fixed IRI misbehaviour(s)

src/ChangeLog

@@ -1,5 +1,11 @@
 2014-10-16  Tim Ruehsen  <tim.ruehsen@gmx.de>
 
+	* url.c (url_parse): little code cleanup
+	* html-url.c (get_urls_html): HTTP header Content-Type charset preceeds http-equiv
+	* iri.c (do_conversion): moved iconv code completely into the function
+	* iri.c (do_conversion): call url_unescape to fix charset conversion
+	* iri.c (remote_to_utf8): use strcasecmp to compare encoding
+
 	* gnutls.c (ssl_connect_wget): do not use SSLv3 except explicitely requested
 	* openssl.c (ssl_init): do not use SSLv3 except explicitely requested
 

src/html-url.c

@@ -748,8 +748,9 @@ get_urls_html (const char *file, const char *url, bool *meta_disallow_follow,
   map_html_tags (fm->content, fm->length, collect_tags_mapper, &ctx, flags,
                  NULL, interesting_attributes);
 
-  /* If meta charset isn't null, override content encoding */
-  if (iri && meta_charset)
+  /* Meta charset is only valid if there was no HTTP header Content-Type charset. */
+  /* This is true for HTTP 1.0 and 1.1. */
+  if (iri && !iri->content_encoding && meta_charset)
     set_content_encoding (iri, meta_charset);
 
   DEBUGP (("no-follow in %s: %d\n", file, ctx.nofollow));

src/iri.c

@@ -38,15 +38,13 @@ as that of the covered work.  */
 #include <errno.h>
 
 #include "utils.h"
+#include "url.h"
 
 /* RFC3987 section 3.1 mandates STD3 ASCII RULES */
 #define IDNA_FLAGS  IDNA_USE_STD3_ASCII_RULES
 
 /* Note: locale encoding is kept in options struct (opt.locale) */
 
-static bool do_conversion (iconv_t cd, char *in, size_t inlen, char **out);
-
-
 /* Given a string containing "charset=XXX", return the encoding if found,
    or NULL otherwise */
 char *
@@ -110,52 +108,34 @@ check_encoding_name (char *encoding)
   return true;
 }
 
-/* Try converting string str from locale to UTF-8. Return a new string
-   on success, or str on error or if conversion isn't needed. */
-const char *
-locale_to_utf8 (const char *str)
-{
-  iconv_t l2u;
-  char *new;
-
-  /* That shouldn't happen, just in case */
-  if (!opt.locale)
-    {
-      logprintf (LOG_VERBOSE, _("locale_to_utf8: locale is unset\n"));
-      opt.locale = find_locale ();
-    }
-
-  if (!opt.locale || !strcasecmp (opt.locale, "utf-8"))
-    return str;
-
-  l2u = iconv_open ("UTF-8", opt.locale);
-  if (l2u == (iconv_t)(-1))
-    {
-      logprintf (LOG_VERBOSE, _("Conversion from %s to %s isn't supported\n"),
-                 quote (opt.locale), quote ("UTF-8"));
-      return str;
-    }
-
-  if (do_conversion (l2u, (char *) str, strlen ((char *) str), &new))
-    return (const char *) new;
-
-  return str;
-}
-
 /* Do the conversion according to the passed conversion descriptor cd. *out
    will contain the transcoded string on success. *out content is
    unspecified otherwise. */
 static bool
-do_conversion (iconv_t cd, char *in, size_t inlen, char **out)
+do_conversion (const char *tocode, const char *fromcode, char *in, size_t inlen, char **out)
 {
+  iconv_t cd;
   /* sXXXav : hummm hard to guess... */
-  size_t len, done, outlen = inlen * 2;
+  size_t len, done, outlen;
   int invalid = 0, tooshort = 0;
-  char *s;
+  char *s, *in_org, *in_save;
 
-  s = xmalloc (outlen + 1);
-  *out = s;
-  len = outlen;
+  cd = iconv_open (tocode, fromcode);
+  if (cd == (iconv_t)(-1))
+    {
+      logprintf (LOG_VERBOSE, _("Conversion from %s to %s isn't supported\n"),
+                 quote (opt.locale), quote ("UTF-8"));
+      return false;
+    }
+
+  /* iconv() has to work on an unescaped string */
+  in_org = in;
+  in_save = in = strndup(in, inlen);
+  url_unescape(in);
+  inlen = strlen(in);
+
+  len = outlen = inlen * 2;
+  *out = s = xmalloc (outlen + 1);
   done = 0;
 
   for (;;)
@@ -164,6 +144,9 @@ do_conversion (iconv_t cd, char *in, size_t inlen, char **out)
         {
           *out = s;
           *(s + len - outlen - done) = '\0';
+          xfree(in_save);
+          iconv_close(cd);
+          logprintf (LOG_VERBOSE, _("converted '%s' (%s) -> '%s' (%s)\n"), in_org, fromcode, *out, tocode);
           return true;
         }
 
