Inserting html on the fly and tabs - firefox-addon

I'm building an add-on that uses a browser overlay.xul, which has included js files. The easiest way I found to modify HTML on a webpage was this bit of code I found on the mdn I believe.
const STATE_START = Components.interfaces.nsIWebProgressListener.STATE_START;
const STATE_STOP = Components.interfaces.nsIWebProgressListener.STATE_STOP;
var myListener = {
QueryInterface: function(aIID) {
if (aIID.equals(Components.interfaces.nsIWebProgressListener) ||
aIID.equals(Components.interfaces.nsISupportsWeakReference) ||
aIID.equals(Components.interfaces.nsISupports))
return this;
throw Components.results.NS_NOINTERFACE;
},
onStateChange: function(aWebProgress, aRequest, aFlag, aStatus) { },
onLocationChange: function(aProgress, aRequest, aURI) {
PageLoad.initialzed = false;
PageLoad.init();
},
onProgressChange: function(aWebProgress, aRequest, curSelf, maxSelf, curTot, maxTot) { },
onStatusChange: function(aWebProgress, aRequest, aStatus, aMessage) { },
onSecurityChange: function(aWebProgress, aRequest, aState) { }
};
gBrowser.addProgressListener(myListener);
var PageLoad = {
initialzed : false,
browser : null,
domain : null,
appcontent : null,
init : function() {
PageLoad.appcontent = document.getElementById("appcontent");
if (PageLoad.appcontent) {
PageLoad.appcontent.addEventListener("DOMContentLoaded", PageLoad.load, true);
}
},
load : function() {
//I do this so it only calls the function once
if (PageLoad.initialzed == false) {
PageLoad.initialzed = true;
PageLoad.browser = gBrowser.contentDocument;
PageLoad.domain = (PageLoad.browser.location.host.match(/([^.]+)\.\w{2,3}(?:\.\w{2})?$/) || [])[1];
myFunctiontoInsertHTML();
}
}
};
Then I can use PageLoad.browser.createElement('div') and so on in my function.. This works good so far but I have come across a problem that with multiple tabs open, my PageLoad.domain variable contains the domain of the last loaded tab or web page (which in turn causes errors inserting html on 2 or more different pages/tabs). So what I want to change is everytime the user clicks on a different tab or reloads the page I want to call the function but I have run into a dead end.

gBrowser.contentDocument is the document of the page loaded in the current browser tab - as soon as the user switches tabs or navigates to a different page its value changes and you shouldn't keep the old value (it is a memory leak). You are observing the progress of all browser tabs, you should look at the one that you got called for. Meaning in your case: aProgress.DOMWindow (see nsIWebProgress.DOMWindow). That's also where you should attach your DOMContentLoaded event listener if you are interested in the load progress of a single tab.

Related

Full-featured autocomplete widget for Dojo

As of now (Dojo 1.9.2) I haven't been able to find a Dojo autocomplete widget that would satisfy all of the following (typical) requirements:
Only executes a query to the server when a predefined number of characters have been entered (without this, big datasets should not be queried)
Does not require a full REST service on the server, only a URL which can be parametrized with a search term and simply returns JSON objects containing an ID and a label to display (so the data-query to the database can be limited just to the required data fields, not loading full data-entities and use only one field thereafter)
Has a configurable time-delay between the key-releases and the start of the server-query (without this excessive number of queries are fired against the server)
Capable of recognizing when there is no need for a new server-query (since the previously executed query is more generic than the current one would be).
Dropdown-stlye (has GUI elements indicating that this is a selector field)
I have created a draft solution (see below), please advise if you have a simpler, better solution to the above requirements with Dojo > 1.9.
The AutoComplete widget as a Dojo AMD module (placed into /gefc/dijit/AutoComplete.js according to AMD rules):
//
// AutoComplete style widget which works together with an ItemFileReadStore
//
// It will re-query the server whenever necessary.
