Bug 865407 - Part 5: Move update display of cues code to TextTrackManager. r=bz, r=rillian

content/html/content/src/HTMLMediaElement.cpp

@@ -3677,12 +3677,12 @@ void HTMLMediaElement::FireTimeUpdate(bool aPeriodic)
     mDecoder->SetFragmentEndTime(mFragmentEnd);
   }
 
-  // Update visible text tracks.
+  // Update the cues displaying on the video.
   // Here mTextTrackManager can be null if the cycle collector has unlinked
   // us before our parent. In that case UnbindFromTree will call us
   // when our parent is unlinked.
   if (mTextTrackManager) {
-    mTextTrackManager->Update(time);
+    mTextTrackManager->UpdateCueDisplay();
   }
 }
 

content/html/content/src/TextTrackManager.cpp

@@ -8,6 +8,14 @@
 #include "mozilla/dom/TextTrackManager.h"
 #include "mozilla/dom/HTMLMediaElement.h"
 #include "mozilla/dom/HTMLTrackElement.h"
+#include "mozilla/dom/TextTrack.h"
+#include "mozilla/dom/TextTrackCue.h"
+#include "mozilla/ClearOnShutdown.h"
+#include "nsComponentManagerUtils.h"
+#include "nsVideoFrame.h"
+#include "nsIFrame.h"
+#include "nsTArrayHelpers.h"
+#include "nsIWebVTTParserWrapper.h"
 
 namespace mozilla {
 namespace dom {
@@ -17,6 +25,8 @@ NS_IMPL_CYCLE_COLLECTION_3(TextTrackManager, mTextTracks,
 NS_IMPL_CYCLE_COLLECTION_ROOT_NATIVE(TextTrackManager, AddRef)
 NS_IMPL_CYCLE_COLLECTION_UNROOT_NATIVE(TextTrackManager, Release)
 
+StaticRefPtr<nsIWebVTTParserWrapper> TextTrackManager::sParserWrapper;
+
 TextTrackManager::TextTrackManager(HTMLMediaElement *aMediaElement)
   : mMediaElement(aMediaElement)
 {
@@ -25,6 +35,13 @@ TextTrackManager::TextTrackManager(HTMLMediaElement *aMediaElement)
   mTextTracks = new TextTrackList(mMediaElement->OwnerDoc()->GetParentObject());
   mPendingTextTracks =
     new TextTrackList(mMediaElement->OwnerDoc()->GetParentObject());
+
+  if (!sParserWrapper) {
+    nsCOMPtr<nsIWebVTTParserWrapper> parserWrapper =
+      do_CreateInstance(NS_WEBVTTPARSERWRAPPER_CONTRACTID);
+    sParserWrapper = parserWrapper;
+    ClearOnShutdown(&sParserWrapper);
+  }
 }
 
 TextTrackManager::~TextTrackManager()
@@ -86,9 +103,42 @@ TextTrackManager::DidSeek()
 }
 
 void
-TextTrackManager::Update(double aTime)
+TextTrackManager::UpdateCueDisplay()
 {
-  mTextTracks->Update(aTime);
+  nsIFrame* frame = mMediaElement->GetPrimaryFrame();
+  if (!frame) {
+    return;
+  }
+
+  nsVideoFrame* videoFrame = do_QueryFrame(frame);
+  if (!videoFrame) {
+    return;
+  }
+
+  nsCOMPtr<nsIContent> overlay = videoFrame->GetCaptionOverlay();
+  if (!overlay) {
+    return;
+  }
+
+  nsTArray<nsRefPtr<TextTrackCue> > activeCues;
+  mTextTracks->GetAllActiveCues(activeCues);
+
+  if (activeCues.Length() > 0) {
+    nsCOMPtr<nsIWritableVariant> jsCues =
+      do_CreateInstance("@mozilla.org/variant;1");
+
+    jsCues->SetAsArray(nsIDataType::VTYPE_INTERFACE,
+                       &NS_GET_IID(nsIDOMEventTarget),
+                       activeCues.Length(),
+                       static_cast<void*>(activeCues.Elements()));
+
+    nsPIDOMWindow* window = mMediaElement->OwnerDoc()->GetWindow();
+    if (window) {
+      sParserWrapper->ProcessCues(window, jsCues, overlay);
+    }
+  } else if (overlay->nsINode::Length() > 0) {
+    nsContentUtils::SetNodeTextContent(overlay, EmptyString(), true);
+  }
 }
 
