Deploying OnSend Outlook Add-ins?

If you have been developing OfficeJS add-ins for Outlook Online (or Exchange 2016 Outlook Web Access) you might not know that you now have the ability to create OnSend add-ins. That is add-ins that can trap the send of an email, test for conditions (such as getBodyAsync() or getSubjectAsync() to look for content) and if they fail, prevent the message from being sent. I recently blogged about it here:

The Most Anticipated OfficeJS Feature is Here

First thing to note about these add-ins is that they cannot be created for or uploaded into the store. The second thing to note is that they MUST be configured first in order to run. The third thing to note is that if you are building an on-prem Exchange based solution, you have to have Exchange 2016 CU6 installed and a bit more. The full details of how to configure the environments (either Office Online or On-prem) are detailed here:

On send feature for Outlook add-ins

If you look this over, there is quite a lot to do. How would you like one single PowerShell command to type? Maybe something like this:


Set-OwaOnSendPolicy

wlEmoticon-hotsmile.png Well, I just published the first version of this script on GitHub:

https://github.com/davecra/Set-OwaOnSendPolicy

All the details you need to know about it are there in the README.MD. However, if you have any questions or issues, please let me know.

 

OfficeJS.dialogs v1.0.7 Published

So, I kind of skipped v1.0.6, which I promised in this post yesterday. There were some unchecked changes on a different laptop that I forgot in v1.0.6, so I published v1.0.7 this morning with all the updates. Some notable update:

  • Moved from CloseDialog() to CloseDialogAsync() on all the dialogs using the original close method. This is to workaround the issue described in the post yesterday.
  • Fixed dialog flashing by hiding the body until the form is ready to be shown. Before you might get a flash of all the controls on the form as they were not hidden before.
  • Internal clean-up and message handling. In the message pump between the dialogs.html and dialogs.js, I cleaned up the pump to use internally standardized messages.
  • Fixed some JSDoc information that was wrong. And there is likely still more. wlEmoticon-disappointedsmile.png
  • Fixed some things that were not working correctly with the progress dialog.
  • Small tweeks and comments, code clean-up

The latest version is now on GitHub. I have also completed all the documentation and updated it to the latest usage information, code examples and screenshots.

I also published the latest version to NPM. Simply type this in you code terminal window:

npm -install officejs.dialogs

CDN Update

I still have a published CDN for OfficeJS.dialogs, but at this point I am going to suggest against using it. I have encountered a number of issues with my message pump when loaded cross-domain. Specifically:

  • you will be able to show a MessageBox, but the Update() method will not work
  • you will be able to show a Progress dialog, but the Update() method will not change the dialog. The progress bar will never advance.

What is happening is that I was going to need for **you** the  developer to provide me your own proxyHTML file (pointing to my CDN/proxy.js library) that I will do the following:

  • Add an iframe, load my html file in that iFrame, post the dialog settings to the iFrame where my dialog in the iFrame gets it, and writes it to localStorage on the CDN domain.
  • Window.replace the contents with my dialogs.html from the CDN. And since the localStorage now has the dialog settings, configure the dialog.
  • So far at this point everything worked. From HERE is where I am pushing my CORS and JavaScript to the edge of sanity…
  • I then create an hidden iFrame in my dialog and then load your proxy html file. I attach to the body change event inside the iFrame.
  • I then start a message pump in the proxy looking for changes to localStorage.
  • If a new message comes from your code with an Update() on MessageBox or Progress, I then get it from the localStorage in a message and write it to the body, therefore updating the iFrame in my dialog, thereby triggering the change event I latched on to, allowing me to write that value to the CDN localStorage where the message pump in dialog will get the message and update the form.

That hurt my brain writing that out. And maybe that is more for my posterity since it took me Binging the Internet to death (I don’t use to the G word folks wlEmoticon-hotsmile.png), just to get that far. My pretzel, I mean head hurts.

