Using Cognos Analtyics 11.1.7IF9.
I have a user who, oddly enough, wants Cognos to make his workflow more efficient. (The nerve!) He thinks that if he can use the TAB button to navigate a prompt page, he'll be faster because he never needs to reach for the mouse.
To test this I created a simple report with a very simple prompt page using only textbox prompts. As I tab I notice it tabs to everything in the browser: browser tabs, the address bar, other objects in Cognos, ...even the labels (text items) I created for the prompts. Oh... and yes, at some point focus lands on a prompt control.
Within Cognos, I see that the tab order generally appears to be from the top down. (I haven't tried multiple columns of prompts in a table yet.) I must tab through the visual elements between the prompts. Also, while value prompts get focus, there is no visible indication of this.
Is there a way to set the tab order for the prompts on a prompt page?
Can I force it to skip the non-prompt elements?
Can the prompts be made to indicate that they have focus?
I tagged this question with javascript because I figure the answer will likely involve a Custom Control or a Page Module.
Of course, then I'll need to figure out how all this will work with cascading prompts and conditional blocks.
I found a similar post complaining about this being a problem in Cognos 8. The answer contains no detail. It just says to go to a non-existent web page.
I had the same frustration as your user and I made a solution a while back that could work for you. It's not the most elegant javascript and I get a weird error in the console but functionally it works so I haven't needed to fix it.
I created a custom control script that does 2 things on a prompt page.
First, it removes the ability to "select" text item elements on the page. If you only have text items and prompts on the page it sets it's "Tabindex" to "-1". This allows you to tab from one prompt field to the next without it selecting invisible elements or text elements between prompts.
Secondly, if you press "Enter" on the keyboard it automatically submits the form. I am pasting the code below which you can save as a .js and call it in a custom control on a prompt page. Set the UI Type to "None"
define( function() {
"use strict";
function AdvancedControl()
{
};
AdvancedControl.prototype.initialize = function( oControlHost, fnDoneInitializing )
{
function enterSubmit (e)
{
if(e.keyCode === 13)
{
try {oControlHost.finish();} catch {}
}
};
function setTab () {
let nL = [...document.querySelectorAll("[specname=textItem]")]
//console.log(nL)
nL.forEach((node) =>{
node.setAttribute('tabindex','-1')
})
};
setTab();
let exec_submit = document.addEventListener("keydown", enterSubmit, false);
try {exec_submit;} catch {}
fnDoneInitializing();
};
return AdvancedControl;
});
When you have saved username and password for some site Chrome will autofill that username and password, but if you try to get the value for the password input field it is empty String even though there is value there ******.
If you click somewhere on the page no mater where the value of the input type="password" will be filled.
This is Fiddle user/pass of the structure of the html and the console.log command. It cannot be seen here but it can be reproduced on every page that has login form and the username and password are autofilled on the load of the page. If you inspect the value of the field before clicking anywhere else on the site it will be empty String.
This is not the case in Firefox or Internet Explorer it will fill the value of the input element with the password.
I am using Windows 7 Ultimate 64-bit OS and Google Chrome version is 48.0.2564.97 m
Is this normal behavior, bug or?
UPDATE:
If you click on F5 to reload the page and inspect the password field the value for password will be there. If you click the reload button in Chrome in top left corner the value for the password field will be empty string.
This seems to be a bug in Chrome. When Chrome auto-fills a password on an initial page load (but not a refresh), the value appears in the form field on-screen, but querying passwordField.value in Javascript returns an empty string. If you depend on seeing that value in Javascript, this prevents you from doing so. Once the user does any other action on the page, such as clicking anywhere on the page, the value suddenly becomes visible to Javascript.
I'm not actually 100% sure if this is a bug, or if there is a security reason for doing this such as preventing a hidden frame from stealing your password by tricking the browser into filling it in.
A workaround that we have used is to detect the background color change that Chrome makes to fields that it has auto-filled. Chrome colors the background of auto-filled fields yellow, and this change is always visible to Javascript even when the value is not. Detecting this in Javascript lets us know that the field was auto-filled with a value, even though we see the value as blank in Javascript. In our case, we have a login form where the submit button is not enabled until you fill in something in the password field, and detecting either a value or the auto-fill background-color is good enough to determine that something is in the field. We can then enable the submit button, and clicking the button (or pressing enter) instantly makes the password field value visible to Javascript because interacting with the page fixes the problem, so we can proceed normally from there.
