Mavention Workspace has the ability to easily change the image of a group (or as we like to call them: Workspace). This will set the new image visible on all platforms including Microsoft Teams (if it is a Teams workspace), Outlook and SharePoint. This sounds like an easy to write code-module. And when I began writing the module, I thought just the same. However, there are some tricky parts in communicating with the Graph API if you want to send files. In this blog, I will elaborate on how to upload images using the Microsoft Graph API.
On this page I have found a way to convert Base64 strings to binary data and return a blob which we can send to the Graph API endpoint. The answer of Jeremy also contains information that helps you understand what is actually going on in the function. However, this function will not work if you are calling it with a base64 string that is gathered by something like this:
This is because the file is rendered with 'data:image/{imageType};base64/' infront of it. So in order to directly gather the base64 string from an HTML element we have to remove this part of the string. This can be done by replacing:
With:
Graph API endpoint
In order to upload or replace an image and make it available on all platforms, we have to update the values that already exist. The endpoint we are looking for is described here. As you can read in the documentation there are some permissions that have to be set according to your needs:- User.ReadWrite -> For the profile picture of the signed in user
- User.ReadWrite.All -> For the profile picture of any user in the organization, on behalf of the app's identity
- Group.ReadWrite.All -> For the profile picture of a group
- Contacts.ReadWrite -> For the photo of a contact
Call the Graph API
This is where the tricky parts starts. We have to give some value to the endpoint, and as you can read in the Microsoft Docs it cannot be a Base64 string, which would be a lot easier. Instead, the endpoint expects binary data of the image. But how to convert a Base64 string to binary data? It's easy to say a Blob will do some magic. But then again; how to put this huge string into a Blob? Firstly the base64 string has to be gathered from the uploaded file. Once a file is selected using a simple <input type="file"/> we can access the data. This can be done with the following code:var file = document.querySelector('#fileInput').files[0];
public uploadImage() { getBase64String(file).then(base64Image => { const groupId = 'The group ID here'; const request = { method: 'PUT', url: 'https://graph.microsoft.com/v1.0//groups/' + groupID + '/photo/$value', responseType: 'application/json', data: base64Image }; this.$http(req).then(result => { // Image has been set }, (err) => { // Image has not been set }); }); }
public getBase64String(file: any) { return new Promise((resolve, reject) => { const reader = new FileReader(); reader.readAsDataURL(file); reader.onload = () => resolve(reader.result); reader.onerror = error => reject(error); }); }
var file = document.querySelector('#fileInput').files[0];
var byteCharacters = atob(base64Image);
var byteCharacters = atob(base64Image.replace(/^data:image\/(png|jpeg|jpg);base64,/, ''));
Performance
As you can read in Jeremy's answer he has roughly tested the performance of the b64toBlob function. I have tried to test it a bit further. As you can see on the same page of Jeremy's answer another developer used 1024 as sliceSize, whereas Jeremy used 512. So, what is the exact difference? And why split the length anyways? Obviously, the performance speed differs per machine, but just for the sake of reference I performed some tests. I have used four slicesizes to see how it affects performance speed. For each sliceSize I uploaded three different images. One of a few kB, one of 1.94Mb and one of 3.91Mb. 4Mb is the maximum size the endpoint accepts, so that makes +/- 2Mb a nice average.No slices
When using no sliceSize, or set sliceSize to base64Image.length we get the following results: We can see some exponential time difference between the three images, which is expected.
Slicesize 512
Let's see how Jeremy's recommendation works out. We see that files around 2Mb are slower than when using a sliceSize of base64Image.length. However, large files seem to be quicker!
Slicesize 1024
Let's see how that other suggestion turns out. Clearly this slicesize is faster than the previous two sizes.
Slicesize 2048
We see that doubling the slicesize has a positive effect between 512 and 1024. Can we speed it up even more when doubling 1024? I guess this is where we have to draw the line. 1024 is faster than 2048, so we can assume that 4096 won't be faster than 1024.