Bottom line, I want to get OfficeJS.dialogs to work cross-domain. I am not sure if the above method is the right approach. I have asked far brighter minds than my own and watched Autie Anne’s Syndrome set in almost immediately. So, if any of you have some ideas on how I can establish 2-way communication between domains like this (without using some crazy huge library), please let me know. wlEmoticon-smile.png

OfficeJS: Second Dialog Does not Display

I have been spending a lot of time in the Officeui.dialog lately. One of my customers has been too and it has been an adventure working out exactly the best way to get messages displayed while running code in the background asynchronously. wlEmoticon-disappointedsmile.png

I am not sure if this problem is limited to the online version of Outlook, but this is where I have been seeing the problem (and where I have spent virtually all of my time). If my code tried to open two dialogs (using Office.context.ui.displayDialogAsync()) one right after the other, the second dialog would not ever be displayed. If I waited a period and then tried again, it would. But we don’t want that. We want boom-boom, dialogs. When I looked at the console, I would see an error like the following:

Uncaught TypeError: Cannot read property ‘addEventHandler’ of undefined

Or, if I read the error object from the displayDialogAsync() function/asyncResult, I would see this:

{name: “Display Dialog Error”, message: “The operation failed because this add-in already has an active dialog.“, code: 12007}

Here is an example of some code that will reproduce the issue:

var i = 0;
function displayDialog() {
  var url = "https://localhost:3000/test.html";
  Office.context.ui.displayDialogAsync(url,{height:20, width:30, displayInIframe:true},
      function (asyncResult) {
          var dialog = asyncResult.value; // get the dialog
          var error = asyncResult.error;
          if(dialog == undefined && error.code > 0) {
            // log the error
            console.log(error.message);
          } else {
            // attache the events
            dialog.addEventHandler(Office.EventType.DialogEventReceived, function (arg) {
              // close this dialog, open the next
              dialog.close();
              i++;
              if(i<4) {
                displayDialog();
              }
            });
            dialog.addEventHandler(Office.EventType.DialogMessageReceived, function (arg) {
              // close this dialog, open the next
              dialog.close();
              i++;
              if(i<4) {
                displayDialog();
              }
            });
          }
      });
}

Notice I had used dialog.close(), but it did not work as designed. What I believe is happening is that the previous dialog is still in memory and has not been cleaned up. What needs to likely happen is a closeAsync().

In order to resolve this, I created the following function: dialogCloseAsync(). This works by issuing the close() and then attempting to add an event handler to the dialog in ansyc (setTimeout) loop. When it errors, we trap the error and issue the async callback. It is a bit ugly as we are trapping an error to get around the problem, but this was the only way I could find a way around the problem. Here is what the function looks like:

/**
 * Closes the currently open dialog asynchronously.
 * This has an ugly workaround which is to try to set a new
 * event handler on the dialog until it fails. When it failed
 * we know the original dialog object was destroyed and we
 * can then proceed. The issue we are working around is that
 * if you call two dialogs back to back, the second one will
 * likely not open at all.
 * @param {Office.context.ui.dialog} dialog The dialog to be closed
 * @param {function()} asyncResult The callback when close is complete
 */
 function dialogCloseAsync(dialog, asyncResult){
    // issue the close
    dialog.close();
    // and then try to add a handler
    // when that fails it is closed
    setTimeout(function() {
        try{
            dialog.addEventHandler(Office.EventType.DialogMessageReceived, function() {});
            dialogCloseAsync(dialog, asyncResult);
        } catch(e) {
            asyncResult(); // done - closed
        }
    }, 0);
}

I had been encountering this issue with different systems when developing the OfficeJS.dialogs library and had tried to set a timeout before I showed each dialog. That worked on some systems, but on others the timeout needed to be longer. So, setting a default timeout did not work. Using this in the original sample, provided above, the code would look like this:

var i = 0;
function displayDialog() {
  var url = "https://localhost:3000/test.html";
  Office.context.ui.displayDialogAsync(url,{height:20, width:30, displayInIframe:true},
      function (asyncResult) {
          var dialog = asyncResult.value; // get the dialog
          var error = asyncResult.error;
          if(dialog == undefined && error.code > 0) {
            // log the error
            console.log(error.message);
          } else {
            // attache the events
            dialog.addEventHandler(Office.EventType.DialogEventReceived, function (arg) {
              // close this dialog, open the next
              dialogCloseAsync(dialog, function() {
                i++;
                if(i<4) {
                  displayDialog();
                }
              });
            });
            dialog.addEventHandler(Office.EventType.DialogMessageReceived, function (arg) {
              // close this dialog, open the next
              dialogCloseAsync(dialog, function() {
                i++;
                if(i<4) {
                  displayDialog();
                }
              });
            });
          }
      });
}