 void

content/html/content/src/TextTrackManager.h

@@ -11,11 +11,16 @@
 #include "mozilla/dom/TextTrack.h"
 #include "mozilla/dom/TextTrackList.h"
 #include "mozilla/dom/TextTrackCueList.h"
+#include "mozilla/StaticPtr.h"
+
+class nsIWebVTTParserWrapper;
 
 namespace mozilla {
 namespace dom {
 
 class HTMLMediaElement;
+class TextTrack;
+class TextTrackCue;
 
 class TextTrackManager
 {
@@ -37,9 +42,35 @@ public:
   void AddCue(TextTrackCue& aCue);
   void AddCues(TextTrack* aTextTrack);
 
-  // Update the display of cues on the video as per the current play back time
-  // of aTime.
-  void Update(double aTime);
+  /**
+   * Overview of WebVTT cuetext and anonymous content setup.
+   *
+   * WebVTT nodes are the parsed version of WebVTT cuetext. WebVTT cuetext is
+   * the portion of a WebVTT cue that specifies what the caption will actually
+   * show up as on screen.
+   *
+   * WebVTT cuetext can contain markup that loosely relates to HTML markup. It
+   * can contain tags like <b>, <u>, <i>, <c>, <v>, <ruby>, <rt>, <lang>,
+   * including timestamp tags.
+   *
+   * When the caption is ready to be displayed the WebVTT nodes are converted
+   * over to anonymous DOM content. <i>, <u>, <b>, <ruby>, and <rt> all become
+   * HTMLElements of their corresponding HTML markup tags. <c> and <v> are
+   * converted to <span> tags. Timestamp tags are converted to XML processing
+   * instructions. Additionally, all cuetext tags support specifying of classes.
+   * This takes the form of <foo.class.subclass>. These classes are then parsed
+   * and set as the anonymous content's class attribute.
+   *
+   * Rules on constructing DOM objects from WebVTT nodes can be found here
+   * http://dev.w3.org/html5/webvtt/#webvtt-cue-text-dom-construction-rules.
+   * Current rules are taken from revision on April 15, 2013.
+   */
+
+  /**
+   * Converts the TextTrackCue's cuetext into a tree of DOM objects and attaches
+   * it to a div on it's owning TrackElement's MediaElement's caption overlay.
+   */
+  void UpdateCueDisplay();
 
   void PopulatePendingList();
 
@@ -55,6 +86,8 @@ private:
   nsRefPtr<TextTrackList> mPendingTextTracks;
   // List of newly introduced Text Track cues.
   nsRefPtr<TextTrackCueList> mNewCues;
+
+  static StaticRefPtr<nsIWebVTTParserWrapper> sParserWrapper;
 };
 
 } // namespace dom

content/media/TextTrack.cpp

@@ -72,14 +72,6 @@ TextTrack::SetDefaultSettings()
   mReadyState = HTMLTrackElement::NONE;
 }
 
-void
-TextTrack::Update(double aTime)
-{
-  if (mCueList) {
-    mCueList->Update(aTime);
-  }
-}
-
 JSObject*
 TextTrack::WrapObject(JSContext* aCx, JS::Handle<JSObject*> aScope)
 {

content/media/TextTrack.h

@@ -99,9 +99,6 @@ public:
   void AddRegion(TextTrackRegion& aRegion);
   void RemoveRegion(const TextTrackRegion& aRegion, ErrorResult& aRv);
 
-  // Time is in seconds.
-  void Update(double aTime);
-
   void AddCue(TextTrackCue& aCue);
   void RemoveCue(TextTrackCue& aCue, ErrorResult& aRv);
   void CueChanged(TextTrackCue& aCue);

content/media/TextTrackCue.cpp

@@ -5,8 +5,6 @@
 
 #include "mozilla/dom/HTMLTrackElement.h"
 #include "mozilla/dom/TextTrackCue.h"
-#include "nsIFrame.h"
-#include "nsVideoFrame.h"
 #include "nsComponentManagerUtils.h"
 #include "mozilla/ClearOnShutdown.h"
 
@@ -101,58 +99,6 @@ TextTrackCue::StashDocument(nsISupports* aGlobal)
   return NS_OK;
 }
 
