
$ctrlByte = \ord(Util::read($this->fileStream, $offset, 1)); |
74
|
|
|
++$offset; |
75
|
|
|
|
76
|
|
|
$type = $ctrlByte >> 5; |
77
|
|
|
|
78
|
|
|
// Pointers are a special case, we don't read the next $size bytes, we |
79
|
|
|
// use the size to determine the length of the pointer and then follow |
80
|
|
|
// it. |
81
|
|
|
if ($type === self::_POINTER) { |
82
|
|
|
list($pointer, $offset) = $this->decodePointer($ctrlByte, $offset); |
83
|
|
|
|
84
|
|
|
// for unit testing |
85
|
|
|
if ($this->pointerTestHack) { |
86
|
|
|
return [$pointer]; |
87
|
|
|
} |
88
|
|
|
|
89
|
|
|
list($result) = $this->decode($pointer); |
90
|
|
|
|
91
|
|
|
return [$result, $offset]; |
92
|
|
|
} |
93
|
|
|
|
94
|
|
|
if ($type === self::_EXTENDED) { |
95
|
|
|
$nextByte = \ord(Util::read($this->fileStream, $offset, 1)); |
96
|
|
|
|
97
|
|
|
$type = $nextByte + 7; |
98
|
|
|
|
99
|
|
|
if ($type < 8) { |
100
|
|
|
throw new InvalidDatabaseException( |
101
|
|
|
'Something went horribly wrong in the decoder. An extended type ' |
102
|
|
|
. 'resolved to a type number < 8 (' |
103
|
|
|
. $type |
104
|
|
|
. ')' |
105
|
|
|
); |
106
|
|
|
} |
107
|
|
|
|
108
|
|
|
++$offset; |
109
|
|
|
} |
110
|
|
|
|
111
|
|
|
list($size, $offset) = $this->sizeFromCtrlByte($ctrlByte, $offset); |
112
|
|
|
|
113
|
|
|
return $this->decodeByType($type, $offset, $size); |
114
|
|
|
} |
115
|
|
|
|
116
|
|
|
private function decodeByType($type, $offset, $size) |
117
|
|
|
{ |
118
|
|
|
switch ($type) { |
119
|
|
|
case self::_MAP: |
120
|
|
|
return $this->decodeMap($size, $offset); |
121
|
|
|
case self::_ARRAY: |
122
|
|
|
return $this->decodeArray($size, $offset); |
123
|
|
|
case self::_BOOLEAN: |
124
|
|
|
return [$this->decodeBoolean($size), $offset]; |
125
|
|
|
} |
126
|
|
|
|
127
|
|
|
$newOffset = $offset + $size; |
128
|
|
|
$bytes = Util::read($this->fileStream, $offset, $size); |
129
|
|
|
switch ($type) { |
130
|
|
|
case self::_BYTES: |
131
|
|
|
case self::_UTF8_STRING: |
132
|
|
|
return [$bytes, $newOffset]; |
133
|
|
|
case self::_DOUBLE: |
134
|
|
|
$this->verifySize(8, $size); |
135
|
|
|
|
136
|
|
|
return [$this->decodeDouble($bytes), $newOffset]; |
137
|
|
|
case self::_FLOAT: |
138
|
|
|
$this->verifySize(4, $size); |
139
|
|
|
|
140
|
|
|
return [$this->decodeFloat($bytes), $newOffset]; |
141
|
|
|
case self::_INT32: |
142
|
|
|
return [$this->decodeInt32($bytes, $size), $newOffset]; |
143
|
|
|
case self::_UINT16: |
144
|
|
|
case self::_UINT32: |
145
|
|
|
case self::_UINT64: |
146
|
|
|
case self::_UINT128: |
147
|
|
|
return [$this->decodeUint($bytes, $size), $newOffset]; |
148
|
|
|
default: |
149
|
|
|
throw new InvalidDatabaseException( |
150
|
|
|
'Unknown or unexpected type: ' . $type |
151
|
|
|
); |
152
|
|
|
} |
153
|
|
|
} |
154
|
|
|
|
155
|
|
|
private function verifySize($expected, $actual) |
156
|
|
|
{ |
157
|
|
|
if ($expected !== $actual) { |
158
|
|
|
throw new InvalidDatabaseException( |
159
|
|
|
"The MaxMind DB file's data section contains bad data (unknown data type or corrupt data)" |
160
|
|
|
); |
161
|
|
|
} |
162
|
|
|
} |
163
|
|
|
|
164
|
|
|
private function decodeArray($size, $offset) |
165
|
|
|
{ |
166
|
|
|
$array = []; |
167
|
|
|
|
168
|
|
|
for ($i = 0; $i < $size; ++$i) { |
