Source

`stdlib/hashlib/hashlib.cpp`

2// Original work; see ACKNOWLEDGMENTS.md for the open-source ideas we build upon.

3#include "hashlib.hpp"

5#include <array>

6#include <cstdint>

7#include <stdexcept>

8#include <vector>

10namespace cheatah::hashlib {

12namespace {

14inline std::uint32_t rotr(std::uint32_t x, std::uint32_t n) {

15 return (x >> n) | (x << (32 - n));

16}

18constexpr std::uint32_t K[64] = {

19 0x428a2f98, 0x71374491, 0xb5c0fbcf, 0xe9b5dba5, 0x3956c25b, 0x59f111f1, 0x923f82a4, 0xab1c5ed5,

20 0xd807aa98, 0x12835b01, 0x243185be, 0x550c7dc3, 0x72be5d74, 0x80deb1fe, 0x9bdc06a7, 0xc19bf174,

21 0xe49b69c1, 0xefbe4786, 0x0fc19dc6, 0x240ca1cc, 0x2de92c6f, 0x4a7484aa, 0x5cb0a9dc, 0x76f988da,

22 0x983e5152, 0xa831c66d, 0xb00327c8, 0xbf597fc7, 0xc6e00bf3, 0xd5a79147, 0x06ca6351, 0x14292967,

23 0x27b70a85, 0x2e1b2138, 0x4d2c6dfc, 0x53380d13, 0x650a7354, 0x766a0abb, 0x81c2c92e, 0x92722c85,

24 0xa2bfe8a1, 0xa81a664b, 0xc24b8b70, 0xc76c51a3, 0xd192e819, 0xd6990624, 0xf40e3585, 0x106aa070,

25 0x19a4c116, 0x1e376c08, 0x2748774c, 0x34b0bcb5, 0x391c0cb3, 0x4ed8aa4a, 0x5b9cca4f, 0x682e6ff3,

26 0x748f82ee, 0x78a5636f, 0x84c87814, 0x8cc70208, 0x90befffa, 0xa4506ceb, 0xbef9a3f7, 0xc67178f2,

27};

29inline std::uint64_t rotr64(std::uint64_t x, unsigned n) {

30 return (x >> n) | (x << (64 - n));

31}

33// SHA-512 round constants (first 64 bits of the fractional parts of the cube roots of

34// the first 80 primes) and initial hash values (square roots of the first 8 primes).

35constexpr std::uint64_t K512[80] = {

36 0x428a2f98d728ae22ULL, 0x7137449123ef65cdULL, 0xb5c0fbcfec4d3b2fULL, 0xe9b5dba58189dbbcULL,

37 0x3956c25bf348b538ULL, 0x59f111f1b605d019ULL, 0x923f82a4af194f9bULL, 0xab1c5ed5da6d8118ULL,

38 0xd807aa98a3030242ULL, 0x12835b0145706fbeULL, 0x243185be4ee4b28cULL, 0x550c7dc3d5ffb4e2ULL,

39 0x72be5d74f27b896fULL, 0x80deb1fe3b1696b1ULL, 0x9bdc06a725c71235ULL, 0xc19bf174cf692694ULL,

40 0xe49b69c19ef14ad2ULL, 0xefbe4786384f25e3ULL, 0x0fc19dc68b8cd5b5ULL, 0x240ca1cc77ac9c65ULL,

41 0x2de92c6f592b0275ULL, 0x4a7484aa6ea6e483ULL, 0x5cb0a9dcbd41fbd4ULL, 0x76f988da831153b5ULL,

42 0x983e5152ee66dfabULL, 0xa831c66d2db43210ULL, 0xb00327c898fb213fULL, 0xbf597fc7beef0ee4ULL,

43 0xc6e00bf33da88fc2ULL, 0xd5a79147930aa725ULL, 0x06ca6351e003826fULL, 0x142929670a0e6e70ULL,

