1 //#define _GLIBCXX_PARALLEL
3 // Usage: ./binloader IN1 IN2 IN3 ... OUT NUM_BUCKETS
12 #include <unordered_set>
18 #define DUMMY_TIMESTAMP 32503680000
24 enum Result { WHITE = 0, DRAW, BLACK };
28 char move[8]; // Na1xc3+
30 int opening_num, white_elo, black_elo;
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);
43 return bpfen_len < other.bpfen_len;
48 int main(int argc, char **argv)
50 int num_buckets = atoi(argv[argc - 1]);
52 vector<vector<Element>> elems;
53 elems.resize(num_buckets);
56 for (int i = 1; i < argc - 2; ++i) {
57 FILE *fp = fopen(argv[i], "rb");
63 int bpfen_len = getc(fp);
64 if (bpfen_len == -1) {
68 char *bpfen = arena.alloc(bpfen_len);
69 if (fread(bpfen, bpfen_len, 1, fp) != 1) {
82 int opening_num, white_elo, black_elo, file_num;
85 if (fread(&white_elo, sizeof(white_elo), 1, fp) != 1) {
90 if (fread(&black_elo, sizeof(black_elo), 1, fp) != 1) {
95 if (fread(&opening_num, sizeof(opening_num), 1, fp) != 1) {
100 if (fread(×tamp, sizeof(timestamp), 1, fp) != 1) {
105 if (fread(&file_num, sizeof(file_num), 1, fp) != 1) {
110 if (fread(&start_position, sizeof(start_position), 1, fp) != 1) {
122 if (l >= int(sizeof(move))) {
123 fprintf(stderr, "Overlong move (%d bytes)\n", l);
127 if (fread(&move[0], l, 1, fp) != 1) {
134 int bucket = hash_key_to_bucket(bpfen, bpfen_len, num_buckets);
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);
151 printf("Read %ld elems\n", num_elems);
154 printf("Sorting...\n");
155 for (int i = 0; i < num_buckets; ++i) {
156 sort(elems[i].begin(), elems[i].end());
159 printf("Writing SSTables...\n");
160 string buf; // Keep allocated.
161 for (int i = 0; i < num_buckets; ++i) {
163 snprintf(filename, sizeof(filename), "%s.part%04d", argv[argc - 2], i);
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);
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);
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);
184 if (!c.has_first_timestamp() || e.timestamp < c.first_timestamp()) {
185 if (e.timestamp != DUMMY_TIMESTAMP) {
186 c.set_first_timestamp(e.timestamp);
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);
192 if (!moves.count(e.move)) {
193 moves.insert(e.move);
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());
207 mtbl_writer_destroy(&mtbl);