Working Answer as of July 8, 2016
Adam correctly stated this is a bug (or intended behavior). However, none of the previous answers actually say how to fix this, so here is a method to force Chrome to treat the autocompleted value as a real value.
Several things need to happen in order, and this needs to only run in Chrome and not Firefox, hence the if.
First we focus on the element. We then create a new TextEvent, and run initTextEvent, which adds in a custom string that we specify (I used "#####") to the beginning of the value. This triggers Chrome to actually start acting like the value is real. We can then remove the custom string that we added, and then we unfocus.
Code:
input.focus();
var event = document.createEvent('TextEvent');
if ( event.initTextEvent ) {
event.initTextEvent('textInput', true, true, window, '#####');
input.dispatchEvent(event);
input.value = input.value.replace('#####','');
}
input.blur();
Edit August 10, 2016
This only works right now in Chrome on Windows and Android. Doesn't work on OSX. Additionally, it will stop working at all in Sept 2016, according to:
https://www.chromestatus.com/features/5718803933560832
Also, I've opened a Chromium ticket.
https://bugs.chromium.org/p/chromium/issues/detail?id=636425
As of August 12, a member of the Chrome team said on the above ticket that the behavior won't be changing because they don't consider it a bug.
Long-term Work-Around Suggestion:
That said, the current behavior has been tweaked from when it was first implemented. The user no longer has to interact with the password input for the value to be reported. The user now just needs to interact (send a mouse or keyboard event) with any part of the page. That means that while running validation on pageload still won't work, clicking on a submit button WILL cause Chrome to correctly report the password value. The work-around then, is to revalidate all inputs that might be autocompleted, if that is what you are trying to do, on submit.
Edit December 13, 2016:
A new Chromium ticket has been opened and is being received better. If interested in changing this behavior of Chrome's, please star this new ticket:
https://bugs.chromium.org/p/chromium/issues/detail?id=669724
Continuing from what Kelderic said, here's my work around. Like a lot of people, I don't need the actual password value. I really just need to know that the password box has been autofilled, so that I can display the proper validation messages.
Personally, I would not use suggested solution to detect the background color change cause by Chrome's autofill. That approach seems brittle. It depends on that yellow color never changing. But that could be changed by an extension and be different in another Blink based browser (ie. Opera). Plus, there's no promise Google wont use a different color in the future. My method works regardless of style.
First, in CSS I set the content of the INPUT when the -webkit-autofil pseudo-class is applied to it:
input:-webkit-autofill {
content: "\feff"
}
Then, I created a routine to check for the content to be set:
const autofillContent = `"${String.fromCharCode(0xFEFF)}"`;
function checkAutofill(input) {
if (!input.value) {
const style = window.getComputedStyle(input);
if (style.content !== autofillContent)
return false;
}
//the autofill was detected
input.classList.add('valid'); //replace this. do want you want to the input
return true;
}
Lastly, I polled the input to allow the autofill time to complete:
const input = document.querySelector("input[type=password]");
if (!checkAutofill(input)) {
let interval = 0;
const intervalId = setInterval(() => {
if (checkAutofill(input) || interval++ >= 20)
clearInterval(intervalId);
}, 100);
}
It is amazing that in 2021 this has not been solved in Chrome yet, I have had issue with autocomplete since 2014 and still nothing.
Chrome functionality autocomplete is misleading for the user, I do not know what are they trying to achieve but does not look good.
As it is now, form appears showing auto-completed text (user/email/pass) to the user, but in the background html - values are not inside of the elements.
As values are not in fields custom validation will disable submit button.
Script that checks fields values will say value is null, which is even more confusing for the user as s/he can see text is there, and can assume it is valid, leading to confusing delete-one insert one character. (Embarrassingly, I have to admit I did not know that you need to click in the body of the HTML, so I wonder how many users don not know the same)
In my case I wanted to have empty field always and then fount out it is just needlessly spent time to make it work.
If we try autocomplete=off we will discover that it is not working. And to validate fields and let say enable button we need to do some trickery.
(Have in mind that I have tried autocomplete=password new-password) and other type of Hocus-Pocus trickery from official resource.
At the end I have done this.
