Added a function to unescape 8bit only, 7bit encoding support.

bug 161479, r=shanjian, sr=jst, a=dbaron.

intl/uconv/idl/nsITextToSubURI.idl

@@ -63,4 +63,17 @@ interface nsITextToSubURI : nsISupports
    *         or error code of nsIUnicodeDecoder in case of covnersion failure
    */
   AString unEscapeURIForUI(in ACString aCharset, in AUTF8String aURIFragment);
+
+  /**
+   * Unescapes only non ASCII characters in the given URI fragment 
+   * note: this method assumes the URI as UTF-8 and fallbacks to the given charset
+   * in case the data is not UTF-8
+   *
+   * @param aCharset the charset to convert from
+   * @param aURIFragment the URI (or URI fragment) to unescape
+   * @return Unescaped aURIFragment  converted to unicode
+   * @throws NS_ERROR_UCONV_NOCONV when there is no decoder for aCharset
+   *         or error code of nsIUnicodeDecoder in case of covnersion failure
+   */
+  AString unEscapeNonAsciiURI(in ACString aCharset, in AUTF8String aURIFragment);
 };

intl/uconv/src/nsTextToSubURI.cpp

@@ -159,6 +159,16 @@ NS_IMETHODIMP  nsTextToSubURI::UnEscapeAndConvert(
   return rv;
 }
 
+static PRBool statefulCharset(const char *charset)
+{
+  if (!nsCRT::strncasecmp(charset, "ISO-2022-", sizeof("ISO-2022-")-1) ||
+      !nsCRT::strcasecmp(charset, "UTF-7") ||
+      !nsCRT::strcasecmp(charset, "HZ-GB-2312"))
+    return PR_TRUE;
+
+  return PR_FALSE;
+}
+
 nsresult nsTextToSubURI::convertURItoUnicode(const nsAFlatCString &aCharset,
                                              const nsAFlatCString &aURI, 
                                              PRBool aIRI, 
@@ -166,12 +176,15 @@ nsresult nsTextToSubURI::convertURItoUnicode(const nsAFlatCString &aCharset,
 {
   nsresult rv = NS_OK;
 
-  if (IsASCII(aURI)) {
+  // check for 7bit encoding the data may not be ASCII after we decode
+  PRBool isStatefulCharset = statefulCharset(aCharset.get());
+
+  if (!isStatefulCharset && IsASCII(aURI)) {
     _retval.Assign(NS_ConvertASCIItoUCS2(aURI));
     return rv;
   }
 
-  if (aIRI) {
+  if (!isStatefulCharset && aIRI) {
     NS_ConvertUTF8toUCS2 ucs2(aURI);
     if (aURI.Equals(NS_ConvertUCS2toUTF8(ucs2))) {
       _retval.Assign(ucs2);
@@ -221,4 +234,15 @@ NS_IMETHODIMP  nsTextToSubURI::UnEscapeURIForUI(const nsACString & aCharset,
   return convertURItoUnicode(PromiseFlatCString(aCharset), unescapedSpec, PR_TRUE, _retval);
 }
 
+NS_IMETHODIMP  nsTextToSubURI::UnEscapeNonAsciiURI(const nsACString & aCharset, 
+                                                   const nsACString &aURIFragment, 
+                                                   nsAString &_retval)
+{
+  nsCAutoString unescapedSpec;
+  NS_UnescapeURL(PromiseFlatCString(aURIFragment),
+                 esc_AlwaysCopy | esc_OnlyNonASCII, unescapedSpec);
+
+  return convertURItoUnicode(PromiseFlatCString(aCharset), unescapedSpec, PR_TRUE, _retval);
+}
+
 //----------------------------------------------------------------------