/**
* Copyright 2009 Google Inc.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*
*/
package org.waveprotocol.wave.client.common.util;
import com.google.gwt.core.client.GWT;
/**
* Collection of constants defining various browser quirky behaviours.
*
* Each constant should be accompanied by a detailed comment, and a "Tested:"
* section detailing which browsers and operating systems the quirk has been
* tested on, so that it's easy to know what's untested as new browser versions
* come out, etc.
*
* Sometimes an "Untested exceptions:" field is appropriate, to note exceptions
* to the "Tested:" field, in particular if they are concerning and represent a
* reasonable doubt as to the correctness of the field's value.
*
* It is preferable to use the constants in this class than to have logic that
* switches on explicit browser checks.
*
* @author danilatos@google.com (Daniel Danilatos)
*/
public final class QuirksConstants {
/**
* Whether we get DOM Mutation events
*
* Tested:
* Safari 3-4, Firefox 3-3.5, Chrome 1-2, IE7, IE8
*
* Will IE9 give us mutation events? probably not.
*/
public static final boolean PROVIDES_MUTATION_EVENTS =
UserAgent.isFirefox() || UserAgent.isWebkit();
/**
* Whether the browser left normalises the caret in most cases (There are
* exceptions, usually to do with links).
*
* Tested:
* Safari 3*, Safari 4 beta, Firefox 3.0, IE7, IE8
*/
public static final boolean USUALLY_LEFT_NORMALISES =
UserAgent.isWebkit() || UserAgent.isWebkit();
/**
* Certain versions of webkit have a specific hack implemented in them, where
* they go against the regular left-normalised behaviour at the end of an
* anchor boundary, if it has an href. They still report the cursor as being
* left-normalised, but if the user types, text goes to the right, outside the
* link.
*
* Tested:
* All OS: Safari 3.2.1, Safari 4 beta, Chrome 1.0, Chrome 2.0 special sauce
*/
public static final boolean LIES_ABOUT_CARET_AT_LINK_END_BOUNDARY =
UserAgent.isWebkit() && UserAgent.isAtLeastVersion(528, 0);
/**
* Similar to {@link #LIES_ABOUT_CARET_AT_LINK_END_BOUNDARY}, but does not
* actually lie, just doesn't like reporting carets as being inside link
* boundaries, and typing occurs outside as well.
*
* Tested: IE8 beta
*
* TODO(danilatos): check IE7
*/
public static final boolean DOES_NOT_LEFT_NORMALISE_AT_LINK_END_BOUNDARY =
UserAgent.isIE();
/**
* If the user is typing, we always get a key event before the browser changes
* the dom.
*
* Tested:
* All OS: Safari 3.2.1, 4 beta, Chrome 1, 2, Firefox 3, IE7, IE8
*
* Untested exceptions: Any IME on Linux!
*/
public static final boolean ALWAYS_GIVES_KEY_EVENT_BEFORE_CHANGING_DOM =
UserAgent.isFirefox() || UserAgent.isIE7();
/**
* Whether we get the magic 229 keycode for IME key events, at least for the
* first one (sometimes we don't get key events for subsequent mutations).
*
* Tested:
* All OS: Safari 3.2.1, 4 beta, Chrome 1, 2, Firefox 3.0
*
* Untested exceptions: Any IME on Linux!
*/
public static final boolean CAN_TELL_WHEN_FIRST_KEY_EVENT_IS_IME =
UserAgent.isIE() || UserAgent.isWebkit() || UserAgent.isWin();
/**
* Whether the old school Ctrl+Insert, Shift+Delete, Shift+Insert shortcuts
* for Copy, Cut, Paste work.
*
* Tested: All OS: Firefox 3, IE 7/8, Safari 3, Chrome 2
*
* Untested exceptions: Safari on Windows
*/
public static final boolean HAS_OLD_SCHOOL_CLIPBOARD_SHORTCUTS =
UserAgent.isWin() || UserAgent.isLinux();
// NOTE(danilatos): These selection constants, unless mentioned otherwise,
// refer to selection boundaries (e.g. the start, or end, of a ranged
// selection, or a collapsed selection).
/**
* Whether the selection is either cleared or correctly transformed by the
* browser in at least the following scenarios:
* - textNode.insertData adds to the cursor location, if it is after the
* deletion point
* - textNode.deleteData subtracts from the cursor location, if it is after
* the deletion point
*
* Tested: Safari 3.2.1, FF 3.0, 3.5, Chrome 1, IE7, IE8
*/
public static final boolean OK_SELECTION_ACROSS_TEXT_NODE_DATA_CHANGES =
UserAgent.isFirefox() || UserAgent.isIE();
/**
* Whether the selection is either cleared or correctly transformed by the
* browser when text nodes are deleted.
*
* Gecko/IE preserve, Webkit clears selection. Both OK behaviours.
