Treat 0-length moves as not having a move.

[remoteglot-book] / binloader.cpp

diff --git a/binloader.cpp b/binloader.cpp
index 2ed60cbe78250ff34eba9f7c358968f90c4ddb97..730f59279eba946af8073826b08cf775564cf994 100644 (file)
--- a/binloader.cpp
+++ b/binloader.cpp
@@ -12,18 +12,36 @@
 #include <unordered_set>
 #include <string.h>
 #include "count.pb.h"
-#include "arena.h"
 #include "hash.h"
 
 #define DUMMY_TIMESTAMP 32503680000
 
 using namespace std;
 
-Arena arena;
+static inline int memcmp_different_len(const void *s1, size_t n1, const void *s2, size_t n2)
+{
+       size_t shared_len = min(n1, n2);
+       if (shared_len >= 8) {
+               uint64_t a1 = *(const uint64_t *)s1;
+               uint64_t a2 = *(const uint64_t *)s2;
+               if (a1 != a2) {
+                       a1 = __builtin_bswap64(a1);
+                       a2 = __builtin_bswap64(a2);
+                       return (a1 < a2) ? -1 : 1;
+               }
+       }
+
+       int s = memcmp(s1, s2, shared_len);
+       if (s != 0) {
+               return s;
+       }
+
+       return n2 - n1;
+}
 
 enum Result { WHITE = 0, DRAW, BLACK };
 struct Element {
-       char *bpfen;
+       char bpfen[32];  // includes prev_board_hash
        int bpfen_len;
        char move[8];   // Na1xc3+
        Result result;
@@ -33,21 +51,13 @@ struct Element {
        long start_position;
 
        bool operator< (const Element& other) const {
-               int shared_len = min(bpfen_len, other.bpfen_len);
-               int s = memcmp(bpfen, other.bpfen, shared_len);
-               if (s < 0) {
-                       return true;
-               } else if (s > 0) {
-                       return false;
-               } else {
-                       return bpfen_len < other.bpfen_len;
-               }
+               int s = memcmp_different_len(bpfen, bpfen_len, other.bpfen, other.bpfen_len);
+               return s < 0;
        }
 };
 
 struct ShardData {
        vector<Element> elems;
-       unique_ptr<Arena> arena;  // Used to allocate bpfen.
        int num_written_subshards = 0;
 };
 
@@ -88,7 +98,7 @@ void write_subshard(const char *basename, ShardData* shard, int bucket)
                        c.set_pgn_file_num(e.file_num);
                        c.set_pgn_start_position(e.start_position);
                }
-               if (!moves.count(e.move)) {
+               if (strlen(e.move) > 0 && !moves.count(e.move)) {
                        moves.insert(e.move);
                        c.add_move(e.move);
                }
@@ -96,9 +106,10 @@ void write_subshard(const char *basename, ShardData* shard, int bucket)
                    e.bpfen_len != shard->elems[i + 1].bpfen_len ||
                    memcmp(e.bpfen, shard->elems[i + 1].bpfen, e.bpfen_len) != 0) {
                        c.SerializeToString(&buf);
-                       mtbl_writer_add(mtbl,
+                       mtbl_res res = mtbl_writer_add(mtbl,
                                (const uint8_t *)e.bpfen, e.bpfen_len,
                                (const uint8_t *)buf.data(), buf.size());
+                       assert(res == mtbl_res_success);
                        c = Count();
                        moves.clear();
                }
@@ -106,7 +117,6 @@ void write_subshard(const char *basename, ShardData* shard, int bucket)
        mtbl_writer_destroy(&mtbl);
 
        shard->elems.clear();
-       shard->arena.reset(new Arena);
 }
 
 int main(int argc, char **argv)
@@ -119,7 +129,6 @@ int main(int argc, char **argv)
 
        for (int i = 0; i < num_buckets; ++i) {
                shards[i].elems.reserve(num_pos_per_subshard);
-               shards[i].arena.reset(new Arena);
        }
 
        size_t num_elems = 0;
@@ -141,6 +150,7 @@ int main(int argc, char **argv)
                        if (bpfen_len == -1) {
                                break;
                        }
+                       assert(bpfen_len <= 32);
                        if (bpfen_len >= int(sizeof(bpfen))) {
                                fprintf(stderr, "Overlong BPFEN (%d bytes)\n", bpfen_len);
                //              exit(1);
@@ -204,7 +214,9 @@ int main(int argc, char **argv)
                //              exit(1);
                                break;
                        }
-                       if (fread(&move[0], l, 1, fp) != 1) {
+                       if (l == 0) {
+                               move[0] = 0;
+                       } else if (fread(&move[0], l, 1, fp) != 1) {
                                perror("fread()");
                //              exit(1);
                                break;
@@ -213,7 +225,6 @@ int main(int argc, char **argv)
 
                        int bucket = hash_key_to_bucket(bpfen, bpfen_len, num_buckets);
                        Element e;
-                       e.bpfen = shards[bucket].arena->alloc(bpfen_len);
                        memcpy(e.bpfen, bpfen, bpfen_len);
                        e.bpfen_len = bpfen_len;
                        strcpy(e.move, move);