Change in uppercase of ß

Author: ScottPJones

Project.toml

@@ -4,13 +4,15 @@ keywords = ["Characters"]
 license = "MIT"
 desc = "Basic functionality for Chr type"
 authors = ["ScottPJones <scottjones@alum.mit.edu>"]
-version = "0.1.5"
+version = "0.1.6"
 
 [deps]
-CharSetEncodings = "cb9422de-a9d8-5b68-86db-ff05833ab307"
-ModuleInterfaceTools = "5cb8414e-7aab-5a03-a681-351269c074bf"
 Pkg = "44cfe95a-1eb2-52ea-b672-e2afdf69b78f"
-Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
-StrAPI = "69e7dfc3-c4d0-5e14-8d95-d6042a05b383"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
+Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 Unicode = "4ec0a83e-493e-50e2-b9ac-8f72acf5a8f5"
+
+ModuleInterfaceTools = "5cb8414e-7aab-5a03-a681-351269c074bf"
+
+StrAPI = "69e7dfc3-c4d0-5e14-8d95-d6042a05b383"
+CharSetEncodings = "cb9422de-a9d8-5b68-86db-ff05833ab307"

src/ChrBase.jl

@@ -23,8 +23,7 @@ using ModuleInterfaceTools
              write_utf8, write_utf16, _write_utf8_2, _write_utf8_3, _write_utf8_4, _write_ucs2,
              _lowercase_l, _uppercase_l, _lowercase_u, _uppercase_u, _titlecase_u,
              _islower_a, _islower_u, _isupper_a, _isupper_l, _isupper_al, _isupper_u,
-             _can_upper_latin, _can_upper_only_latin, _can_upper_ch, _can_lower_ch,
-             _can_upper, _can_upper_l
+             _can_upper_ch, _can_lower_ch, _can_upper, _can_upper_l
 
 @api develop! _isvalid_chr
 

src/casefold.jl

@@ -28,15 +28,21 @@ uppercase(ch::LatinChr) = LatinChr(_uppercase_l(codepoint(ch)))
 function uppercase(ch::_LatinChr)
     cb = codepoint(ch)
     _can_upper(cb) && return _LatinChr(cb - 0x20)
-    # We don't uppercase 0xdf, the ß character
-    cb == 0xb5 ? UCS2Chr(0x39c) : (cb == 0xff ? UCS2Chr(0x178) : ch)
+    # We didn't used to uppercase 0xdf, the ß character, now we do
+    !V6_COMPAT && cb == 0xdf && return UCS2Chr(0x1e9e)
+    cb == 0xb5 ? UCS2Chr(0x39c) : cb == 0xff ? UCS2Chr(0x178) : ch
 end
 titlecase(ch::LatinChars) = uppercase(ch)
 
-_can_upper_latin(ch)      = _can_upper(ch) | (ch == 0xb5) | (ch == 0xff)
-_can_upper_only_latin(ch) = _can_upper_l(ch) | (ch == 0xb5) | (ch == 0xff)
+@static if V6_COMPAT
+    @inline _can_upper_ch(ch) =
+        (ch <= 0x7f
+         ? _islower_a(ch)
+         : (ch > 0xff ? _islower_u(ch) : ifelse(c > 0xdf, c != 0xf7, c == 0xb5)))
+else
+    @inline _can_upper_ch(ch) =
+        ch <= 0x7f ? _islower_a(ch) : (ch <= 0xff ? _is_lower_l(ch) : _islower_u(ch))
+end
 
-@inline _can_upper_ch(ch) =
-    ch <= 0x7f ? _islower_a(ch) : (ch <= 0xff ? _can_upper_only_latin(ch) : _islower_u(ch))
 @inline _can_lower_ch(ch) =
     ch <= 0x7f ? _isupper_a(ch) : (ch <= 0xff ? _isupper_l(ch) : _isupper_u(ch))