Summary:
Now HTMLEditor::IsInLink returns boolean and nsINode if returning true.
I would like to return Element instead of boolean for reducing refcounting
and QI like HTMLEditRules::IsInListItem.
Reviewers: masayuki
Tags: #secure-revision
Bug #: 1487305
Differential Revision: https://phabricator.services.mozilla.com/D4640
--HG--
extra : rebase_source : 275629d7e1108b4e56fb39fa745f491bfb55963d
Even though HTMLEditor::GetIsCSSEnabled() is an implementation of an XPCOM
method, it uses nsresult as its return type. We should change it to
NS_IMETHODIMP.
Additionally, SetDocumentStateCommand::GetCommandStateParams() calls this,
but HTMLEditor can expose non-virtual method, HTMLEditor::IsCSSEnabled().
Therefore, this patch makes it public and makes SetDocumentStateCommand use
HTMLEditor::IsCSSEnabled().
Differential Revision: https://phabricator.services.mozilla.com/D4304
--HG--
extra : moz-landing-system : lando
Neither comm-central nor BlueGriffon uses nsIHTMLEditor.getFontColorSetate().
So, we can get rid of this from nsIHTMLEditor. However, we need to keep it
as a non-virtual public method since it's used by FontColorStateCommand.
Differential Revision: https://phabricator.services.mozilla.com/D4303
--HG--
extra : moz-landing-system : lando
nsITableEditor::GetNextCellElement() is an XPCOM method but it's used internally
a lot. So, HTMLEditor should implement it with non-virtual method and
internal users should use the non-virtual method.
Therefore, this patch creates HTMLEditor::GetNextSelectedTableCellElement().
Differential Revision: https://phabricator.services.mozilla.com/D4194
--HG--
extra : moz-landing-system : lando
There are surprisingly many of them.
(Plus a couple of unnecessary checks after `new` calls that were nearby.)
--HG--
extra : rebase_source : 47b6d5d7c5c99b1b50b396daf7a3b67abfd74fc1
Oddly, on 63 Beta simulation, nsIDocument::GetWindow() may return nullptr
when HTMLEditor is being destroyed by unload of the page. I'm not sure if
this is an expected change. However, HTMLEditor::HideResizers() should
not stop cleaning up even if it meets unexpected situation.
Additionally, this patch moves all HTMLEditor members related to resizers
to local variables since while HideResizers() is cleaning up old resizers,
the members may be overwritten by ShowResizers() if mutation event listener
or something does something.
Differential Revision: https://phabricator.services.mozilla.com/D4057
--HG--
extra : moz-landing-system : lando
HTMLEditor::GetFirstSelectedCell() is an XPCOM method, but used internally a
lot. Therefore, we should create a non-virtual method for internal use.
This patch creates HTMLEditor::GetFirstSelectedTableCellElement(), and it
won't return NS_SUCCESS_EDITOR_ELEMENT_NOT_FOUND since nobody needs the
value. It's enough to check whether the result is nullptr without error for
any callers.
Differential Revision: https://phabricator.services.mozilla.com/D4060
--HG--
extra : moz-landing-system : lando
Oddly, on Android, size of resized objects may be 2px different from ideal
value. I don't know the reason, could be zoom level or something is affected.
However, fortunately, this difference is not important for this test because
this test checks whether resizers actually works with specific elements.
So, even if the result is 2px smaller or bigger than ideal value, we succeeded
to check the resizer makes the element bigger or smaller as expected.
Therefore, this patch makes the test allow 2px differences of the result.
Additionally, on Android, this test is always timed out if TV (even opt build).
So, this patch disables TV on Android.
Differential Revision: https://phabricator.services.mozilla.com/D4058
--HG--
extra : moz-landing-system : lando
nsITableEditor::GetCellAt() is an XPCOM method, but this is called internally
a lot. So, we should create non-virtual method for internal use.
The XPCOM method retrieves a <table> element if given element is nullptr.
However, no internal user needs this feature. So, we can create
GetTableCellElementAt() simply. Then, we can get rid of nsresult and
ErrorResult from it.
Differential Revision: https://phabricator.services.mozilla.com/D3956
--HG--
extra : moz-landing-system : lando
HTMLEditor::GetTableSize() is an XPCOM method but used internally a lot.
Therefore, it shouldn't be called for internal use. Additionally, the
callers need to declare two int32_t variables, but this causes the code
messy. Therefore, this patch creates TableSize struct and it implements
HTMLEditor::GetTableSize(). Then, all callers of it is replaced with
TableSize struct.
New TableSize struct does not support computes <table> element from anchor
of Selection since there is no user of this in C++ code.
