Disable CPU extensions (SSE, AVX)

Author: ollydev

.github/workflows/build.yml

@@ -18,7 +18,6 @@ jobs:
             output: 'libtesseract32.dll'
             cmake_args: '-G "Visual Studio 16 2019" -A Win32'
             sw_args: '-platform x86 -win-mt'
-            platform: 'x32'
 
           - os: windows-latest
             name: 'Windows 64'
@@ -40,37 +39,43 @@ jobs:
 
     steps:
       - uses: actions/checkout@v2
+        with:  
+          submodules: true
+      
       - uses: egorpugin/sw-action@master
-
+      
       - name: Cache
         id: cache
         uses: actions/[email protected]
         with:
           path: storage
-          key: ${{ matrix.config.name }}${{ matrix.config.sw_args }}
+          key: ${{ matrix.config.os }}${{ matrix.config.sw_args }}
 
       - name: Build Tesseract
         if: steps.cache.outputs.cache-hit != 'true'
         run: |
-          ./sw build ${{ matrix.config.sw_args }} -storage-dir=storage -static -static-dependencies -config r org.sw.demo.google.tesseract.tesseract-master
-
+          mv -f sw.cpp tesseract/sw.cpp
+          ./sw ${{ matrix.config.sw_args }} -d tesseract -storage-dir ../storage -static -static-dependencies -config r build
+          mv tesseract storage/pkg
+          
       - name: Extract Headers & Libraries
+        continue-on-error: true
         run: |
           mkdir -p libs
+          mkdir -p headers/tesseract
+          mkdir -p headers/leptonica
+          
           find "storage/pkg/" -name "*.a" -exec cp "{}" "libs" \;
           find "storage/pkg/" -name "*.lib" -exec cp "{}" "libs" \;
           
-          mkdir -p leptonica
-          leptonica=$(find "storage/pkg/" -name "allheaders.h" -exec dirname {} \; -quit)
-          find "$leptonica" -name "*.h*" -exec cp "{}" leptonica/ \;
-          find "storage/pkg/" -name "*endianness.h*" -exec cp "{}" leptonica/ \;
+          find "storage/pkg/tesseract/include" -name "*.h" -exec cp "{}" headers/tesseract/ \;
+          find "storage/pkg/tesseract" -name "version.h" -exec cp "{}" headers/tesseract/ \; -quit
           
-          mkdir -p tesseract
-          tesseract=$(find "storage/pkg/" -name "baseapi.h" -exec dirname {} \; -quit)
-          find "$tesseract" -name "*.h*" -exec cp "{}" tesseract/ \;
-          find "storage/pkg/" -wholename "*tesseract/version.h" -exec cp "{}" tesseract/ \;
+          leptonica=$(find "storage/pkg/" -name "*leptonica*" -exec dirname {} \; -quit)/../../../../
+          find "$leptonica" -name "*.h" -exec cp "{}" headers/leptonica/ \;
 
       - name: Build Plugin
+        continue-on-error: true
         run: |
           cmake -S . ${{ matrix.config.cmake_args }}
           cmake --build .  --config Release

.gitmodules

@@ -0,0 +1,3 @@
+[submodule "tesseract"]
+	path = tesseract
+	url = https://github.com/tesseract-ocr/tesseract

CMakeLists.txt

@@ -1,15 +1,15 @@
 cmake_minimum_required(VERSION 3.15)
 project(Tesseract DESCRIPTION "tesseract")
 
-SET(CMAKE_BUILD_TYPE Release)
 SET(CMAKE_CXX_STANDARD 20)
 SET(CMAKE_OSX_DEPLOYMENT_TARGET "10.13" CACHE STRING "Minimum OS X deployment version" FORCE)
 
 # ----------------------------- LINKER -----------------------------
 
-file(GLOB TESSERACT_LIBRARIES
-    "libs/*.a"
-    "libs/*.lib"
+file(
+	GLOB TESSERACT_LIBRARIES
+        "libs/*.a"
+        "libs/*.lib"
     )
 
 IF(WIN32)
@@ -31,7 +31,7 @@ set(SOURCE_LIST
    )
 
 set(INCLUDE_DIRECTORIES
-        .
+        headers
    )
 
