feat: md5 wrapper implement

CMakeLists.txt

@@ -1,14 +1,14 @@
-cmake_minimum_required(VERSION 3.5)
+cmake_minimum_required(VERSION 3.12)
 project(md5sum LANGUAGES CXX)
 
 option(MD5_ENABLE_TESTING "Enable testing of the md5sum library." ON)
 option(MD5_ENABLE_BENCHMARK "Enable benchmark of the md5sum library." ON)
 
 set(CMAKE_CXX_STANDARD 20)
+add_compile_options(-Wall -Wextra)
 
-add_compile_options(-fno-rtti -fno-exceptions -Wall -Wextra)
-
-add_library(md5sum STATIC src/md5_core.cc src/md5_wrapper.cc)
+add_library(md5sum STATIC src/core.cc src/wrapper.cc)
+target_compile_options(md5sum PRIVATE -fno-rtti -fno-exceptions)
 target_include_directories(md5sum INTERFACE src/)
 
 include(third_party/ThirdParty.cmake)
@@ -23,4 +23,5 @@ endif()
 if (MD5_ENABLE_BENCHMARK)
     add_executable(md5_benchmark benchmark.cc)
     target_link_libraries(md5_benchmark PRIVATE md5sum::md5sum benchmark::benchmark_main)
+    target_compile_options(md5_benchmark PRIVATE -fno-rtti -fno-exceptions)
 endif()

benchmark.cc

@@ -10,22 +10,16 @@ std::string test_data() {
     return {data, data + 64};
 }
 
-static void BM_MD5_NEXT(benchmark::State &state) {
+static void MD5_Update(benchmark::State &state) {
     const auto data = test_data();
 
-//    md5::MD5::md5_ctx c;
-//    md5::MD5::md5_reset(&c);
-
     md5::MD5 kk;
 
     for (auto _ : state) {
-
         kk.Update(data.c_str(), 64);
-
-//        md5::MD5::md5_update(&c, data.c_str(), 64);
     }
 }
 
-BENCHMARK(BM_MD5_NEXT);
+BENCHMARK(MD5_Update);
 
 BENCHMARK_MAIN();

src/md5_core.cc → src/core.cc

@@ -35,13 +35,12 @@
 
 namespace md5 {
 
-static const unsigned char Padding[64] {0x80};
+static const unsigned char Padding[64] {0x80, /* 0x00, ... */};
 
 consteval int K(int index) { // index -> [0, 64)
-    auto i = index >> 4;
     const int step[] = {1, 5, 3, 7};
     const int begin[] = {0, 1, 5, 0};
-    return (begin[i] + step[i] * index) & 0b1111;
+    return (begin[index >> 4] + step[index >> 4] * index) & 0b1111;
 }
 
 consteval int S(int index) { // index -> [0, 64)

src/md5.h

@@ -3,27 +3,29 @@
 #include <string>
 #include <cstdint>
 
+static_assert(sizeof(uintptr_t) == 8,
+    "Project only works on 64-bits architecture.");
+
 static_assert(__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__,
     "Project only works on little-endian architecture.");
 
-// TODO: static_assert for 64-bit arch
-
 namespace md5 {
 
 class MD5 {
 public:
     MD5() = default;
 
+    MD5& Final();
     MD5& Reset();
-    // TODO: using std::string_view
+
+    MD5& Update(const std::string_view &data);
     MD5& Update(const void *buffer, uint64_t len);
-    MD5& Final();
 
-    [[nodiscard]] std::string Digest();
-    [[nodiscard]] std::string String();
+    [[nodiscard]] std::string Digest() const;
 
 public:
     static std::string Hash(const std::string_view &data);
+    static std::string Hash(const void *data, uint64_t len);
 
 private:
     static constexpr uint32_t MD5_A = 0x67452301;
@@ -42,7 +44,7 @@ private:
 private:
     md5_ctx ctx_;
     char buffer_[64] {};
-    char buffer_size_ = 0;
+    uint64_t buffer_size_ = 0;
 
 private:
     /// Update md5 ctx with specified data, note that `len` is a multiple of 64.

