Merge branch 'tarlocvec_serialize' into 'master'

serialize target_location buckets as vectors

See merge request muellan/metacache!7

src/hash_multimap.h

@@ -922,6 +922,8 @@ class hash_multimap
     void deserialize(std::istream& is)
     {
         using len_t = std::uint64_t;
+        using target_id = typename value_type::target_id_t;
+        using window_id = typename value_type::window_id_t;
 
         clear();
 
@@ -937,17 +939,25 @@ class hash_multimap
             reserve_values(nvalues);
             reserve_keys(nkeys);
 
+            std::vector<target_id> target_id_buf;
+            std::vector<window_id> window_id_buf;
+
             for(len_t i = 0; i < nkeys; ++i) {
                 key_type key;
                 bucket_size_type nvals = 0;
                 read_binary(is, key);
                 read_binary(is, nvals);
                 if(nvals > 0) {
+                    read_binary(is, target_id_buf);
+                    read_binary(is, window_id_buf);
+
                     auto it = insert_into_slot(std::move(key), nullptr, 0, 0);
+                    if(it == buckets_.end()) continue;
+
                     it->resize(alloc_, nvals);
 
-                    for(auto v = it->values_, e = v+nvals; v < e; ++v) {
-                        read_binary(is, *v);
+                    for(size_t i = 0; i < nvals; ++i) {
+                        it->values_[i] = {target_id_buf[i], window_id_buf[i]};
                     }
                 }
             }
@@ -964,18 +974,32 @@ class hash_multimap
     void serialize(std::ostream& os) const
     {
         using len_t = std::uint64_t;
+        using target_id = typename value_type::target_id_t;
+        using window_id = typename value_type::window_id_t;
 
         write_binary(os, len_t(non_empty_bucket_count()));
         write_binary(os, len_t(value_count()));
 
+        std::vector<target_id> target_id_buf;
+        std::vector<window_id> window_id_buf;
+
         for(const auto& bucket : buckets_) {
             if(!bucket.empty()) {
                 write_binary(os, bucket.key());
                 write_binary(os, bucket.size());
 
+                target_id_buf.clear();
+                target_id_buf.reserve(bucket.size());
+                window_id_buf.clear();
+                window_id_buf.reserve(bucket.size());
+
                 for(const auto& v : bucket) {
-                    write_binary(os, v);
+                    target_id_buf.emplace_back(v.tgt);
+                    window_id_buf.emplace_back(v.win);
                 }
+
+                write_binary(os, target_id_buf);
+                write_binary(os, window_id_buf);
             }
         }
     }

src/io_serialize.h

@@ -62,9 +62,8 @@ write_binary(std::ostream& os, const std::vector<T>& v)
 {
     std::uint64_t n = v.size();
     os.write(reinterpret_cast<const char*>(&n), sizeof(n));
-    for(const auto& x : v) {
-        write_binary(os, x);
-    }
+    if(n > 0)
+        os.write(reinterpret_cast<const char*>(v.data()), n * sizeof(T));
 }
 
 
@@ -116,13 +115,11 @@ inline void
 read_binary(std::istream& is, std::vector<T>& v)
 {
 
-    std::uint64_t l = 0;
-    is.read(reinterpret_cast<char*>(&l), sizeof(l));
-    v.clear();
-    v.resize(l);
-    for(auto& x : v) {
-        read_binary(is, x);
-    }
+    std::uint64_t n = 0;
+    is.read(reinterpret_cast<char*>(&n), sizeof(n));
+    v.resize(n);
+    if(n > 0)
+        is.read(reinterpret_cast<char*>(v.data()), n * sizeof(T));
 }
 
 

src/sketch_database.h

@@ -158,6 +158,9 @@ class sketch_database
     //avoid padding bits
     struct target_location
     {
+        using target_id_t = target_id;
+        using window_id_t = window_id;
+
         constexpr
         target_location(target_id g = 0, window_id w = 0) noexcept :
             tgt{g}, win{w}
@@ -871,10 +874,6 @@ class sketch_database
 
         //target insertion parameters
         read_binary(is, maxLocsPerFeature_);
-        //dummy for legacy db format compatibility
-        //could be re-used in the future
-        float dummy;
-        read_binary(is, dummy);
 
         //taxon metadata
         read_binary(is, taxa_);
@@ -944,10 +943,6 @@ class sketch_database
 
         //target insertion parameters
         write_binary(os, maxLocsPerFeature_);
-        //dummy for legacy db format compatibility
-        //could be re-used in the future
-        float dummy = 0.0f;
-        write_binary(os, dummy);
 
         //taxon & target metadata
         write_binary(os, taxa_);

src/version.h

@@ -26,7 +26,7 @@
 
 #define MC_VERSION 20180830
 
-#define MC_DB_VERSION 20180227
+#define MC_DB_VERSION 20180831
 
 #define MC_VERSION_STRING "0.3.0"