Added unescape function which takes a charset to uconv,

changed uriloader to use the uconv unescape to handle unescape non Ascii URI correctly.
bug 155569, r=ftang, sr=bzbarsky.

intl/uconv/idl/nsITextToSubURI.idl

@@ -50,4 +50,17 @@ interface nsITextToSubURI : nsISupports
 {
 	string ConvertAndEscape(in string charset, in wstring text);
 	wstring UnEscapeAndConvert(in string charset, in string text);
+
+  /**
+   * Unescapes the given URI fragment (for UI purpose only)
+   * note: escaping back the result unescaped string is not guaranteed to be
+   * identical to the original escaped string
+   *
+   * @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 unEscapeURIForUI(in ACString aCharset, in AUTF8String aURIFragment);
 };

intl/uconv/src/nsTextToSubURI.cpp

@@ -39,6 +39,8 @@
 #include "nsString.h"
 #include "nsIUnicodeEncoder.h"
 #include "nsICharsetConverterManager.h"
+#include "nsICharsetConverterManager2.h"
+#include "nsReadableUtils.h"
 #include "nsITextToSubURI.h"
 #include "nsIServiceManager.h"
 #include "nsUConvDll.h"
@@ -157,4 +159,66 @@ NS_IMETHODIMP  nsTextToSubURI::UnEscapeAndConvert(
   return rv;
 }
 
+nsresult nsTextToSubURI::convertURItoUnicode(const nsAFlatCString &aCharset,
+                                             const nsAFlatCString &aURI, 
+                                             PRBool aIRI, 
+                                             nsAString &_retval)
+{
+  nsresult rv = NS_OK;
+
+  if (IsASCII(aURI)) {
+    _retval.Assign(NS_ConvertASCIItoUCS2(aURI));
+    return rv;
+  }
+
+  if (aIRI) {
+    NS_ConvertUTF8toUCS2 ucs2(aURI);
+    if (aURI.Equals(NS_ConvertUCS2toUTF8(ucs2))) {
+      _retval.Assign(ucs2);
+      return rv;
+    }
+  }
+
+  nsCOMPtr<nsICharsetConverterManager2> charsetConverterManager;
+
+  charsetConverterManager = do_GetService(NS_CHARSETCONVERTERMANAGER_CONTRACTID, &rv);
+  NS_ENSURE_SUCCESS(rv, rv);
+
+  nsCOMPtr<nsIAtom> charsetAtom;
+  rv = charsetConverterManager->GetCharsetAtom2(aCharset.get(), getter_AddRefs(charsetAtom));
+  NS_ENSURE_SUCCESS(rv, rv);
+
+  nsCOMPtr<nsIUnicodeDecoder> unicodeDecoder;
+  rv = charsetConverterManager->GetUnicodeDecoder(charsetAtom, 
+                                                  getter_AddRefs(unicodeDecoder));
+  NS_ENSURE_SUCCESS(rv, rv);
+
+  PRInt32 srcLen = aURI.Length();
+  PRInt32 dstLen;
+  rv = unicodeDecoder->GetMaxLength(aURI.get(), srcLen, &dstLen);
+  NS_ENSURE_SUCCESS(rv, rv);
+
+  PRUnichar *ustr = (PRUnichar *) nsMemory::Alloc(dstLen * sizeof(PRUnichar));
+  NS_ENSURE_TRUE(ustr, NS_ERROR_OUT_OF_MEMORY);
+
+  rv = unicodeDecoder->Convert(aURI.get(), &srcLen, ustr, &dstLen);
+
+  if (NS_SUCCEEDED(rv))
+    _retval.Assign(ustr, dstLen);
+  
+  nsMemory::Free(ustr);
+
+  return rv;
+}
+
+NS_IMETHODIMP  nsTextToSubURI::UnEscapeURIForUI(const nsACString & aCharset, 
+                                                const nsACString &aURIFragment, 
+                                                nsAString &_retval)
+{
+  nsCAutoString unescapedSpec(aURIFragment);
+  NS_UnescapeURL(unescapedSpec);
+
+  return convertURItoUnicode(PromiseFlatCString(aCharset), unescapedSpec, PR_TRUE, _retval);
+}
+
 //----------------------------------------------------------------------

intl/uconv/src/nsTextToSubURI.h

@@ -48,6 +48,21 @@ class nsTextToSubURI: public nsITextToSubURI {
 public:
   nsTextToSubURI();
   virtual ~nsTextToSubURI();
+
+private:
+  // IRI is "Internationalized Resource Identifiers"
+  // http://www.ietf.org/internet-drafts/draft-duerst-iri-01.txt
+  // 
+  // if the IRI option is true then we assume that the URI is encoded as UTF-8
+  // note: there is no definite way to distinguish between IRI and a URI encoded 
+  // with a non-UTF-8 charset
+  // Use this option carefully -- it may cause dataloss
+  // (recommended to set to true for UI purpose only)
+  //
+  nsresult convertURItoUnicode(const nsAFlatCString &aCharset,
+                               const nsAFlatCString &aURI, 
+                               PRBool aIRI, 
+                               nsAString &_retval);
 };
 
 #endif // nsTextToSubURI_h__

uriloader/build/Makefile.in

@@ -40,6 +40,7 @@ REQUIRES	= xpcom \
 		  rdf \
 		  helperAppDlg \
 		  dom \
+		  uconv \
 		  $(NULL)
 
 CPPSRCS		= nsURILoaderModule.cpp

uriloader/exthandler/Makefile.in

@@ -52,6 +52,7 @@ REQUIRES	= xpcom \
 		  plugin \
 		  pref \
 		  intl \
+		  uconv \
 		  windowwatcher \
 		  $(NULL)
 

uriloader/exthandler/nsExternalHelperAppService.cpp

@@ -80,6 +80,8 @@
 #include "nsIPromptService.h"
 #include "nsIDOMWindow.h"
 
+#include "nsITextToSubURI.h"
+
 const char *FORCE_ALWAYS_ASK_PREF = "browser.helperApps.alwaysAsk.force";
 
 static NS_DEFINE_CID(kRDFServiceCID, NS_RDFSERVICE_CID);
@@ -1046,8 +1048,21 @@ nsresult nsExternalAppHandler::SetUpTempFile(nsIChannel * aChannel)
     url->GetFileName(leafName);
     if (!leafName.IsEmpty())
     {
-      NS_UnescapeURL(leafName);
-      mSuggestedFileName = NS_ConvertUTF8toUCS2(leafName); // XXX leafName may not be UTF-8
+
+      nsCOMPtr<nsITextToSubURI> textToSubURI = do_GetService(NS_ITEXTTOSUBURI_CONTRACTID, &rv);
+      if (NS_SUCCEEDED(rv))
+      {
+        nsCAutoString originCharset;
+        url->GetOriginCharset(originCharset);
+        rv = textToSubURI->UnEscapeURIForUI(originCharset, leafName, 
+                                            mSuggestedFileName);
+      }
+
+      if (NS_FAILED(rv))
+      {
+        mSuggestedFileName = NS_ConvertUTF8toUCS2(leafName); // use escaped name
+        rv = NS_OK;
+      }
 
 #ifdef XP_WIN
       // Make sure extension is still correct.