-void
-TextTrackCue::CreateCueOverlay()
-{
-  nsCOMPtr<nsINodeInfo> nodeInfo =
-    mDocument->NodeInfoManager()->GetNodeInfo(nsGkAtoms::div,
-                                              nullptr,
-                                              kNameSpaceID_XHTML,
-                                              nsIDOMNode::ELEMENT_NODE);
-  mDisplayState = NS_NewHTMLDivElement(nodeInfo.forget());
-  mDisplayState->SetClassName(NS_LITERAL_STRING("caption-text"));
-}
-
-void
-TextTrackCue::RenderCue()
-{
-  nsRefPtr<DocumentFragment> frag = GetCueAsHTML();
-  if (!frag || !mTrackElement) {
-    return;
-  }
-
-  if (!mDisplayState) {
-    CreateCueOverlay();
-  }
-
-  HTMLMediaElement* parent = mTrackElement->mMediaParent;
-  if (!parent) {
-    return;
-  }
-
-  nsIFrame* frame = parent->GetPrimaryFrame();
-  if (!frame) {
-    return;
-  }
-
-  nsVideoFrame* videoFrame = do_QueryFrame(frame);
-  if (!videoFrame) {
-    return;
-  }
-
-  nsIContent* overlay =  videoFrame->GetCaptionOverlay();
-  if (!overlay) {
-    return;
-  }
-
-  ErrorResult rv;
-  nsContentUtils::SetNodeTextContent(overlay, EmptyString(), true);
-  nsContentUtils::SetNodeTextContent(mDisplayState, EmptyString(), true);
-
-  mDisplayState->nsINode::AppendChild(*frag, rv);
-  overlay->AppendChild(*mDisplayState, rv);
-}
-
 already_AddRefed<DocumentFragment>
 TextTrackCue::GetCueAsHTML()
 {

content/media/TextTrackCue.h

@@ -319,36 +319,6 @@ public:
     return mId;
   }
 
-  /**
-   * Overview of WebVTT cuetext and anonymous content setup.
-   *
-   * WebVTT nodes are the parsed version of WebVTT cuetext. WebVTT cuetext is
-   * the portion of a WebVTT cue that specifies what the caption will actually
-   * show up as on screen.
-   *
-   * WebVTT cuetext can contain markup that loosely relates to HTML markup. It
-   * can contain tags like <b>, <u>, <i>, <c>, <v>, <ruby>, <rt>, <lang>,
-   * including timestamp tags.
-   *
-   * When the caption is ready to be displayed the WebVTT nodes are converted
-   * over to anonymous DOM content. <i>, <u>, <b>, <ruby>, and <rt> all become
-   * HTMLElements of their corresponding HTML markup tags. <c> and <v> are
-   * converted to <span> tags. Timestamp tags are converted to XML processing
-   * instructions. Additionally, all cuetext tags support specifying of classes.
-   * This takes the form of <foo.class.subclass>. These classes are then parsed
-   * and set as the anonymous content's class attribute.
-   *
-   * Rules on constructing DOM objects from WebVTT nodes can be found here
-   * http://dev.w3.org/html5/webvtt/#webvtt-cue-text-dom-construction-rules.
-   * Current rules are taken from revision on April 15, 2013.
-   */
-
-  /**
-   * Converts the TextTrackCue's cuetext into a tree of DOM objects and attaches
-   * it to a div on it's owning TrackElement's MediaElement's caption overlay.
-   */
-  void RenderCue();
-
   /**
    * Produces a tree of anonymous content based on the tree of the processed
    * cue text.
@@ -363,7 +333,6 @@ public:
 private:
   void CueChanged();
   void SetDefaultCueSettings();
-  void CreateCueOverlay();
   nsresult StashDocument(nsISupports* aGlobal);
 
   nsRefPtr<nsIDocument> mDocument;

content/media/TextTrackCueList.cpp

@@ -38,17 +38,6 @@ TextTrackCueList::TextTrackCueList(nsISupports* aParent) : mParent(aParent)
   SetIsDOMBinding();
 }
 