//
define([
"dojo/_base/declare",
"dijit/form/FilteringSelect"
],
function(declare, _FilteringSelect) {
return declare(
[_FilteringSelect], {
// minimum number of input characters to trigger search
minKeyCount: 2,
// the term for which we have queried the server for the last time
lastServerQueryTerm: null,
// The query URL which will be set on the store when a server query
// is needed
queryURL: null,
//------------------------------------------------------------------------
postCreate: function() {
this.inherited(arguments);
// Setting defaults
if (this.searchDelay == null)
this.searchDelay = 500;
if (this.searchAttr == null)
this.searchAttr = "label";
if (this.autoComplete == null)
this.autoComplete = true;
if (this.minKeyCount == null)
this.minKeyCount = 2;
},
escapeRegExp: function (str) {
return str.replace(/[\-\[\]\/\{\}\(\)\*\+\?\.\\\^\$\|]/g, "\\$&");
},
replaceAll: function (find, replace, str) {
return str.replace(new RegExp(this.escapeRegExp(find), 'g'), replace);
},
startsWith: function (longStr, shortStr) {
return (longStr.match("^" + shortStr) == shortStr)
},
// override search method, count the input length
_startSearch: function (/*String*/ key) {
// If there is not enough text entered, we won't start querying
if (!key || key.length < this.minKeyCount) {
this.closeDropDown();
return;
}
// Deciding if the server needs to be queried
var serverQueryNeeded = false;
if (this.lastServerQueryTerm == null)
serverQueryNeeded = true;
else if (!this.startsWith(key, this.lastServerQueryTerm)) {
// the key does not start with the server queryterm
serverQueryNeeded = true;
}
if (serverQueryNeeded) {
// Creating a query url templated with the autocomplete term
var url = this.replaceAll('${autoCompleteTerm}', key, this.queryURL);
this.store.url = url
// We need to close the store in order to allow the FilteringSelect
// to re-open it with the new query term
this.store.close();
this.lastServerQueryTerm = key;
}
// Calling the super start search
this.inherited(arguments);
}
}
);
});
Notes:
I included some string functions to make it standalone, these should go to their proper places in your JS library.
The JavaScript embedded into the page which uses teh AutoComplete widget:
require([
"dojo/ready",
"dojo/data/ItemFileReadStore",
"gefc/dijit/AutoComplete",
"dojo/parser"
],
function(ready, ItemFileReadStore, AutoComplete) {
ready(function() {
// The initially displayed data (current value, possibly null)
// This makes it possible that the widget does not fire a query against
// the server immediately after initialization for getting a label for
// its current value
var dt = null;
<g:if test="${tenantInstance.technicalContact != null}">
dt = {identifier:"id", items:[
{id: "${tenantInstance.technicalContact.id}",
label:"${tenantInstance.technicalContact.name}"
}
]};
</g:if>
// If there is no current value, this will have no data
var partnerStore = new ItemFileReadStore(
{ data: dt,
urlPreventCache: true,
clearOnClose: true
}
);
var partnerSelect = new AutoComplete({
id: "technicalContactAC",
name: "technicalContact.id",
value: "${tenantInstance?.technicalContact?.id}",
displayValue: "${tenantInstance?.technicalContact?.name}",
queryURL: '<g:createLink controller="partner"
action="listForAutoComplete"
absolute="true"/>?term=\$\{autoCompleteTerm\}',
store: partnerStore,
searchAttr: "label",
autoComplete: true
},
"technicalContactAC"
);
})
})
Notes:
This is not standalone JavaScript, but generated with Grails on the server side, thus you see <g:if... and other server-side markup in the code). Replace those sections with your own markup.
<g:createLink will result in something like this after server-side page generation: /Limes/partner/listForAutoComplete?term=${autoCompleteTerm}
As of dojo 1.9, I would start by recommending that you replace your ItemFileReadStore by a store from the dojo/store package.
Then, I think dijit/form/FilteringSelect already has the features you need.
Given your requirement to avoid a server round-trip at the initial page startup, I would setup 2 different stores :
a dojo/store/Memory that would handle your initial data.
a dojo/store/JsonRest that queries your controller on subsequent requests.
Then, to avoid querying the server at each keystroke, set the FilteringSelect's intermediateChanges property to false, and implement your logic in the onChange extension point.
For the requirement of triggering the server call after a delay, implement that in the onChange as well. In the following example I did a simple setTimeout, but you should consider writing a better debounce method. See this blog post and the utility functions of dgrid.