<script>
$('#user').value = ' '; //one space
$('#pass').value = ' '; // one space - if this is empty/null it will autopopulate regardless of on load event
window.addEventListener('load', () => {
$('#user').value = ''; // empty string
$('#pass').value = ''; // empty string
});
</script>
So, it will blink for a split second in some cases in password field with * not ideal but :/ ...
Here's my solution to this issue:
$(document).ready(function(){
if ( $("input:-webkit-autofill").length ){
$(".error").text("Chrome autofill detected. Please click anywhere.");
}
});
$(document).click(function(){
$(".error").text("");
});
Basically, clicking makes the input visible to the user, so I ask the user to click and when they do, I hide the message.
Not the most elegant solution but probably the quickest.
$(document).ready
does not wait for autofill of browser, it should be replaced by
$(window).on("load", checkforAutoFill())
Another option as of Dec. 16 / Chrome 54
I can't get the value of the password field, but, after "a short while", I can get the length of the password by selecting it, which is sufficient for me to enable the submit button.
setTimeout(function() {
// get the password field
var pwd = document.getElementById('pwd');
pwd.focus();
pwd.select();
var noChars = pwd.selectionEnd;
// move focus to username field for first-time visitors
document.getElementById('username').focus()
if (noChars > 0) {
document.getElementById('loginBtn').disabled = false;
}
}, 100);
The workaround specified by Adam:
... detect the background color change that Chrome makes to fields that it has auto-filled. Chrome colors the background of auto-filled fields yellow, and this change is always visible to Javascript even when the value is not. Detecting this in Javascript lets us know that the field was auto-filled with a value, even though we see the value as blank in Javascript
I did like this:-
getComputedStyle(element).backgroundColor === "rgb(250, 255, 189)"
where rgb(250, 255, 189) is the yellow color Chrome applies to auto filled inputs.
I have found a solution to this issue that works for my purposes at least.
I have a login form that I just want to hit enter on as soon as it loads but I was running into the password blank issue in Chrome.
The following seems to work, allowing the initial enter key to fail and retrying again once Chrome wakes up and provides the password value.
$(function(){
// bind form submit loginOnSubmit
$('#loginForm').submit(loginOnSubmit);
// submit form when enter pressed on username or password inputs
$('#username,#password').keydown(function(e) {
if (e.keyCode == 13) {
$('#loginForm').submit(e);
return false;
}
});
});
function loginOnSubmit(e, passwordRetry) {
// on submit check if password is blank, if so run this again in 100 milliseconds
// passwordRetry flag prevents an infinite loop
if(password.value == "" && passwordRetry != true)
{
setTimeout(function(){loginOnSubmit(e,true);},100);
return false;
}
// login logic here
}
Just wrote an angular directive related to this. Ended up with the following code:
if ('password' == $attrs.type) {
const _interval = $interval(() => { //interval required, chrome takes some time to autofill
if ($element.is(':-webkit-autofill')) { //jQuery.is()
//your code
$interval.cancel(_interval);
}
}, 500, 10); //0.5s, 10 times
}
ps: it wont detect 100% of the times, chrome might take longer than 5 seconds to fill the input.
Chrome's intended behavior is that an auto-filled password has an empty value in the DOM until the user interacts with the frame in some way, at which point chrome actually populates the value. Until this point any client side validation or attempt to ajax submit the form will see the password as empty.
This 'populate password value on frame interaction' behavior is inconsistent. I've found when the form is hosted in a same-origin iframe it only operates on the first load, and never on subsequent loads.
This is most evident on ajax forms where the autocomplete password populates on first load, however if that password is invalid and the ajax submission re-renders the form DOM, the autocompleted password re-appears visually but the value is never populated, irrespective of interaction.
None of the workarounds mentioned such as triggering blur or input events worked in this scenario. The only workaround I've found is to reset the password field value after the ajax process re-renders the form, e.g.:
$('input[type="password"]').val("");
After the above, Chrome actually autocompletes the password again but with the value actually populated.
In my current use case I'm using ASP.NET's Ajax.BeginForm and use the above workaround in the AjaxOptions.OnSuccess callback.
$element.is("*:-webkit-autofill")
works for me
With Angular, the new behaviour in Chrome (only allowing autofilled values to be read after the user has interaction with the page) manifests itself as an issue when you're using Angular's validation functionality in certain scenarios (for e.g using standard method/action attributes on the form). As the submit handler is executed immediately, it does not allow the form validators to capture the autofilled values from Chrome.