169
|
|
|
list($value, $offset) = $this->decode($offset); |
170
|
|
|
array_push($array, $value); |
171
|
|
|
} |
172
|
|
|
|
173
|
|
|
return [$array, $offset]; |
174
|
|
|
} |
175
|
|
|
|
176
|
|
|
private function decodeBoolean($size) |
177
|
|
|
{ |
178
|
|
|
return $size === 0 ? false : true; |
179
|
|
|
} |
180
|
|
|
|
181
|
|
|
private function decodeDouble($bits) |
182
|
|
|
{ |
183
|
|
|
// This assumes IEEE 754 doubles, but most (all?) modern platforms |
184
|
|
|
// use them. |
185
|
|
|
// |
186
|
|
|
// We are not using the "E" format as that was only added in |
187
|
|
|
// 7.0.15 and 7.1.1. As such, we must switch byte order on |
188
|
|
|
// little endian machines. |
189
|
|
|
list(, $double) = unpack('d', $this->maybeSwitchByteOrder($bits)); |
190
|
|
|
|
191
|
|
|
return $double; |
192
|
|
|
} |
193
|
|
|
|
194
|
|
|
private function decodeFloat($bits) |
195
|
|
|
{ |
196
|
|
|
// This assumes IEEE 754 floats, but most (all?) modern platforms |
197
|
|
|
// use them. |
198
|
|
|
// |
199
|
|
|
// We are not using the "G" format as that was only added in |
200
|
|
|
// 7.0.15 and 7.1.1. As such, we must switch byte order on |
201
|
|
|
// little endian machines. |
202
|
|
|
list(, $float) = unpack('f', $this->maybeSwitchByteOrder($bits)); |
203
|
|
|
|
204
|
|
|
return $float; |
205
|
|
|
} |
206
|
|
|
|
207
|
|
|
private function decodeInt32($bytes, $size) |
208
|
|
|
{ |
209
|
|
|
switch ($size) { |
210
|
|
|
case 0: |
211
|
|
|
return 0; |
212
|
|
|
case 1: |
213
|
|
|
case 2: |
214
|
|
|
case 3: |
215
|
|
|
$bytes = str_pad($bytes, 4, "\x00", STR_PAD_LEFT); |
216
|
|
|
break; |
217
|
|
|
case 4: |
218
|
|
|
break; |
219
|
|
|
default: |
220
|
|
|
throw new InvalidDatabaseException( |
221
|
|
|
"The MaxMind DB file's data section contains bad data (unknown data type or corrupt data)" |
222
|
|
|
); |
223
|
|
|
} |
224
|
|
|
|
225
|
|
|
list(, $int) = unpack('l', $this->maybeSwitchByteOrder($bytes)); |
226
|
|
|
|
227
|
|
|
return $int; |
228
|
|
|
} |
229
|
|
|
|
230
|
|
|
private function decodeMap($size, $offset) |
231
|
|
|
{ |
232
|
|
|
$map = []; |
233
|
|
|
|
234
|
|
|
for ($i = 0; $i < $size; ++$i) { |
235
|
|
|
list($key, $offset) = $this->decode($offset); |
236
|
|
|
list($value, $offset) = $this->decode($offset); |
237
|
|
|
$map[$key] = $value; |
238
|
|
|
} |
239
|
|
|
|
240
|
|
|
return [$map, $offset]; |
241
|
|
|
} |
242
|
|
|
|
243
|
|
|
private function decodePointer($ctrlByte, $offset) |
244
|
|
|
{ |
245
|
|
|
$pointerSize = (($ctrlByte >> 3) & 0x3) + 1; |
246
|
|
|
|
247
|
|
|
$buffer = Util::read($this->fileStream, $offset, $pointerSize); |
248
|
|
|
$offset = $offset + $pointerSize; |
249
|
|
|
|
250
|
|
|
switch ($pointerSize) { |
251
|
|
|
case 1: |
252
|
|
|
$packed = \chr($ctrlByte & 0x7) . $buffer; |
253
|
|
|
list(, $pointer) = unpack('n', $packed); |
254
|
|
|
$pointer += $this->pointerBase; |
255
|
|
|
break; |
256
|
|
|
case 2: |
257
|
|
|
$packed = "\x00" . \chr($ctrlByte & 0x7) . $buffer; |
258
|
|
|
list(, $pointer) = unpack('N', $packed); |
259
|
|
|
$pointer += $this->pointerBase + 2048; |
260
|
|
|
break; |
261
|
|
|
case 3: |
262
|
|
|
$packed = \chr($ctrlByte & 0x7) . $buffer; |
263
|
|
|
|
264
|
|
|
// It is safe to use 'N' here, even on 32 bit machines as the |