I would do this in your GSP page :
require(["dojo/store/Memory", "dojo/store/JsonRest", "dijit/form/FilteringSelect", "dojo/_base/lang"],
function(Memory, JsonRest, FilteringSelect, lang) {
var initialPartnerStore = undefined;
<g:if test="${tenantInstance.technicalContact != null}">
dt = {identifier:"id", items:[
{id: "${tenantInstance.technicalContact.id}",
label:"${tenantInstance.technicalContact.name}"
}
]};
initialPartnerStore = new Memory({
data : dt
});
</g:if>
var partnerStore = new JsonRest({
target : '<g:createLink controller="partner" action="listForAutoComplete" absolute="true"/>',
});
var queryDelay = 500;
var select = new FilteringSelect({
id: "technicalContactAC",
name: "technicalContact.id",
value: "${tenantInstance?.technicalContact?.id}",
displayValue: "${tenantInstance?.technicalContact?.name}",
store: initialPartnerStore ? initialPartnerStore : partnerStore,
query : { term : ${autoCompleteTerm} },
searchAttr: "label",
autoComplete: true,
intermediateChanges : false,
onChange : function(newValue) {
// Change to the JsonRest store to query the server
if (this.store !== partnerStore) {
this.set("store", partnerStore);
}
// Only query after your desired delay
setTimeout(lang.hitch(this, function(){
this.set('query', { term : newValue }
}), queryDelay);
}
}).startup();
});
This code is untested, but you get the idea...

Knockout bindings are not working with jsTree

I'm trying to get jsTree (1.0-rc3) working with Knockout.js (2.2.1).
See example jsFiddle: http://jsfiddle.net/adeconsulting/qfr6A/
Note: I've included several JS resources in the Fiddle to match as close as possible my Visual Studio project, in case there's a conflict between libraries which might be causing this problem.
Run the Fiddle and navigate through the jsTree, it's a list of servers by their physical location and type. It helps to have Firebug's console open so you can see the ajax calls and responses. When you click a leaf node, an ajax call is made to retrieve the server details and display a form whose values use Knockout bindings. I hide the form when a non-leaf node is selected.
It works the first time you click a leaf node. After that, Knockout does not update the form for leaf node clicks. However, if you happen to click the Edit button, then all of a sudden the most recent server details ARE displayed.
I'm thinking that there's a conflict between jsTree and Knockout bindings, but don't know where to start troubleshooting what that might be.
Since stackoverflow apparently requires at least one code block, here's the JavaScript portion of the Fiddle:
// Global vars:
var prevJsTreeNodeId = null;
var serverModelBindingsApplied = false;
var serverLoadInProgress = false;
/*
* The knockout.js view model
*/
var ServerViewModel = function () {
// Data
var self = this;
self.IsReadOnly = ko.observable(true); // the form's input mode
self.btnEditSave = ko.observable("Edit"); // the Edit/Save button text
self.Server = ko.observable({}); // the Server object
// Operations
self.setEditable = function () {
self.IsReadOnly(false);
self.btnEditSave("Save");
};
self.setReadOnly = function () {
self.IsReadOnly(true);
self.btnEditSave("Edit");
};
self.doEditSave = function () {
var flag = self.IsReadOnly();
if (flag) {
// switch to Edit mode
self.setEditable();
}
else {
// switch back to readOnly
self.setReadOnly();
}
};
// use ajax to update the knockout.js view model's Server object for the specified server name
self.load = function (serverName) {
if (!serverLoadInProgress) {
serverLoadInProgress = true;
// use ajax to retrieve the server's details
var data = {
json: JSON.stringify({
ServerName: serverName,
PrimaryIP: "1.2.3.4",
BrandDesc: "Dell",
OSDesc: "Windows 2003 Server",
Location: "xyz"
}),
delay: 1
};
$.ajax({
url:"/echo/json/",
data:data,
type:"POST",
success:function(response)
{
console.log(response);
window.ServerViewModelInstance.Server = ko.mapping.fromJS(response);
// apply bindings the first time we retrieve a Server object
if (!serverModelBindingsApplied) {
ko.applyBindings(window.ServerViewModelInstance,
document.getElementById('servercontent'));
serverModelBindingsApplied = true;
}
else {
// hmmm... updating the view model's .Server property doesn't trigger the
// form to be updated, yet if we click the Edit button, the new values
// suddenly appear, so try emulating that here...