44 0x27b70a8546d22ffcULL, 0x2e1b21385c26c926ULL, 0x4d2c6dfc5ac42aedULL, 0x53380d139d95b3dfULL,

45 0x650a73548baf63deULL, 0x766a0abb3c77b2a8ULL, 0x81c2c92e47edaee6ULL, 0x92722c851482353bULL,

46 0xa2bfe8a14cf10364ULL, 0xa81a664bbc423001ULL, 0xc24b8b70d0f89791ULL, 0xc76c51a30654be30ULL,

47 0xd192e819d6ef5218ULL, 0xd69906245565a910ULL, 0xf40e35855771202aULL, 0x106aa07032bbd1b8ULL,

48 0x19a4c116b8d2d0c8ULL, 0x1e376c085141ab53ULL, 0x2748774cdf8eeb99ULL, 0x34b0bcb5e19b48a8ULL,

49 0x391c0cb3c5c95a63ULL, 0x4ed8aa4ae3418acbULL, 0x5b9cca4f7763e373ULL, 0x682e6ff3d6b2b8a3ULL,

50 0x748f82ee5defb2fcULL, 0x78a5636f43172f60ULL, 0x84c87814a1f0ab72ULL, 0x8cc702081a6439ecULL,

51 0x90befffa23631e28ULL, 0xa4506cebde82bde9ULL, 0xbef9a3f7b2c67915ULL, 0xc67178f2e372532bULL,

52 0xca273eceea26619cULL, 0xd186b8c721c0c207ULL, 0xeada7dd6cde0eb1eULL, 0xf57d4f7fee6ed178ULL,

53 0x06f067aa72176fbaULL, 0x0a637dc5a2c898a6ULL, 0x113f9804bef90daeULL, 0x1b710b35131c471bULL,

54 0x28db77f523047d84ULL, 0x32caab7b40c72493ULL, 0x3c9ebe0a15c9bebcULL, 0x431d67c49c100d4cULL,

55 0x4cc5d4becb3e42b6ULL, 0x597f299cfc657e2aULL, 0x5fcb6fab3ad6faecULL, 0x6c44198c4a475817ULL,

56};

58// to_hex / from_hex are this module's PUBLIC canonical byte<->hex helpers (defined after this

59// anonymous namespace, declared in hashlib.hpp). The digest paths below and the tls / ed25519 /

60// x509 modules all share that one implementation instead of re-rolling their own.

62// SHA-256 core: returns the raw 32-byte digest.

63std::array<std::uint8_t, 32> sha256_raw(std::string_view data) {

64 std::uint32_t h[8] = {0x6a09e667, 0xbb67ae85, 0x3c6ef372, 0xa54ff53a,

65 0x510e527f, 0x9b05688c, 0x1f83d9ab, 0x5be0cd19};

67 // Pad: append 0x80, then zeros, then the 64-bit big-endian bit length.

68 const std::uint64_t bitlen = static_cast<std::uint64_t>(data.size()) * 8;

69 std::vector<std::uint8_t> msg(data.begin(), data.end());

70 msg.push_back(0x80);

71 while (msg.size() % 64 != 56) msg.push_back(0x00);

72 for (int i = 7; i >= 0; --i) msg.push_back(static_cast<std::uint8_t>(bitlen >> (i * 8)));

74 for (std::size_t off = 0; off < msg.size(); off += 64) {

75 std::uint32_t w[64];

76 for (int i = 0; i < 16; ++i) {

77 w[i] = (static_cast<std::uint32_t>(msg[off + i * 4]) << 24) |

78 (static_cast<std::uint32_t>(msg[off + i * 4 + 1]) << 16) |

79 (static_cast<std::uint32_t>(msg[off + i * 4 + 2]) << 8) |

80 (static_cast<std::uint32_t>(msg[off + i * 4 + 3]));

81 }

82 for (int i = 16; i < 64; ++i) {

83 const std::uint32_t s0 = rotr(w[i - 15], 7) ^ rotr(w[i - 15], 18) ^ (w[i - 15] >> 3);