A solution I found for this to explicitly call the form controllers $commitViewValue function in the submit handler to trigger a revalidation before checking form.$valid or form.invalid etc.
Example:
function submit ($event) {
// Allow model to be updated by Chrome autofill
// #see http://stackoverflow.com/questions/35049555/chrome-autofill-autocomplete-no-value-for-password
$scope.loginModule.$commitViewValue();
if ($scope.loginModule.$invalid) {
// Disallow login
$scope.loginModule.$submitted = true;
$event.preventDefault();
} else {
// Allow login
}
}
Although this is working for us so far, I would be very interested if someone has found another, more elegant work around for the issue.
var txtInput = $(sTxt);
txtInput.focus();
txtInput.select();
This solution worked in my case.
Using jQuery 3.1.1.
If you want make input to be seen as fulfilled, try to trigger blur on it:
$('input[type="password"]').blur();
The autocomplete feature has successfully disabled.
It Works!
[HTML]
<div id="login_screen" style="min-height: 45px;">
<input id="password_1" type="text" name="password">
</div>
[JQuery]
$("#login_screen").on('keyup keydown mousedown', '#password_1', function (e) {
let elem = $(this);
if (elem.val().length > 0 && elem.attr("type") === "text") {
elem.attr("type", "password");
} else {
setTimeout(function () {
if (elem.val().length === 0) {
elem.attr("type", "text");
elem.hide();
setTimeout(function () {
elem.show().focus();
}, 1);
}
}, 1);
}
if (elem.val() === "" && e.type === "mousedown") {
elem.hide();
setTimeout(function () {
elem.show().focus();
}, 1);
}
});
To me none of this solutions seemed to work.
I think this is worth mentioning that if you want to use it for CSS styling you sould use -webkit-autofill property like this:
input:-webkit-autofill~.label,
input:-webkit-autofill:hover~.label,
input:-webkit-autofill:focus~.label
input:focus~.label,
input:not(.empty)~.label {
top: -12px;
font-size: 12px;
color: rgba(0, 0, 0, .4);
font-weight: 600
}
My solution comparing my css to the chrome autocomplete color...
$('input, select, textarea').each(function(){
var inputValue = $(this).val();
if ( inputValue != "" || $(this).css("background-color") != "rgba(255, 255, 255, 0)") {
$(this).parents('.form-group').addClass('focused');
}
});
I tried all the solutions and wasn't working for me so i came up with this.
My problem is i have an input that move the placeholder top when it is filled, off course this is not working when Chrome autofill it.
Only tested in Chrome :
setTimeout(function () {
var autofilled = document.querySelectorAll('input:-webkit-autofill');
for (var i = 0; i < autofilled.length; i++) {
Do something with your input autofilled
}
}, 200);
My version is 95.0.4638.69
I'm facing a similar issue and I solved it by changing my form's name from "login-form" to another name which does not mean anything and solve it. Reason why I didn't remove name attribute is because if I remove name attribute Chrome will look up to id attribute and do the same thing.
Option using onanimationstart event (ReactJs) - Mar 22
I could avoid the needing of verifying periodically if the input was autofilled, as described above using setInterval, by taking advantage of the onanimationstart event. I don't know if it will work in every case, but definitely did the trick for me.
I'll provide a code sample in ReactJs, it may be explanatory enough to be transposed to another context.
First of all, is necessary to add in your input the onAnimationStart property, in such a way that the event is passed as parameter to your function, as following below.
<input
className={componentClass}
placeholder={placeholder}
onChange={handleChange}
onFocus={onFocus}
onMouseEnter={onHover}
onMouseLeave={onHover}
onBlur={onBlur}
disabled={disabled}
name={name}
value={value}
onAnimationStart={e => this.onAnimationStart(e)}
/>
Then let's proceed to the onAnimationStart function body.
onAnimationStart(event) {
// on autofill animation
if (event.animationName === 'onAutoFillStart') {
event.target?.labels[0].classList.add('grm-form__isAutofilled');
}
}
First I verified if the animation name was actually the auto-fill animation, and then I added a class to the first label of my input, this is my use case but can be adapted to solve different problems.
