Refactoring (#5)

Initial code cleanups

.editorconfig

@@ -0,0 +1,17 @@
+root = true
+
+[!*.{h,cpp}]
+indent_style = space
+indent_size = 2
+end_of_line = lf
+charset = utf-8
+trim_trailing_whitespace = true
+insert_final_newline = true
+
+[*.{h,cpp}]
+indent_style = tab
+trim_trailing_whitespace = true
+insert_final_newline = true
+
+[*.md]
+trim_trailing_whitespace = false

CMakeLists.txt

@@ -1,20 +1,33 @@
-cmake_minimum_required(VERSION 3.10)
-project(snowboy)
+cmake_minimum_required(VERSION 3.12)
+project(snowman VERSION 1.0.0 DESCRIPTION "Snowman hotword detection library")
 list(APPEND CMAKE_MODULE_PATH ${CMAKE_CURRENT_LIST_DIR}/cmake)
 
-#option(SNOWBOY_CXX11_COMPAT "Build library with C++11 strings disable to be binary compatible with the original release." OFF)
-set(SNOWBOY_CXX11_COMPAT ON)
+option(SNOWMAN_CXX11_COMPAT "Build library with C++11 strings disabled to be binary compatible with the original release." ON)
+option(SNOWMAN_BUILD_APPS "Build helper applications like enroll or cut" ON)
+option(SNOWMAN_BUILD_TESTS "Build unit tests (requires gtest and openssl)" ON)
+option(SNOWMAN_BUILD_SHARED "Build library as a shared library instead of a static library" OFF)
+# According to steam ~99.17% of users have at least ssse3
+# and 100% have SSE, SSE2 and SSE3, so this is on by default
+option(SNOWMAN_BUILD_WITH_SSE3 "Enable sse3 optimizations" ON)
+# ~98.3%, so this is on by default as well
+option(SNOWMAN_BUILD_WITH_SSE4 "Enable sse4 optimizations" ON)
+# ~94.7%
+option(SNOWMAN_BUILD_WITH_AVX "Enable avx optimizations" OFF)
+# ~82%
+option(SNOWMAN_BUILD_WITH_AVX2 "Enable avx2 optimizations" OFF)
+option(SNOWMAN_BUILD_NATIVE "Build library for the current cpu. This makes sure it uses every instruction set available, but the resulting binary probably won't run on older hardware." OFF)
 
-if(SNOWBOY_CXX11_COMPAT)
-add_compile_options(-D_GLIBCXX_USE_CXX11_ABI=0)
-endif()
-
-add_compile_options(-std=c++0x -Wall -Wno-sign-compare -Wno-unused-local-typedefs -Winit-self -rdynamic)
-add_compile_options(-DHAVE_POSIX_MEMALIGN -I. -fno-omit-frame-pointer -fPIC -msse -msse2)
 add_compile_options("$<$<CONFIG:DEBUG>:-fsanitize=address>")
-add_link_options(-rdynamic)
 add_link_options("$<$<CONFIG:DEBUG>:-fsanitize=address>")
 
+if(SNOWMAN_CXX11_COMPAT)
+add_compile_options(-D_GLIBCXX_USE_CXX11_ABI=0)
+endif()
+
 add_subdirectory(lib)
-add_subdirectory(apps)
-add_subdirectory(test)
+if(SNOWMAN_BUILD_APPS)
+    add_subdirectory(apps)
+endif()
+if(SNOWMAN_BUILD_TESTS)
+    add_subdirectory(test)
+endif()

apps/cut.cpp

@@ -1,13 +1,9 @@
 #include <cstring>
-#include <framer-stream.h>
 #include <fstream>
 #include <helper.h>
 #include <iostream>
-#include <pipeline-personal-enroll.h>
 #include <snowboy-detect.h>
-#include <template-container.h>
-#include <template-enroll-stream.h>
-#include <types.h>
+#include <wave-header.h>
 
 const static auto root = detect_project_root();
 

apps/enroll.cpp

@@ -1,14 +1,7 @@
 #include <cstring>
-#include <framer-stream.h>
-#include <fstream>
 #include <helper.h>
 #include <iostream>
-#include <openssl/crypto.h>
-#include <openssl/md5.h>
-#include <pipeline-personal-enroll.h>
 #include <snowboy-detect.h>
-#include <template-container.h>
-#include <template-enroll-stream.h>
 
 const static auto root = detect_project_root();
 
@@ -20,9 +13,6 @@ int main(int argc, const char** argv) {
 	bool cut_recordings;
 	if (!parse_args(argc, argv, output, recordings, lang, cut_recordings)) return -1;
 
