Check generated locale for non-ASCII 8-bit characters with case conversion.

If a locale does not have 8-bit characters with case conversion which
are different from the ASCII conversion (±0x20) then we can perform
some optimizations.  These will follow later.

diff --git a/ChangeLog b/ChangeLog
index da91742d216dfd2a1979e3a5c2af021f511a64fa..9d6b6d3d103db16d32c5db02c97261d590eb3a69 100644 (file)
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,16 @@
+2009-07-20  Ulrich Drepper  <drepper@redhat.com>
+
+       * locale/localeinfo.h (LIMAGIC): Update value for LC_CTYPE.
+       * locale/langinfo.h: Define _NL_CTYPE_NONASCII_CASE.
+       * locale/C-ctype.c (_nl_C_LC_CTYPE): Add initializer for
+       _NL_CTYPE_NONASCII_CASE.
+       * locale/programs/ld-ctype.c (locale_ctype_t): Add nonascii_case
+       field.
+       (ctype_finish): Check whether there are any 8-bit characters outside
+       the range ASCII has or whether the mapping isn't the same as for
+       ASCII (±0x20).  Set nonascii_case appropriately.
+       (ctype_output): Add output handler for nonascii_case.
+
 2009-07-17  Ulrich Drepper  <drepper@redhat.com>
 
        * sysdeps/generic/sysdep.h: Define cfi_personality, cfi_lsda,

diff --git a/locale/C-ctype.c b/locale/C-ctype.c
index 85f3d2addbde12e1bf892da50892d57a3139f736..2044fb77b6f00745c32fde0da02dcbabc6ba3697 100644 (file)
--- a/locale/C-ctype.c
+++ b/locale/C-ctype.c
@@ -1,4 +1,4 @@
-/* Copyright (C) 1995-2002, 2003 Free Software Foundation, Inc.
+/* Copyright (C) 1995-2002, 2003, 2009 Free Software Foundation, Inc.
    This file is part of the GNU C Library.
    Contributed by Ulrich Drepper <drepper@cygnus.com>, 1995.
 
@@ -682,6 +682,8 @@ const struct locale_data _nl_C_LC_CTYPE attribute_hidden =
     { .string = (const char *) _nl_C_LC_CTYPE_class_alnum.header },
     /* NR_MAPS wctrans_tables */
     { .string = (const char *) _nl_C_LC_CTYPE_map_toupper.header },
-    { .string = (const char *) _nl_C_LC_CTYPE_map_tolower.header }
+    { .string = (const char *) _nl_C_LC_CTYPE_map_tolower.header },
+    /* _NL_CTYPE_NONASCII_CASE */
+    { .word = 0 }
   }
 };

diff --git a/locale/langinfo.h b/locale/langinfo.h
index 59017b31c89763997d73bbd5a39a431685b739f7..d7ef6f60f471bc4f6bbc07a736f08005faf5dcd2 100644 (file)
--- a/locale/langinfo.h
+++ b/locale/langinfo.h
@@ -348,6 +348,7 @@ enum
   _NL_CTYPE_EXTRA_MAP_12,
   _NL_CTYPE_EXTRA_MAP_13,
   _NL_CTYPE_EXTRA_MAP_14,
+  _NL_CTYPE_NONASCII_CASE,
   _NL_NUM_LC_CTYPE,
 
   /* LC_MONETARY category: formatting of monetary quantities.

diff --git a/locale/localeinfo.h b/locale/localeinfo.h
index 3661080bb256c04eb78e26e03105bdc714359755..19ea41ae6dd719e58273e83531e974f3d7b2562d 100644 (file)
--- a/locale/localeinfo.h
+++ b/locale/localeinfo.h
@@ -1,5 +1,5 @@
 /* Declarations for internal libc locale interfaces
-   Copyright (C) 1995-2003, 2005, 2006, 2007, 2008
+   Copyright (C) 1995-2003, 2005, 2006, 2007, 2008, 2009
    Free Software Foundation, Inc.
    This file is part of the GNU C Library.
 
@@ -35,6 +35,8 @@
 #define        LIMAGIC(category) \
   (category == LC_COLLATE                                              \
    ? ((unsigned int) (0x20051014 ^ (category)))                                \
+   : category == LC_CTYPE                                              \
+   ? ((unsigned int) (0x20090720 ^ (category)))                                \
    : ((unsigned int) (0x20031115 ^ (category))))
 
 /* Two special weight constants for the collation data.  */

diff --git a/locale/programs/ld-ctype.c b/locale/programs/ld-ctype.c
index d4474bf1a2dba0e15f8c789fbd3a39fb58b69389..376a02c2f00eb50d7bad6ff2d9d9ecd45d729843 100644 (file)
--- a/locale/programs/ld-ctype.c
+++ b/locale/programs/ld-ctype.c
@@ -1,4 +1,4 @@
-/* Copyright (C) 1995-2006, 2007 Free Software Foundation, Inc.
+/* Copyright (C) 1995-2006, 2007, 2009 Free Software Foundation, Inc.
    This file is part of the GNU C Library.
    Contributed by Ulrich Drepper <drepper@gnu.org>, 1995.
 
@@ -181,6 +181,7 @@ struct locale_ctype_t
   size_t default_missing_lineno;
 
   uint32_t to_nonascii;
+  uint32_t nonascii_case;
 
   /* The arrays for the binary representation.  */
   char_class_t *ctype_b;
@@ -625,6 +626,27 @@ character <SP> not defined in character map")));
   else
     ctype->class256_collection[space_seq->bytes[0]] |= BIT (tok_print);
 
+  /* Check whether all single-byte characters make to their upper/lowercase
+     equivalent according to the ASCII rules.  */
+  for (cnt = 'A'; cnt <= 'Z'; ++cnt)
+    {
+      uint32_t uppval = ctype->map256_collection[0][cnt];
+      uint32_t lowval = ctype->map256_collection[1][cnt];
+      uint32_t lowuppval = ctype->map256_collection[0][lowval];
+      uint32_t lowlowval = ctype->map256_collection[1][lowval];
+
+      if (uppval != cnt
+         || lowval != cnt + 0x20
+         || lowuppval != cnt
+         || lowlowval != cnt + 0x20)
+       ctype->nonascii_case = 1;
+    }
+  for (cnt = 0; cnt < 256; ++cnt)
+    if (cnt < 'A' || (cnt > 'Z' && cnt < 'a') || cnt > 'z')
+      if (ctype->map256_collection[0][cnt] != cnt
+         || ctype->map256_collection[1][cnt] != cnt)
+       ctype->nonascii_case = 1;
+
   /* Now that the tests are done make sure the name array contains all
      characters which are handled in the WIDTH section of the
      character set definition file.  */
@@ -1045,6 +1067,9 @@ ctype_output (struct localedef_t *locale, const struct charmap_t *charmap,
          CTYPE_DATA (_NL_CTYPE_MAP_TO_NONASCII,
                      &ctype->to_nonascii, sizeof (uint32_t));
 
+         CTYPE_DATA (_NL_CTYPE_NONASCII_CASE,
+                     &ctype->nonascii_case, sizeof (uint32_t));
+
          case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS_MB_LEN):
            iov[2 + elem + offset].iov_base = alloca (sizeof (uint32_t));
            iov[2 + elem + offset].iov_len = sizeof (uint32_t);