git.sesse.net Git - remoteglot-book/blob - binloader.cpp

   1 //#define _GLIBCXX_PARALLEL
   2
   3 // Usage: ./binloader IN1 IN2 IN3 ... OUT NUM_BUCKETS
   4
   5 #include <stdio.h>
   6 #include <vector>
   7 #include <mtbl.h>
   8 #include <algorithm>
   9 #include <utility>
  10 #include <memory>
  11 #include <string>
  12 #include <unordered_set>
  13 #include <string.h>
  14 #include "count.pb.h"
  15 #include "arena.h"
  16 #include "hash.h"
  17
  18 #define DUMMY_TIMESTAMP 32503680000
  19
  20 using namespace std;
  21
  22 Arena arena;
  23
  24 enum Result { WHITE = 0, DRAW, BLACK };
  25 struct Element {
  26         char *bpfen;
  27         int bpfen_len;
  28         char move[8];   // Na1xc3+
  29         Result result;
  30         int opening_num, white_elo, black_elo;
  31         int file_num;
  32         time_t timestamp;
  33         long start_position;
  34
  35         bool operator< (const Element& other) const {
  36                 int shared_len = min(bpfen_len, other.bpfen_len);
  37                 int s = memcmp(bpfen, other.bpfen, shared_len);
  38                 if (s < 0) {
  39                         return true;
  40                 } else if (s > 0) {
  41                         return false;
  42                 } else {
  43                         return bpfen_len < other.bpfen_len;
  44                 }
  45         }
  46 };
  47
  48 int main(int argc, char **argv)
  49 {
  50         int num_buckets = atoi(argv[argc - 1]);
  51
  52         vector<vector<Element>> elems;
  53         elems.resize(num_buckets);
  54
  55         size_t num_elems = 0;
  56         for (int i = 1; i < argc - 2; ++i) {
  57                 FILE *fp = fopen(argv[i], "rb");
  58                 if (fp == NULL) {
  59                         perror(argv[i]);
  60                         exit(1);
  61                 }
  62                 for ( ;; ) {
  63                         int bpfen_len = getc(fp);
  64                         if (bpfen_len == -1) {
  65                                 break;
  66                         }
  67
  68                         char *bpfen = arena.alloc(bpfen_len);
  69                         if (fread(bpfen, bpfen_len, 1, fp) != 1) {
  70                                 perror("fread()");
  71                 //              exit(1);
  72                                 break;
  73                         }
  74
  75                         int r = getc(fp);
  76                         if (r == -1) {
  77                                 perror("getc()");
  78                                 //exit(1);
  79                                 break;
  80                         }
  81
  82                         int opening_num, white_elo, black_elo, file_num;
  83                         time_t timestamp;
  84                         long start_position;
  85                         if (fread(&white_elo, sizeof(white_elo), 1, fp) != 1) {
  86                                 perror("fread()");
  87                                 //exit(1);
  88                                 break;
  89                         }
  90                         if (fread(&black_elo, sizeof(black_elo), 1, fp) != 1) {
  91                                 perror("fread()");
  92                                 //exit(1);
  93                                 break;
  94                         }
  95                         if (fread(&opening_num, sizeof(opening_num), 1, fp) != 1) {
  96                                 perror("fread()");
  97                                 //exit(1);
  98                                 break;
  99                         }
 100                         if (fread(&timestamp, sizeof(timestamp), 1, fp) != 1) {
 101                                 perror("fread()");
 102                                 //exit(1);
 103                                 break;
 104                         }
 105                         if (fread(&file_num, sizeof(file_num), 1, fp) != 1) {
 106                                 perror("fread()");
 107                                 //exit(1);
 108                                 break;
 109                         }
 110                         if (fread(&start_position, sizeof(start_position), 1, fp) != 1) {
 111                                 perror("fread()");
 112                                 //exit(1);
 113                                 break;
 114                         }
 115
 116
 117                         char move[8];
 118                         int l = getc(fp);
 119                         if (l == -1) {
 120                                 break;