-void
-TextTrackCueList::Update(double aTime)
-{
-  const uint32_t length = mList.Length();
-  for (uint32_t i = 0; i < length; i++) {
-    if (aTime > mList[i]->StartTime() && aTime < mList[i]->EndTime()) {
-      mList[i]->RenderCue();
-    }
-  }
-}
-
 JSObject*
 TextTrackCueList::WrapObject(JSContext* aCx, JS::Handle<JSObject*> aScope)
 {

content/media/TextTrackCueList.h

@@ -41,9 +41,6 @@ public:
     return mList.Length();
   }
 
-  // Time is in seconds.
-  void Update(double aTime);
-
   TextTrackCue* IndexedGetter(uint32_t aIndex, bool& aFound);
   TextTrackCue* operator[](uint32_t aIndex);
   TextTrackCue* GetCueById(const nsAString& aId);

content/media/TextTrackList.cpp

@@ -27,15 +27,6 @@ TextTrackList::TextTrackList(nsISupports* aGlobal) : mGlobal(aGlobal)
   SetIsDOMBinding();
 }
 
-void
-TextTrackList::Update(double aTime)
-{
-  uint32_t length = Length(), i;
-  for (i = 0; i < length; i++) {
-    mTextTracks[i]->Update(aTime);
-  }
-}
-
 void
 TextTrackList::GetAllActiveCues(nsTArray<nsRefPtr<TextTrackCue> >& aCues)
 {

content/media/TextTrackList.h

@@ -38,9 +38,6 @@ public:
     return mTextTracks.Length();
   }
 
-  // Time is in seconds.
-  void Update(double aTime);
-
   // Get all the current active cues.
   void GetAllActiveCues(nsTArray<nsRefPtr<TextTrackCue> >& aCues);
 

content/media/webvtt/WebVTTParserWrapper.js

@@ -50,6 +50,11 @@ WebVTTParserWrapper.prototype =
     return WebVTTParser.convertCueToDOMTree(window, cue.text);
   },
 
+  processCues: function(window, cues, overlay)
+  {
+    // TODO: Call prcoess cues on vtt.js
+  },
+
   classDescription: "Wrapper for the JS WebVTTParser (vtt.js)",
   classID: Components.ID(WEBVTTPARSERWRAPPER_CID),
   QueryInterface: XPCOMUtils.generateQI([Ci.nsIWebVTTParserWrapper]),

content/media/webvtt/nsIWebVTTParserWrapper.idl

@@ -7,6 +7,7 @@
 interface nsIDOMHTMLElement;
 interface nsIWebVTTListener;
 interface nsIDOMWindow;
+interface nsIVariant;
 
 /**
  * Interface for a wrapper of a JS WebVTT parser (vtt.js).
@@ -61,6 +62,24 @@ interface nsIWebVTTParserWrapper : nsISupports
    */
   nsIDOMHTMLElement convertCueToDOMTree(in nsIDOMWindow window,
                                         in nsISupports cue);
+
+
+  /**
+   * Compute the display state of the VTTCues in cues along with any VTTRegions
+   * that they might be in. First, it computes the positioning and styling of
+   * the cues and regions passed and converts them into a DOM tree rooted at
+   * a containing HTMLDivElement. It then adjusts those computed divs for
+   * overlap avoidance using the dimensions of 'overlay'. Finally, it adds the
+   * computed divs to the VTTCues display state property for use later.
+   *
+   * @param window  A window object with which it will create the DOM tree
+   *                and containing div element.
+   * @param cues    An array of VTTCues who need there display state to be
+   *                computed.
+   * @param overlay The HTMLElement that the cues will be displayed within.
+   */
+  void processCues(in nsIDOMWindow window, in nsIVariant cues,
+                   in nsISupports overlay);
 };
 
 %{C++

layout/style/html.css

@@ -728,25 +728,9 @@ audio:not([controls]) {
 
 video > .caption-box {
   position: relative;
-  pointer-events: none;
-}
-
-video > div .caption-text {
-  position: absolute;
-  bottom: 24px;
-  padding: 5px;
-  left: 0;
-  right: 0;
-  text-align: center;
-  pointer-events: auto;
-  font-family: "Helvetica Neue", Helvetica, Arial, sans-serif;
-  font-size: 20px;
-  color: white;
-  text-shadow:
-    -2px -2px 1px #000,
-    2px -2px 1px #000,
-    -2px 2px 1px #000,
-    2px 2px 1px #000;
+  overflow: hidden;
+  padding: 1.5%;
+  box-sizing: border-box;
 }
 
 /* emulation of non-standard HTML <marquee> tag */