Bug 844461 - Perform alias resolution on the fallback encoding pref value. r=smontagu.

content/base/test/test_bug431701.html

@@ -62,7 +62,7 @@ function xhrDoc(idx) {
 // inputEncoding expected for that document.
 
 var tests = [
- [ frameDoc("one"), "ISO-8859-1", "ISO-8859-1" ],
+ [ frameDoc("one"), "windows-1252", "windows-1252" ],
  [ frameDoc("two"), "UTF-8", "UTF-8" ],
  [ frameDoc("three"), "windows-1252", "windows-1252" ],
  [ frameDoc("four"), "UTF-8", "UTF-8" ],

content/html/document/src/nsHTMLDocument.cpp

@@ -82,6 +82,7 @@
 //AHMED 12-2
 #include "nsBidiUtils.h"
 
+#include "mozilla/dom/EncodingUtils.h"
 #include "nsIEditingSession.h"
 #include "nsIEditor.h"
 #include "nsNodeInfoManager.h"
@@ -328,7 +329,7 @@ nsHTMLDocument::TryHintCharset(nsIMarkupDocumentViewer* aMarkupDV,
       if(requestCharsetSource <= aCharsetSource)
         return;
 
-      if(NS_SUCCEEDED(rv) && IsAsciiCompatible(requestCharset)) {
+      if(NS_SUCCEEDED(rv) && EncodingUtils::IsAsciiCompatible(requestCharset)) {
         aCharsetSource = requestCharsetSource;
         aCharset = requestCharset;
 
@@ -352,7 +353,7 @@ nsHTMLDocument::TryUserForcedCharset(nsIMarkupDocumentViewer* aMarkupDV,
     return;
 
   // mCharacterSet not updated yet for channel, so check aCharset, too.
-  if (WillIgnoreCharsetOverride() || !IsAsciiCompatible(aCharset)) {
+  if (WillIgnoreCharsetOverride() || !EncodingUtils::IsAsciiCompatible(aCharset)) {
     return;
   }
 
@@ -364,7 +365,7 @@ nsHTMLDocument::TryUserForcedCharset(nsIMarkupDocumentViewer* aMarkupDV,
 
   if(NS_SUCCEEDED(rv) &&
      !forceCharsetFromDocShell.IsEmpty() &&
-     IsAsciiCompatible(forceCharsetFromDocShell)) {
+     EncodingUtils::IsAsciiCompatible(forceCharsetFromDocShell)) {
     aCharset = forceCharsetFromDocShell;
     aCharsetSource = kCharsetFromUserForced;
     return;
@@ -377,7 +378,7 @@ nsHTMLDocument::TryUserForcedCharset(nsIMarkupDocumentViewer* aMarkupDV,
     if (csAtom) {
       nsAutoCString charset;
       csAtom->ToUTF8String(charset);
-      if (!IsAsciiCompatible(charset)) {
+      if (!EncodingUtils::IsAsciiCompatible(charset)) {
         return;
       }
       aCharset = charset;
@@ -400,12 +401,12 @@ nsHTMLDocument::TryCacheCharset(nsICachingChannel* aCachingChannel,
 
   nsCString cachedCharset;
   rv = aCachingChannel->GetCacheTokenCachedCharset(cachedCharset);
-  // Check IsAsciiCompatible() even in the cache case, because the value
+  // Check EncodingUtils::IsAsciiCompatible() even in the cache case, because the value
   // might be stale and in the case of a stale charset that is not a rough
   // ASCII superset, the parser has no way to recover.
   if (NS_SUCCEEDED(rv) &&
       !cachedCharset.IsEmpty() &&
-      IsAsciiCompatible(cachedCharset))
+      EncodingUtils::IsAsciiCompatible(cachedCharset))
   {
     aCharset = cachedCharset;
     aCharsetSource = kCharsetFromCache;
@@ -425,16 +426,6 @@ CheckSameOrigin(nsINode* aNode1, nsINode* aNode2)
     equal;
 }
 