self.setReadOnly();
}
}
});
serverLoadInProgress = false;
}
};
}; // ServerViewModel
/*
* document load
*/
$(function () {
// configure the jsTree
$("#divtree")
.jstree({
"themes": { "theme": "default", "dots": true, "icons": true },
"plugins": ["themes", "html_data", "ui", "types"],
"types": {
"type_attr": "tag", // the attribute which contains the type name
"max_depth": -2, // disable depth check
"max_children": -2, // disable max children check
"valid_children": ["root"],
"types": {
"root": {
"valid_children": ["level1"]
},
"level1": {
"valid_children": ["level2"],
"start_drag": false,
"move_node": false,
"delete_node": false,
"remove": false
},
"level2": {
"valid_children": ["leaf"],
// use the theme icon for the level2 nodes
"start_drag": false,
"move_node": false,
"delete_node": false,
"remove": false
},
"leaf": {
"valid_children": "none"
}
}
}
});
// register to receive notifications from #divtree when a jsTree node is selected
$("#divtree").bind("select_node.jstree", function (event, data) {
// data.rslt.obj is the jquery extended node that was clicked
var key = data.rslt.obj.attr("key");
var id = data.rslt.obj.attr("id");
if (id == prevJsTreeNodeId) {
// user clicked the same node, nothing to do
return;
}
prevJsTreeNodeId = id;
// when a jsTree node is selected, reset the knockout.js view model to read only
window.ServerViewModelInstance.setReadOnly();
var idx = key.indexOf("Server");
if (idx === 0) { // "Server|servername"
// show the "servercontent" div
$("#servercontent").show();
// display the server details
var serverName = key.substr(idx + 7, key.length);
window.ServerViewModelInstance.load(serverName);
}
else {
// hide the "servercontent" div
$("#servercontent").hide();
}
});
// hide the "servercontent" div
$("#servercontent").hide();
// instantiate the knockout.js view model
window.ServerViewModelInstance = new ServerViewModel();
}); // document ready
// initialization timer routine to select the main jsTree node
setTimeout(function () {
// open the root node
$.jstree._reference("#divtree").open_node("#root");
}, 500);
Sorry for my bad formatting below - this editor is not my friend... :-/
If I understand you right, the detail panel for a clicked tree node isn't updated with the correct data - right?
Try to do the following:
change:
window.ServerViewModelInstance.Server = ko.mapping.fromJS(response);
to: window.ServerViewModelInstance.Server(response);
(e.g. not overwriting the initial ko.observable which you only binds once, instead updating its values)
and in view where you bind to the observables..
for instance instead of: ... "value: Server.ServerName, ...
change it to: ... "value: Server().ServerName, ...
(e.g executing the function before accessing the property)
It works and updates the form when clicking on a new server name node in the tree (tried in firefox)
A copy of your example with the modified code can be found at: http://jsfiddle.net/RZ92g/2/

Latency issue with Primefaces overlayPanel - loads to lazy

I am using Primefaces 3.2 with jsf 2 and glassfish 3.1.2.
I have a p:dataTable of users containing avatars of the user. Whenever the user moves the mouse over the avatar a p:overlayPanel appears with more information (lazy loaded) on the user, and disappears when the user moves the cursor away - like this:
<p:overlayPanel for="avatar" dynamic="true" showEvent="mouseover" hideEvent="mouseout" ...>
This works very well - as long as the user is "slowhanded". Whenever an user moves the cursor fast above many avatars many of the overlayPanels stay visible.
For example when the user has the cursor over the position where user avatars are displayed and uses the scroll wheel of his mouse to scroll the usertable down or up.
I believe that the overlaypanel starts to load the information dynamically (dynamic="true") from the server when showEvent="mouseover" is dispatched and displays the overlaypanel after the response from the server arrives.
This way it is not possible to detect whether the cursor is already away when the overlaypanel becomes visible - so the hideEvent="mouseout" is never dispatched.
Is there a way to make the primefaces overlaypanel appear directly on mousover, showing a loading gif and update the content into the overlaypanel when the response comes from the server.
Is this a good appraoch or does anyone know any other way to solve this nasty problem?
Thanks Pete
As my first answer is already very long and contains valid information, I decided to open a new answer presenting my final approach.