src/md5_math.h

@@ -44,7 +44,7 @@ consteval double sin(double x) {
     if (std::abs(x) > PI / 2) {
         x = ((x > 0) ? 1 : -1) * PI - x; // -PI / 2 < x < PI / 2
     }
-    return sin_core(x);
+    return sin_core(x); // closer to 0 for better accuracy
 }
 
 } // namespace md5::math

src/md5_wrapper.cc

@@ -1,64 +0,0 @@
-#include "md5.h"
-
-namespace md5 {
-
-MD5& MD5::Reset() {
-    ctx_.A = MD5_A;
-    ctx_.B = MD5_B;
-    ctx_.C = MD5_C;
-    ctx_.D = MD5_D;
-
-    ctx_.size = 0;
-    buffer_size_ = 0;
-    return *this;
-}
-
-MD5& MD5::Update(const void *buffer, uint64_t len) {
-    // TODO: handle buffer_
-
-    if (buffer_size_ == 0) {
-        // TODO: len % 64 == 0
-        md5_update(&ctx_, buffer, len);
-    }
-
-    // TODO: save extra data into buffer_
-
-    return *this;
-}
-
-MD5& MD5::Final() {
-    // TODO: call md5_final with buffer_
-    return *this;
-}
-
-std::string MD5::Digest() {
-    // TODO: return md5 digest in std::string (length = 16)
-    return "";
-}
-
-std::string MD5::String() {
-    // TODO: return md5 string (length = 32)
-    return "";
-}
-
-std::string MD5::Hash(const std::string_view &data) {
-
-    md5_ctx ctx;
-    md5_final(&ctx, data.data(), data.size());
-
-//    std::cout << std::hex << __builtin_bswap32(ctx.A);
-//    std::cout << std::hex << __builtin_bswap32(ctx.B);
-//    std::cout << std::hex << __builtin_bswap32(ctx.C);
-//    std::cout << std::hex << __builtin_bswap32(ctx.D) << std::endl;
-
-    // TODO: perf convert speed
-    char tmp[33];
-    sprintf(tmp, "%08x%08x%08x%08x",
-            __builtin_bswap32(ctx.A),
-            __builtin_bswap32(ctx.B),
-            __builtin_bswap32(ctx.C),
-            __builtin_bswap32(ctx.D));
-    return {tmp};
-}
-
-} // namespace md5

src/wrapper.cc

@@ -0,0 +1,85 @@
+#include <cstring>
+
+#include "md5.h"
+
+namespace md5 {
+
+// TODO: inline impl
+MD5& MD5::Reset() {
+    ctx_.A = MD5_A;
+    ctx_.B = MD5_B;
+    ctx_.C = MD5_C;
+    ctx_.D = MD5_D;
+
+    ctx_.size = 0;
+    buffer_size_ = 0;
+    return *this;
+}
+
+MD5& MD5::Update(const void *data, uint64_t len) {
+    if (buffer_size_ != 0) {
+        if (buffer_size_ + len < 64) { // buffer not filled
+            std::memcpy(buffer_ + buffer_size_, data, len);
+            buffer_size_ += len;
+            return *this; // save into buffer and return
+        }
+
+        auto size = 64 - buffer_size_;
+        std::memcpy(buffer_ + buffer_size_, data, size);
+        md5_update(&ctx_, buffer_, 64); // fill and update with buffer
+        data = reinterpret_cast<const char*>(data) + size;
+        buffer_size_ = 0;
+        len -= size;
+    } // buffer is empty for now
+
+    auto size = len & ~(uint64_t)0b111111;
+    md5_update(&ctx_, data, size);
+    data = reinterpret_cast<const char*>(data) + size;
+    len &= 0b111111; // len -> [0, 64)
+
+    if (len != 0) {
+        std::memcpy(buffer_, data, len); // save remain data into buffer
+        buffer_size_ = len;
+    }
+    return *this;
+}
+
+MD5& MD5::Update(const std::string_view &data) {
+    return Update(data.data(), data.size());
+}
+
+MD5& MD5::Final() {
+    md5_final(&ctx_, buffer_, buffer_size_);
+    return *this;
+}
+
+std::string MD5::Digest() const {
+    // TODO: perf convert speed
+    char tmp[33];
+    sprintf(tmp, "%08x%08x%08x%08x",
+            __builtin_bswap32(ctx_.A),
+            __builtin_bswap32(ctx_.B),
+            __builtin_bswap32(ctx_.C),
+            __builtin_bswap32(ctx_.D));
+    return {tmp};
+}
+
+std::string MD5::Hash(const void *data, uint64_t len) {
+    md5_ctx ctx;
+    md5_final(&ctx, data, len);
+
+    // TODO: perf convert speed
+    char tmp[33];
+    sprintf(tmp, "%08x%08x%08x%08x",
+            __builtin_bswap32(ctx.A),
+            __builtin_bswap32(ctx.B),
+            __builtin_bswap32(ctx.C),
+            __builtin_bswap32(ctx.D));
+    return {tmp};
+}
+
+std::string MD5::Hash(const std::string_view &data) {
+    return Hash(data.data(), data.size());
+}
+
+} // namespace md5

