fix 5 unicode 15 non-case mapping exceptions

Fixes GH #145 fixup perl cross-checks with 5.40. regen unicode tables (no changes with 15.0.0, only whitespace)
rurban · Dec 27, 2024 · fef1aa2 · fef1aa2
1 parent 8bd6568
commit fef1aa2
Show file tree

Hide file tree

Showing 14 changed files with 2,396 additions and 2,379 deletions.
diff --git a/src/extwchar/towctrans.c b/src/extwchar/towctrans.c
@@ -39,7 +39,7 @@ SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
     { (u1), (l) - (u1), (u2) - (u1) + 1 }
 #define CASELACE(u1, u2) CASEMAP((u1), (u2), (u1) + 1)
 
-/* Unicode 14.0 */
+/* Unicode 15.0 */
 
 /* must be sorted */
 static const struct {
@@ -167,7 +167,8 @@ static const unsigned short pairs[][2] = {
     /* upper - lower */
     {'I', 0x0131},
     {'S', 0x017f},
-    {0x00b5, 0x03bc},
+    {0x00b5, 0x3bc},
+    {0x0101, 0x100}, // Unicode 15
     {0x0130, 'i'},
     {0x0178, 0x00ff},
     {0x017f, 0x73},

diff --git a/src/extwchar/towfc_s.c b/src/extwchar/towfc_s.c
@@ -44,7 +44,7 @@
  *    returning the number of new wide character codepoints needed.
  *    The usual \c iswupper(wc) case returns 1, and the special 104 full
  *    folding cases as specified in Unicode 10.0 \c CaseFolding.txt return either
- *    2 or 3. This implements Unicode 14.0
+ *    2 or 3. This implements Unicode 14.0. (5 errors for Unicode 15)
  *
  * @param[in]   wc  unicode character codepoint
  *
@@ -229,7 +229,7 @@ static const struct {
 
 /* Return the number of wide lower-case characters needed to full fold-case
    the given uppercase character. Returns 0, 1, 2 or 3.
-   0 if the charcater stays the same, 1 if one character changes,
+   0 if the character stays the same, 1 if one character changes,
    2 or 3 if the character will be replaced with 2 or 3.
 
    Note that accents expand to more characters than 1 via NFD decomposition.
@@ -239,8 +239,11 @@ static const struct {
 int iswfc(const uint32_t wc) {
     /* the slow variant would walk the 2 loops */
     if (likely((wc < 0xdf) || (wc > 0x0587 && wc < 0x1e96) ||
-               (wc > 0x1FFC && wc < 0xFB00) || (wc > 0xFB17)))
+               (wc > 0x1FFC && wc < 0xFB00) || (wc > 0xFB17))) {
+        if (wc == 0x1cbb || wc == 0x1cbc)
+            return 0;
         goto single;
+    }
     if (wc < 0x1e96) {
         if (wc == 0xdf || wc == 0x130 || wc == 0x149 || wc == 0x1f0 ||
             wc == 0x587)
@@ -305,7 +308,7 @@ int iswfc(const uint32_t wc) {
    May return 2 on sizeof(wchar_t)==2 if >0xffff, i.e. converted to surrogate
    pair
 
-   perl5.27.3 -E'no warnings; for (0..0x10ffff){
+   perl -E'no warnings; for (0..0x10ffff){
      my ($lc,$fc) = (lc(pack"W",$_), fc(pack"W",$_));
      printf "U+%04X: fc: %X, lc: %X\n", $_, unpack("W",$fc), unpack("W",$lc)
        if $lc ne $fc and length($fc)==1;

diff --git a/src/extwchar/unw16ifcan.h b/src/extwchar/unw16ifcan.h
@@ -283,24 +283,24 @@ typedef struct { const uint32_t cp; const wchar_t* v; } UNWIF_canon_exc_t;
 /* sorted for binary search */
 #define UNWIF_canon_exc_size 9
 static const UNWIF_canon_exc_t UNWIF_canon_exc [9] = {
-        { 0x1d160, L"\xd834\xdd58\xd834\xdd65\xd834\xdd6e" },
-        { 0x1d161, L"\xd834\xdd58\xd834\xdd65\xd834\xdd6f" },
-        { 0x1d162, L"\xd834\xdd58\xd834\xdd65\xd834\xdd70" },
-        { 0x1d163, L"\xd834\xdd58\xd834\xdd65\xd834\xdd71" },
-        { 0x1d164, L"\xd834\xdd58\xd834\xdd65\xd834\xdd72" },
-        { 0x1d1bd, L"\xd834\xddb9\xd834\xdd65\xd834\xdd6e" },
-        { 0x1d1be, L"\xd834\xddba\xd834\xdd65\xd834\xdd6e" },
-        { 0x1d1bf, L"\xd834\xddb9\xd834\xdd65\xd834\xdd6f" },
-        { 0x1d1c0, L"\xd834\xddba\xd834\xdd65\xd834\xdd6f" }
+	{ 0x1d160, L"\xd834\xdd58\xd834\xdd65\xd834\xdd6e" },
+	{ 0x1d161, L"\xd834\xdd58\xd834\xdd65\xd834\xdd6f" },
+	{ 0x1d162, L"\xd834\xdd58\xd834\xdd65\xd834\xdd70" },
+	{ 0x1d163, L"\xd834\xdd58\xd834\xdd65\xd834\xdd71" },
+	{ 0x1d164, L"\xd834\xdd58\xd834\xdd65\xd834\xdd72" },
+	{ 0x1d1bd, L"\xd834\xddb9\xd834\xdd65\xd834\xdd6e" },
+	{ 0x1d1be, L"\xd834\xddba\xd834\xdd65\xd834\xdd6e" },
+	{ 0x1d1bf, L"\xd834\xddb9\xd834\xdd65\xd834\xdd6f" },
+	{ 0x1d1c0, L"\xd834\xddba\xd834\xdd65\xd834\xdd6f" }
 };
 
 static const wchar_t* UNWIF_canon_tbl [6] = {
-        (const wchar_t*) UNWIF_canon_tbl_1,
-        (const wchar_t*) UNWIF_canon_tbl_2,
-        (const wchar_t*) UNWIF_canon_tbl_3,
-        (const wchar_t*) UNWIF_canon_tbl_4,
-        NULL,
-        (const wchar_t*) UNWIF_canon_tbl_6
+	(const wchar_t*) UNWIF_canon_tbl_1,
+	(const wchar_t*) UNWIF_canon_tbl_2,
+	(const wchar_t*) UNWIF_canon_tbl_3,
+	(const wchar_t*) UNWIF_canon_tbl_4,
+	NULL,
+	(const wchar_t*) UNWIF_canon_tbl_6
 };
 
 /* the rows */