@@ -202,9 +185,35 @@ do_conversion (iconv_t cd, char *in, size_t inlen, char **out)
         }
     }
 
+    xfree(in_save);
+    iconv_close(cd);
+    logprintf (LOG_VERBOSE, _("converted '%s' (%s) -> '%s' (%s)\n"), in_org, fromcode, *out, tocode);
     return false;
 }
 
+/* Try converting string str from locale to UTF-8. Return a new string
+   on success, or str on error or if conversion isn't needed. */
+const char *
+locale_to_utf8 (const char *str)
+{
+  char *new;
+
+  /* That shouldn't happen, just in case */
+  if (!opt.locale)
+    {
+      logprintf (LOG_VERBOSE, _("locale_to_utf8: locale is unset\n"));
+      opt.locale = find_locale ();
+    }
+
+  if (!opt.locale || !strcasecmp (opt.locale, "utf-8"))
+    return str;
+
+  if (do_conversion ("UTF-8", opt.locale, (char *) str, strlen ((char *) str), &new))
+    return (const char *) new;
+
+  return str;
+}
+
 /* Try to "ASCII encode" UTF-8 host. Return the new domain on success or NULL
    on error. */
 char *
@@ -258,7 +267,6 @@ idn_decode (char *host)
 bool
 remote_to_utf8 (struct iri *iri, const char *str, const char **new)
 {
-  iconv_t cd;
   bool ret = false;
 
   if (!iri->uri_encoding)
@@ -267,7 +275,7 @@ remote_to_utf8 (struct iri *iri, const char *str, const char **new)
   /* When `i->uri_encoding' == "UTF-8" there is nothing to convert.  But we must
      test for non-ASCII symbols for correct hostname processing in `idn_encode'
      function. */
-  if (!strcmp (iri->uri_encoding, "UTF-8"))
+  if (!strcasecmp (iri->uri_encoding, "UTF-8"))
     {
       const char *p = str;
       for (p = str; *p; p++)
@@ -279,15 +287,9 @@ remote_to_utf8 (struct iri *iri, const char *str, const char **new)
       return false;
     }
 
-  cd = iconv_open ("UTF-8", iri->uri_encoding);
-  if (cd == (iconv_t)(-1))
-    return false;
-
-  if (do_conversion (cd, (char *) str, strlen ((char *) str), (char **) new))
+  if (do_conversion ("UTF-8", iri->uri_encoding, (char *) str, strlen (str), (char **) new))
     ret = true;
 
-  iconv_close (cd);
-
   /* Test if something was converted */
   if (!strcmp (str, *new))
     {

src/url.c

@@ -681,7 +681,6 @@ url_parse (const char *url, int *error, struct iri *iri, bool percent_encode)
   char *user = NULL, *passwd = NULL;
 
   const char *url_encoded = NULL;
-  char *new_url = NULL;
 
   int error_code;
 
@@ -695,29 +694,29 @@ url_parse (const char *url, int *error, struct iri *iri, bool percent_encode)
       goto error;
     }
 
+  url_encoded = url;
+
   if (iri && iri->utf8_encode)
     {
+      char *new_url = NULL;
+
       iri->utf8_encode = remote_to_utf8 (iri, iri->orig_url ? iri->orig_url : url, (const char **) &new_url);
       if (!iri->utf8_encode)
         new_url = NULL;
       else
         {
           iri->orig_url = xstrdup (url);
-          percent_encode = true;
+          url_encoded = reencode_escapes (new_url);
+          if (url_encoded != new_url)
+            xfree (new_url);
+          percent_encode = false;
         }
     }
 
-  /* XXX XXX Could that change introduce (security) bugs ???  XXX XXX*/
   if (percent_encode)
-    url_encoded = reencode_escapes (new_url ? new_url : url);
-  else
-    url_encoded = new_url ? new_url : url;
+    url_encoded = reencode_escapes (url);
 
   p = url_encoded;
-
-  if (new_url && url_encoded != new_url)
-    xfree (new_url);
-
   p += strlen (supported_schemes[scheme].leading_string);
   uname_b = p;
   p = url_skip_credentials (p);