test/md5_update.cc

@@ -4,46 +4,6 @@
 
 using namespace md5;
 
-std::string test_data() {
-    char data[64];
-    for (char i = 0; i < 64; ++i) {
-        data[i] = i;
-    }
-    return {data, data + 64};
-}
-
-//void dump_ctx(const md5::MD5::md5_ctx *c) {
-//    std::cout << std::hex << c->A << std::endl;
-//    std::cout << std::hex << c->B << std::endl;
-//    std::cout << std::hex << c->C << std::endl;
-//    std::cout << std::hex << c->D << std::endl;
-//    std::cout << std::dec << c->size << std::endl;
-//}
-
-TEST(md5sum, main) {
-    auto data = test_data() + test_data() + test_data() + test_data();
-
-//    md5::MD5::md5_ctx c;
-//    dump_ctx(&c);
-
-    // md5::md5_update(&c, data.c_str(), data.size());
-    // md5::md5_update(&c, data.c_str(), data.size());
-    // dump_ctx(&c);
-
-    // md5::md5_reset(&c);
-    // md5::md5_update(&c, data.c_str(), data.size());
-    // md5::md5_update(&c, data.c_str(), data.size());
-    // dump_ctx(&c);
-
-//    md5::MD5::md5_final(&c, data.c_str(), 0);
-//    dump_ctx(&c);
-//
-//    md5::MD5::md5_reset(&c);
-//    md5::MD5::md5_final(&c, data.c_str(), data.size());
-//    dump_ctx(&c);
-
-}
-
 TEST(md5sum, hash) {
     auto test_data = [](uint8_t size) -> std::string {
         std::string data {};
@@ -326,3 +286,23 @@ TEST(md5sum, hash) {
     EXPECT_EQ(MD5::Hash(test_data(0xfe)), "7bdac450b9343317aa89895d4dda181e");
     EXPECT_EQ(MD5::Hash(test_data(0xff)), "11b7aaa64c413d2f0fccf893881c46a2");
 }
+
+
+TEST(md5sum, update) {
+
+    std::string test_data {};
+    test_data.resize(256 * 256);
+
+    for (int i = 0; i < test_data.size(); ++i) {
+        test_data[i] = i & 0xff;
+    }
+
+    for (int round = 1; round <= 256; ++round) {
+        MD5 md5_obj;
+        for (int i = 0; i < 256; ++i) {
+            md5_obj.Update(test_data.data() + i * round, round);
+        }
+        EXPECT_EQ(md5_obj.Final().Digest(), MD5::Hash(test_data.data(), 256 * round));
+    }
+
+}