Build Reactive Signals for Bluesky's AT Protocol Firehose in Laravel
Add CID (Content Identifier) implementation for IPLD
+371
+271
src/Core/CID.php
+271
src/Core/CID.php
···
1
+
<?php
2
+
3
+
declare(strict_types=1);
4
+
5
+
namespace SocialDept\Signal\Core;
6
+
7
+
use RuntimeException;
8
+
use SocialDept\Signal\Binary\Reader;
9
+
use SocialDept\Signal\Binary\Varint;
10
+
11
+
/**
12
+
* Content Identifier (CID) parser for IPLD.
13
+
*
14
+
* Supports CIDv0 (base58btc) and CIDv1 (multibase).
15
+
* Minimal implementation for reading CIDs from CBOR and CAR data.
16
+
*/
17
+
class CID
18
+
{
19
+
private const BASE32_CHARSET = 'abcdefghijklmnopqrstuvwxyz234567';
20
+
private const BASE58BTC_CHARSET = '123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz';
21
+
22
+
public function __construct(
23
+
public readonly int $version,
24
+
public readonly int $codec,
25
+
public readonly string $hash,
26
+
) {}
27
+
28
+
/**
29
+
* Parse CID from binary data.
30
+
*
31
+
* @param string $data Binary CID data
32
+
* @return self
33
+
* @throws RuntimeException If CID is malformed
34
+
*/
35
+
public static function fromBinary(string $data): self
36
+
{
37
+
$reader = new Reader($data);
38
+
39
+
// Read version
40
+
$version = $reader->readVarint();
41
+
42
+
if ($version === 0x12) {
43
+
// CIDv0 (legacy format - starts with multihash directly)
44
+
// Reset and read as v0
45
+
$reader = new Reader($data);
46
+
$hashType = $reader->readVarint(); // 0x12 = sha256
47
+
$hashLength = $reader->readVarint(); // typically 32
48
+
$hash = $reader->readBytes($hashLength);
49
+
50
+
return new self(
51
+
version: 0,
52
+
codec: 0x70, // dag-pb
53
+
hash: chr($hashType) . chr($hashLength) . $hash,
54
+
);
55
+
}
56
+
57
+
if ($version !== 1) {
58
+
throw new RuntimeException("Unsupported CID version: {$version}");
59
+
}
60
+
61
+
// Read codec
62
+
$codec = $reader->readVarint();
63
+
64
+
// Read multihash (hash type + length + hash bytes)
65
+
$hashType = $reader->readVarint();
66
+
$hashLength = $reader->readVarint();
67
+
$hashBytes = $reader->readBytes($hashLength);
68
+
69
+
// Store complete multihash
70
+
$hash = chr($hashType) . chr($hashLength) . $hashBytes;
71
+
72
+
return new self(
73
+
version: $version,
74
+
codec: $codec,
75
+
hash: $hash,
76
+
);
77
+
}
78
+
79
+
/**
80
+
* Parse CID from string (base32 or base58btc).
81
+
*
82
+
* @param string $str CID string
83
+
* @return self
84
+
* @throws RuntimeException If CID string is invalid
85
+
*/
86
+
public static function fromString(string $str): self
87
+
{
88
+
if (empty($str)) {
89
+
throw new RuntimeException('Empty CID string');
90
+
}
91
+
92
+
// Check multibase prefix
93
+
$prefix = $str[0];
94
+
95
+
if ($prefix === 'b') {
96
+
// base32 (CIDv1)
97
+
$binary = self::decodeBase32(substr($str, 1));
98
+
99
+
return self::fromBinary($binary);
100
+
}
101
+
102
+
if ($prefix === 'Q' || $prefix === '1') {
103
+
// base58btc (likely CIDv0)
104
+
$binary = self::decodeBase58($str);
105
+
106
+
return self::fromBinary($binary);
107
+
}
108
+
109
+
throw new RuntimeException("Unsupported multibase prefix: {$prefix}");
110
+
}
111
+
112
+
/**
113
+
* Convert CID to string representation.
114
+
*/
115
+
public function toString(): string
116
+
{
117
+
if ($this->version === 0) {
118
+
// CIDv0 is always base58btc without prefix
119
+
return self::encodeBase58($this->hash);
120
+
}
121
+
122
+
// CIDv1 uses base32 with 'b' prefix
123
+
$binary = chr($this->version) . $this->encodeVarint($this->codec) . $this->hash;
124
+
125
+
return 'b' . self::encodeBase32($binary);
126
+
}
127
+
128
+
/**
129
+
* Get binary representation.