Im now using Primefaces inheritance pattern making the code alot cleaner. Also I noticed that replacing/overwriting the whole bindEvents function isnt necessary, as we can remove the old event handlers. Finally this code fixs the latest issue experienced: A hide event before ajax arrival.
PrimeFaces.widget.OverlayPanel = PrimeFaces.widget.OverlayPanel
.extend({
bindEvents : function() {
this._super();
var showEvent = this.cfg.showEvent + '.ui-overlay', hideEvent = this.cfg.hideEvent
+ '.ui-overlay';
$(document).off(showEvent + ' ' + hideEvent, this.targetId).on(
showEvent, this.targetId, this, function(e) {
var _self = e.data;
clearTimeout(_self.timer);
_self.timer = setTimeout(function() {
_self.hidden = false;
_self.show();
}, 300);
}).on(hideEvent, this.targetId, this, function(e) {
var _self = e.data;
clearTimeout(_self.timer);
_self.hidden = true;
_self.hide();
});
},
_show : function() {
if (!this.cfg.dynamic || !this.hidden) {
this._super();
}
}
});
Im sorry for the poor formatting: Eclipses fault ;)
Wow, finally after a long debuging session and testing various approaches i recognized that the problem isnt the ajax request but the event handlers itself:
.on(hideEvent, this.targetId, this, function(e) {
var _self = e.data;
if(_self.isVisible()) {
_self.hide();
}
});
As you can see, the widget is just hidden if its visible before. If your moving your mouse out too fast, now two things can happen:
The widget isnt visible at all
The animation is still going on
In this case the event is discarded and the panel stays visible. As animations are queued, one simply has to remove the if statement to fix the issue. I did this by replacing the whole bindEvents method:
PrimeFaces.widget.OverlayPanel.prototype.bindEvents = function() {
//mark target and descandants of target as a trigger for a primefaces overlay
this.target.data('primefaces-overlay-target', this.id).find('*').data('primefaces-overlay-target', this.id);
//show and hide events for target
if(this.cfg.showEvent == this.cfg.hideEvent) {
var event = this.cfg.showEvent;
$(document).off(event, this.targetId).on(event, this.targetId, this, function(e) {
e.data.toggle();
});
}
else {
var showEvent = this.cfg.showEvent + '.ui-overlay',
hideEvent = this.cfg.hideEvent + '.ui-overlay';
$(document).off(showEvent + ' ' + hideEvent, this.targetId).on(showEvent, this.targetId, this, function(e) {
var _self = e.data;
if(!_self.isVisible()) {
_self.show();
}
})
.on(hideEvent, this.targetId, this, function(e) {
var _self = e.data;
_self.hide();
});
}
//enter key support for mousedown event
this.bindKeyEvents();
var _self = this;
//hide overlay when mousedown is at outside of overlay
$(document.body).bind('mousedown.ui-overlay', function (e) {
if(_self.jq.hasClass('ui-overlay-hidden')) {
return;
}
//do nothing on target mousedown
var target = $(e.target);
if(_self.target.is(target)||_self.target.has(target).length > 0) {
return;
}
//hide overlay if mousedown is on outside
var offset = _self.jq.offset();
if(e.pageX < offset.left ||
e.pageX > offset.left + _self.jq.outerWidth() ||
e.pageY < offset.top ||
e.pageY > offset.top + _self.jq.outerHeight()) {
_self.hide();
}
});
//Hide overlay on resize
var resizeNS = 'resize.' + this.id;
$(window).unbind(resizeNS).bind(resizeNS, function() {
if(_self.jq.hasClass('ui-overlay-visible')) {
_self.hide();
}
});
};
Execute this code on load and the issue should be gone.
As your replacing the js code nevertheless, you can use this oppurtunity to implement quite a nice feature. By using timeouts in the event handlers one can easily implement a little delay not just improving usability (no more thousands of popups appear) but also reducing network traffic:
$(document).off(showEvent + ' ' + hideEvent, this.targetId).on(showEvent, this.targetId, this, function(e) {
var _self = e.data;
_self.timer = setTimeout( function(){
if(!_self.isVisible()) {
_self.show();
}
}, 300);
})
.on(hideEvent, this.targetId, this, function(e) {
var _self = e.data;
clearTimeout(_self.timer);
_self.hide();
});
Ofcourse you can use a global variable to control the delay time. If you want a more flexible approach youll have to overwrite the encodeScript method in the OverlayPanelRender to transmit an additional property. You could access it then with _self.cfg.delay. Notice though that youll have to replace the component model OverlayPanel too providing it with an extra attribute.
