Add Mac OS ARM support

.github/workflows/build-cld2.yml

@@ -0,0 +1,83 @@
+name: Build CLD2 Native Libraries
+
+on:
+  push:
+
+jobs:
+  build:
+    strategy:
+      matrix:
+        include:
+          - os: ubuntu-24.04
+            os_name: linux
+            arch: x86-64
+            artifact_suffix: linux-x86-64
+          - os: ubuntu-24.04-arm
+            os_name: linux
+            arch: aarch64
+            artifact_suffix: linux-aarch64
+          - os: macos-15-intel
+            os_name: mac
+            arch: x86-64
+            artifact_suffix: darwin-x86-64
+          - os: macos-latest
+            os_name: mac
+            arch: aarch64
+            artifact_suffix: darwin-aarch64
+
+    runs-on: ${{ matrix.os }}
+
+    steps:
+      - name: Checkout this project
+        uses: actions/checkout@v4
+
+      - name: Build libcld2
+        working-directory: internal
+        run: |
+          chmod +x compile_libs.sh
+          ./compile_libs.sh
+
+      - name: Create output directory
+        run: |
+          mkdir -p output/${{ matrix.artifact_suffix }}
+
+      - name: Copy library to output
+        run: |
+          cp internal/libcld2.${{ matrix.os_name == 'mac' && 'dylib' || 'so' }} output/${{ matrix.artifact_suffix }}/ 
+          cp internal/libcld2_full.${{ matrix.os_name == 'mac' && 'dylib' || 'so' }} output/${{ matrix.artifact_suffix }}/
+
+      - name: Save artifact
+        uses: actions/upload-artifact@v4
+        with:
+          name: cld2-bin-${{ matrix.artifact_suffix }}
+          path: output
+
+  assemble:
+    needs: build
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - name: Download all artifacts
+        uses: actions/download-artifact@v4
+        with:
+          path: output
+          pattern: cld2-bin-*
+          merge-multiple: true
+
+      - name: Display structure
+        run: ls -R
+
+      - name: Get short commit ID
+        id: extract_rev_id
+        run: |
+          SHORT_COMMIT_ID=$(echo "${{ github.sha }}" | cut -c1-7)
+          echo "short_commit_id=${SHORT_COMMIT_ID}" >> $GITHUB_OUTPUT
+
+      - name: Save consolidated artifact
+        uses: actions/upload-artifact@v4
+        with:
+          name: cld2-bin-all-${{ steps.extract_rev_id.outputs.short_commit_id }}
+          path: output