-bool
-nsHTMLDocument::IsAsciiCompatible(const nsACString& aPreferredName)
-{
-  return !(aPreferredName.LowerCaseEqualsLiteral("utf-16") ||
-           aPreferredName.LowerCaseEqualsLiteral("utf-16be") ||
-           aPreferredName.LowerCaseEqualsLiteral("utf-16le") ||
-           aPreferredName.LowerCaseEqualsLiteral("utf-7") ||
-           aPreferredName.LowerCaseEqualsLiteral("x-imap4-modified-utf7"));
-}
-
 void
 nsHTMLDocument::TryParentCharset(nsIDocShell*  aDocShell,
                                  nsIDocument* aParentDocument,
@@ -460,8 +451,8 @@ nsHTMLDocument::TryParentCharset(nsIDocShell*  aDocShell,
   if (kCharsetFromParentForced == parentSource ||
       kCharsetFromUserForced == parentSource) {
     if (WillIgnoreCharsetOverride() ||
-        !IsAsciiCompatible(aCharset) || // if channel said UTF-16
-        !IsAsciiCompatible(parentCharset)) {
+        !EncodingUtils::IsAsciiCompatible(aCharset) || // if channel said UTF-16
+        !EncodingUtils::IsAsciiCompatible(parentCharset)) {
       return;
     }
     aCharset.Assign(parentCharset);
@@ -477,7 +468,7 @@ nsHTMLDocument::TryParentCharset(nsIDocShell*  aDocShell,
     // Make sure that's OK
     if (!aParentDocument ||
         !CheckSameOrigin(this, aParentDocument) ||
-        !IsAsciiCompatible(parentCharset)) {
+        !EncodingUtils::IsAsciiCompatible(parentCharset)) {
       return;
     }
 
@@ -496,7 +487,7 @@ nsHTMLDocument::TryParentCharset(nsIDocShell*  aDocShell,
     // Make sure that's OK
     if (!aParentDocument ||
         !CheckSameOrigin(this, aParentDocument) ||
-        !IsAsciiCompatible(parentCharset)) {
+        !EncodingUtils::IsAsciiCompatible(parentCharset)) {
       return;
     }
 
@@ -517,10 +508,12 @@ nsHTMLDocument::TryWeakDocTypeDefault(int32_t& aCharsetSource,
 
   // Don't let the user break things by setting intl.charset.default to
   // not a rough ASCII superset
-  if (!defCharset.IsEmpty() && IsAsciiCompatible(defCharset)) {
-    aCharset = defCharset;
+  nsAutoCString canonical;
+  if (EncodingUtils::FindEncodingForLabel(defCharset, canonical) &&
+      EncodingUtils::IsAsciiCompatible(canonical)) {
+    aCharset = canonical;
   } else {
-    aCharset.AssignLiteral("ISO-8859-1");
+    aCharset.AssignLiteral("windows-1252");
   }
   aCharsetSource = kCharsetFromWeakDocTypeDefault;
   return;
@@ -538,7 +531,7 @@ nsHTMLDocument::TryDefaultCharset( nsIMarkupDocumentViewer* aMarkupDV,
   if (aMarkupDV) {
     nsresult rv =
       aMarkupDV->GetDefaultCharacterSet(defaultCharsetFromDocShell);
-    if(NS_SUCCEEDED(rv) && IsAsciiCompatible(defaultCharsetFromDocShell)) {
+    if(NS_SUCCEEDED(rv) && EncodingUtils::IsAsciiCompatible(defaultCharsetFromDocShell)) {
       aCharset = defaultCharsetFromDocShell;
       aCharsetSource = kCharsetFromUserDefault;
     }
@@ -3808,7 +3801,7 @@ nsHTMLDocument::WillIgnoreCharsetOverride()
   if (mCharacterSetSource == kCharsetFromByteOrderMark) {
     return true;
   }
-  if (!IsAsciiCompatible(mCharacterSet)) {
+  if (!EncodingUtils::IsAsciiCompatible(mCharacterSet)) {
     return true;
   }
   nsCOMPtr<nsIWyciwygChannel> wyciwyg = do_QueryInterface(mChannel);

content/html/document/src/nsHTMLDocument.h

@@ -297,8 +297,6 @@ protected:
 
   static uint32_t gWyciwygSessionCnt;
 
-  static bool IsAsciiCompatible(const nsACString& aPreferredName);
-
   static void TryHintCharset(nsIMarkupDocumentViewer* aMarkupDV,
                              int32_t& aCharsetSource,
                              nsACString& aCharset);

dom/encoding/EncodingUtils.cpp

@@ -68,5 +68,15 @@ EncodingUtils::FindEncodingForLabel(const nsACString& aLabel,
       labelsEncodings, ArrayLength(labelsEncodings), label, aOutEncoding));
 }
 