-	{
-		std::ofstream t{output, std::ios::binary | std::ios::trunc};
-	}
 	snowboy::SnowboyPersonalEnroll enroll{root + "resources/pmdl/" + lang + "/personal_enroll.res", output};
 	snowboy::SnowboyTemplateCut cut{root + "resources/pmdl/" + lang + "/personal_enroll.res"};
 	for (auto& e : recordings) {

lib/CMakeLists.txt

@@ -1,5 +1,5 @@
 
-add_library(snowboy_reimpl
+set(SNOWMAN_SRC
     ${CMAKE_CURRENT_SOURCE_DIR}/agc.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/audio-lib.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/dtw-lib.cpp
@@ -28,6 +28,7 @@ add_library(snowboy_reimpl
     ${CMAKE_CURRENT_SOURCE_DIR}/raw-nnet-vad-stream.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/snowboy-debug.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/snowboy-detect.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/snowboy-detect-c.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/snowboy-io.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/snowboy-math.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/snowboy-options.cpp
@@ -41,11 +42,35 @@ add_library(snowboy_reimpl
     ${CMAKE_CURRENT_SOURCE_DIR}/vad-state-stream.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/vector-wrapper.cpp
 )
-target_include_directories(snowboy_reimpl PUBLIC ${CMAKE_CURRENT_SOURCE_DIR})
-target_link_directories(snowboy_reimpl PRIVATE /usr/lib/atlas-base)
-target_link_libraries(snowboy_reimpl dl m rt pthread f77blas cblas lapack_atlas atlas)
+if(SNOWMAN_BUILD_SHARED)
+    add_library(snowman SHARED ${SNOWMAN_SRC})
+    set_target_properties(snowman PROPERTIES VERSION ${PROJECT_VERSION})
+    set_target_properties(mylib PROPERTIES SOVERSION ${CMAKE_PROJECT_VERSION_MAJOR})
+else()
+    add_library(snowman STATIC ${SNOWMAN_SRC})
+endif()
+target_compile_features(snowman PRIVATE cxx_std_11)
+target_compile_options(snowman PRIVATE -Wall -Wno-sign-compare -Winit-self -rdynamic)
+target_compile_options(snowman PRIVATE -DHAVE_POSIX_MEMALIGN -fno-omit-frame-pointer)
+target_include_directories(snowman PUBLIC ${CMAKE_CURRENT_SOURCE_DIR})
+target_link_directories(snowman PRIVATE /usr/lib/atlas-base)
+target_link_libraries(snowman m pthread f77blas cblas lapack_atlas atlas)
+
+if(SNOWMAN_BUILD_WITH_SSE3)
+target_compile_options(snowman PRIVATE -msse -msse2 -msse3 -mssse3)
+endif()
+if(SNOWMAN_BUILD_WITH_SSE4)
+target_compile_options(snowman PRIVATE -msse4.2)
+endif()
+if(SNOWMAN_BUILD_WITH_AVX)
+target_compile_options(snowman PRIVATE -mavx)
+endif()
+if(SNOWMAN_BUILD_WITH_AVX2)
+target_compile_options(snowman PRIVATE -mavx2)
+endif()
+if(SNOWMAN_BUILD_NATIVE)
+target_compile_options(snowman PRIVATE -march=native -mtune=native)
+endif()
+
+add_library(snowboy ALIAS snowman)
 
-file(TOUCH ${CMAKE_CURRENT_BINARY_DIR}/dummy.cpp)
-add_library(snowboy ${CMAKE_CURRENT_BINARY_DIR}/dummy.cpp)
-target_link_libraries(snowboy -Wl,--start-group snowboy_reimpl -Wl,--end-group)
-#target_link_libraries(snowboy -Wl,--start-group snowboy_reimpl ${CMAKE_CURRENT_SOURCE_DIR}/../resources/nnet-lib.o -Wl,--end-group)

lib/audio-lib.cpp

@@ -1,7 +1,7 @@
 #include <audio-lib.h>
 #include <matrix-wrapper.h>
 #include <snowboy-debug.h>
-#include <types.h>
+#include <wave-header.h>
 
