git.sesse.net Git - plocate/blob - db.h

   1 #ifndef DB_H
   2 #define DB_H 1
   3
   4 #include <stdint.h>
   5
   6 struct Header {
   7         char magic[8];  // "\0plocate";
   8         uint32_t version;  // 1.
   9         uint32_t hashtable_size;
  10         uint32_t extra_ht_slots;
  11         uint32_t num_docids;
  12         uint64_t hash_table_offset_bytes;
  13         uint64_t filename_index_offset_bytes;
  14
  15         // Version 1 and up only.
  16         uint32_t max_version;  // Nominally 1 or 2, but can be increased if more features are added in a backward-compatible way.
  17         uint32_t zstd_dictionary_length_bytes;
  18         uint64_t zstd_dictionary_offset_bytes;
  19
  20         // Only if max_version >= 2, and only relevant for updatedb.
  21         uint64_t directory_data_length_bytes;
  22         uint64_t directory_data_offset_bytes;
  23         uint64_t next_zstd_dictionary_length_bytes;
  24         uint64_t next_zstd_dictionary_offset_bytes;
  25         uint64_t conf_block_length_bytes;
  26         uint64_t conf_block_offset_bytes;
  27 };
  28
  29 struct Trigram {
  30         uint32_t trgm;
  31         uint32_t num_docids;
  32         uint64_t offset;
  33
  34         bool operator==(const Trigram &other) const
  35         {
  36                 return trgm == other.trgm;
  37         }
  38         bool operator<(const Trigram &other) const
  39         {
  40                 return trgm < other.trgm;
  41         }
  42 };
  43
  44 inline uint32_t hash_trigram(uint32_t trgm, uint32_t ht_size)
  45 {
  46         // CRC-like computation.
  47         uint32_t crc = trgm;
  48         for (int i = 0; i < 32; i++) {
  49                 bool bit = crc & 0x80000000;
  50                 crc <<= 1;
  51                 if (bit) {
  52                         crc ^= 0x1edc6f41;
  53                 }
  54         }
  55         return crc % ht_size;
  56 }
  57
  58 #endif  // !defined(DB_H)