CMakeLists.txt

@@ -0,0 +1,155 @@
+cmake_minimum_required(VERSION 3.10)
+project (cld2)
+enable_language(CXX)
+
+set (VERSION "0.0.197")
+set (common_SOURCE_FILES
+    internal/cldutil.cc
+    internal/cldutil_shared.cc
+    internal/compact_lang_det.cc
+    internal/compact_lang_det_hint_code.cc
+    internal/compact_lang_det_impl.cc
+    internal/debug.cc
+    internal/fixunicodevalue.cc 
+    internal/generated_entities.cc
+    internal/generated_language.cc
+    internal/generated_ulscript.cc
+    internal/getonescriptspan.cc
+    internal/lang_script.cc
+    internal/offsetmap.cc
+    internal/scoreonescriptspan.cc
+    internal/tote.cc
+    internal/utf8statetable.cc
+    )
+
+set (cld2_SOURCE_FILES
+    internal/generated_distinct_bi_0.cc
+    internal/cld_generated_cjk_uni_prop_80.cc
+    internal/cld2_generated_cjk_compatible.cc
+    internal/cld_generated_cjk_delta_bi_4.cc
+    internal/cld2_generated_quadchrome_2.cc
+    internal/cld2_generated_deltaoctachrome.cc
+    internal/cld2_generated_distinctoctachrome.cc
+    internal/cld_generated_score_quad_octa_2.cc
+    )
+
+set (cld2_full_SOURCE_FILES
+    internal/generated_distinct_bi_0.cc
+    internal/cld_generated_cjk_uni_prop_80.cc
+    internal/cld2_generated_cjk_compatible.cc
+    internal/cld_generated_cjk_delta_bi_32.cc
+    internal/cld2_generated_quad0122.cc
+    internal/cld2_generated_deltaocta0122.cc
+    internal/cld2_generated_distinctocta0122.cc
+    internal/cld_generated_score_quad_octa_0122.cc
+    )
+
+set (cld2_dynamic_SOURCE_FILES
+    internal/cld2_dynamic_data.cc
+    internal/cld2_dynamic_data_loader.cc
+    )
+
+add_library(cld2 SHARED ${common_SOURCE_FILES} ${cld2_SOURCE_FILES})
+set_target_properties(cld2 PROPERTIES
+  ENABLE_EXPORTS On
+  OUTPUT_NAME cld2
+  VERSION ${VERSION}
+  SOVERSION 0
+  )
+add_library(cld2_full SHARED ${cld2_full_SOURCE_FILES})
+set_target_properties(cld2_full PROPERTIES
+  ENABLE_EXPORTS On
+  OUTPUT_NAME cld2_full
+  VERSION ${VERSION}
+  SOVERSION 0
+  )
+
+add_library(cld2_dynamic SHARED ${cld2_dynamic_SOURCE_FILES})
+set_target_properties(cld2_dynamic PROPERTIES
+  ENABLE_EXPORTS On
+  OUTPUT_NAME cld2_dynamic
+  VERSION ${VERSION}
+  SOVERSION 0
+  COMPILE_FLAGS "-DCLD2_DYNAMIC_MODE"
+  )
+install(TARGETS cld2 DESTINATION lib/${CMAKE_LIBRARY_ARCHITECTURE})
+install(TARGETS cld2_full DESTINATION lib/${CMAKE_LIBRARY_ARCHITECTURE})
+install(TARGETS cld2_dynamic DESTINATION lib/${CMAKE_LIBRARY_ARCHITECTURE})
+
+set (cld2_internal_HEADERS
+    internal/cld2_dynamic_compat.h
+    internal/cld2_dynamic_data_extractor.h
+    internal/cld2_dynamic_data.h
+    internal/cld2_dynamic_data_loader.h
+    internal/cld2tablesummary.h
+    internal/cldutil.h
+    internal/cldutil_offline.h
+    internal/cldutil_shared.h
+    internal/compact_lang_det_hint_code.h
+    internal/compact_lang_det_impl.h
+    internal/debug.h
+    internal/fixunicodevalue.h
+    internal/generated_language.h
+    internal/generated_ulscript.h
+    internal/getonescriptspan.h
+    internal/integral_types.h
+    internal/lang_script.h
+    internal/langspan.h
+    internal/offsetmap.h
+    internal/port.h
+    internal/scoreonescriptspan.h
+    internal/stringpiece.h
+    internal/tote.h
+    internal/unittest_data.h
+    internal/utf8acceptinterchange.h
+    internal/utf8prop_lettermarkscriptnum.h
+    internal/utf8repl_lettermarklower.h
+    internal/utf8scannot_lettermarkspecial.h
+    internal/utf8statetable.h
+    )
+
+install(FILES ${cld2_internal_HEADERS} DESTINATION include/cld2/internal)
+set (cld2_public_HEADERS
+    public/compact_lang_det.h
+    public/encodings.h
+    )
+install(FILES ${cld2_public_HEADERS} DESTINATION include/cld2/public)
+
+set (full_SOURCE_FILES
+    internal/cld_generated_cjk_uni_prop_80.cc
+    internal/cld2_generated_cjk_compatible.cc
+    internal/cld_generated_cjk_delta_bi_32.cc
+    internal/generated_distinct_bi_0.cc
+    internal/cld2_generated_quad0122.cc
+    internal/cld2_generated_deltaocta0122.cc
+    internal/cld2_generated_distinctocta0122.cc
+    internal/cld_generated_score_quad_octa_0122.cc
+    )
+
+add_executable(compact_lang_det_test_full ${full_SOURCE_FILES} internal/compact_lang_det_test.cc)
+add_executable(cld2_unittest_full         ${full_SOURCE_FILES} internal/cld2_unittest_full.cc)
+add_executable(cld2_unittest_full_avoid   ${full_SOURCE_FILES} internal/cld2_unittest_full.cc)
+set_target_properties(cld2_unittest_full_avoid PROPERTIES COMPILE_FLAGS "-Davoid_utf8_string_constants")
+
+add_executable(cld2_dynamic_data_tool               internal/cld2_dynamic_data_extractor.cc internal/cld2_dynamic_data_tool.cc)
+add_executable(compact_lang_det_dynamic_test_chrome ${common_SOURCE_FILES} internal/cld2_dynamic_data_extractor.cc internal/compact_lang_det_test.cc)
+add_executable(cld2_dynamic_unittest                ${common_SOURCE_FILES} internal/cld2_unittest.cc)
+set_target_properties(compact_lang_det_dynamic_test_chrome PROPERTIES COMPILE_FLAGS "-DCLD2_DYNAMIC_MODE")
+set_target_properties(cld2_dynamic_unittest PROPERTIES COMPILE_FLAGS "-DCLD2_DYNAMIC_MODE")
+
+add_executable(compact_lang_det_test_chrome_2  internal/compact_lang_det_test.cc)
+add_executable(compact_lang_det_test_chrome_16 internal/compact_lang_det_test.cc)
+add_executable(cld2_unittest_chrome_2          internal/cld2_unittest.cc)
+add_executable(cld2_unittest_avoid_chrome_2    internal/cld2_unittest.cc)
+set_target_properties(cld2_unittest_avoid_chrome_2 PROPERTIES COMPILE_FLAGS "-Davoid_utf8_string_constants")
+
+target_link_libraries(compact_lang_det_test_full cld2)
+target_link_libraries(cld2_unittest_full cld2)
+target_link_libraries(cld2_unittest_full_avoid cld2)
+target_link_libraries(cld2_dynamic_data_tool cld2 cld2_dynamic)
+target_link_libraries(compact_lang_det_dynamic_test_chrome cld2_dynamic)
+target_link_libraries(cld2_dynamic_unittest cld2_dynamic)
+target_link_libraries(compact_lang_det_test_chrome_2 cld2)
+target_link_libraries(compact_lang_det_test_chrome_16 cld2)
+target_link_libraries(cld2_unittest_chrome_2 cld2)
+target_link_libraries(cld2_unittest_avoid_chrome_2 cld2)