From ae2fb1d5717f896168aeafd895adacf38733f384 Mon Sep 17 00:00:00 2001
From: battlmonstr <battlmonstr@users.noreply.github.com>
Date: Sat, 16 Mar 2024 19:04:07 +0100
Subject: [PATCH] RecSplit.add_key: use ByteView

---
 silkworm/db/snapshots/index.cpp               |  8 +++----
 silkworm/db/snapshots/rec_split/rec_split.hpp | 24 ++++++++++---------
 2 files changed, 17 insertions(+), 15 deletions(-)

diff --git a/silkworm/db/snapshots/index.cpp b/silkworm/db/snapshots/index.cpp
index 5b9b514fa6..30e70c6fd9 100644
--- a/silkworm/db/snapshots/index.cpp
+++ b/silkworm/db/snapshots/index.cpp
@@ -73,14 +73,14 @@ bool HeaderIndex::walk(RecSplit8& rec_split, uint64_t i, uint64_t offset, ByteVi
     const ethash::hash256 hash = keccak256(rlp_encoded_header);
     ensure(hash.bytes[0] == first_hash_byte,
            [&]() { return "HeaderIndex: invalid prefix=" + to_hex(first_hash_byte) + " hash=" + to_hex(hash.bytes); });
-    rec_split.add_key(hash.bytes, kHashLength, offset);
+    rec_split.add_key(ByteView{hash.bytes}, offset);
     return true;
 }
 
 bool BodyIndex::walk(RecSplit8& rec_split, uint64_t i, uint64_t offset, ByteView /*word*/) {
     Bytes uint64_buffer;
     seg::varint::encode(uint64_buffer, i);
-    rec_split.add_key(uint64_buffer.data(), uint64_buffer.size(), offset);
+    rec_split.add_key(uint64_buffer, offset);
     return true;
 }
 
@@ -249,7 +249,7 @@ void TransactionIndex::build() {
                         throw std::runtime_error{error.str()};
                     }
 
-                    tx_hash_to_block_rs.add_key(tx_hash.bytes, kHashLength, block_number);
+                    tx_hash_to_block_rs.add_key(tx_hash, block_number);
                     i++;
                 }
 
@@ -278,7 +278,7 @@ bool TransactionIndex::walk(RecSplit8& rec_split, uint64_t i, uint64_t offset, B
         throw std::runtime_error{error.str()};
     }
 
-    rec_split.add_key(tx_hash.bytes, kHashLength, offset);
+    rec_split.add_key(tx_hash, offset);
     return true;
 }
 
diff --git a/silkworm/db/snapshots/rec_split/rec_split.hpp b/silkworm/db/snapshots/rec_split/rec_split.hpp
index 61c88475fe..be89854af6 100644
--- a/silkworm/db/snapshots/rec_split/rec_split.hpp
+++ b/silkworm/db/snapshots/rec_split/rec_split.hpp
@@ -65,6 +65,8 @@
 #include <gsl/util>
 
 #include <silkworm/core/common/assert.hpp>
+#include <silkworm/core/common/bytes.hpp>
+#include <silkworm/core/common/bytes_to_string.hpp>
 #include <silkworm/core/common/endian.hpp>
 #include <silkworm/core/common/math.hpp>
 #include <silkworm/core/common/util.hpp>
@@ -389,17 +391,17 @@ class RecSplit {
         building_strategy_->add_key(bucket_id, bucket_key, offset);
     }
 
-    void add_key(const void* key_data, const size_t key_length, uint64_t offset) {
+    void add_key(ByteView key, uint64_t offset) {
         if (built_) {
             throw std::logic_error{"cannot add key after perfect hash function has been built"};
         }
 
-        const auto key_hash = murmur_hash_3(key_data, key_length);
+        const auto key_hash = murmur_hash_3(key);
         add_key(key_hash, offset);
     }
 
     void add_key(const std::string& key, uint64_t offset) {
-        add_key(key.c_str(), key.size(), offset);
+        add_key(string_view_to_byte_view(key), offset);
     }
 
     //! Build the MPHF using the RecSplit algorithm and save the resulting index file
@@ -608,17 +610,17 @@ class RecSplit {
     }
 
     //! Return the value associated with the given key within the MPHF mapping
-    std::size_t operator()(const std::string& key) const { return operator()(murmur_hash_3(key.c_str(), key.size())); }
+    std::size_t operator()(ByteView key) const { return operator()(murmur_hash_3(key)); }
 
-    //! Return the value associated with the given key within the index
-    [[nodiscard]] std::size_t lookup(ByteView key) const { return lookup(key.data(), key.size()); }
+    //! Return the value associated with the given key within the MPHF mapping
+    std::size_t operator()(const std::string& key) const { return operator()(string_view_to_byte_view(key)); }
 
     //! Return the value associated with the given key within the index
-    [[nodiscard]] std::size_t lookup(const std::string& key) const { return lookup(key.data(), key.size()); }
+    [[nodiscard]] std::size_t lookup(const std::string& key) const { return lookup(string_view_to_byte_view(key)); }
 
     //! Return the value associated with the given key within the index
-    std::size_t lookup(const void* key, const size_t length) const {
-        const auto record = operator()(murmur_hash_3(key, length));
+    std::size_t lookup(ByteView key) const {
+        const auto record = operator()(key);
         const auto position = 1 + 8 + bytes_per_record_ * (record + 1);
 
         const auto region = encoded_file_->region();
@@ -842,9 +844,9 @@ class RecSplit {
         }
     }
 
-    hash128_t inline murmur_hash_3(const void* data, const size_t length) const {
+    hash128_t inline murmur_hash_3(ByteView data) const {
         hash128_t h{};
-        hasher_->hash_x64_128(data, length, &h);
+        hasher_->hash_x64_128(data.data(), data.size(), &h);
         return h;
     }