 namespace snowboy {
 	float GetMaxWaveAmplitude(const WaveHeader& hdr) {

lib/dtw-lib.cpp

@@ -109,7 +109,7 @@ namespace snowboy {
 			auto local_22c = std::numeric_limits<float>::max();
 			for (auto row = 0; row < param_2.m_rows; row++) {
 				/* try { // try from 00101d18 to 00101d74 has its CatchHandler @ 00102283 */
-				int local_1e8, local_1e4, local_1e0, local_1dc;
+				int local_1e8 = 0, local_1e4 = 0, local_1e0 = 0, local_1dc = 0;
 				snowboy::SlidingDtw::ComputeBandBoundary(row, &local_1e8, &local_1e4);
 				if (0 < row) {
 					snowboy::SlidingDtw::ComputeBandBoundary(row - 1, &local_1e0, &local_1dc);
@@ -200,22 +200,16 @@ namespace snowboy {
 		Matrix local_1d8;
 		local_1d8.Resize(param_2.m_rows, param_3.m_rows);
 		for (auto row = 0; row != local_1d8.m_rows; row++) {
-			const int iVar7 = (int(row) - 1) * (int)local_1d8.m_stride;
-			const int iVar14 = row * local_1f8.m_stride;
-			const int local_284 = row * local_1f8.m_stride;
-			const int iVar11 = row * local_1d8.m_stride;
-			const int iVar15 = row * local_1d8.m_stride;
 			if (0 < local_1d8.m_cols) {
-				auto pfVar9 = local_1d8.m_data + (long)iVar7 + -1;
-				auto pfVar8 = local_1d8.m_data + (long)iVar11 + -1;
-				if (row == 0) {
-					auto lVar12 = 0;
-					do {
+				auto pfVar9 = local_1d8.m_data + (row - 1) * local_1d8.m_stride - 1;
+				auto pfVar8 = local_1d8.m_data + (row * local_1d8.m_stride) - 1;
+				for (auto lVar12 = 0; lVar12 < local_1d8.m_cols; lVar12++) {
+					if (row == 0) {
 						while (((int)lVar12 == 0 || (row == 0))) {
-							pfVar9 = pfVar9 + 1;
 							local_1d8.m_data[lVar12] = local_1f8.m_data[lVar12];
-							lVar12 += 1;
+							pfVar9 = pfVar9 + 1;
 							pfVar8 = pfVar8 + 1;
+							lVar12++;
 							if (local_1d8.m_cols <= (int)lVar12) goto LAB_00186e9d;
 						}
 						auto fVar16 = pfVar9[1];
@@ -225,18 +219,12 @@ namespace snowboy {
 						if (*pfVar9 <= fVar16) {
 							fVar16 = *pfVar9;
 						}
-						auto lVar6 = iVar14 + lVar12;
-						lVar12 += 1;
-						pfVar8[1] = fVar16 + local_1f8.m_data[lVar6];
-						pfVar9 = pfVar9 + 1;
-						pfVar8 = pfVar8 + 1;
-					} while ((int)lVar12 < local_1d8.m_cols);
-				} else {
-					auto lVar12 = 0;
-					do {
+						pfVar8[1] = fVar16 + local_1f8.m_data[row * local_1f8.m_stride + lVar12];
+
+					} else {
 						if (((int)lVar12 == 0) || (row == 0)) {
 							if ((int)lVar12 == 0) {
-								local_1d8.m_data[iVar15] = local_1f8.m_data[local_284] + local_1d8.m_data[iVar15 - local_1d8.m_stride];
+								local_1d8.m_data[row * local_1d8.m_stride] = local_1f8.m_data[row * local_1f8.m_stride] + local_1d8.m_data[(row - 1) * local_1d8.m_stride];
 							}
 						} else {
 							auto fVar16 = pfVar9[1];
@@ -246,40 +234,37 @@ namespace snowboy {
 							if (*pfVar9 <= fVar16) {
 								fVar16 = *pfVar9;
 							}
-							pfVar8[1] = fVar16 + local_1f8.m_data[iVar14 + lVar12];
+							pfVar8[1] = fVar16 + local_1f8.m_data[row * local_1f8.m_stride + lVar12];
 						}
-						lVar12 += 1;
-						pfVar9 = pfVar9 + 1;
-						pfVar8 = pfVar8 + 1;
-					} while ((int)lVar12 < local_1d8.m_cols);
+					}
+					pfVar9 = pfVar9 + 1;
+					pfVar8 = pfVar8 + 1;
 				}
 			}
 		LAB_00186e9d:
 			[]() {}(); // TODO: This is just here cause for some reason a label directly before the closing bracket does not work
 		}
-		auto local_228 = -1;
-		int iVar11 = local_1d8.m_rows - 1;
-		SubVector{local_1d8, iVar11}.Min(&local_228);
-		auto fVar16 = local_1d8.m_data[local_1d8.m_stride * iVar11 + local_228];
+		auto min_index = -1;
