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         // Only if max_version >= 2.
  29         bool check_visibility;
  30 };
  31
  32 struct Trigram {
  33         uint32_t trgm;
  34         uint32_t num_docids;
  35         uint64_t offset;
  36
  37         bool operator==(const Trigram &other) const
  38         {
  39                 return trgm == other.trgm;
  40         }
  41         bool operator<(const Trigram &other) const
  42         {
  43                 return trgm < other.trgm;
  44         }
  45 };
  46
  47 inline uint32_t hash_trigram(uint32_t trgm, uint32_t ht_size)
  48 {
  49         // CRC-like computation.
  50         uint32_t crc = trgm;
  51         for (int i = 0; i < 32; i++) {
  52                 bool bit = crc & 0x80000000;
  53                 crc <<= 1;
  54                 if (bit) {
  55                         crc ^= 0x1edc6f41;
  56                 }
  57         }
  58         return crc % ht_size;
  59 }
  60
  61 #endif  // !defined(DB_H)