url: Don't pass UTF-8 hostname to libidn unless it's valid UTF-8. draft2

- Fix bug in utf8len: Return error when another character in the
sequence is expected but that character is null. That is what I had
initally intended, but I made a mistake with the logic.

- Convert locale to UTF-8 then check if UTF-8 is valid.

My understanding of the conversion from locale to UTF-8 is that if the
locale string which is eventually passed from libidn to libiconv as
'from_codeset' is determined to be UTF-8 then (according to the reporter
Thijs Alkemade) no conversion will happen since the 'to_codeset' is
"UTF-8".

So, we can only check for valid UTF-8 after we're sure we have UTF-8.
The alternative here I guess would be to second guess iconv which could
lead to bugs as Daniel noted on the wget mailing list, or just call it
directly like Alessandro has done.

1 files changed, 8 insertions, 5 deletions

diff --git a/lib/url.c b/lib/url.c
index 0facadcbf..880297b51 100644
--- a/lib/url.c
+++ b/lib/url.c
@@ -3656,7 +3656,7 @@ static curl_off_t utf8len(const char *str)
       continue;
     if(*ch < 0xC2 || *ch > 0xF4)
       return error;
-    if(*++ch)
+    if(!*++ch)
       return error;
     /* second byte */
     if(first == 0xE0) {
@@ -3681,14 +3681,14 @@ static curl_off_t utf8len(const char *str)
       if(first <= 0xDF)
         continue;
     }
-    if(*++ch)
+    if(!*++ch)
       return error;
     /* third byte */
     if(*ch < 0x80 || *ch > 0xBF)
       return error;
     if(first <= 0xEF)
       continue;
-    if(*++ch)
+    if(!*++ch)
       return error;
     /* fourth byte */
     if(*ch < 0x80 || *ch > 0xBF)
@@ -3775,12 +3775,15 @@ static void fix_hostname(struct SessionHandle *data,
     char *ace_hostname = NULL;
     int rc;
     /* Don't pass UTF-8 hostname to libidn unless it's valid UTF-8 */
-    if(codepage_is_utf8() && utf8len(host->name) < 0){
+    char *utf8 = stringprep_locale_to_utf8(host->name);
+    if(utf8len(utf8) < 0) {
       infof(data, "Hostname contains invalid UTF-8 sequence\n");
       rc = IDNA_STRINGPREP_ERROR;
     }
     else
-      rc = idna_to_ascii_lz(host->name, &ace_hostname, 0);
+      rc = idna_to_ascii_8z(utf8, &ace_hostname, 0);
+    idn_free(utf8);
+    utf8 = NULL;
     infof (data, "Input domain encoded as `%s'\n",
            stringprep_locale_charset ());
     if(rc != IDNA_SUCCESS)