84 const std::uint32_t s1 = rotr(w[i - 2], 17) ^ rotr(w[i - 2], 19) ^ (w[i - 2] >> 10);

85 w[i] = w[i - 16] + s0 + w[i - 7] + s1;

86 }

88 std::uint32_t a = h[0], b = h[1], c = h[2], d = h[3];

89 std::uint32_t e = h[4], f = h[5], g = h[6], hh = h[7];

90 for (int i = 0; i < 64; ++i) {

91 const std::uint32_t S1 = rotr(e, 6) ^ rotr(e, 11) ^ rotr(e, 25);

92 const std::uint32_t ch = (e & f) ^ (~e & g);

93 const std::uint32_t t1 = hh + S1 + ch + K[i] + w[i];

94 const std::uint32_t S0 = rotr(a, 2) ^ rotr(a, 13) ^ rotr(a, 22);

95 const std::uint32_t maj = (a & b) ^ (a & c) ^ (b & c);

96 const std::uint32_t t2 = S0 + maj;

97 hh = g; g = f; f = e; e = d + t1;

98 d = c; c = b; b = a; a = t1 + t2;

99 }

100 h[0] += a; h[1] += b; h[2] += c; h[3] += d;

101 h[4] += e; h[5] += f; h[6] += g; h[7] += hh;

102 }

103

104 std::array<std::uint8_t, 32> out{};

105 for (int i = 0; i < 8; ++i)

106 for (int j = 0; j < 4; ++j) out[i * 4 + j] = static_cast<std::uint8_t>(h[i] >> ((3 - j) * 8));

107 return out;

108}

109

110// SHA-512-family core (FIPS 180-4): the 80-round compression over 128-byte blocks, started

111// from the caller's initial hash values. SHA-512 and SHA-384 share everything but the IV and

112// how much of the final state is kept, so both run through this one engine.

113std::array<std::uint8_t, 64> sha512_core(std::string_view data, const std::uint64_t iv[8]) {

114 std::uint64_t h[8];

115 for (int i = 0; i < 8; ++i) h[i] = iv[i];

116

117 // Pad: 0x80, zeros to a 128-byte block leaving 16 bytes, then the 128-bit big-endian

118 // bit length (the high 64 bits are always 0 for our inputs).

119 const std::uint64_t bitlen = static_cast<std::uint64_t>(data.size()) * 8;

120 std::vector<std::uint8_t> msg(data.begin(), data.end());

121 msg.push_back(0x80);

122 while (msg.size() % 128 != 112) msg.push_back(0x00);

123 for (int i = 0; i < 8; ++i) msg.push_back(0x00); // high 64 bits of the length

124 for (int i = 7; i >= 0; --i) msg.push_back(static_cast<std::uint8_t>(bitlen >> (i * 8)));

125

126 for (std::size_t off = 0; off < msg.size(); off += 128) {

127 std::uint64_t w[80];

128 for (int i = 0; i < 16; ++i) {

129 w[i] = 0;

130 for (int j = 0; j < 8; ++j)

131 w[i] = (w[i] << 8) | msg[off + i * 8 + j];

132 }

133 for (int i = 16; i < 80; ++i) {

134 const std::uint64_t s0 = rotr64(w[i - 15], 1) ^ rotr64(w[i - 15], 8) ^ (w[i - 15] >> 7);

135 const std::uint64_t s1 = rotr64(w[i - 2], 19) ^ rotr64(w[i - 2], 61) ^ (w[i - 2] >> 6);

136 w[i] = w[i - 16] + s0 + w[i - 7] + s1;

137 }

138

139 std::uint64_t a = h[0], b = h[1], c = h[2], d = h[3];

140 std::uint64_t e = h[4], f = h[5], g = h[6], hh = h[7];

141 for (int i = 0; i < 80; ++i) {

142 const std::uint64_t S1 = rotr64(e, 14) ^ rotr64(e, 18) ^ rotr64(e, 41);