At the same time I thank you for this brilliant solution I take the opportunity to update it for Primefaces 5.2. In our application the code broke after that upgrade.
Follows the updated code for Primefaces 5.2:
PrimeFaces.widget.OverlayPanel.prototype.bindTargetEvents = function() {
var $this = this;
//mark target and descandants of target as a trigger for a primefaces overlay
this.target.data('primefaces-overlay-target', this.id).find('*').data('primefaces-overlay-target', this.id);
//show and hide events for target
if(this.cfg.showEvent === this.cfg.hideEvent) {
var event = this.cfg.showEvent;
this.target.on(event, function(e) {
$this.toggle();
});
}
else {
var showEvent = this.cfg.showEvent + '.ui-overlaypanel',
hideEvent = this.cfg.hideEvent + '.ui-overlaypanel';
this.target
.off(showEvent + ' ' + hideEvent)
.on(showEvent, function(e) {
clearTimeout($this.timer);
$this.timer = setTimeout(function() {
$('.ui-overlaypanel').hide();
$this.hidden = false;
$this.show();
}, 500);
})
.on(hideEvent, function(e) {
clearTimeout($this.timer);
$this.timer = setTimeout(function() {
// don't hide if hovering overlay
if(! $this.jq.is(":hover")) {
$this.hide();
}
}, 100);
});
}
$this.target.off('keydown.ui-overlaypanel keyup.ui-overlaypanel').on('keydown.ui-overlaypanel', function(e) {
var keyCode = $.ui.keyCode, key = e.which;
if(key === keyCode.ENTER||key === keyCode.NUMPAD_ENTER) {
e.preventDefault();
}
})
.on('keyup.ui-overlaypanel', function(e) {
var keyCode = $.ui.keyCode, key = e.which;
if(key === keyCode.ENTER||key === keyCode.NUMPAD_ENTER) {
$this.toggle();
e.preventDefault();
}
});
};
I also added an extra feature which allows the user to move the mouse over the overlay without hiding it. It should hide when you move the mouse out of it then which I accomplished through:
<p:overlayPanel .... onShow="onShowOverlayPanel(this)" ...>
function onShowOverlayPanel(ovr) {
ovr.jq.on("mouseleave", function(e) {
ovr.jq.hide();
});
}
Hope you enjoy!
It's been a long time, but in case anyone bumps into this problem, a showDelay attribute was added to the overlayPanel to solve this problem starting from Primefaces 6.2. However, it is not in the official documentation for some reason.

Loading script after page fully loaded

I am building a firefox addon that loads javascript at every page load. I'm using progress listener function I found on this page: https://developer.mozilla.org/en/Code_snippets/Progress_Listeners
My problem is that the code seems to execute to early before the page is fully loaded which causes my script to not run. Here is my code.
var PageLoad = {
browser: null,
domain: null,
oldURL: null,
init: function() {
gBrowser.addProgressListener(urlBarListener,Components.interfaces.nsIWebProgress.NOTIFY_LOCATION);
},
uninit: function() {
gBrowser.removeProgressListener(urlBarListener);
},
processNewURL: function(aURI) {
//if (aURI.spec == this.oldURL)
//return;
MyObject.function();
this.oldURL = aURI.spec;
}
};
var urlBarListener = {
locChange: false,
QueryInterface: function(aIID) {
if (aIID.equals(Components.interfaces.nsIWebProgressListener) ||
aIID.equals(Components.interfaces.nsISupportsWeakReference) ||
aIID.equals(Components.interfaces.nsISupports))
return this;
throw Components.results.NS_NOINTERFACE;
},
onLocationChange: function(aProgress, aRequest, aURI) {
PageLoad.processNewURL(aURI);
},
onStateChange: function(aWebProgress, aRequest, aFlag, aStatus) {},
onProgressChange: function(a, b, c, d, e, f) {},
onStatusChange: function(a, b, c, d) {},
onSecurityChange: function(a, b, c) {}
};
window.addEventListener("load",
function() {
PageLoad.init()
}, false);
var MyObject = {
function : function() {
var script = PageLoad.browser.createElement('script');
script.src = 'url_to_script.js';
PageLoad.browser.getElementsByTagName('head')[0].appendChild(script);
}
};
With this code I get this error message on the console:
PageLoad.browser.getElementByTagName("head")[0] is undefined
If I add a timeout then the script does work intermittently but if the page loads slow I get the same error, here's what works sometimes setTimeout(MyObject.function, 1000);
I need a more reliable way of making sure it's executing the script after the page is loaded.