As I found this workaround, I have updated OfficeJS.dialogs to use dialogCloseAsync(). Now, the MessageBox, Wait and Form objects will use closeDialogAsync() commands to replace the original closeDialog() commands I provided previously. I will be blogging about the updates to v1.0.6, shortly. wlEmoticon-hotsmile.png

Documentation, documentation, documentation…

I have published working documentation in GitHub for OfficeJS.dialogswlEmoticon-hotsmile.png

If you click that link, you will see the README.md at the bottom of the repository has all the usage details, some sample code and example images of each dialog type supported in the library. But, you can also see a prettier version here:

https://davecra.github.io/OfficeJS.dialogs/

I have never used GitHub for documentation before. It uses a format called Markdown. It is actually very easy to learn and use. I spent about 3 hours and knocked out that document you see and it has:

  • Linked images from this blog
  • A Table of Contents with hyperlinks to different sections
  • Inline hyperlinks to different places in the documentation
  • A sorts of formatting including Headings.

Anyway, as I was building that documentation, I found a few minor bugs – and actually one really big one. I will likely be posting v1.0.6 pretty soon.

The big one: If you try to use the CDN for OfficeJS.dialogs, you will not see anything. The reason is that the HTML page that I use is not in “your” domain. For the Office.context.ui.displayDialogAsync() method to work, the URL to show the INITIAL file, needs to be from the same domain as your code. I totally face palmed wlEmoticon-confusedsmile.png on this one. So, I am going to provide a method:

OfficeJS.dialogs.Initialize(url);

You will call this first if you will be using the CDN. The URL will be an HTML page on your site that has some specific JavaScript (I will provide) to initially load your page, but then update with the OfficeJS.dialogs.html from the CDN. Or, at least that is the idea.

In the meantime, I suggest using NPM to install locally to your solution as per my blog post here:

npm load officejs.dialogs

OfficeJS.dialogs Updated (v1.0.5)

I have been working hard on my OfficeJS.Dialogs library and just published version 1.0.5. You can get it from NPM and GitHub. See my previous post for more information on how to do this.

I have added a few new features:

  • A simple Alert.Show() method that displays a simple OK box. For those times you want to just simply pop up a quick notification to the user.
  • A Progress.Show() that displays a progress bar. This allows for you to show the progress bar and then issue Progress.Update() to move the progress bar along. When you are done you call Progress.Complete().
  • A Wait.Show() dialog that will show an indeterminate spinner. This form will remain up until you issue a Wait.CloseDialog().
  • New UpdateMessage() and Update() methods were added to the MessageBox. This was done to allow you to quickly ask a lot of questions of the user in one instance of the dialog, without giving the user back to the application for a second while the new dialog is rendered. UpdateMessage() will just update the message but keep all the buttons the same, but you will specify a new callback. Update() will allow you to fundamentally change all the settings the MessageBox (buttons, icon, caption, text and all), plus a new callback function.
  • Behind the scenes I made some improvements/bug fixes:
    • If you try to show two dialogs too quickly, nothing will happen. So I added a half-second delay between dialog displays to make sure you never get an overlap.
    • You will get an error message in your callback if more than one dialog is attempted to be opened at once.
    • “Window Messaging” has been setup with Progress and MessageBox to allow the parent and the dialog to pass messages back and forth. It involves using setTimeout().

For those interested in the last item, here is what that look like:

        /**
        * Handles messages coming from the parent
        */
        function startMessageHandler() {
            setTimeout(function() {
                var message = localStorage.getItem("dialogMessage");
                localStorage.setItem("dialogMessage", ""); // clear the message
                if(message !== undefined && message !== null && message != "")
                {
                    var msg = JSON.parse(message);
                    if(msg.message == "update") {
                        // update the form
                        updateForm(msg.settings);
                    } else if(msg.message == "close") {
                        // do nothing special here
                        return; // stops the message pump
                    } else if(msg.message == "progress") {
                        if(msg.settings.Number > 100) return;
                        $("#bar").prop("value",msg.settings.Number);
                    }
                }
                startMessageHandler(); // call again
            }, 0);
        }