Differential Revision: https://phabricator.services.mozilla.com/D3953
--HG--
extra : moz-landing-system : lando
HTMLEditor::GetCellIndexes() is an XPCOM method and used a lot internally.
So, we need alternative way to retrieve indexes of a cell without virtual
calls. In a lot of places, receiving indexes with 2 int32_t variables causes
the code messy and that causes making it harder to understand which are
index for same cell and where they come from. So, making both of them stored
one variable makes the callers simpler. Therefore, this patch creates
CellIndexes stack class to get and store the result simply. Then, this makes
all callers of GetCellIndexes() use this new class and makes GetCellIndexes()
also use this new class.
FYI: This patch does NOT put ErrorResult instances in small block scope as far as
possible. The reason is, I see its destructor in profile sometimes. I don't think
that we should use nsresult& instead of ErrorResult& only for this performance
reason, but I think that creating each instance in loops does not make sense.
Differential Revision: https://phabricator.services.mozilla.com/D3849
--HG--
extra : moz-landing-system : lando
Nobody uses nsITableEditor.getNextRow(). Therefore, this patch removes this
XPCOM API.
Differential Revision: https://phabricator.services.mozilla.com/D3949
--HG--
extra : moz-landing-system : lando
nsITableEditor::GetNextRow() is an XPCOM method. Therefore, we should have
a non-virtual method for internal use of it.
This changes the definition in nsITableEditor. First, it allows only <tr>
element as what HTMLEditor::GetNextRow() has actually done. Then, changes
the return type to Element since it always returns an element node.
Differential Revision: https://phabricator.services.mozilla.com/D3948
--HG--
extra : moz-landing-system : lando
There are 2 bugs:
One is a simple mistake. kTest is each item of the tests, kTests is array of
all tests. When it needs to refer kTest.isAbsolutePosition, it referred
kTests.isAbsolutePosiiton. Therefore, the test always failed to enable
editing UI for absolute positioned element.
The other is, this test requires to disable inline-table-editing UI (which is
add or remove rows and columns). Note that even if the UI is disabled,
resizers is available for <table> elements.
Differential Revision: https://phabricator.services.mozilla.com/D3954
--HG--
extra : moz-landing-system : lando
nsITableEditor::GetFirstRow() is an XPCOM method, so, for internal use,
we should create non-virtual method, that is GetFirstTableRowElement().
This patch makes it never return NS_SUCCESS_EDITOR_ELEMENT_NOT_FOUND since
nobody refers it and it's detectable. If the method returns nullptr without
error, it's the case of NS_SUCCESS_EDITOR_ELEMENT_NOT_FOUND.
Additionally, this patch changes the return type of GetFirstRow() from
Node to Element since it always return an Element node if not null.
Differential Revision: https://phabricator.services.mozilla.com/D3780
--HG--
extra : moz-landing-system : lando
Android's reftest of Bug 1443902 is failed rarely (Bug 1476914, Bug 1475049
Bug 1477502 and Bug 1476129) due to "image comparison, max difference: 1,
number of differing pixels: 1". But I don't know why difference is 1 and
I think that this might be reftest framework for Android or Android emulator
issue.
So I would like to add fuzzy-if as workaround.
Differential Revision: https://phabricator.services.mozilla.com/D3859
--HG--
extra : moz-landing-system : lando
HTMLEditRules::WillDeleteSelection is complex since some variables is reused.
So I would like to clean up this to use block scope and EditorDOMPoint
before fixing bug 685799.
Differential Revision: https://phabricator.services.mozilla.com/D3761
--HG--
extra : moz-landing-system : lando
RemoveList doesn't use aListType parameter, but c-c still call this method.
So I keep aListType parameter even if unused.
Differential Revision: https://phabricator.services.mozilla.com/D3762
--HG--
extra : moz-landing-system : lando
HTMLEditor::RefereshEditingUI() works only with enabled UIs. Therefore, if
UI is disabled while it's visible, it keeps shown. This is too bad if web
apps tries to disable the Gecko specific UIs after we show some of them.
This patch adds HTMLEditor::HideAnonymousEditingUIsIfUnnecessary() to hide
unnecessary UIs and makes RefereshEditingUI() call it always.
First, HTMLEditor::HideInlineTableEditingUI() always returns NS_OK. So, we
can change its return type to void.
Additionally, it removes each UI from the DOM tree one by one. However, each
mutation could cause showing same UI again. In such case,
ShowInlineTableEditingUI() overwrites each UI with newly created element.
Then, HTMLEditor cannot remove the old UI anymore. Therefore, this patch
moves all members of the UI into local variables first.
HTMLEditor::CheckSelectionStateForAnonymousButtons() is called a lot internally.
