Merge pull request #33 from chrisduerr/unicode_15_1

Manishearth · web-flow · commit 24651a204dcb · 2023-09-19T10:54:32.000-06:00
Update to Unicode 15.1
diff --git a/scripts/unicode.py b/scripts/unicode.py
@@ -106,9 +106,9 @@ def load_east_asian_widths() -> "list[EffectiveWidth]":
     `Ambiguous` chracters are assigned `EffectiveWidth.AMBIGUOUS`."""
     with fetch_open("EastAsianWidth.txt") as eaw:
         # matches a width assignment for a single codepoint, i.e. "1F336;N  # ..."
-        single = re.compile(r"^([0-9A-F]+);(\w+) +# (\w+)")
+        single = re.compile(r"^([0-9A-F]+)\s+;\s+(\w+) +# (\w+)")
         # matches a width assignment for a range of codepoints, i.e. "3001..3003;W  # ..."
-        multiple = re.compile(r"^([0-9A-F]+)\.\.([0-9A-F]+);(\w+) +# (\w+)")
+        multiple = re.compile(r"^([0-9A-F]+)\.\.([0-9A-F]+)\s+;\s+(\w+) +# (\w+)")
         # map between width category code and condensed width
         width_codes = {
             **{c: EffectiveWidth.NARROW for c in ["N", "Na", "H"]},
diff --git a/src/tables.rs b/src/tables.rs
@@ -12,7 +12,7 @@
 
 /// The version of [Unicode](http://www.unicode.org/)
 /// that this version of unicode-width is based on.
-pub const UNICODE_VERSION: (u8, u8, u8) = (15, 0, 0);
+pub const UNICODE_VERSION: (u8, u8, u8) = (15, 1, 0);
 
 pub mod charwidth {
     use core::option::Option::{self, None, Some};
@@ -395,13 +395,13 @@ pub mod charwidth {
         0x00, 0x00, 0x00, 0x00, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0x9A, 0xAA, 0xAA, 0xAA, 0xAA,
         0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA,
         0xAA, 0xAA, 0xAA, 0x55, 0x55, 0x55, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0x5A, 0x55, 0x55, 0x55,
-        0x55, 0x55, 0x55, 0xAA, 0xAA, 0xAA, 0x55, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA,
+        0x55, 0x55, 0x55, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA,
         0xAA, 0xAA, 0x0A, 0xA0, 0xAA, 0xAA, 0xAA, 0x6A, 0xA9, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA,
         0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0x6A,
         0x81, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0x55, 0xA9, 0xAA, 0xAA, 0xAA,
         0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xA9, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0x6A,
         0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA,
-        0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0x55, 0x55, 0x55, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA,
+        0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0x55, 0x55, 0x95, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA,
         0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0x6A, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA,
         0xAA, 0xFF, 0xFF, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA,
         0xAA, 0xAA, 0xAA, 0x56, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA,