265
|
|
|
// first bit is 0. |
266
|
|
|
list(, $pointer) = unpack('N', $packed); |
267
|
|
|
$pointer += $this->pointerBase + 526336; |
268
|
|
|
break; |
269
|
|
|
case 4: |
270
|
|
|
// We cannot use unpack here as we might overflow on 32 bit |
271
|
|
|
// machines |
272
|
|
|
$pointerOffset = $this->decodeUint($buffer, $pointerSize); |
273
|
|
|
|
274
|
|
|
$byteLength = $pointerSize + $this->pointerBaseByteSize; |
275
|
|
|
|
276
|
|
|
if ($byteLength <= _MM_MAX_INT_BYTES) { |
277
|
|
|
$pointer = $pointerOffset + $this->pointerBase; |
278
|
|
|
} elseif (\extension_loaded('gmp')) { |
279
|
|
|
$pointer = gmp_strval(gmp_add($pointerOffset, $this->pointerBase)); |
280
|
|
|
} elseif (\extension_loaded('bcmath')) { |
281
|
|
|
$pointer = bcadd($pointerOffset, $this->pointerBase); |
282
|
|
|
} else { |
283
|
|
|
throw new RuntimeException( |
284
|
|
|
'The gmp or bcmath extension must be installed to read this database.' |
285
|
|
|
); |
286
|
|
|
} |
287
|
|
|
} |
288
|
|
|
|
289
|
|
|
return [$pointer, $offset]; |
|
|
|
|
290
|
|
|
} |
291
|
|
|
|
292
|
|
|
private function decodeUint($bytes, $byteLength) |
293
|
|
|
{ |
294
|
|
|
if ($byteLength === 0) { |
295
|
|
|
return 0; |
296
|
|
|
} |
297
|
|
|
|
298
|
|
|
$integer = 0; |
299
|
|
|
|
300
|
|
|
for ($i = 0; $i < $byteLength; ++$i) { |
301
|
|
|
$part = \ord($bytes[$i]); |
302
|
|
|
|
303
|
|
|
// We only use gmp or bcmath if the final value is too big |
304
|
|
|
if ($byteLength <= _MM_MAX_INT_BYTES) { |
305
|
|
|
$integer = ($integer << 8) + $part; |
306
|
|
|
} elseif (\extension_loaded('gmp')) { |
307
|
|
|
$integer = gmp_strval(gmp_add(gmp_mul($integer, 256), $part)); |
308
|
|
|
} elseif (\extension_loaded('bcmath')) { |
309
|
|
|
$integer = bcadd(bcmul($integer, 256), $part); |
310
|
|
|
} else { |
311
|
|
|
throw new RuntimeException( |
312
|
|
|
'The gmp or bcmath extension must be installed to read this database.' |
313
|
|
|
); |
314
|
|
|
} |
315
|
|
|
} |
316
|
|
|
|
317
|
|
|
return $integer; |
318
|
|
|
} |
319
|
|
|
|
320
|
|
|
private function sizeFromCtrlByte($ctrlByte, $offset) |
321
|
|
|
{ |
322
|
|
|
$size = $ctrlByte & 0x1f; |
323
|
|
|
|
324
|
|
|
if ($size < 29) { |
325
|
|
|
return [$size, $offset]; |
326
|
|
|
} |
327
|
|
|
|
328
|
|
|
$bytesToRead = $size - 28; |
329
|
|
|
$bytes = Util::read($this->fileStream, $offset, $bytesToRead); |
330
|
|
|
|
331
|
|
|
if ($size === 29) { |
332
|
|
|
$size = 29 + \ord($bytes); |
333
|
|
|
} elseif ($size === 30) { |
334
|
|
|
list(, $adjust) = unpack('n', $bytes); |
335
|
|
|
$size = 285 + $adjust; |
336
|
|
|
} elseif ($size > 30) { |
337
|
|
|
list(, $adjust) = unpack('N', "\x00" . $bytes); |
338
|
|
|
$size = $adjust + 65821; |
339
|
|
|
} |
340
|
|
|
|
341
|
|
|
return [$size, $offset + $bytesToRead]; |
342
|
|
|
} |
343
|
|
|
|
344
|
|
|
private function maybeSwitchByteOrder($bytes) |
345
|
|
|
{ |
346
|
|
|
return $this->switchByteOrder ? strrev($bytes) : $bytes; |
347
|
|
|
} |
348
|
|
|
|
349
|
|
|
private function isPlatformLittleEndian() |
350
|
|
|
{ |
351
|
|
|
$testint = 0x00FF; |
352
|
|
|
$packed = pack('S', $testint); |
353
|
|
|
|
354
|
|
|
return $testint === current(unpack('v', $packed)); |
|
|
|
|
355
|
|
|
} |
356
|
|
|
} |
357
|
|
|
|