143 const std::uint64_t ch = (e & f) ^ (~e & g);

144 const std::uint64_t t1 = hh + S1 + ch + K512[i] + w[i];

145 const std::uint64_t S0 = rotr64(a, 28) ^ rotr64(a, 34) ^ rotr64(a, 39);

146 const std::uint64_t maj = (a & b) ^ (a & c) ^ (b & c);

147 const std::uint64_t t2 = S0 + maj;

148 hh = g; g = f; f = e; e = d + t1;

149 d = c; c = b; b = a; a = t1 + t2;

150 }

151 h[0] += a; h[1] += b; h[2] += c; h[3] += d;

152 h[4] += e; h[5] += f; h[6] += g; h[7] += hh;

153 }

154

155 std::array<std::uint8_t, 64> out{};

156 for (int i = 0; i < 8; ++i)

157 for (int j = 0; j < 8; ++j) out[i * 8 + j] = static_cast<std::uint8_t>(h[i] >> ((7 - j) * 8));

158 return out;

159}

160

161// SHA-512: the core with its own initial hash values (square roots of the first 8 primes).

162std::array<std::uint8_t, 64> sha512_raw(std::string_view data) {

163 static constexpr std::uint64_t kIv[8] = {

164 0x6a09e667f3bcc908ULL, 0xbb67ae8584caa73bULL, 0x3c6ef372fe94f82bULL, 0xa54ff53a5f1d36f1ULL,

165 0x510e527fade682d1ULL, 0x9b05688c2b3e6c1fULL, 0x1f83d9abfb41bd6bULL, 0x5be0cd19137e2179ULL};

166 return sha512_core(data, kIv);

167}

168

169// SHA-384: the same core with the SHA-384 initial hash values (square roots of the 9th..16th

170// primes), keeping the leftmost 48 bytes of the final state.

171std::array<std::uint8_t, 48> sha384_raw(std::string_view data) {

172 static constexpr std::uint64_t kIv[8] = {

173 0xcbbb9d5dc1059ed8ULL, 0x629a292a367cd507ULL, 0x9159015a3070dd17ULL, 0x152fecd8f70e5939ULL,

174 0x67332667ffc00b31ULL, 0x8eb44a8768581511ULL, 0xdb0c2e0d64f98fa7ULL, 0x47b5481dbefa4fa4ULL};

175 const std::array<std::uint8_t, 64> full = sha512_core(data, kIv);

176 std::array<std::uint8_t, 48> out{};

177 for (int i = 0; i < 48; ++i) out[i] = full[i];

178 return out;

179}

180

181} // namespace

182

183std::string sha256(std::string_view data) {

184 const auto d = sha256_raw(data);

185 return to_hex(d.data(), d.size());

186}

187

188std::string sha384(std::string_view data) {

189 const auto d = sha384_raw(data);

190 return to_hex(d.data(), d.size());

191}

192

193std::string sha512(std::string_view data) {

194 const auto d = sha512_raw(data);

195 return to_hex(d.data(), d.size());

196}

197

198std::string sha256_digest(std::string_view data) {

199 const auto d = sha256_raw(data);

200 return std::string(reinterpret_cast<const char*>(d.data()), d.size());

201}

202

203std::string sha384_digest(std::string_view data) {

204 const auto d = sha384_raw(data);

205 return std::string(reinterpret_cast<const char*>(d.data()), d.size());

206}

207

208std::string sha512_digest(std::string_view data) {

209 const auto d = sha512_raw(data);

210 return std::string(reinterpret_cast<const char*>(d.data()), d.size());

211}

212

213// ---- HMAC + HKDF (RFC 2104 / RFC 5869) over the SHA-256 above ----------------------

214

215std::string hmac_sha256(std::string_view key, std::string_view data) {

216 constexpr std::size_t kBlock = 64; // SHA-256 block size

217 std::string k(key);

218 if (k.size() > kBlock) k = sha256_digest(k); // long keys hash down first

219 k.resize(kBlock, '\0');

