Merge commit 'a39d4083cab589d7cd6a13e8a4b8db8875261d75'

[gnulib.git] / lib / unistr / u-strcoll.h

diff --git a/lib/unistr/u-strcoll.h b/lib/unistr/u-strcoll.h
index af404a0..04c8a14 100644 (file)
--- a/lib/unistr/u-strcoll.h
+++ b/lib/unistr/u-strcoll.h
@@ -1,6 +1,6 @@
 /* Compare UTF-8/UTF-16/UTF-32 strings using the collation rules of the current
    locale.
-   Copyright (C) 2009 Free Software Foundation, Inc.
+   Copyright (C) 2009-2014 Free Software Foundation, Inc.
    Written by Bruno Haible <bruno@clisp.org>, 2009.
 
    This program is free software: you can redistribute it and/or modify it
@@ -23,57 +23,68 @@ FUNC (const UNIT *s1, const UNIT *s2)
      When it fails, it sets errno, but also returns a meaningful return value,
      for the sake of callers which ignore errno.  */
   int final_errno = errno;
+  const char *encoding = locale_charset ();
   char *sl1;
   char *sl2;
   int result;
 
-  sl1 = U_STRCONV_TO_LOCALE (s1);
+  /* Pass iconveh_error here, not iconveh_question_mark.  Otherwise the
+     conversion to locale encoding can do transliteration or map some
+     characters to question marks, leading to results that depend on the
+     iconv() implementation and are not obvious.  */
+  sl1 = U_STRCONV_TO_ENCODING (s1, encoding, iconveh_error);
   if (sl1 != NULL)
     {
-      sl2 = U_STRCONV_TO_LOCALE (s2);
+      sl2 = U_STRCONV_TO_ENCODING (s2, encoding, iconveh_error);
       if (sl2 != NULL)
-       {
-         /* Compare sl1 and sl2.  */
-         errno = 0;
-         result = strcoll (sl1, sl2);
-         if (errno == 0)
-           {
-             /* strcoll succeeded.  */
-             free (sl1);
-             free (sl2);
-           }
-         else
-           {
-             /* strcoll failed.  */
-             final_errno = errno;
-             free (sl1);
-             free (sl2);
-             result = U_STRCMP (s1, s2);
-           }
-       }
+        {
+          /* Compare sl1 and sl2.  */
+          errno = 0;
+          result = strcoll (sl1, sl2);
+          if (errno == 0)
+            {
+              /* strcoll succeeded.  */
+              free (sl1);
+              free (sl2);
+              /* The conversion to locale encoding can drop Unicode TAG
+                 characters.  Therefore sl1 and sl2 may be equal when s1
+                 and s2 were in fact different.  Return a nonzero result
+                 in this case.  */
+              if (result == 0)
+                result = U_STRCMP (s1, s2);
+            }
+          else
+            {
+              /* strcoll failed.  */
+              final_errno = errno;
+              free (sl1);
+              free (sl2);
+              result = U_STRCMP (s1, s2);
+            }
+        }
       else
-       {
-         /* s1 could be converted to locale encoding, s2 not.  */
-         final_errno = errno;
-         free (sl1);
-         result = -1;
-       }
+        {
+          /* s1 could be converted to locale encoding, s2 not.  */
+          final_errno = errno;
+          free (sl1);
+          result = -1;
+        }
     }
   else
     {
       final_errno = errno;
-      sl2 = U_STRCONV_TO_LOCALE (s2);
+      sl2 = U_STRCONV_TO_ENCODING (s2, encoding, iconveh_error);
       if (sl2 != NULL)
-       {
-         /* s2 could be converted to locale encoding, s1 not.  */
-         free (sl2);
-         result = 1;
-       }
+        {
+          /* s2 could be converted to locale encoding, s1 not.  */
+          free (sl2);
+          result = 1;
+        }
       else
-       {
-         /* Neither s1 nor s2 could be converted to locale encoding.  */
-         result = U_STRCMP (s1, s2);
-       }
+        {
+          /* Neither s1 nor s2 could be converted to locale encoding.  */
+          result = U_STRCMP (s1, s2);
+        }
     }
 
   errno = final_errno;