Merge branch 'master' of github.com:simdutf/simdutf

simdutf · Oct 24, 2023 · b82a978 · b82a978
2 parents ee29bbd + af4858e
commit b82a978
Show file tree

Hide file tree

Showing 7 changed files with 12 additions and 19 deletions.
diff --git a/.github/workflows/vs16-ci.yml b/.github/workflows/vs16-ci.yml
@@ -31,7 +31,7 @@ jobs:
     - name: Run Debug tests
       run: |
         cd build
-        ctest -C Debug  --output-on-failure
+        ctest -C Debug  --output-on-failure -E basic_fuzzer
     - name: Install
       run: |
         cmake --install build --config Release

diff --git a/.github/workflows/vs17-ci-cxx20.yml b/.github/workflows/vs17-ci-cxx20.yml
@@ -31,7 +31,7 @@ jobs:
     - name: Run Debug tests
       run: |
         cd build
-        ctest -C Debug  --output-on-failure
+        ctest -C Debug  --output-on-failure  -E basic_fuzzer
     - name: Install
       run: |
         cmake --install build --config Release

diff --git a/.github/workflows/vs17-ci.yml b/.github/workflows/vs17-ci.yml
@@ -31,7 +31,7 @@ jobs:
     - name: Run Debug tests
       run: |
         cd build
-        ctest -C Debug  --output-on-failure
+        ctest -C Debug  --output-on-failure  -E basic_fuzzer
     - name: Install
       run: |
         cmake --install build --config Release

diff --git a/.github/workflows/vs17-clang-ci.yml b/.github/workflows/vs17-clang-ci.yml
@@ -31,7 +31,7 @@ jobs:
     - name: Run Debug tests
       run: |
         cd build
-        ctest -C Debug  --output-on-failure
+        ctest -C Debug  --output-on-failure  -E basic_fuzzer
     - name: Install
       run: |
         cmake --install build --config Release

diff --git a/src/simdutf/haswell/simd.h b/src/simdutf/haswell/simd.h
@@ -348,9 +348,7 @@ namespace simd {
 
       return  simd8x64<bool>(
         (this->chunks[0] <= mask_high) & (this->chunks[0] >= mask_low),
-        (this->chunks[1] <= mask_high) & (this->chunks[1] >= mask_low),
-        (this->chunks[2] <= mask_high) & (this->chunks[2] >= mask_low),
-        (this->chunks[3] <= mask_high) & (this->chunks[3] >= mask_low)
+        (this->chunks[1] <= mask_high) & (this->chunks[1] >= mask_low)
       ).to_bitmask();
     }
     simdutf_really_inline uint64_t not_in_range(const T low, const T high) const {

diff --git a/src/simdutf/haswell/simd16-inl.h b/src/simdutf/haswell/simd16-inl.h
@@ -243,9 +243,7 @@ struct simd16<uint16_t>: base16_numeric<uint16_t>  {
 
       return  simd16x32<bool>(
         (this->chunks[0] <= mask_high) & (this->chunks[0] >= mask_low),
-        (this->chunks[1] <= mask_high) & (this->chunks[1] >= mask_low),
-        (this->chunks[2] <= mask_high) & (this->chunks[2] >= mask_low),
-        (this->chunks[3] <= mask_high) & (this->chunks[3] >= mask_low)
+        (this->chunks[1] <= mask_high) & (this->chunks[1] >= mask_low)
       ).to_bitmask();
     }
     simdutf_really_inline uint64_t not_in_range(const T low, const T high) const {

diff --git a/tests/bele_tests.cpp b/tests/bele_tests.cpp
@@ -5,19 +5,16 @@
 #include <tests/helpers/test.h>
 
 // We use explicit arrays so that no funny business is possible.
-
-//
-// s = "@\u00A7\u2208\U0001D4AA"
 const unsigned char utf8_string[] = {0x40,0xc2,0xa7,0xe2,0x88,0x88,0xf0,0x9d,0x92,0xaa};
 const char *utf8 = reinterpret_cast<const char*>(utf8_string);
 const size_t utf8_size = sizeof(utf8_string)/sizeof(uint8_t);
 
-const unsigned char utf16le_string[] = {0x40,0x00,0xa7,0x00,0x08,0x22,0x35,0xd8,0xaa,0xdc};
+alignas(char16_t) const unsigned char utf16le_string[] = {0x40,0x00,0xa7,0x00,0x08,0x22,0x35,0xd8,0xaa,0xdc};
 const char16_t *utf16le = reinterpret_cast<const char16_t*>(utf16le_string);
 const size_t utf16_size = sizeof(utf16le_string)/sizeof(uint16_t);
 
 
-const unsigned char utf16be_string[] = {0x00,0x40,0x00,0xa7,0x22,0x08,0xd8,0x35,0xdc,0xaa};
+alignas(char16_t) const unsigned char utf16be_string[] = {0x00,0x40,0x00,0xa7,0x22,0x08,0xd8,0x35,0xdc,0xaa};
 const char16_t *utf16be = reinterpret_cast<const char16_t*>(utf16be_string);
 #if SIMDUTF_IS_BIG_ENDIAN
 const char16_t *utf16 = utf16be;
@@ -27,11 +24,11 @@ const char16_t *utf16 = utf16le;
 
 // Native order
 #if SIMDUTF_IS_BIG_ENDIAN
-const unsigned char utf32_string[] = {0x00,0x00,0x00,0x40,0x00,0x00,0x00,0xa7,0x00,0x00,0x22,0x08,0x00,0x01,0xd4,0xaa};
-const char32_t *utf32 = reinterpret_cast<const char32_t*>(utf32_string); // Technically undefined behavior.
+alignas(char32_t) const unsigned char utf32_string[] = {0x00,0x00,0x00,0x40,0x00,0x00,0x00,0xa7,0x00,0x00,0x22,0x08,0x00,0x01,0xd4,0xaa};
+const char32_t *utf32 = reinterpret_cast<const char32_t*>(utf32_string);
 #else
-const unsigned char utf32_string[] = {0x40,0x00,0x00,0x00,0xa7,0x00,0x00,0x00,0x08,0x22,0x00,0x00,0xaa,0xd4,0x01,0x00};
-const char32_t *utf32 = reinterpret_cast<const char32_t*>(utf32_string); // Technically undefined behavior.
+alignas(char32_t) const unsigned char utf32_string[] = {0x40,0x00,0x00,0x00,0xa7,0x00,0x00,0x00,0x08,0x22,0x00,0x00,0xaa,0xd4,0x01,0x00};
+const char32_t *utf32 = reinterpret_cast<const char32_t*>(utf32_string);
 #endif
 const size_t utf32_size = sizeof(utf32_string)/sizeof(char32_t);
 const size_t number_of_code_points = utf32_size;