220 std::string inner(kBlock, '\0');

221 std::string outer(kBlock, '\0');

222 for (std::size_t i = 0; i < kBlock; ++i) {

223 inner[i] = static_cast<char>(k[i] ^ 0x36);

224 outer[i] = static_cast<char>(k[i] ^ 0x5c);

225 }

226 return sha256_digest(outer + sha256_digest(inner + std::string(data)));

227}

228

229std::string hmac_sha384(std::string_view key, std::string_view data) {

230 constexpr std::size_t kBlock = 128; // SHA-384 shares SHA-512's 128-byte block

231 std::string k(key);

232 if (k.size() > kBlock) k = sha384_digest(k); // long keys hash down first

233 k.resize(kBlock, '\0');

234 std::string inner(kBlock, '\0');

235 std::string outer(kBlock, '\0');

236 for (std::size_t i = 0; i < kBlock; ++i) {

237 inner[i] = static_cast<char>(k[i] ^ 0x36);

238 outer[i] = static_cast<char>(k[i] ^ 0x5c);

239 }

240 return sha384_digest(outer + sha384_digest(inner + std::string(data)));

241}

242

243std::string hmac_sha512(std::string_view key, std::string_view data) {

244 constexpr std::size_t kBlock = 128; // SHA-512 block size

245 std::string k(key);

246 if (k.size() > kBlock) k = sha512_digest(k); // long keys hash down first

247 k.resize(kBlock, '\0');

248 std::string inner(kBlock, '\0');

249 std::string outer(kBlock, '\0');

250 for (std::size_t i = 0; i < kBlock; ++i) {

251 inner[i] = static_cast<char>(k[i] ^ 0x36);

252 outer[i] = static_cast<char>(k[i] ^ 0x5c);

253 }

254 return sha512_digest(outer + sha512_digest(inner + std::string(data)));

255}

256

257// ---- Hex (the one canonical byte<->hex for the whole crypto stack) -----------------

258

259std::string to_hex(const std::uint8_t* data, std::size_t n) {

260 static const char hexd[] = "0123456789abcdef";

261 std::string out;

262 out.reserve(n * 2);

263 for (std::size_t i = 0; i < n; ++i) {

264 out.push_back(hexd[data[i] >> 4]);

265 out.push_back(hexd[data[i] & 0xF]);

266 }

267 return out;

268}

269

270std::string to_hex(std::string_view bytes) {

271 return to_hex(reinterpret_cast<const std::uint8_t*>(bytes.data()), bytes.size());

272}

273

274std::string from_hex(std::string_view hex) {

275 if (hex.size() % 2 != 0) throw std::invalid_argument("hashlib::from_hex: odd-length hex");

276 auto nibble = [](char c) -> int {

277 if (c >= '0' && c <= '9') return c - '0';

278 if (c >= 'a' && c <= 'f') return c - 'a' + 10;

279 if (c >= 'A' && c <= 'F') return c - 'A' + 10;

280 return -1;

281 };

282 std::string out;

283 out.reserve(hex.size() / 2);

284 for (std::size_t i = 0; i < hex.size(); i += 2) {

285 const int hi = nibble(hex[i]), lo = nibble(hex[i + 1]);

286 if (hi < 0 || lo < 0) throw std::invalid_argument("hashlib::from_hex: non-hex character");

287 out.push_back(static_cast<char>((hi << 4) | lo));

288 }

289 return out;

290}

291

292// ---- Base64 (RFC 4648, standard alphabet) -----------------------------------------

293

294namespace {

295constexpr char kB64[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";

296} // namespace

297

298std::string base64_encode(std::string_view data) {

299 const auto* p = reinterpret_cast<const unsigned char*>(data.data());

300 const std::size_t n = data.size();

301 std::string out;

302 out.reserve(((n + 2) / 3) * 4);

303 std::size_t i = 0;