Just set the autocomplete attribute to username for the username field and new-password for the password field;
<input type="text" id="username" autocomplete="username">
<input type="password" id="password" autocomplete="new-password" >
You mentioned:
If you click somewhere on the page no matter where the value of the input type="password" will be filled.
Which is why I simply use $('body').click(); to simulate this first click, after which the value is available in JavaScript.
Also, I set autocomplete="new-password" on my signup form password field, so that the field is not autofilled and users have to fill in a new password.
See this Google Developers page for more information.
It's not a bug. It's a security issue. Imagine if one could just use javascript to retrieve autofilled passwords without the users' acknowledgment.
My goal is to show a different first page depending on whether the user is logged in or not. The login check happens using a synchronous Ajax call, the outcome of which decides whether to show a login dialog or the first user page.
The normal way to do this would be to set $.mobile.autoInitialize = false and then later on initialize programmatically, as described in the answer to this question. For some reason this won't work, instead another page gets loaded every single time.
I decided to give up on this way and try out a different parcour. I now use a placeholder, empty startup page that should be shown for as long as the login check takes. After the login check it should automatically change. This is done by calling a function that performs the ajax call needed for authentication on the pagechange event that introduces this startup page. The function takes care of changing to the outcome page as well.
The trick is that it doesn't quite do that.. Instead it shows the correct page for just a short time, then changes back to the placeholder. Calling preventDefault in pagechange didn't prevent this, as described in the tutorial on dynamic pages. Adding a timer fixed this, leading me to think that the placeholder wasn't quite finished when pageshow got fired (as per this page on page events), or some side-effect of the initial page load still lingered.
I'm really clueless as to how to fix this seemingly trivial, yet burdensome problem. What causes this extra change back to the initial page? Also, if my approach to intercepting the initial page load is wrong, what would be the correct approach instead?
I use jQuery Mobile 1.4.0 and jQuery 1.10.2 (1.8.3 before).
EDIT: Below is the code to my last try before I posted the question here. It does not work: preventDefault does not prevent the transition to the placeholder page.
$(document).on("pagebeforechange", function(e, data) {
if (typeof(data.options.fromPage) === "undefined" && data.toPage[0].id === "startup") {
e.preventDefault();
initLogin();
}
});
function initLogin() {
// ... Login logic
if (!loggedIn) // Pseudo
$('body').pagecontainer("change", "#login", {});
}
If you're using a Multi-page model, you can prevent showing any page on pagebeforechange event. This event fires twice for each page, once for the page which is about to be hidden and once for the page which is about to be shown; however, no changes is commenced in this stage.
It collects data from both pages and pass them to change page function. The collected data is represented as a second argument object, that can be retrieved once the event is triggered.
What you need from this object is two properties, .toPage and .options.fromPage. Based on these properties values, you decide whether to show first page or immediately move to another one.
var logged = false; /* for demo */
$(document).on("pagebeforechange", function (e, data) {
if (!logged && data.toPage[0].id == "homePage" && typeof data.options.fromPage == "undefined") {
/* immediately show login dialig */
$.mobile.pageContainer.pagecontainer("change", "#loginDialog", {
transition: "flip"
});
e.preventDefault(); /* this will stop showing first page */
}
});
data.toPage[0].id value is first page in DOM id.
data.options.fromPage should be undefined as it shouldn't be redirected from another page within the same webapp.
Demo
I'm undergoing the same problem as the one described by #RubenVereecken, that is, a coming back to the initial page once the cange to my second page has completed. In fact, he posed the question "What causes this extra change back to the initial page?" and it hasn't been replied yet.
Unfortunately, I don't know the reason since I haven't found how the page-event order works in JQM-1.4.2 yet, but fortunately, the workaround suggested by #Omar is working for me.
It's not exactly the same code but the general idea works at the time of preventing a coming back to the initial page. My code is as follows:
$(document).on("pagebeforechange", function(event, data) {
if ( typeof (data.toPage) == "string") {
if (data.toPage.indexOf("#") == -1 && typeof (data.options.fromPage[0].id) == string") {
event.preventDefault();
}
}});
The condition data.toPage.indexOf("#") == -1 is because I checked that all the undesired coming-backs to the initial page were happening when the property '.toPage' was set to something like [http://localhost/.../index.html].