+		auto min_value = SubVector{local_1d8, local_1d8.m_rows - 1}.Min(&min_index);
 		if (param_4 != nullptr) {
-			while (iVar11 != 0) {
+			for (int iVar11 = local_1d8.m_rows - 1; iVar11 != 0;) {
 				// TODO: This is wrong
 				// If I look at the code it should only be
-				// param_4->at(iVar11).push_back(local_228);
+				// param_4->at(iVar11).push_back(min_index);
 				// But that produces different results from what it should
 				if (param_4->at(iVar11).empty())
-					param_4->at(iVar11).push_back(local_228);
+					param_4->at(iVar11).push_back(min_index);
 				else
-					param_4->at(iVar11).at(0) = local_228;
-				if (0 >= local_228) {
+					param_4->at(iVar11).at(0) = min_index;
+				if (0 >= min_index) {
 					iVar11--;
 					continue;
 				}
-				auto fVar18 = local_1d8.m_data[local_1d8.m_stride * iVar11 + local_228] - local_1f8.m_data[local_1f8.m_stride * iVar11 + local_228];
-				float pfVar8[3] = {fVar18, fVar18, fVar18};
-				pfVar8[0] = std::abs(fVar18 - local_1d8.m_data[(iVar11 + -1) * local_1d8.m_stride + (local_228 - 1)]);
-				pfVar8[1] = std::abs(fVar18 - local_1d8.m_data[(local_228 - 1) + iVar11 * local_1d8.m_stride]);
-				pfVar8[2] = std::abs(fVar18 - local_1d8.m_data[(iVar11 + -1) * local_1d8.m_stride + local_228]);
+				auto fVar18 = local_1d8.m_data[local_1d8.m_stride * iVar11 + min_index] - local_1f8.m_data[local_1f8.m_stride * iVar11 + min_index];
+				float pfVar8[3];
+				pfVar8[0] = std::abs(fVar18 - local_1d8.m_data[(iVar11 + -1) * local_1d8.m_stride + (min_index - 1)]);
+				pfVar8[1] = std::abs(fVar18 - local_1d8.m_data[(min_index - 1) + iVar11 * local_1d8.m_stride]);
+				pfVar8[2] = std::abs(fVar18 - local_1d8.m_data[(iVar11 + -1) * local_1d8.m_stride + min_index]);
 				auto pfVar9 = pfVar8 + 1;
 				if (pfVar8[0] <= pfVar8[1]) {
 					pfVar9 = pfVar8;
@@ -290,26 +275,26 @@ namespace snowboy {
 				auto iVar10 = (int)((long)((long)pfVar9 - (long)pfVar8) >> 2);
 				if (iVar10 != 0) {
 					if (iVar10 == 1) {
-						local_228 -= 1;
+						min_index -= 1;
 					} else {
 						if (iVar10 == 2) {
-							iVar11 = iVar11 + -1;
+							iVar11--;
 						}
 					}
 				} else {
-					local_228 -= 1;
-					iVar11 = iVar11 + -1;
+					min_index -= 1;
+					iVar11--;
 				}
 			}
 			// TODO: This is wrong
 			// If I look at the code it should only be
-			// param_4->at(0).push_back(local_228);
+			// param_4->at(0).push_back(min_index);
 			// But that produces different results from what it should
 			if (param_4->at(0).empty())
-				param_4->at(0).push_back(local_228);
+				param_4->at(0).push_back(min_index);
 			else
-				param_4->at(0).at(0) = local_228;
+				param_4->at(0).at(0) = min_index;
 		}
-		return fVar16 / param_2.m_rows;
+		return min_value / param_2.m_rows;
 	}
 } // namespace snowboy

lib/dtw-lib.h

@@ -16,7 +16,6 @@ namespace snowboy {
 		// TODO: This could be replaced with enum DistanceType
 		std::string distance_metric;
 	};
-	static_assert(sizeof(SlidingDtwOptions) == 0x10);
 	struct SlidingDtw {
 		SlidingDtwOptions m_options;
 		std::deque<std::deque<float>> field_x18;
@@ -38,7 +37,6 @@ namespace snowboy {
 		void ComputeBandBoundary(int, int*, int*) const;
 		virtual ~SlidingDtw();
 	};
-	static_assert(sizeof(SlidingDtw) == 0x78);
 
 	float DtwAlign(DistanceType, const MatrixBase&, const MatrixBase&, std::vector<std::vector<int>>*);
 } // namespace snowboy

lib/eavesdrop-stream.h

@@ -14,5 +14,4 @@ namespace snowboy {
 		virtual std::string Name() const override;
 		virtual ~EavesdropStream();
 	};
-	static_assert(sizeof(EavesdropStream) == 0x28);
 } // namespace snowboy