130
+
*/
131
+
public function toBinary(): string
132
+
{
133
+
if ($this->version === 0) {
134
+
return $this->hash;
135
+
}
136
+
137
+
return chr($this->version) . $this->encodeVarint($this->codec) . $this->hash;
138
+
}
139
+
140
+
/**
141
+
* Decode base32 string to binary.
142
+
*/
143
+
private static function decodeBase32(string $str): string
144
+
{
145
+
$str = strtolower($str);
146
+
$result = '';
147
+
$bits = 0;
148
+
$value = 0;
149
+
150
+
for ($i = 0; $i < strlen($str); $i++) {
151
+
$char = $str[$i];
152
+
$pos = strpos(self::BASE32_CHARSET, $char);
153
+
154
+
if ($pos === false) {
155
+
throw new RuntimeException("Invalid base32 character: {$char}");
156
+
}
157
+
158
+
$value = ($value << 5) | $pos;
159
+
$bits += 5;
160
+
161
+
if ($bits >= 8) {
162
+
$result .= chr(($value >> ($bits - 8)) & 0xFF);
163
+
$bits -= 8;
164
+
}
165
+
}
166
+
167
+
return $result;
168
+
}
169
+
170
+
/**
171
+
* Encode binary to base32 string.
172
+
*/
173
+
private static function encodeBase32(string $data): string
174
+
{
175
+
$result = '';
176
+
$bits = 0;
177
+
$value = 0;
178
+
179
+
for ($i = 0; $i < strlen($data); $i++) {
180
+
$value = ($value << 8) | ord($data[$i]);
181
+
$bits += 8;
182
+
183
+
while ($bits >= 5) {
184
+
$result .= self::BASE32_CHARSET[($value >> ($bits - 5)) & 0x1F];
185
+
$bits -= 5;
186
+
}
187
+
}
188
+
189
+
if ($bits > 0) {
190
+
$result .= self::BASE32_CHARSET[($value << (5 - $bits)) & 0x1F];
191
+
}
192
+
193
+
return $result;
194
+
}
195
+
196
+
/**
197
+
* Decode base58btc string to binary.
198
+
*/
199
+
private static function decodeBase58(string $str): string
200
+
{
201
+
$decoded = gmp_init(0);
202
+
$base = gmp_init(58);
203
+
204
+
for ($i = 0; $i < strlen($str); $i++) {
205
+
$char = $str[$i];
206
+
$pos = strpos(self::BASE58BTC_CHARSET, $char);
207
+
208
+
if ($pos === false) {
209
+
throw new RuntimeException("Invalid base58 character: {$char}");
210
+
}
211
+
212
+
$decoded = gmp_add(gmp_mul($decoded, $base), gmp_init($pos));
213
+
}
214
+
215
+
$hex = gmp_strval($decoded, 16);
216
+
if (strlen($hex) % 2) {
217
+
$hex = '0' . $hex;
218
+
}
219
+
220
+
// Add leading zeros
221
+
for ($i = 0; $i < strlen($str) && $str[$i] === '1'; $i++) {
222
+
$hex = '00' . $hex;
223
+
}
224
+
225
+
return hex2bin($hex);
226
+
}
227
+
228
+
/**
229
+
* Encode binary to base58btc string.
230
+
*/
231
+
private static function encodeBase58(string $data): string
232
+
{
233
+
$num = gmp_init('0x' . bin2hex($data));
234
+
$base = gmp_init(58);
235
+
$result = '';
236
+
237
+
while (gmp_cmp($num, 0) > 0) {
238
+
[$num, $remainder] = gmp_div_qr($num, $base);
239
+
$result = self::BASE58BTC_CHARSET[gmp_intval($remainder)] . $result;
240
+
}
241
+
242
+
// Add leading '1's for leading zero bytes
243
+
for ($i = 0; $i < strlen($data) && ord($data[$i]) === 0; $i++) {
244
+
$result = '1' . $result;
245
+
}
246
+
247
+
return $result;
248
+
}
249
+
250
+
/**
251
+
* Encode varint for CID binary format.