Unrelated, and it doesn't seem to cause any problems but I also see this error message:
gBrowser.addProgressListener was called with a second argument, which is not supported. See bug 608628.
If you want to load javascript at every page load - the best way is subscribing to DOMContentLoaded event:
var MyObject = {
processDOMContentLoaded: function(doc) {
var script = doc.createElement('script');
script.src = 'url_to_script.js';
script.type = 'text/javascript';
doc.getElementsByTagName('head')[0].appendChild(script);
}
};
window.addEventListener('load', function() {
var appcontent = document.getElementById('appcontent');
if(appcontent != null) {
appcontent.addEventListener('DOMContentLoaded', function(event) {
var doc = event.originalTarget;
if(doc instanceof HTMLDocument) {
MyObject.processDOMContentLoaded(doc);
}
}, true);
}
}, false);
Have not tested, but should work.
You are using onLocationChange method - but if you look at how the browser behaves, the location in the address bar changes as soon as a connection with the server is established. You should look at state changes instead:
onStateChange: function(aWebProgress, aRequest, aFlag, aStatus)
{
if ((aFlag & Components.interfaces.nsIWebProgressListener.STATE_STOP) &&
(aFlag & Components.interfaces.nsIWebProgressListener.STATE_IS_WINDOW))
{
// A window finished loading
PageLoad.windowLoaded(aWebProgress.DOMWindow);
}
},
Note that the window that finished loading is explicitly passed to PageLoad.windowLoaded() - you will be receiving notifications from different tabs and you cannot assume that the notification comes from the foreground tab.
As to the warning you are getting, just leave out the second parameter in the call to gBrowser.addProgressListener():
gBrowser.addProgressListener(urlBarListener);
tabbrowser.addProgressListener() only accepts one parameter, unlike nsIWebProgress.addProgressListener() which has a second parameter.
Actually its a great question.
You should use event listener, but carefully, because if you trigger for every page load its can trigger you more than one time (in the worst case about dozens of times).
So how you can do that?
window.addEventListener("load", function load(event){
window.removeEventListener("load", load, false); //remove listener, no longer needed
myExtension.init();
},false);
var myExtension = {
init: function() {
var appcontent = document.getElementById("appcontent"); // browser
if(appcontent){
appcontent.addEventListener("DOMContentLoaded", myExtension.onPageLoad, true);
}
},
onPageLoad: function(aEvent) {
var doc = aEvent.originalTarget; // doc is document that triggered the event
var win = doc.defaultView; // win is the window for the doc
if (doc.nodeName != "#document") return;
if (win != win.top) return;
if (win.frameElement) return;
alert("the main page has been loaded");
},
};
get notice that we check for the type of the trigger every pageload triggering to prevent multi load.
The answers that were provided were acceptable but I found yet another solution that works perfectly.
var PageLoad = {
init: function() {
if(gBrowser) gBrowser.addEventListener("DOMContentLoaded", this.onPageLoad, false);
},
onPageLoad: function(aEvent) {
var doc = aEvent.originalTarget; // doc is document that triggered the event
var win = doc.defaultView; // win is the window for the doc
if (doc.nodeName != "#document") return;
if (win != win.top) return;
if (win.frameElement) return;
MyAddon.function();
}
}
window.addEventListener("load", function load(event){
window.removeEventListener("load", load, false); //remove listener, no longer needed
PageLoad.init();
},false);

How to make "jQuery UI tab" blink/ flash

I want to make "jQuery UI TAB" blink (like notification).
I have diffrent tabs (Inbox | Sent | Important). My timer function checks if there is a new message in inbox, if so, I want the Inbox tab to start blinking/ flashing unless its clicked open.