Here is an example of how to use the Progress dialog. The method signature is like this:

Progress.Show( [message], [start], [max], [completeCallback], [cancelCallback] )

  • The message is the message the user see’s when the dialog is opened.
  • The start is the number you want the progress bar to start at. Usually this should just be zero (0).
  • The max is the number you want the Progress bar to end at. Usually this should be 100. But it can be any number you want. If you have 5 steps to perform in the background while this dialog is up, you can set this to 5.
  • The completeCallback is your callback function to be called when your code calls the Progress.Compelte().
  • The cancelCallback is what gets called when the user presses the Cancel button on the form.

By itself, this will do nothing. You will have to call the Progress.Update() command in order to move the progress bar, or update the message to the user. Here is the method signature for the Update method:

Progress.Update( [amount], [message] )

  • The amount is how much you want the progress bar to move. If you do not specify an amount, an amount of 1 is assumed.
  • The message is a new message to provide the progress bar. If you want to update the message and do not want to increment the progress bar, specify an amount of zero (0).

Once you are all done with the Progress dialog, you issue a Progress.Complete() call. There are no parameters to it. Once called, your completeCallback in the Progress.Show() call will then be executed.

Here is an example:

// reset first to make sure we get a fresh object
Progress.Reset();
// display a progress bar form and set it from 0 to 100
Progress.Show("Please wait while this happens...", 0, 100, function() {
    // once we are done - when your code
    // calls Progress.Complete()
    Alert.Show("All done folks!");
  }, function() {
    // this is only going to be called if the user cancels
    Alert.Show("The user cancelled");
});
doProgress();

function doProgress() {
  // increment by one, the result that comes back is
  // two pieces of information: Cancelled and Value
  var result = Progress.Update(1);
  // if we are not cancelled and the value is not 100%
  // we will keep going, but in your code you will
  // likely just be incrementing and making sure
  // at each stage that the user has not cancelled
  if(!result.Cancelled && result.Value <= 100) {     setTimeout(function() {       // this is only for our example to       // cause the progress bar to move       doProgress();     },100);   } else if(result.Value >= 100) {
    Progress.Compelte(); // done
  }
};

That example also uses the new Alert dialog. This one is very simple:

Alert.Show ( [message] )

This next example uses the new Wait dialog, which is much simpler to implement. Here is the method signature:

Wait.Show( [message], [showCancel], [cancelCallback] )

  • The message is the message you want to show the user. If you specify null, it will appear as simply “Please wait…”
  • The showCancel flag if set will allow the user to see a Cancel button.
  • The cancelCallback function is only valid if the showCancel option is true. When the user presses cancel, this function gets called.

When you are ready to close the Wait dialog, you issues a Wait.CloseDialog(). Here is an example:

  Wait.Show(null, true, function() {
    Alert.Show("The user cancelled.");
  });
  setTimeout(function(){
    Wait.CloseDialog();
    Alert.Show("Done!");
  }, 15000);

If you have some suggestions for some things you would like to see added to this library, please add a comment below or reach out to me on LinkedIn or Twitter. Some ideas I will be working on:

  • Allow you to call another dialog type without having the close the dialog.
  • A selection dialog, where you have a dropdown list of a listbox where you wan select (or multi-select) items.
  • An option to resize forms.
  • An option to use the message handler in your own custom form – minimal code

VSCode: error TS5042: Option ‘project’ cannot be mixed with source files on a command line.

I have been using VSCode a lot lately. I have recently started delving into TypeScript and wanted to use VSCode to begin developing my first project. I was following a tutorial on Lynda.com and for the life of me could not get it to “transpile” my TS project. I setup everything according to this site and it failed. What I kept getting was this error:

error TS5042: Option ‘project’ cannot be mixed with source files on a command line.