+bool
+EncodingUtils::IsAsciiCompatible(const nsACString& aPreferredName)
+{
+  return !(aPreferredName.LowerCaseEqualsLiteral("utf-16") ||
+           aPreferredName.LowerCaseEqualsLiteral("utf-16be") ||
+           aPreferredName.LowerCaseEqualsLiteral("utf-16le") ||
+           aPreferredName.LowerCaseEqualsLiteral("utf-7") ||
+           aPreferredName.LowerCaseEqualsLiteral("x-imap4-modified-utf7"));
+}
+
 } // namespace dom
 } // namespace mozilla

dom/encoding/EncodingUtils.h

@@ -70,6 +70,15 @@ public:
     aString.Trim(" \t\n\f\r");
   }
 
+  /**
+   * Check is the encoding is ASCII-compatible in the sense that Basic Latin
+   * encodes to ASCII bytes. (The reverse may not be true!)
+   *
+   * @param aPreferredName a preferred encoding label
+   * @return whether the encoding is ASCII-compatible
+   */
+  static bool IsAsciiCompatible(const nsACString& aPreferredName);
+
 private:
   EncodingUtils() MOZ_DELETE;
 };

extensions/universalchardet/tests/CharsetDetectionTests.js

@@ -50,8 +50,11 @@ function InitDetectorTests()
                 .getComplexValue("intl.charset.default",
                                  Ci.nsIPrefLocalizedString)
                 .data;
+            if (gExpectedCharset == "ISO-8859-1") {
+                gExpectedCharset = "windows-1252";
+            }
         } catch (e) {
-            gExpectedCharset = "ISO-8859-8";
+            gExpectedCharset = "windows-1252";
         }
     }
 

extensions/universalchardet/tests/test_bug638318.html

@@ -23,7 +23,7 @@ https://bugzilla.mozilla.org/show_bug.cgi?id=638318
 /** Test for Bug 638318 **/
 /* Note! This test uses the chardet test harness but doesn't test chardet! */
 CharsetDetectionTests("bug638318_text.html",
-		      "ISO-8859-1",
+		      "windows-1252",
 		      new Array(""));
 </script>
 </pre>

extensions/universalchardet/tests/test_bug811363-1-1.html

@@ -22,7 +22,7 @@ https://bugzilla.mozilla.org/show_bug.cgi?id=811363
 <script class="testbody" type="text/javascript">
 /** Test for Bug 811363 **/
 CharsetDetectionTests("bug811363-invalid-1.text",
-		      "ISO-8859-1",
+		      "windows-1252",
 		      new Array("ja_parallel_state_machine",
 				"zh_parallel_state_machine",
 				"zhtw_parallel_state_machine",

extensions/universalchardet/tests/test_bug811363-1-5.html

@@ -22,7 +22,7 @@ https://bugzilla.mozilla.org/show_bug.cgi?id=811363
 <script class="testbody" type="text/javascript">
 /** Test for Bug 811363 **/
 CharsetDetectionTests("bug811363-invalid-5.text",
-		      "ISO-8859-1",
+		      "windows-1252",
 		      new Array("ja_parallel_state_machine",
 				"zh_parallel_state_machine",
 				"zhtw_parallel_state_machine",

layout/base/nsDocumentViewer.cpp

@@ -39,6 +39,7 @@
 #include "nsContentUtils.h"
 #include "nsLayoutStylesheetCache.h"
 #include "mozilla/Preferences.h"
+#include "mozilla/dom/EncodingUtils.h"
 
 #include "nsIDeviceContextSpec.h"
 #include "nsViewManager.h"
@@ -155,6 +156,7 @@ static const char sPrintOptionsContractID[]         = "@mozilla.org/gfx/printset
 #include "jsfriendapi.h"
 
 using namespace mozilla;
+using namespace mozilla::dom;
 
 #ifdef DEBUG
 
@@ -2970,10 +2972,14 @@ nsDocumentViewer::GetDefaultCharacterSet(nsACString& aDefaultCharacterSet)
     const nsAdoptingCString& defCharset =
       Preferences::GetLocalizedCString("intl.charset.default");
 
-    if (!defCharset.IsEmpty()) {
-      mDefaultCharacterSet = defCharset;
+    // Don't let the user break things by setting intl.charset.default to
+    // not a rough ASCII superset
+    nsAutoCString canonical;
+    if (EncodingUtils::FindEncodingForLabel(defCharset, canonical) &&
+        EncodingUtils::IsAsciiCompatible(canonical)) {
+      mDefaultCharacterSet = canonical;
     } else {
-      mDefaultCharacterSet.AssignLiteral("ISO-8859-1");
+      mDefaultCharacterSet.AssignLiteral("windows-1252");
     }
   }
   aDefaultCharacterSet = mDefaultCharacterSet;