*
* Tested: Safari 3.2.1, FF 3.0, 3.5, Chrome 1, IE7, IE8
*/
public static final boolean OK_SELECTION_ACROSS_NODE_REMOVALS = true;
/**
* Whether the browser moves the selection into the neighbouring text node
* after a text node split before the selection point, or at least clears the
* selection.
*
* Tested: Safari 3.2.1, FF 3.0, 3.5, Chrome 1, IE7, IE8
*/
public static final boolean OK_SELECTION_ACROSS_TEXT_NODE_SPLITS =
UserAgent.isIE();
/**
* In this case, only clearing occurs by Webkit. Other two browsers move the
* selection to the point where the moved node was. Which is BAD for wrapping!
*
* Tested: Safari 3.2.1, FF 3.0, 3.5, Chrome 1, IE7, IE8
*
* @see #PRESERVES_SEMANTIC_SELECTION_ACROSS_MUTATIONS_OR_CLEARS_IT
*/
public static final boolean OK_SELECTION_ACROSS_MOVES =
UserAgent.isWebkit();
/**
* Preserves changes to text nodes made by calling methods on the text nodes
* directly (i.e. not moving or deleting the text nodes).
*/
public static final boolean PRESERVES_SEMANTIC_SELECTION_ACROSS_INTRINSIC_TEXT_NODE_CHANGES =
OK_SELECTION_ACROSS_TEXT_NODE_DATA_CHANGES &&
OK_SELECTION_ACROSS_TEXT_NODE_SPLITS;
/**
* Whether the selection preservation is completely reliable across mutations
* in terms of correctness. It might get cleared in some circumstances, but
* that's OK, we can just check if the selection is gone and restore it. We
* don't need to transform it for correctness.
*
* The biggest problem here is wrap. Currently implemented with insertBefore,
* it breaks selections in all browsers, even IE, AND IE doesn't clear the
* selection, just moves it. Damn! Anyway, there might be a smarter way to
* implement wrap - perhaps using an exec command to apply a styling to a
* region and using the resulting wrapper nodes... but that's a long way off.
*
* Tested: Safari 3.2.1, FF 3.0, 3.5, Chrome 1, IE7, IE8
*/
public static final boolean PRESERVES_SEMANTIC_SELECTION_ACROSS_MUTATIONS_OR_CLEARS_IT =
OK_SELECTION_ACROSS_TEXT_NODE_DATA_CHANGES &&
OK_SELECTION_ACROSS_TEXT_NODE_SPLITS &&
OK_SELECTION_ACROSS_MOVES;
/**
* Whether changing stuff in the middle of a ranged selection (that doesn't
* affect the selection end points in any way), such as splitting some
* entirely contained text node, affects the ranged selection. With firefox,
* it appears that the selection is still "correct", but visually new text
* nodes inserted don't get highlighted as selected, which is bad.
*
* Tested: Safari 3.2.1, FF 3.0, IE8
*/
public static final boolean RANGED_SELECTION_AFFECTED_BY_INTERNAL_CHANGED =
UserAgent.isFirefox();
/**
* True if IME input in not totally munged by adjacent text node mutations
*
* Tested:
* Safari 3.2.1, FF 3.0, 3.5, Chrome 1, IE7, IE8
*/
public static final boolean PRESERVES_IME_STATE_ACROSS_ADJACENT_CHANGES =
UserAgent.isIE();
/**
* True if we can do the __proto__ override trick to remove defaults method
* in a JSO.
* WARNING(dnailatos/reuben) Should be kept as static constant for
* speed reasons in string map implementation.
*
* Tested: Safari 4, FF 3.0, 3.5, Chrome 3-4, IE8
*/
public static final boolean DOES_NOT_SUPPORT_JSO_PROTO_FIELD = UserAgent.isIE();
/**
* It appears that in some browsers, if you stopPropagation() an IME
* composition or contextmenu event, the default action for the event is not
* executed (as if you had cancelled it)
*
* TODO(danilatos): File a bug
*
* Tested:
* FF 3.0
*
* Untested:
* Everything else (at time of writing, nothing else has composition
* events...)
*/
public static final boolean CANCEL_BUBBLING_CANCELS_IME_COMPOSITION_AND_CONTEXTMENU =
UserAgent.isFirefox();
/**
* True if mouse events have rangeParent and rangeOffset fields.
*
* Tested:
* FF 3.0, 3.6
* Safari 4
* Chrome 5
*/
public static final boolean SUPPORTS_EVENT_GET_RANGE_METHODS =
UserAgent.isFirefox();
/**
* True if preventDefault stops a native context menu from being shown. In
* firefox this is not the case when dom.event.contextmenu.enabled is set.
