commit 3ac191bd5558758c3ca13f0fca245e241f1a899b
author René Kijewski <rene.kijewski@fu-berlin.de> Tue Apr 16 00:06:53 2024 +0200
committer René Kijewski <rene.kijewski@fu-berlin.de> Tue Apr 16 00:06:53 2024 +0200
tree 6f38fd56a4431cd35556fe9f8ef08382d43783a8
parent b49ac3b3d610d80ea7702d0ae2b4e8baae5dbcd7

Optimize ASCII case

scripts/make_unicode_categories.py

diff --git a/scripts/make_unicode_categories.py b/scripts/make_unicode_categories.py
index 3ef477e..51d89d9 100755
--- a/scripts/make_unicode_categories.py
+++ b/scripts/make_unicode_categories.py
@@ -137,8 +137,11 @@
     print("    };", file=output_file)
     print(file=output_file)
 
-    print("    if (codepoint > 0x110000) codepoint = 0x110000;", file=output_file)
+    print("    if (JSON5EncoderCpp_expect(codepoint < 256, true)) {", file=output_file)
+    print("        return (demiplane_data[0][codepoint / 4] >> (2 * (codepoint % 4))) % 4;", file=output_file)
+    print("    }", file=output_file)
     print(file=output_file)
+    print("    if (codepoint > 0x110000) codepoint = 0x110000;", file=output_file)
     print("    std::uint32_t fst_row = codepoint / 0x100;", file=output_file)
     print("    std::uint32_t fst_col = codepoint % 0x100;", file=output_file)
     print("    std::uint32_t snd_row = fst_row / 64;", file=output_file)

src/_unicode_cat_of.hpp

diff --git a/src/_unicode_cat_of.hpp b/src/_unicode_cat_of.hpp
index ae3becc..3eae7dc 100644
--- a/src/_unicode_cat_of.hpp
+++ b/src/_unicode_cat_of.hpp
@@ -817,8 +817,11 @@
         0x0cu, 0x0cu, 0x0cu, 0x0cu, 0x0eu, 0x0cu, 0x0cu, 0x0cu,
     };
 
-    if (codepoint > 0x110000) codepoint = 0x110000;
+    if (JSON5EncoderCpp_expect(codepoint < 256, true)) {
+        return (demiplane_data[0][codepoint / 4] >> (2 * (codepoint % 4))) % 4;
+    }
 
+    if (codepoint > 0x110000) codepoint = 0x110000;
     std::uint32_t fst_row = codepoint / 0x100;
     std::uint32_t fst_col = codepoint % 0x100;
     std::uint32_t snd_row = fst_row / 64;