252
+
*/
253
+
private function encodeVarint(int $value): string
254
+
{
255
+
$result = '';
256
+
257
+
while ($value >= 0x80) {
258
+
$result .= chr(($value & 0x7F) | 0x80);
259
+
$value >>= 7;
260
+
}
261
+
262
+
$result .= chr($value);
263
+
264
+
return $result;
265
+
}
266
+
267
+
public function __toString(): string
268
+
{
269
+
return $this->toString();
270
+
}
271
+
}
+100
tests/Unit/CIDTest.php
+100
tests/Unit/CIDTest.php
···
1
+
<?php
2
+
3
+
declare(strict_types=1);
4
+
5
+
namespace SocialDept\Signal\Tests\Unit;
6
+
7
+
use PHPUnit\Framework\TestCase;
8
+
use SocialDept\Signal\Core\CID;
9
+
10
+
class CIDTest extends TestCase
11
+
{
12
+
public function test_parse_binary_cidv1(): void
13
+
{
14
+
// CIDv1: version=1, codec=0x71 (dag-cbor), sha256 hash
15
+
$hash = str_repeat("\x00", 32);
16
+
$binary = "\x01\x71\x12\x20" . $hash;
17
+
18
+
$cid = CID::fromBinary($binary);
19
+
20
+
$this->assertSame(1, $cid->version);
21
+
$this->assertSame(0x71, $cid->codec);
22
+
$this->assertSame("\x12\x20" . $hash, $cid->hash);
23
+
}
24
+
25
+
public function test_parse_binary_cidv0(): void
26
+
{
27
+
// CIDv0: starts with 0x12 (sha256) 0x20 (32 bytes)
28
+
$hash = str_repeat("\x00", 32);
29
+
$binary = "\x12\x20" . $hash;
30
+
31
+
$cid = CID::fromBinary($binary);
32
+
33
+
$this->assertSame(0, $cid->version);
34
+
$this->assertSame(0x70, $cid->codec); // dag-pb
35
+
$this->assertSame("\x12\x20" . $hash, $cid->hash);
36
+
}
37
+
38
+
public function test_to_string_cidv1(): void
39
+
{
40
+
$hash = str_repeat("\x00", 32);
41
+
$binary = "\x01\x71\x12\x20" . $hash;
42
+
$cid = CID::fromBinary($binary);
43
+
44
+
$str = $cid->toString();
45
+
46
+
// Should start with 'b' (base32 prefix)
47
+
$this->assertStringStartsWith('b', $str);
48
+
49
+
// Should be able to parse it back
50
+
$parsed = CID::fromString($str);
51
+
$this->assertSame($cid->version, $parsed->version);
52
+
$this->assertSame($cid->codec, $parsed->codec);
53
+
}
54
+
55
+
public function test_to_binary_cidv1(): void
56
+
{
57
+
$hash = str_repeat("\x00", 32);
58
+
$binary = "\x01\x71\x12\x20" . $hash;
59
+
$cid = CID::fromBinary($binary);
60
+
61
+
$this->assertSame($binary, $cid->toBinary());
62
+
}
63
+
64
+
public function test_round_trip_binary(): void
65
+
{
66
+
$hash = hash('sha256', 'test', true);
67
+
$binary = "\x01\x71\x12\x20" . $hash;
68
+
69
+
$cid = CID::fromBinary($binary);
70
+
$encoded = $cid->toBinary();
71
+
$decoded = CID::fromBinary($encoded);
72
+
73
+
$this->assertSame($cid->version, $decoded->version);
74
+
$this->assertSame($cid->codec, $decoded->codec);
75
+
$this->assertSame($cid->hash, $decoded->hash);
76
+
}
77
+
78
+
public function test_round_trip_string(): void
79
+
{
80
+
$hash = hash('sha256', 'test', true);
81
+
$binary = "\x01\x71\x12\x20" . $hash;
82
+
$cid = CID::fromBinary($binary);
83
+
84
+
$str = $cid->toString();
85
+
$parsed = CID::fromString($str);
86
+
87
+
$this->assertSame($cid->version, $parsed->version);
88
+
$this->assertSame($cid->codec, $parsed->codec);
89
+
$this->assertSame($cid->hash, $parsed->hash);
90
+
}
91
+
92
+
public function test_to_string_magic_method(): void
93
+
{
94
+
$hash = str_repeat("\x00", 32);
95
+
$binary = "\x01\x71\x12\x20" . $hash;
96
+
$cid = CID::fromBinary($binary);
97
+
98
+
$this->assertSame($cid->toString(), (string) $cid);
99
+
}
100
+
}