*
* Tested:
* FF 3.0, 3.6
*/
public static final boolean PREVENT_DEFAULT_STOPS_CONTEXTMENT =
!UserAgent.isFirefox();
/**
* True if displaying a context menu updates the current selection. Safari selects
* the word clicked unless you click on the current selection, Firefox does not
* change the selection and Chrome and IE clears the selection unless you click
* on the current selection.
*
* The selection that is active on mousedown will, in all browsers, be the
* original selection and the selection on the contextmenu event will be the new
* one.
*
* Tested:
* FF 3.0, 3.5
* Chrome 5
* Safari 4
* IE 8
*/
public static final boolean CONTEXTMENU_SETS_SELECTION =
!UserAgent.isFirefox();
/**
* True if the browser has the setBaseAndExtent JS method to allow better setting of
* the selection within the browser.
*
* So far, only webkit browsers have this in their API, and documentation is scarce. See:
* http://developer.apple.com/DOCUMENTATION/AppleApplications/Reference/WebKitDOMRef
* /DOMSelection_idl/Classes/DOMSelection/index.html#//apple_ref/idl/instm
* /DOMSelection/setBaseAndExtent/void/(inNode,inlong,inNode,inlong)
*/
public static final boolean HAS_BASE_AND_EXTENT =
UserAgent.isWebkit();
/**
* Chrome on Mac generates doesn't keypress for command combos, only keydown.
*
* In general, it makes sense to only fire the keypress event if the combo
* generates content. https://bugs.webkit.org/show_bug.cgi?id=30397
*
* However since it is the odd one out here, it is listed as a
* quirk.
*/
public static final boolean COMMAND_COMBO_DOESNT_GIVE_KEYPRESS =
UserAgent.isMac() && UserAgent.isChrome();
/**
* True if the browser has native support for getElementsByClassName.
*
* Tested:
* Chrome, Safari 3.1, Firefox 3.0
*/
public static final boolean SUPPORTS_GET_ELEMENTS_BY_CLASSNAME =
GWT.isScript() && checkGetElementsByClassNameSupport();
/**
* True if the browser supports composition events.
*
* (This does not differentiate between the per-browser composition event quirks,
* such as whether they provide a text vs a compositionupdate event, or other
* glitches).
*
* Tested:
* Chrome 3.0, 4.0; Safari 4; FF 3.0, 3.5; IE 7,8
*/
public static final boolean SUPPORTS_COMPOSITION_EVENTS =
UserAgent.isFirefox() || (UserAgent.isWebkit()
&& UserAgent.isAtLeastVersion(532, 5));
/**
* True if the browser does an extra modification of the DOM after the
* compositionend event, and also fires a text input event if the composition
* was not cancelled.
*
* Tested: Chrome 4.0; FF 3.0, 3.5;
*/
public static final boolean MODIFIES_DOM_AND_FIRES_TEXTINPUT_AFTER_COMPOSITION =
UserAgent.isWebkit(); // Put an upper bound on the version here when it's fixed...
/**
* True if the browser keeps the selection in an empty span after the
* app has programmatically set it there.
*
* Tested:
* Chrome 3.0, 4.0; Safari 3, 4; FF 3.0, 3.5; IE 7,8
*/
public static final boolean SUPPORTS_CARET_IN_EMPTY_SPAN =
UserAgent.isFirefox();
/**
* True if the browser automatically scrolls a contentEditable element
* into view when we set focus on the element
*
* Tested:
* Chrome 5.0.307.11 beta / linux, Safari 4.0.4 / mac, Firefox 3.0.7 + 3.6 / linux
*/
public static final boolean ADJUSTS_SCROLL_TOP_WHEN_FOCUSING =
UserAgent.isWebkit();
/**
* True if the browser does not emit a paste event for plaintext paste.
* This was a bug on Webkit and has been fixed and pushed to Chrome 4+
*
* Tested:
* Chrome 4.0.302.3; Safari 4.05 Mac
*/
public static final boolean PLAINTEXT_PASTE_DOES_NOT_EMIT_PASTE_EVENT =
UserAgent.isSafari();
/**
* True if the browser supports input type 'search'.
*
* Tested:
* Chrome 9.0, Chrome 4.0
*/
public static final boolean SUPPORTS_SEARCH_INPUT =
UserAgent.isWebkit();
/**
* True if the browser sanitizes pasted content to contenteditable to
* prevent script execution.
*
* Tested:
* Chrome 9.0, Safari 5, FF 3.5, FF 4.0
*/
public static final boolean SANITIZES_PASTED_CONTENT =
(UserAgent.isWebkit() && UserAgent.isAtLeastVersion(533, 16)) ||
(UserAgent.isFirefox() && UserAgent.isAtLeastVersion(4, 0));
private static native boolean checkGetElementsByClassNameSupport() /*-{
return !!document.body.getElementsByClassName;
}-*/;
private QuirksConstants(){}
}