From f4fefec343f42e1c5856965e22663f7d770f1719 Mon Sep 17 00:00:00 2001
From: Benny Halevy <bhalevy@scylladb.com>
Date: Mon, 17 Apr 2023 12:29:20 +0300
Subject: [PATCH] utils: hashing: add simple_xx_hasher

And a respective unit test.

Signed-off-by: Benny Halevy <bhalevy@scylladb.com>
---
 test/boost/hashers_test.cc | 15 +++++++++++++++
 utils/hashers.cc           |  3 +++
 utils/simple_hashers.hh    | 32 ++++++++++++++++++++++++++++++++
 3 files changed, 50 insertions(+)
 create mode 100644 utils/simple_hashers.hh

diff --git a/test/boost/hashers_test.cc b/test/boost/hashers_test.cc
index 169e641130..b8a38f839a 100644
--- a/test/boost/hashers_test.cc
+++ b/test/boost/hashers_test.cc
@@ -13,6 +13,7 @@
 #include <seastar/testing/thread_test_case.hh>
 #include "utils/hashers.hh"
 #include "utils/xx_hasher.hh"
+#include "utils/simple_hashers.hh"
 #include "gc_clock.hh"
 #include "test/lib/simple_schema.hh"
 #include "reader_concurrency_semaphore.hh"
@@ -102,3 +103,17 @@ SEASTAR_THREAD_TEST_CASE(mutation_fragment_sanity_check) {
         check_hash(f, 0x5092daca1b27ea26ull);
     }
 }
+
+BOOST_AUTO_TEST_CASE(basic_xx_hasher_sanity_check) {
+    simple_xx_hasher hasher1;
+    hasher1.update(reinterpret_cast<const char*>(std::data(text_part1)), std::size(text_part1));
+    hasher1.update(reinterpret_cast<const char*>(std::data(text_part2)), std::size(text_part2));
+    auto hash1 = hasher1.finalize();
+
+    bytes_view_hasher hasher2;
+    hasher2.update(reinterpret_cast<const char*>(std::data(text_part1)), std::size(text_part1));
+    hasher2.update(reinterpret_cast<const char*>(std::data(text_part2)), std::size(text_part2));
+    auto hash2 = hasher2.finalize();
+
+    BOOST_CHECK_EQUAL(hash1, hash2);
+}
diff --git a/utils/hashers.cc b/utils/hashers.cc
index 1961efa29f..d89a8372d4 100644
--- a/utils/hashers.cc
+++ b/utils/hashers.cc
@@ -8,6 +8,7 @@
 
 #include "utils/hashers.hh"
 #include "utils/xx_hasher.hh"
+#include "utils/simple_hashers.hh"
 
 #define CRYPTOPP_ENABLE_NAMESPACE_WEAK 1
 #include <cryptopp/md5.h>
@@ -19,6 +20,8 @@ static_assert(HasherReturningBytes<md5_hasher>);
 static_assert(HasherReturningBytes<sha256_hasher>);
 static_assert(HasherReturningBytes<xx_hasher>);
 
+static_assert(SimpleHasher<simple_xx_hasher>);
+
 template <typename T> struct hasher_traits;
 template <> struct hasher_traits<md5_hasher> { using impl_type = CryptoPP::Weak::MD5; };
 template <> struct hasher_traits<sha256_hasher> { using impl_type = CryptoPP::SHA256; };
diff --git a/utils/simple_hashers.hh b/utils/simple_hashers.hh
new file mode 100644
index 0000000000..06aff002d3
--- /dev/null
+++ b/utils/simple_hashers.hh
@@ -0,0 +1,32 @@
+/*
+ * Copyright (C) 2023-present ScyllaDB
+ */
+
+/*
+ * SPDX-License-Identifier: AGPL-3.0-or-later
+ */
+
+#pragma once
+
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Warray-bounds"
+#include <xxhash.h>
+#pragma GCC diagnostic pop
+
+#include "utils/hashing.hh"
+
+template<typename H>
+concept SimpleHasher = HasherReturning<H, size_t>;
+
+struct simple_xx_hasher : public hasher {
+    XXH64_state_t _state;
+    simple_xx_hasher(uint64_t seed = 0) noexcept {
+        XXH64_reset(&_state, seed);
+    }
+    void update(const char* ptr, size_t length) noexcept override {
+        XXH64_update(&_state, ptr, length);
+    }
+    size_t finalize() {
+        return static_cast<size_t>(XXH64_digest(&_state));
+    }
+};