This error is all over the web and not a single solution helped. I am guessing it is caused by the fact I have VS2015, VS2017, and VSCode installed on my box. But, this is the weird thing, if I go to the PowerShell in the Terminal window and paste the EXACT command that VSCode uses… IT WORKS!!! What? I started puling my hair out at this point…

Ok, now that really sucked. So, I got fed up and just build my own tasks.json using the “Other” option and here is what I entered:

{
    // See https://go.microsoft.com/fwlink/?LinkId=733558
    // for the documentation about the tasks.json format
    "version": "2.0.0",
    "tasks": [
        {
            "taskName": "DaveCra TSC Task for VSCode",
            "command": "tsc -p tsconfig.json",
            "type": "shell",
            "problemMatcher": [
                "$tsc"
            ],
            "group": {
                "kind": "build",
                "isDefault": true
            }
        }
    ]
}

Once I set this up, it was the default build task and a CTRL+SHIFT+B ran fine. Right after that I was able to go to the PowerShell terminal and type “node ” and it ran like a charm. wlEmoticon-hotsmile.png

Goodbye ModNotebooks, Hello OfficeLens

I like taking notes, the old fashioned way (by hand in a notebook, on… paper), but I like finding them the new way (using search on my PC). I love OneNote and how easy it is to keep an online notebook with all sorts of data. For example, due to archival rules some of my email starts to “disappear” from my work account after a couple of years. But some emails I like to keep – little nuggets of wisdom, notices, personal information and such I like to keep around. So, I export them to OneNote, where I can keep them safe.

So, I was living this duplicitous life of daily note taking with pen and paper but also filling my online notebook with all sort of information. I wanted to find a way to bridge these two. A few years back I was at a Microsoft convention and ran into a member of the OneNote team that turned me on to a small startup firm called: ModNotebooks. Their website is now gone, this is all you see:

mod.PNG

What they did was to send you a very high quality notebook, you would fill it up and ship it off to them. In the back of the notebook was a folder with a mail pouch and prepaid postage. They would receive it, scan it in for you and then dump the contents into your OneNote file behind the scenes. I filled 7 notebooks with them. But alas, their business model did not work, I am guessing, and they have gone out of business.

Now, I am left with my seeming duplicity again. wlEmoticon-disappointedsmile.png How do I get back to having my handwritten notes in OneNote again? Enter, OfficeLens.

I have actually been using this app for some time to scan travel receipts to PDF to turn in for expense reimbursement. And while I have always known I can use it for OneNote notes, I never tried it – namely because I had a better thing in ModNotebooks – or so I thought. Being forced to use something is sometimes what it takes. So, I gave OfficeLens a shot.

First, you download it to your phone. It comes on all three platforms (Windows Phone, iOS, and Android). Next, you hook up your Microsoft account (Live/Hotmail/MSN) and then you start scanning. It automatically finds the page and draws a border around it:

20170717_145127000_iOS.png

When you click the button at the bottom, it shows you what it got:

20170717_143239000_iOS.png

In the lower left, you can click the (+1), to add more page (up to 10 at a time – my only complaint – more on that later)* Once complete with your scanning, you click Done at the top right. This will take you to the “Export To” page.

20170717_143248000_iOS.png

From here I select OneNote and it asks me where I want to put it in my Notebook and when I click Save, it begins to upload it to my OneNote notebook:

And once it is in OneNote, it is fully text searchable, based on my handwriting. Yes, I said “fully text searchable” from my handwriting. Here is what it looks like once it gets to the final destination and I perform a search:

OneNote.PNG

Amazing, eh?

So, my only complaint is that the tool only allows you to scan in 10 pages at once. I wish there was a way to override this, for two reasons:

  1. I take a lot of notes and I fill 10 pages quickly. I have to get into the habit of scanning every week at this point to keep up. It is not impossible, but it would be nice to skip a few weeks and then scan them all in – in bulk. Right now, I have to upload it in several different batches of 10.
  2. I recently was working with legal documents and needed to print them out, sign them, scan them in and then send them off. The document had 11 pages. 11. 11! So, I had to scan 10 pages, then scan the last one and then find PDF merge software to merge them all together. That was NOT fun.

With that said, Au Revoir ModNotebooks. You were great while you lasted. I will likely be going back to using Moleskin’s and then using OfficeLens to digitize. We shall see… TUL (the “u” is long, so “tool”) has some nice notebooks too – I love their fine-point pen:

20170717_152827116_iOS

If you have any suggestions, please leave comments below. wlEmoticon-hotsmile.png

Going back to taking notes by hand…