X-Git-Url: http://erislabs.net/gitweb/?a=blobdiff_plain;f=lib%2Fstriconveha.c;h=fdb84216547da155db5832a678ec01beae6239d9;hb=ccd54782ae73477aa3f3987ab04db9e6fde025d6;hp=9da18c93ef8e5ff07a10ca7632bfce623922cd27;hpb=36a6f6825953d52a32710a6c38d3ef3a5870d3ac;p=gnulib.git

diff --git a/lib/striconveha.c b/lib/striconveha.c
index 9da18c93e..fdb842165 100644
--- a/lib/striconveha.c
+++ b/lib/striconveha.c
@@ -25,6 +25,9 @@
 #include <stdlib.h>
 #include <string.h>
 
+#include "allocsa.h"
+#include "c-strcase.h"
+
 #define SIZEOF(a) (sizeof(a)/sizeof(a[0]))
 
 
@@ -143,14 +146,16 @@ uniconv_register_autodetect (const char *name,
     }
 }
 
-int
-mem_iconveha (const char *src, size_t srclen,
-	      const char *from_codeset, const char *to_codeset,
-	      enum iconv_ilseq_handler handler,
-	      char **resultp, size_t *lengthp)
+/* Like mem_iconveha, except no handling of transliteration.  */
+static int
+mem_iconveha_notranslit (const char *src, size_t srclen,
+			 const char *from_codeset, const char *to_codeset,
+			 enum iconv_ilseq_handler handler,
+			 size_t *offsets,
+			 char **resultp, size_t *lengthp)
 {
   int retval = mem_iconveh (src, srclen, from_codeset, to_codeset, handler,
-			    resultp, lengthp);
+			    offsets, resultp, lengthp);
   if (retval >= 0 || errno != EINVAL)
     return retval;
   else
@@ -162,13 +167,32 @@ mem_iconveha (const char *src, size_t srclen,
       for (alias = autodetect_list; alias != NULL; alias = alias->next)
 	if (strcmp (from_codeset, alias->name) == 0)
 	  {
-	    const char * const *encodings = alias->encodings_to_try;
+	    const char * const *encodings;
 
+	    if (handler != iconveh_error)
+	      {
+		/* First try all encodings without any forgiving.  */
+		encodings = alias->encodings_to_try;
+		do
+		  {
+		    retval = mem_iconveha_notranslit (src, srclen,
+						      *encodings, to_codeset,
+						      iconveh_error, offsets,
+						      resultp, lengthp);
+		    if (!(retval < 0 && errno == EILSEQ))
+		      return retval;
+		    encodings++;
+		  }
+		while (*encodings != NULL);
+	      }
+
+	    encodings = alias->encodings_to_try;
 	    do
 	      {
-		retval = mem_iconveha (src, srclen,
-				       from_codeset, to_codeset, handler,
-				       resultp, lengthp);
+		retval = mem_iconveha_notranslit (src, srclen,
+						  *encodings, to_codeset,
+						  handler, offsets,
+						  resultp, lengthp);
 		if (!(retval < 0 && errno == EILSEQ))
 		  return retval;
 		encodings++;
@@ -185,10 +209,52 @@ mem_iconveha (const char *src, size_t srclen,
     }
 }
 
-char *
-str_iconveha (const char *src,
+int
+mem_iconveha (const char *src, size_t srclen,
 	      const char *from_codeset, const char *to_codeset,
-	      enum iconv_ilseq_handler handler)
+	      bool transliterate,
+	      enum iconv_ilseq_handler handler,
+	      size_t *offsets,
+	      char **resultp, size_t *lengthp)
+{
+  if (srclen == 0)
+    {
+      /* Nothing to convert.  */
+      *lengthp = 0;
+      return 0;
+    }
+
+  /* When using GNU libc >= 2.2 or GNU libiconv >= 1.5,
+     we want to use transliteration.  */
+#if (__GLIBC__ == 2 && __GLIBC_MINOR__ >= 2) || __GLIBC__ > 2 || _LIBICONV_VERSION >= 0x0105
+  if (transliterate)
+    {
+      int retval;
+      size_t len = strlen (to_codeset);
+      char *to_codeset_suffixed = (char *) allocsa (len + 10 + 1);
+      memcpy (to_codeset_suffixed, to_codeset, len);
+      memcpy (to_codeset_suffixed + len, "//TRANSLIT", 10 + 1);
+
+      retval = mem_iconveha_notranslit (src, srclen,
+					from_codeset, to_codeset_suffixed,
+					handler, offsets, resultp, lengthp);
+
+      freesa (to_codeset_suffixed);
+
+      return retval;
+    }
+  else
+#endif
+    return mem_iconveha_notranslit (src, srclen,
+				    from_codeset, to_codeset,
+				    handler, offsets, resultp, lengthp);
+}
+
+/* Like str_iconveha, except no handling of transliteration.  */
+static char *
+str_iconveha_notranslit (const char *src,
+			 const char *from_codeset, const char *to_codeset,
+			 enum iconv_ilseq_handler handler)
 {
   char *result = str_iconveh (src, from_codeset, to_codeset, handler);
 
@@ -203,11 +269,30 @@ str_iconveha (const char *src,
       for (alias = autodetect_list; alias != NULL; alias = alias->next)
 	if (strcmp (from_codeset, alias->name) == 0)
 	  {
-	    const char * const *encodings = alias->encodings_to_try;
+	    const char * const *encodings;
+
+	    if (handler != iconveh_error)
+	      {
+		/* First try all encodings without any forgiving.  */
+		encodings = alias->encodings_to_try;
+		do
+		  {
+		    result = str_iconveha_notranslit (src,
+						      *encodings, to_codeset,
+						      iconveh_error);
+		    if (!(result == NULL && errno == EILSEQ))
+		      return result;
+		    encodings++;
+		  }
+		while (*encodings != NULL);
+	      }
 
+	    encodings = alias->encodings_to_try;
 	    do
 	      {
-		result = str_iconveha (src, *encodings, to_codeset, handler);
+		result = str_iconveha_notranslit (src,
+						  *encodings, to_codeset,
+						  handler);
 		if (!(result == NULL && errno == EILSEQ))
 		  return result;
 		encodings++;
@@ -223,3 +308,41 @@ str_iconveha (const char *src,
       return NULL;
     }
 }
+
+char *
+str_iconveha (const char *src,
+	      const char *from_codeset, const char *to_codeset,
+	      bool transliterate,
+	      enum iconv_ilseq_handler handler)
+{
+  if (*src == '\0' || c_strcasecmp (from_codeset, to_codeset) == 0)
+    {
+      char *result = strdup (src);
+
+      if (result == NULL)
+	errno = ENOMEM;
+      return result;
+    }
+
+  /* When using GNU libc >= 2.2 or GNU libiconv >= 1.5,
+     we want to use transliteration.  */
+#if (__GLIBC__ == 2 && __GLIBC_MINOR__ >= 2) || __GLIBC__ > 2 || _LIBICONV_VERSION >= 0x0105
+  if (transliterate)
+    {
+      char *result;
+      size_t len = strlen (to_codeset);
+      char *to_codeset_suffixed = (char *) allocsa (len + 10 + 1);
+      memcpy (to_codeset_suffixed, to_codeset, len);
+      memcpy (to_codeset_suffixed + len, "//TRANSLIT", 10 + 1);
+
+      result = str_iconveha_notranslit (src, from_codeset, to_codeset_suffixed,
+					handler);
+
+      freesa (to_codeset_suffixed);
+
+      return result;
+    }
+  else
+#endif
+    return str_iconveha_notranslit (src, from_codeset, to_codeset, handler);
+}