change: add base32 decode, initial deconstruction of hashed

laumann

Aug 25, 2022, 3:36 PM

VKLGQREYOZDV46F672RFE5XJO3OEOP4EHTCWZYOJY24HVPQX3L6QC

Dependencies

• [2] X36ICMJN Initial import for blake3
• [3] OBKF6SII change: decompress the hashed section too
• [4] RIWSVVAS change: decompress the 'contents' with zstd_seekable
• [5] Q7TKZCJP Add initial support for reading the offsets from a (fixed) change
• [*] QEFCNNVC change: display offsets for given change file
• [*] PEUS54XQ
• [*] B3XLVPNC Add ani.c and initial Makefile

Change contents

• edit in change.h at line 35
  [4.810]

  [4.810]
  + struct change_header {

  + char *message; /* the "commit" message */

  + char *description; /* the other undescribed part of a change */

  + char *timestamp; /* yes, as a string */

  + void *authors; /* something... */

  + };

  +
• edit in change.h at line 43
  [4.814]

  [4.814]
  + * The parts of a given change that is blake3-hashed.

  + *

  + * The actual hashing happens on the serialized (w/bincode) version of this

  + * struct. Decoding the 'changes' field will take quite a bit of doing, it's a

  + * fairly convoluted field.

  + */

  + struct hashed {

  + uint64_t version; /* same as in offsets? here to be hashed! */

  + struct change_header header;

  + void *dependencies; /* array of hashes */

  + void *extra_known; /* another array of hashes */

  + };

  +

  + /**
• edit in change.c at line 11
  [4.66]

  [4.1192]
  + #include "blake3.h"
• edit in change.c at line 13
  [4.1212]

  [3.0]
  + #include "base32.h"
• replacement in change.c at line 25
  [3.130]→[3.130:167](∅→∅)
  − printf("%s = [%02x", name, buf[0]);
  [3.130]

  [3.167]
  + printf("%s = [%u", name, buf[0]);
• replacement in change.c at line 27
  [3.194]→[3.194:222](∅→∅)
  − printf(", %02x", buf[i]);
  [3.194]

  [3.222]
  + printf(", %u", buf[i]);
• replacement in change.c at line 144
  [3.1083]→[3.1083:1155](∅→∅)
  − change_decode_hashed(int fd, size_t comp_hashed_len, size_t hashed_len)
  [3.1083]

  [3.1155]
  + change_decode_hashed(int fd, size_t comp_hashed_len, size_t hashed_len, uint8_t *expected_hash)
• edit in change.c at line 147
  [3.1172]

  [4.2591]
  + struct hashed hashed;

  + blake3_hasher b3sum;

  + uint8_t output[BLAKE3_OUT_LEN];

  + size_t i;
• edit in change.c at line 158
  [3.1373]

  [3.1373]
  +

  + /* TODO Decode into 'struct hashed' */

  +

  + /* Step 1. Compute the hash of the result buf! */

  +

  + blake3_hasher_init(&b3sum);

  + blake3_hasher_update(&b3sum, buf, hashed_len);

  + blake3_hasher_finalize(&b3sum, output, BLAKE3_OUT_LEN);

  +

  + for (i = 0; i < BLAKE3_OUT_LEN; i++)

  + printf("%02x", output[i]);

  + printf("\n");

  +

  + if (expected_hash != NULL) {

  + for (i = 0; i < BLAKE3_OUT_LEN; i++)

  + printf("%02x", expected_hash[i]);

  + printf("\n");

  + }

  +
• edit in change.c at line 207
  [4.2620]

  [7.122]
  + uint8_t contents_hash[32];
• edit in change.c at line 227
  [4.2474]

  [4.2474]
  + base32_decode_hash(contents_hash, hash);
• replacement in change.c at line 235
  [3.1874]→[3.1874:1956](∅→∅)
  − err = change_decode_hashed(fd, off.unhashed_off - OFFSETS_SIZE, off.hashed_len);
  [3.1874]

  [3.1956]
  + err = change_decode_hashed(fd, off.unhashed_off - OFFSETS_SIZE, off.hashed_len, contents_hash);
• file addition: base32.h (----------)
  [8.1]
  + #ifndef BASE32_H

  + #define BASE32_H

  +

  + /**

  + * Takes care of decoding a base32-encoded blake3 hash sum.

  + *

  + * The input is assumed to be an array of 53 base32 characters, dst is assumed

  + * to be able to hold at least 32 bytes.

  + */

  + void base32_decode_hash(uint8_t *dst, char const *in);

  +

  + #endif
• file addition: base32.c (----------)
  [8.1]
  + #include <stdint.h>

  +

  + /**

  + * Inspired by the base32 implementation here:

  + *

  + * https://sourceforge.net/p/cyoencode/code/HEAD/tree/trunk/src/CyoDecode.c

  + *

  + * Basically, the idea is to allocate a much too large array, fill it with 0xff

  + * and plug in the values from the alphabet in their corresponding holes.

  + */

  + static uint8_t BASE32[0x80] = {

  + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,

  + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,

  + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,

  + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,

  + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,

  + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,

  + 0xff, 0xff, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f,

  + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,

  + 0xff, 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06,

  + 0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e,

  + 0x0f, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16,

  + 0x17, 0x18, 0x19, 0xff, 0xff, 0xff, 0xff, 0xff,

  + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,

  + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,

  + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,

  + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff

  + };

  +

  + void

  + base32_decode_hash(uint8_t *dst, char const *in)

  + {

  + int i, r, off;

  + uint8_t v[8];

  +

  + for (r = 0; r < 6; r++) {

  + off = 8 * r;

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

  + v[i] = BASE32[(int)in[off + i]];

  +

  + off = 5 * r;

  + dst[off] = v[0] << 3 | v[1] >> 2;

  + dst[off+1] = v[1] << 6 | v[2] << 1 | v[3] >> 4;

  + dst[off+2] = v[3] << 4 | v[4] >> 1;

  + dst[off+3] = v[4] << 7 | v[5] << 2 | v[6] >> 3;

  + dst[off+4] = v[6] << 5 | v[7];

  + }

  + /* decode the remaining two bytes */

  + off = 48; /* 6×8 */

  + for (i = 0; i < 5; i++)

  + v[i] = BASE32[(int)in[off+i]];

  +

  + off = 30; /* 6×5 */

  + dst[off] = v[0] << 3 | v[1] >> 2;

  + dst[off+1] = v[1] << 6 | v[2] << 1 | v[3] >> 4;

  + }
• edit in Makefile at line 23
  [2.53469]

  [2.53469]
  + OBJS += base32.o