Bug 1130891 part 2 - Add option in document encoder for including ruby annotation in plain text. r=roc

dom/base/nsDocumentEncoder.cpp

@@ -124,6 +124,11 @@ protected:
             // We have already checked that our parent is visible.
             return true;
           }
+          if (aNode->IsHTMLElement(nsGkAtoms::rp)) {
+            // Ruby parentheses are part of ruby structure, hence
+            // shouldn't be stripped out even if it is not displayed.
+            return true;
+          }
           return false;
         }
         bool isVisible = frame->StyleVisibility()->IsVisible();

dom/base/nsIDocumentEncoder.idl

@@ -35,7 +35,7 @@ interface nsIDocumentEncoderNodeFixup : nsISupports
   nsIDOMNode fixupNode(in nsIDOMNode aNode, out boolean aSerializeCloneKids);
 };
 
-[scriptable, uuid(1158bd7e-a08b-4ff6-9417-6f99144cfccc)]
+[scriptable, uuid(e5ec69d7-eaa7-4de7-986b-455e17c7f71a)]
 interface nsIDocumentEncoder : nsISupports
 {
   // Output methods flag bits. There are a frightening number of these,
@@ -234,6 +234,12 @@ interface nsIDocumentEncoder : nsISupports
    */
   const unsigned long OutputForPlainTextClipboardCopy = (1 << 25);
 
+  /**
+   * Include ruby annotations and ruby parentheses in the output.
+   * PlainText output only.
+   */
+  const unsigned long OutputRubyAnnotation = (1 << 26);
+
   /**
    * Initialize with a pointer to the document and the mime type.
    * @param aDocument Document to encode.

dom/base/nsPlainTextSerializer.cpp

@@ -29,6 +29,7 @@ using namespace mozilla::dom;
 
 #define PREF_STRUCTS "converter.html2txt.structs"
 #define PREF_HEADER_STRATEGY "converter.html2txt.header_strategy"
+#define PREF_ALWAYS_INCLUDE_RUBY "converter.html2txt.always_include_ruby"
 
 static const  int32_t kTabSize=4;
 static const  int32_t kIndentSizeHeaders = 2;  /* Indention of h1, if
@@ -92,6 +93,7 @@ nsPlainTextSerializer::nsPlainTextSerializer()
   mStartedOutput = false;
 
   mPreformattedBlockBoundary = false;
+  mWithRubyAnnotation = false;  // will be read from pref and flag later
 
   // initialize the tag stack to zero:
   // The stack only ever contains pointers to static atoms, so they don't
@@ -188,6 +190,13 @@ nsPlainTextSerializer::Init(uint32_t aFlags, uint32_t aWrapColumn,
     }
   }
 
+  // The pref is default inited to false in libpref, but we use true
+  // as fallback value because we don't want to affect behavior in
+  // other places which use this serializer currently.
+  mWithRubyAnnotation =
+    Preferences::GetBool(PREF_ALWAYS_INCLUDE_RUBY, true) ||
+    (mFlags & nsIDocumentEncoder::OutputRubyAnnotation);
+
   // XXX We should let the caller decide whether to do this or not
   mFlags &= ~nsIDocumentEncoder::OutputNoFramesContent;
 
@@ -255,6 +264,19 @@ nsPlainTextSerializer::ShouldReplaceContainerWithPlaceholder(nsIAtom* aTag)
     (aTag == nsGkAtoms::video);
 }
 
+bool
+nsPlainTextSerializer::IsIgnorableRubyAnnotation(nsIAtom* aTag)
+{
+  if (mWithRubyAnnotation) {
+    return false;
+  }
+
+  return
+    aTag == nsGkAtoms::rp ||
+    aTag == nsGkAtoms::rt ||
+    aTag == nsGkAtoms::rtc;
+}
+
 NS_IMETHODIMP 
 nsPlainTextSerializer::AppendText(nsIContent* aText,
                                   int32_t aStartOffset,
@@ -440,6 +462,12 @@ nsPlainTextSerializer::DoOpenContainer(nsIAtom* aTag)
     mIgnoredChildNodeLevel++;
     return NS_OK;
   }
+  if (IsIgnorableRubyAnnotation(aTag)) {
+    // Ignorable ruby annotation shouldn't be replaced by a placeholder
+    // character, neither any of its descendants.
+    mIgnoredChildNodeLevel++;
+    return NS_OK;
+  }
 
   if (mFlags & nsIDocumentEncoder::OutputForPlainTextClipboardCopy) {
     if (mPreformattedBlockBoundary && DoOutput()) {
@@ -780,6 +808,10 @@ nsPlainTextSerializer::DoCloseContainer(nsIAtom* aTag)
     mIgnoredChildNodeLevel--;
     return NS_OK;
   }
+  if (IsIgnorableRubyAnnotation(aTag)) {
+    mIgnoredChildNodeLevel--;
+    return NS_OK;
+  }
 
   if (mFlags & nsIDocumentEncoder::OutputForPlainTextClipboardCopy) {
     if (DoOutput() && IsInPre() && IsElementBlock(mElement)) {

dom/base/nsPlainTextSerializer.h

@@ -113,6 +113,7 @@ private:
   bool PopBool(nsTArray<bool>& aStack);
 
   bool ShouldReplaceContainerWithPlaceholder(nsIAtom* aTag);
+  bool IsIgnorableRubyAnnotation(nsIAtom* aTag);
 
   bool IsElementPreformatted(mozilla::dom::Element* aElement);
   bool IsElementBlock(mozilla::dom::Element* aElement);
@@ -175,6 +176,9 @@ private:
 
   bool             mPreformattedBlockBoundary;
 
+  // Whether the output should include ruby annotations.
+  bool             mWithRubyAnnotation;
+
   nsString         mURL;
   int32_t          mHeaderStrategy;    /* Header strategy (pref)
                                           0 = no indention

modules/libpref/init/all.js

@@ -1717,6 +1717,11 @@ pref("network.stricttransportsecurity.preloadlist", true);
 
 pref("converter.html2txt.structs",          true); // Output structured phrases (strong, em, code, sub, sup, b, i, u)
 pref("converter.html2txt.header_strategy",  1); // 0 = no indention; 1 = indention, increased with header level; 2 = numbering and slight indention
+// Whether we include ruby annotation in the text despite whether it
+// is requested. This was true because we didn't explicitly strip out
+// annotations. Set false by default to provide a better behavior, but
+// we want to be able to pref-off it if user doesn't like it.
+pref("converter.html2txt.always_include_ruby", false);
 
 pref("intl.accept_languages",               "chrome://global/locale/intl.properties");
 pref("intl.menuitems.alwaysappendaccesskeys","chrome://global/locale/intl.properties");