Add method internal_memory_builder_partitioned_phf::build_from_hashes

This mirrors internal_memory_builder_single_phf::build_from_hashes
jermp · Apr 30, 2024 · f3b3f4d · f3b3f4d
1 parent 28aedcb
commit f3b3f4d
Show file tree

Hide file tree

Showing 3 changed files with 39 additions and 29 deletions.
diff --git a/include/builders/internal_memory_builder_partitioned_phf.hpp b/include/builders/internal_memory_builder_partitioned_phf.hpp
@@ -12,6 +12,17 @@ struct internal_memory_builder_partitioned_phf {
     template <typename Iterator>
     build_timings build_from_keys(Iterator keys, uint64_t num_keys,
                                   build_configuration const& config) {
+        build_configuration actual_config = config;
+        if (config.seed == constants::invalid_seed)
+            actual_config.seed = random_value();
+
+        return build_from_hashes(hash_generator<Iterator, hasher_type>(keys, actual_config.seed),
+                                 num_keys, actual_config);
+    }
+
+    template <typename Iterator>
+    build_timings build_from_hashes(Iterator hashes, uint64_t num_keys,
+                                    build_configuration const& config) {
         assert(num_keys > 1);
         util::check_hash_collision_probability<Hasher>(num_keys);
 
@@ -25,7 +36,7 @@ struct internal_memory_builder_partitioned_phf {
         uint64_t num_partitions = config.num_partitions;
         if (config.verbose_output) std::cout << "num_partitions " << num_partitions << std::endl;
 
-        m_seed = config.seed == constants::invalid_seed ? random_value() : config.seed;
+        m_seed = config.seed;
         m_num_keys = num_keys;
         m_table_size = 0;
         m_num_partitions = num_partitions;
@@ -41,9 +52,8 @@ struct internal_memory_builder_partitioned_phf {
         for (auto& partition : partitions) partition.reserve(1.5 * average_partition_size);
 
         progress_logger logger(num_keys, " == partitioned ", " keys", config.verbose_output);
-        for (uint64_t i = 0; i != num_keys; ++i, ++keys) {
-            auto const& key = *keys;
-            auto hash = hasher_type::hash(key, m_seed);
+        for (uint64_t i = 0; i != num_keys; ++i, ++hashes) {
+            auto hash = *hashes;
             auto b = m_bucketer.bucket(hash.mix());
             partitions[b].push_back(hash);
             logger.log();

diff --git a/include/builders/internal_memory_builder_single_phf.hpp b/include/builders/internal_memory_builder_single_phf.hpp
@@ -19,7 +19,7 @@ struct internal_memory_builder_single_phf {
             for (auto attempt = 0; attempt < 10; ++attempt) {
                 m_seed = random_value();
                 try {
-                    return build_from_hashes(hash_generator<RandomAccessIterator>(keys, m_seed),
+                    return build_from_hashes(hash_generator<RandomAccessIterator, hasher_type>(keys, m_seed),
                                              num_keys, config);
                 } catch (seed_runtime_error const& error) {
                     std::cout << "attempt " << attempt + 1 << " failed" << std::endl;
@@ -28,8 +28,8 @@ struct internal_memory_builder_single_phf {
             throw seed_runtime_error();
         }
         m_seed = config.seed;
-        return build_from_hashes(hash_generator<RandomAccessIterator>(keys, m_seed), num_keys,
-                                 config);
+        return build_from_hashes(hash_generator<RandomAccessIterator, hasher_type>(keys, m_seed),
+                                 num_keys, config);
     }
 
     template <typename RandomAccessIterator>
@@ -205,27 +205,6 @@ struct internal_memory_builder_single_phf {
     std::vector<uint64_t> m_pilots;
     std::vector<uint64_t> m_free_slots;
 
-    template <typename RandomAccessIterator>
-    struct hash_generator {
-        hash_generator(RandomAccessIterator keys, uint64_t seed) : m_iterator(keys), m_seed(seed) {}
-
-        inline auto operator*() {
-            return hasher_type::hash(*m_iterator, m_seed);
-        }
-
-        inline void operator++() {
-            ++m_iterator;
-        }
-
-        inline hash_generator operator+(uint64_t offset) const {
-            return hash_generator(m_iterator + offset, m_seed);
-        }
-
-    private:
-        RandomAccessIterator m_iterator;
-        uint64_t m_seed;
-    };
-
     typedef std::vector<bucket_payload_pair> pairs_t;
 
     struct buckets_iterator_t {

diff --git a/include/builders/util.hpp b/include/builders/util.hpp
@@ -286,4 +286,25 @@ void fill_free_slots(bit_vector_builder const& taken, uint64_t num_keys, FreeSlo
     assert(next_used_slot == table_size);
 }
 
-}  // namespace pthash
+template <typename RandomAccessIterator, typename Hasher>
+struct hash_generator {
+    hash_generator(RandomAccessIterator keys, uint64_t seed) : m_iterator(keys), m_seed(seed) {}
+
+    inline auto operator*() {
+        return Hasher::hash(*m_iterator, m_seed);
+    }
+
+    inline void operator++() {
+        ++m_iterator;
+    }
+
+    inline hash_generator operator+(uint64_t offset) const {
+        return hash_generator(m_iterator + offset, m_seed);
+    }
+
+private:
+    RandomAccessIterator m_iterator;
+    uint64_t m_seed;
+};
+
+}  // namespace pthash