Especially, its virtual call cost may make damage to our performance since
it's called from a selection listener.
So, we should create non-virtual method, RefereshEditingUI() for internal use.
Even after we disable Gecko specific editing UIs by default, web apps can
enable them with execCommand. However, until such web apps change their
behavior, users cannot use Gecko specific UIs. At least for now, we should
make users can enable them by default.
MozReview-Commit-ID: AuAdw4FQ4He
--HG--
extra : rebase_source : a1f88f2928df0d7afb4361c425d75c74872ac9d5
Currently, absolute position editor listens to mouse events at the default
event group to handle dragging of positioner. However, this is blocked by
a call of Event.stopPropagation() in web apps unexpectedly. Therefore,
we should make it listen to the events at the system event group instead.
MozReview-Commit-ID: Hoa8c9QvMuG
--HG--
extra : rebase_source : 77500356fd1a65e8d81da131e09bc48229a208f9
We have another built-in UI of editor which is not implemented by any other
browsers. That is a draggable handler to move absolute positioned elements.
So, we should disable it in default for compatibility with the other browsers.
However, different from resizers and inline table editor, we don't have
command to enable/disable this feature but for backward compatibility, we
should have it. Therefore, this patch adds new command
"enableAbsolutePositionEditor".
Note that whether resizing UI is available only with enableObjectResizing
state is different from enableInlineTableEditing command. Resizers for
absolute positioned elements are NOT available both enableObjectResizing
and enableAbsolutePositionEditor are enabled.
Additionally, this adds automated tests to check basic functions of absolute
positioned editor.
MozReview-Commit-ID: 9ZSGB8tLpFw
--HG--
rename : editor/libeditor/tests/test_resizers_appearance.html => editor/libeditor/tests/test_abs_positioner_appearance.html
rename : editor/libeditor/tests/test_resizers_resizing_elements.html => editor/libeditor/tests/test_abs_positioner_positioning_elements.html
extra : rebase_source : d516f3f3ef36d4ad13938f214cb6e3868d7ff407
Gecko supports resizers of <img> elements and <table>, <td>, <th> elements and
has UI to remove existing table row or column in default. However, the other
browsers don't have such UI and web apps need to disable this feature with
calling both:
document.execCommand("enableObjectResizing", false, false);
document.execCommand("enableInlineTableEditing", false, false);
for avoiding conflicting with their own features to edit such elements.
Therefore, it doesn't make sense to keep enabling them in default only on
Gecko. If web apps want to keep using these features, they should call:
document.execCommand("enableObjectResizing", false, true);
document.execCommand("enableInlineTableEditing", false, true);
at initializing the editor.
And also this patch fixes bugs of
document.queryCommandState("enableObjectResizing") and
document.queryCommandState("enableInlineTableEditing"). They always return
false even after calling document.execCommand(..., false, true) since
nsSetDocumentStateCommand::GetCommandStateParams() sets bool value as
STATE_ATTRIBUTE. However, nsHTMLDocument::QueryCommandValue() which is the
caller referring STATE_ATTRIBUTE doesn't treat it as bool value. And also
those commands are related to state of document. Therefore, they should be
return as bool value of STATE_ALL instead. Then,
nsHTMLDocument::QueryCommandState() returns the state as expected. Note that
those commands are supported only by Gecko. So, we don't need to worry about
the compatibility.
Finally, this patch rewrites 2 existing tests to check basic behavior of
resizers and appearance of resizers.
Note that this patch does not add new tests to test inline table editor
since it's difficult to test the behavior with current API. Perhaps, we
should add an API to nsIHTMLEditor to retrieve each anonymous elements in
another bug since it requires to add wrapping API of SpecialPowers.
MozReview-Commit-ID: 1FhYo5vcV60
--HG--
rename : editor/libeditor/tests/test_objectResizing.html => editor/libeditor/tests/test_resizers_appearance.html
rename : editor/libeditor/tests/test_bug640321.html => editor/libeditor/tests/test_resizers_resizing_elements.html
extra : rebase_source : a707de5a64ef1f8ce974cdf1be093d1b4f61c7bc
The methods compared with const characters since we've supported "namedanchor"
which is not in nsGkAtoms. Now, it's dropped so that we can compare given
atom with nsGkAtoms.
Differential Revision: https://phabricator.services.mozilla.com/D3586
--HG--
extra : moz-landing-system : lando
Nobody (including comm-central and BlueGriffon) does not use "namedanchor"
special element name with those XPCOMs. Of course, our internal callers too.
Therefore, we can drop.
Note that there is no static Atom for this, so, keeping it makes unnecessary
runtime cost for Firefox users.
This could cause breaking some legacy add-ons for Thunderbird. However,
they can use "anchor" special element name for same purpose.