 # ---------------------------- COMPILE ----------------------------

sw.cpp

@@ -0,0 +1,101 @@
+void build(Solution &s)
+{
+    auto &tess = s.addProject("google.tesseract", "master");
+    tess += Git("https://github.com/tesseract-ocr/tesseract", "", "{v}");
+
+    auto &libtesseract = tess.addTarget<LibraryTarget>("libtesseract");
+    {
+        libtesseract.setChecks("libtesseract");
+
+        libtesseract.ExportAllSymbols = true;
+        libtesseract.PackageDefinitions = true;
+
+        libtesseract += cpp14;
+
+        libtesseract += "include/.*"_rr;
+        libtesseract += "src/.*"_rr;
+        libtesseract -= "src/lstm/.*\\.cc"_rr;
+        libtesseract -= "src/training/.*"_rr;
+
+        libtesseract -=
+            "src/api/tesseractmain.cpp",
+            "src/viewer/svpaint.cpp";
+
+        libtesseract.Public += "include"_idir;
+        libtesseract.Protected +=
+            "src/opencl"_id,
+            "src/ccmain"_id,
+            "src/api"_id,
+            "src/dict"_id,
+            "src/viewer"_id,
+            "src/wordrec"_id,
+            "src/ccstruct"_id,
+            "src/cutil"_id,
+            "src/textord"_id,
+            "src/ccutil"_id,
+            "src/lstm"_id,
+            "src/classify"_id,
+            "src/arch"_id,
+            "src/training"_id;
+
+        auto win_or_mingw =
+            libtesseract.getBuildSettings().TargetOS.Type == OSType::Windows ||
+            libtesseract.getBuildSettings().TargetOS.Type == OSType::Mingw
+            ;
+
+        libtesseract -= "src/arch/dotproductfma.cpp";
+        libtesseract -= "src/arch/dotproductavx.cpp";
+        libtesseract -= "src/arch/dotproductsse.cpp";
+        libtesseract -= "src/arch/intsimdmatrixsse.cpp";
+        libtesseract -= "src/arch/intsimdmatrixavx2.cpp";
+		
+        if (!win_or_mingw)
+            libtesseract += "pthread"_slib;
+
+        libtesseract.Public += "HAVE_CONFIG_H"_d;
+        libtesseract.Public += "_SILENCE_STDEXT_HASH_DEPRECATION_WARNINGS=1"_d;
+        libtesseract.Public += "HAVE_LIBARCHIVE"_d;
+        libtesseract.Interface += sw::Shared, "TESS_IMPORTS"_d;
+        libtesseract.Private += sw::Shared, "TESS_EXPORTS"_d;
+
+        libtesseract.Public += "org.sw.demo.danbloomberg.leptonica"_dep;
+        libtesseract.Public += "org.sw.demo.libarchive.libarchive"_dep;
+
+        if (win_or_mingw)
+        {
+            libtesseract.Public += "ws2_32.lib"_slib;
+            libtesseract.Protected += "NOMINMAX"_def;
+        }
+
+        libtesseract.Variables["TESSERACT_MAJOR_VERSION"] = libtesseract.Variables["PACKAGE_MAJOR_VERSION"];
+        libtesseract.Variables["TESSERACT_MINOR_VERSION"] = libtesseract.Variables["PACKAGE_MINOR_VERSION"];
+        libtesseract.Variables["TESSERACT_MICRO_VERSION"] = libtesseract.Variables["PACKAGE_PATCH_VERSION"];
+        libtesseract.Variables["TESSERACT_VERSION_STR"] = "master";
+        libtesseract.configureFile("include/tesseract/version.h.in", "tesseract/version.h");
+    }
+}
+
+void check(Checker &c)
+{
+    auto &s = c.addSet("libtesseract");
+    s.checkFunctionExists("getline");
+    s.checkIncludeExists("dlfcn.h");
+    s.checkIncludeExists("inttypes.h");
+    s.checkIncludeExists("memory.h");
+    s.checkIncludeExists("stdint.h");
+    s.checkIncludeExists("stdlib.h");
+    s.checkIncludeExists("string.h");
+    s.checkIncludeExists("sys/stat.h");
+    s.checkIncludeExists("sys/types.h");
+    s.checkIncludeExists("tiffio.h");
+    s.checkIncludeExists("unistd.h");
+    s.checkTypeSize("long long int");
+    s.checkTypeSize("size_t");
+    s.checkTypeSize("void *");
+    s.checkTypeSize("wchar_t");
+    {
+        auto &c = s.checkSymbolExists("snprintf");
+        c.Parameters.Includes.push_back("stdio.h");
+    }
+}
+