 121                         }
 122                         if (l >= int(sizeof(move))) {
 123                                 fprintf(stderr, "Overlong move (%d bytes)\n", l);
 124                 //              exit(1);
 125                                 break;
 126                         }
 127                         if (fread(&move[0], l, 1, fp) != 1) {
 128                                 perror("fread()");
 129                 //              exit(1);
 130                                 break;
 131                         }
 132                         move[l] = 0;
 133
 134                         int bucket = hash_key_to_bucket(bpfen, bpfen_len, num_buckets);
 135                         Element e;
 136                         e.bpfen = bpfen;
 137                         e.bpfen_len = bpfen_len;
 138                         strcpy(e.move, move);
 139                         e.result = Result(r);
 140                         e.opening_num = opening_num;
 141                         e.white_elo = white_elo;
 142                         e.black_elo = black_elo;
 143                         e.file_num = file_num;
 144                         e.timestamp = timestamp;
 145                         e.start_position = start_position;
 146                         elems[bucket].push_back(e);
 147                         ++num_elems;
 148                 }
 149                 fclose(fp);
 150
 151                 printf("Read %ld elems\n", num_elems);
 152         }
 153
 154         printf("Sorting...\n");
 155         for (int i = 0; i < num_buckets; ++i) {
 156                 sort(elems[i].begin(), elems[i].end());
 157         }
 158
 159         printf("Writing SSTables...\n");
 160         string buf;  // Keep allocated.
 161         for (int i = 0; i < num_buckets; ++i) {
 162                 char filename[256];
 163                 snprintf(filename, sizeof(filename), "%s.part%04d", argv[argc - 2], i);
 164
 165                 mtbl_writer_options* wopt = mtbl_writer_options_init();
 166                 mtbl_writer_options_set_compression(wopt, MTBL_COMPRESSION_SNAPPY);
 167                 mtbl_writer* mtbl = mtbl_writer_init(filename, wopt);
 168                 Count c;
 169                 unordered_set<string> moves;
 170                 for (size_t j = 0; j < elems[i].size(); ++j) {
 171                         const Element &e = elems[i][j];
 172                         if (e.result == WHITE) {
 173                                 c.set_white(c.white() + 1);
 174                         } else if (e.result == DRAW) {
 175                                 c.set_draw(c.draw() + 1);
 176                         } else if (e.result == BLACK) {
 177                                 c.set_black(c.black() + 1);
 178                         }
 179                         if (e.white_elo >= 100 && e.black_elo >= 100) {
 180                                 c.set_sum_white_elo(c.sum_white_elo() + e.white_elo);
 181                                 c.set_sum_black_elo(c.sum_black_elo() + e.black_elo);
 182                                 c.set_num_elo(c.num_elo() + 1);
 183                         }
 184                         if (!c.has_first_timestamp() || e.timestamp < c.first_timestamp()) {
 185                                 if (e.timestamp != DUMMY_TIMESTAMP) {
 186                                         c.set_first_timestamp(e.timestamp);
 187                                 }
 188                                 c.set_opening_num(e.opening_num);
 189                                 c.set_pgn_file_num(e.file_num);
 190                                 c.set_pgn_start_position(e.start_position);
 191                         }
 192                         if (!moves.count(e.move)) {
 193                                 moves.insert(e.move);
 194                                 c.add_move(e.move);
 195                         }
 196                         if (j == elems[i].size() - 1 ||
 197                             e.bpfen_len != elems[i][j + 1].bpfen_len ||
 198                             memcmp(e.bpfen, elems[i][j + 1].bpfen, e.bpfen_len) != 0) {
 199                                 c.SerializeToString(&buf);
 200                                 mtbl_writer_add(mtbl,
 201                                         (const uint8_t *)e.bpfen, e.bpfen_len,
 202                                         (const uint8_t *)buf.data(), buf.size());
 203                                 c = Count();
 204                                 moves.clear();
 205                         }
 206                 }
 207                 mtbl_writer_destroy(&mtbl);
 208         }
 209 }