Differential Revision: https://phabricator.services.mozilla.com/D3585
--HG--
extra : moz-landing-system : lando
HTMLElementOrParentByTagName() is the last user of IsLinkTag(const nsAString&)
and IsNamedAnchorTag(const nsAString&). For making their maintenance easier,
let's make GetElementOrParentByTagName() take const nsAtom& for tag name.
GetElementOrParentByTagName() has two functions, one is looking for an element
starting from a node. The other is, if the start node is nullptr, it retrieves
anchor node of Selection as start node. Therefore, this patch splits the
first part to GetElementOrParentByTagNameInternal(). Then, creates its
wrapper which retrieves anchor of Selection automatically,
GetElementOrParentByTagNameAtSelection().
Additionally, this patch makes all internal callers of HTMLEditor use
GetElementOrParentByTagNameInternal() or
GetElementOrParentByTagNameAtSelection() directly. Then, public method,
GetElementOrParentByTagName() is called only by outer classes.
Note that some callers use both GetElementOrParentByTagNameInternal()
and GetElementOrParentByTagNameAtSelection() since they don't check whether
setting node is nullptr. They may be bug of them. We should investigate
the API callers later.
Differential Revision: https://phabricator.services.mozilla.com/D3584
--HG--
extra : moz-landing-system : lando
Nobody (including comm-central and BlueGriffon) does not use "namedanchor"
special element name with those XPCOMs. Of course, our internal callers too.
Therefore, we can drop.
Note that there is no static Atom for this, so, keeping it makes unnecessary
runtime cost for Firefox users.
This could cause breaking some legacy add-ons for Thunderbird. However,
they can use "anchor" special element name for same purpose.
Differential Revision: https://phabricator.services.mozilla.com/D3585
--HG--
extra : moz-landing-system : lando
HTMLElementOrParentByTagName() is the last user of IsLinkTag(const nsAString&)
and IsNamedAnchorTag(const nsAString&). For making their maintenance easier,
let's make GetElementOrParentByTagName() take const nsAtom& for tag name.
GetElementOrParentByTagName() has two functions, one is looking for an element
starting from a node. The other is, if the start node is nullptr, it retrieves
anchor node of Selection as start node. Therefore, this patch splits the
first part to GetElementOrParentByTagNameInternal(). Then, creates its
wrapper which retrieves anchor of Selection automatically,
GetElementOrParentByTagNameAtSelection().
Additionally, this patch makes all internal callers of HTMLEditor use
GetElementOrParentByTagNameInternal() or
GetElementOrParentByTagNameAtSelection() directly. Then, public method,
GetElementOrParentByTagName() is called only by outer classes.
Note that some callers use both GetElementOrParentByTagNameInternal()
and GetElementOrParentByTagNameAtSelection() since they don't check whether
setting node is nullptr. They may be bug of them. We should investigate
the API callers later.
Differential Revision: https://phabricator.services.mozilla.com/D3584
--HG--
extra : moz-landing-system : lando
HTMLEditor::GetSelectionContainer() is a public method, but it's not used by
outer classes. So, we can make it a protected member.
Additionally, this patch cleans up the method.
- Renames to GetSelectionContainerElement() for making clearer what will be
returned.
- Makes it const.
- Makes it take Selection reference since most callers already have Selection.
- Makes it use RangeBoundary to access start point and end point of range since nsRange::StartOffset() and nsRange::EndOffset() may be slow.
- Makes it not use GetSelectedElement() since it requires unnecessary additional cost and the condition to call it means it uses only the first path in GetSelectedElement() which just returns start node of the range.
- Makes it output warning when it returns nullptr since it reaches nullptr only when illegal cases, e.g., Selection is in orphan node.
Differential Revision: https://phabricator.services.mozilla.com/D3461
--HG--
extra : moz-landing-system : lando
EditorBase::AreNodesSameType() is overridden only by HTMLEditor and the
implementation is enough simple to re-implement in EditorBase.
Additionally, this is called from condition of a loop in
JoinNodesDeepWithTransaction(). So, the virtual call cost may make damage
to the performance.
Differential Revision: https://phabricator.services.mozilla.com/D3460
--HG--
extra : moz-landing-system : lando
HTMLEditor::SetIsCSSEnabled() is an XPCOM but it's defined with nsresult.
Differential Revision: https://phabricator.services.mozilla.com/D3459
--HG--
extra : moz-landing-system : lando
Makoto Kato
Margareta Eliza Balazs
Masayuki Nakano
arthur.iakab
Nicholas Nethercote
Ehsan Akhgari
Cosmin Sabou
Dorel Luca
Daniel Varga
Noemi Erli