304 for (; i + 3 <= n; i += 3) {

305 const unsigned v = (static_cast<unsigned>(p[i]) << 16) |

306 (static_cast<unsigned>(p[i + 1]) << 8) | static_cast<unsigned>(p[i + 2]);

307 out.push_back(kB64[(v >> 18) & 63]);

308 out.push_back(kB64[(v >> 12) & 63]);

309 out.push_back(kB64[(v >> 6) & 63]);

310 out.push_back(kB64[v & 63]);

311 }

312 if (n - i == 1) {

313 const unsigned v = static_cast<unsigned>(p[i]) << 16;

314 out.push_back(kB64[(v >> 18) & 63]);

315 out.push_back(kB64[(v >> 12) & 63]);

316 out.push_back('=');

317 out.push_back('=');

318 } else if (n - i == 2) {

319 const unsigned v =

320 (static_cast<unsigned>(p[i]) << 16) | (static_cast<unsigned>(p[i + 1]) << 8);

321 out.push_back(kB64[(v >> 18) & 63]);

322 out.push_back(kB64[(v >> 12) & 63]);

323 out.push_back(kB64[(v >> 6) & 63]);

324 out.push_back('=');

325 }

326 return out;

327}

328

329std::string base64_decode(std::string_view text, bool strict) {

330 std::array<int, 256> rev;

331 rev.fill(-1);

332 for (int i = 0; i < 64; ++i) rev[static_cast<unsigned char>(kB64[i])] = i;

333 std::string out;

334 out.reserve((text.size() / 4) * 3);

335 int buf = 0;

336 int bits = 0;

337 for (unsigned char c : text) {

338 if (c == '=') break; // padding -> end of data

339 const int d = rev[c];

340 if (d < 0) {

341 // Whitespace is always skipped. Any OTHER non-alphabet byte is silently ignored in the

342 // default (lenient) mode, but REJECTED — empty result — in strict mode. X.509 PEM parsing

343 // uses strict so a malformed body fails closed instead of decoding to garbage.

344 if (strict && c != ' ' && c != '\n' && c != '\r' && c != '\t') return std::string();

345 continue; // skip whitespace / (lenient) non-alphabet bytes

346 }

347 buf = (buf << 6) | d;

348 bits += 6;

349 if (bits >= 8) {

350 bits -= 8;

351 out.push_back(static_cast<char>((buf >> bits) & 0xff));

352 }

353 }

354 return out;

355}

356

357// HKDF (RFC 5869) generic over its HMAC and hash-output length, so the SHA-256 and SHA-384 key

358// schedules (the latter for the TLS_AES_256_GCM_SHA384 cipher suite) share one implementation.

359namespace {

360using HmacFn = std::string (*)(std::string_view, std::string_view);

361

362std::string hkdf_expand_impl(std::string_view prk, std::string_view info, long long length,

363 HmacFn hmac, std::size_t hlen) {

364 if (length <= 0 || length > 255 * static_cast<long long>(hlen)) return "";

365 std::string okm;

366 std::string t; // T(0) = empty

367 for (unsigned char counter = 1; static_cast<long long>(okm.size()) < length; ++counter) {

368 t = hmac(prk, t + std::string(info) + static_cast<char>(counter));

369 okm += t;

370 }

371 okm.resize(static_cast<std::size_t>(length));

372 return okm;

373}

374} // namespace

375

376std::string hkdf_extract(std::string_view salt, std::string_view ikm) {

377 return hmac_sha256(salt, ikm);

378}

379

380std::string hkdf_expand(std::string_view prk, std::string_view info, long long length) {

381 return hkdf_expand_impl(prk, info, length, hmac_sha256, 32);

382}

383

384std::string hkdf_extract_sha384(std::string_view salt, std::string_view ikm) {

385 return hmac_sha384(salt, ikm);

386}

387

388std::string hkdf_expand_sha384(std::string_view prk, std::string_view info, long long length) {

389 return hkdf_expand_impl(prk, info, length, hmac_sha384, 48);

390}

391

392} // namespace cheatah::hashlib