1 //#define _GLIBCXX_PARALLEL
3 // Usage: ./binloader IN1 IN2 IN3 ... OUT NUM_BUCKETS
12 #include <unordered_set>
17 #define DUMMY_TIMESTAMP 32503680000
21 enum Result { WHITE = 0, DRAW, BLACK };
26 int opening_num, white_elo, black_elo;
31 bool operator< (const Element& other) const {
32 return bpfen < other.bpfen;
36 int main(int argc, char **argv)
38 int num_buckets = atoi(argv[argc - 1]);
40 vector<vector<Element>> elems;
41 elems.resize(num_buckets);
44 for (int i = 1; i < argc - 2; ++i) {
45 FILE *fp = fopen(argv[i], "rb");
58 if (fread(&bpfen[0], l, 1, fp) != 1) {
71 int opening_num, white_elo, black_elo, file_num;
74 if (fread(&white_elo, sizeof(white_elo), 1, fp) != 1) {
79 if (fread(&black_elo, sizeof(black_elo), 1, fp) != 1) {
84 if (fread(&opening_num, sizeof(opening_num), 1, fp) != 1) {
89 if (fread(×tamp, sizeof(timestamp), 1, fp) != 1) {
94 if (fread(&file_num, sizeof(file_num), 1, fp) != 1) {
99 if (fread(&start_position, sizeof(start_position), 1, fp) != 1) {
112 if (fread(&move[0], l, 1, fp) != 1) {
118 int bucket = hash_key_to_bucket(bpfen.data(), bpfen.size(), num_buckets);
119 elems[bucket].emplace_back(Element {std::move(bpfen), std::move(move), Result(r), opening_num, white_elo, black_elo, timestamp, file_num, start_position});
124 printf("Read %ld elems\n", num_elems);
127 printf("Sorting...\n");
128 for (int i = 0; i < num_buckets; ++i) {
129 sort(elems[i].begin(), elems[i].end());
132 printf("Writing SSTables...\n");
133 string buf; // Keep allocated.
134 for (int i = 0; i < num_buckets; ++i) {
136 snprintf(filename, sizeof(filename), "%s.part%04d", argv[argc - 2], i);
138 mtbl_writer_options* wopt = mtbl_writer_options_init();
139 mtbl_writer_options_set_compression(wopt, MTBL_COMPRESSION_SNAPPY);
140 mtbl_writer* mtbl = mtbl_writer_init(filename, wopt);
142 unordered_set<string> moves;
143 for (size_t j = 0; j < elems[i].size(); ++j) {
144 const Element &e = elems[i][j];
145 if (e.result == WHITE) {
146 c.set_white(c.white() + 1);
147 } else if (e.result == DRAW) {
148 c.set_draw(c.draw() + 1);
149 } else if (e.result == BLACK) {
150 c.set_black(c.black() + 1);
152 if (e.white_elo >= 100 && e.black_elo >= 100) {
153 c.set_sum_white_elo(c.sum_white_elo() + e.white_elo);
154 c.set_sum_black_elo(c.sum_black_elo() + e.black_elo);
155 c.set_num_elo(c.num_elo() + 1);
157 if (!c.has_first_timestamp() || e.timestamp < c.first_timestamp()) {
158 if (e.timestamp != DUMMY_TIMESTAMP) {
159 c.set_first_timestamp(e.timestamp);
161 c.set_opening_num(e.opening_num);
162 c.set_pgn_file_num(e.file_num);
163 c.set_pgn_start_position(e.start_position);
165 if (!moves.count(e.move)) {
166 moves.insert(e.move);
169 if (j == elems[i].size() - 1 || e.bpfen != elems[i][j + 1].bpfen) {
170 c.SerializeToString(&buf);
171 mtbl_writer_add(mtbl,
172 (const uint8_t *)e.bpfen.data(), e.bpfen.size(),
173 (const uint8_t *)buf.data(), buf.size());
178 mtbl_writer_destroy(&mtbl);