Don't use conversion with transliteration in u{8,16,32}_strcoll.

4 files changed, 15 insertions, 10 deletions

diff --git a/lib/unistr/u-strcoll.h b/lib/unistr/u-strcoll.h
index 9ec5c60170..62addbaf1a 100644
--- a/lib/unistr/u-strcoll.h
+++ b/lib/unistr/u-strcoll.h
@@ -23,14 +23,19 @@ 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.  */
@@ -41,10 +46,10 @@ FUNC (const UNIT *s1, const UNIT *s2)
               /* strcoll succeeded.  */
               free (sl1);
               free (sl2);
-              /* The conversion to locale encoding can do transliteration or
-                 map some characters to question marks.  Therefore sl1 and sl2
-                 may be equal when s1 and s2 were in fact different.  Return a
-                 nonzero result in this case.  */
+              /* 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);
             }
@@ -68,7 +73,7 @@ FUNC (const UNIT *s1, const UNIT *s2)
   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.  */
diff --git a/lib/unistr/u16-strcoll.c b/lib/unistr/u16-strcoll.c
index 5a504bf4d9..b75ba34a3d 100644
--- a/lib/unistr/u16-strcoll.c
+++ b/lib/unistr/u16-strcoll.c
@@ -29,5 +29,5 @@
 #define FUNC u16_strcoll
 #define UNIT uint16_t
 #define U_STRCMP u16_strcmp
-#define U_STRCONV_TO_LOCALE u16_strconv_to_locale
+#define U_STRCONV_TO_ENCODING u16_strconv_to_encoding
 #include "u-strcoll.h"
diff --git a/lib/unistr/u32-strcoll.c b/lib/unistr/u32-strcoll.c
index 9748855342..95bab07575 100644
--- a/lib/unistr/u32-strcoll.c
+++ b/lib/unistr/u32-strcoll.c
@@ -29,5 +29,5 @@
 #define FUNC u32_strcoll
 #define UNIT uint32_t
 #define U_STRCMP u32_strcmp
-#define U_STRCONV_TO_LOCALE u32_strconv_to_locale
+#define U_STRCONV_TO_ENCODING u32_strconv_to_encoding
 #include "u-strcoll.h"
diff --git a/lib/unistr/u8-strcoll.c b/lib/unistr/u8-strcoll.c
index 9ffa135e68..b4d01061cf 100644
--- a/lib/unistr/u8-strcoll.c
+++ b/lib/unistr/u8-strcoll.c
@@ -29,5 +29,5 @@
 #define FUNC u8_strcoll
 #define UNIT uint8_t
 #define U_STRCMP u8_strcmp
-#define U_STRCONV_TO_LOCALE u8_strconv_to_locale
+#define U_STRCONV_TO_ENCODING u8_strconv_to_encoding
 #include "u-strcoll.h"