Have tried diffrent options like .effect(..), .tabs(fx: {..}) but nothing seems to work :(
Any idea if its possible or not?
Yes it's definitely possible.
To give me some practice, I've written a jQuery blinker plugin for you:
jQuery:
(function($){
// **********************************
// ***** Start: Private Members *****
var pluginName = 'blinker';
var blinkMain = function(data){
var that = this;
this.css(data.settings.css_1);
clearTimeout(data.timeout);
data.timeout = setTimeout(function(){
that.css(data.settings.css_0);
}, data.settings.cycle * data.settings.ratio);
};
// ***** Fin: Private Members *****
// ********************************
// *********************************
// ***** Start: Public Methods *****
var methods = {
init : function(options) {
//"this" is a jquery object on which this plugin has been invoked.
return this.each(function(index){
var $this = $(this);
var data = $this.data(pluginName);
// If the plugin hasn't been initialized yet
if (!data){
var settings = {
css_0: {
color: $this.css('color'),
backgroundColor: $this.css('backgroundColor')
},
css_1: {
color: '#000',
backgroundColor: '#F90'
},
cycle: 2000,
ratio: 0.5
};
if(options) { $.extend(true, settings, options); }
$this.data(pluginName, {
target : $this,
settings: settings,
interval: null,
timeout: null,
blinking: false
});
}
});
},
start: function(){
return this.each(function(index){
var $this = $(this);
var data = $this.data(pluginName);
if(!data.blinking){
blinkMain.call($this, data);
data.interval = setInterval(function(){
blinkMain.call($this, data);
}, data.settings.cycle);
data.blinking = true;
}
});
},
stop: function(){
return this.each(function(index){
var $this = $(this);
var data = $this.data(pluginName);
clearInterval(data.interval);
clearTimeout(data.timeout);
data.blinking = false;
this.style = '';
});
}
};
// ***** Fin: Public Methods *****
// *******************************
// *****************************
// ***** Start: Supervisor *****
$.fn[pluginName] = function( method ) {
if ( methods[method] ) {
return methods[method].apply( this, Array.prototype.slice.call( arguments, 1 ));
} else if ( typeof method === 'object' || !method ) {
return methods.init.apply( this, arguments );
} else {
$.error( 'Method ' + method + ' does not exist in jQuery.' + pluginName );
}
};
// ***** Fin: Supervisor *****
// ***************************
})( jQuery );
See it in action here
The plugin and the fiddle are pretty raw in that I haven't tried to integrate with jQuery-ui-tabs. This may be easy or hard, I don't know, but providing each tab is addressable by class or id then it shouldn't be too difficult.
Something you may need to consider is stopping a blinking tab when it is clicked. For this you may wish to call the .blinker('stop') method directly (with a .on('click') handler) or from an appropriate jQuery-ui-tabs callback.
API
The plugin is properly written in jQuery's preferred pattern. It puts just one member in the jQuery.fn namespace and .blinker(...) will chain like standard jQuery methods.
Methods :
.blinker('init' [,options]) : Initialises selected element(s) with blinker behaviour. Called automatically with .blinker(options), or just .blinker() in its simplest form.
.blinker('start') : causes selected element(s) to start blinking between two styles as determined by plugin defaults and/or options.
.blinker('stop') : causes selected element(s) to stop blinking and return to their natural CSS style(s).
Options : a map of properties, which determine blinker styles and timing:
css_0 : (optional) a map of css properties representing the blink OFF-state.
css_1 : a map of CSS properties representing the blink ON-state.
cycle : the blink cycle time in milliseconds (default 2000).
ratio : ON time as a proportion of cycle time (default 0.5).
By omitting css_0 from the options map, the OFF state is determined by the element(s)' natural CSS styling defined elsewhere (typically in a stylesheet).
Default values are hard-coded for css_1.color, css_1.backgroundColor, cycle time and ratio. Changing the default settings programmatically is not handled, so for different default styling the plugin will need to be edited.
jQuery comes by default with a slew of effects to pick from. You can easily use them wherever you see the need for them and they can be applied like so:
$('#newmsg').effect("pulsate", {}, 1000);
Demo
yes... this is what you need...!!!
this is javascript
if(newmessage==true){
$('#chat-86de45de47-tab').effect("pulsate", {}, 1000);
}
i think it's

Resources