Way back in June (wait, that's only two months ago?) I wrote up a blog post showing how to use the AWS SDK for Signed URLs: "Quick example using AWS Node.js SDK V3 for Signed URLs". The idea for this was to cover a very specific set of functionality I needed to use along with Adobe's Firefly Services. Specifically my needs are:

• Create a readable URL for a cloud storage asset
• Create a writable URL for a cloud storage asset

And on top of that - also I needed to upload directly to cloud storage. I worked with Azure Storage Blob SDK and came up with the following functions. Honestly, use this with a grain of salt as it "worked for me", but I can't make any promises about how reliable/safe/etc this code is. That being said, I'd love any comments or suggestions.

Imports and Connecting

Once I installed the SDK, I began by importing what I needed:

import { BlobServiceClient, BlobSASPermissions, generateBlobSASQueryParameters, StorageSharedKeyCredential } from "@azure/storage-blob";

Next, I loaded in my credentials as well as an account and container name. So to be clear, for credentials it's an Azure key that I got from my portal and a connections string. The account name was also from the portal, and finally the container name is the 'bucket' where I'm working. I feel like the connection string could be constructed dynamically, but I hard coded it. All of these values are in my environment:

// Credentials for Azure
const AZURE_ACCOUNTNAME = process.env.AZURE_ACCOUNTNAME;
const AZURE_KEY = process.env.AZURE_KEY;
const AZURE_CONTAINERNAME = process.env.AZURE_CONTAINERNAME;
const AZURE_CONNECTIONSTRING = process.env.AZURE_CONNECTIONSTRING;

And finally, I created my client objects:

const blobServiceClient = BlobServiceClient.fromConnectionString(AZURE_CONNECTIONSTRING);
const containerClient = blobServiceClient.getContainerClient(AZURE_CONTAINERNAME);

Creating Read URLs

To create readable URLs, I used two functions.

function createSASReadString(key, accountName, containerName, duration=5) {
	
	let permissions = new BlobSASPermissions();
	permissions.read = true;

	let currentDateTime = new Date();
	let expiryDateTime = new Date(currentDateTime.setMinutes(currentDateTime.getMinutes()+duration));
	let blobSasModel = {
		containerName,
		permissions,
		expiresOn: expiryDateTime
	};

	let credential = new StorageSharedKeyCredential(accountName,key);
	return generateBlobSASQueryParameters(blobSasModel,credential);

}

function getSignedDownloadUrl(name, key, accountName, containerName) {
	let b = containerClient.getBlockBlobClient(name);
	return b.url + '?' + createSASReadString(key, accountName, containerName);
}

Note that getSignedDownloadUrl chains to createSASReadString and doesn't modify the duration, I could update that. And honestly, looking at this now, I think it should be one function. When I was building this, I thought I'd be reusing createSASReadString a few times but I don't think I did. You could easily wrap those two together and I may do so in the future.

Using it then is as simple as:

let inputURL = await getSignedDownloadUrl(fileName, AZURE_KEY, AZURE_ACCOUNTNAME, AZURE_CONTAINERNAME);

Note that I'm passing in my auth stuff. In that previous blog post the methods I wrote used the global s3 objects which is "bad", but is simpler as well. I thought the approach above was a bit more generic and pure.

I don't want to get that caught up in it though - feel free to modify what I build. ;)

Creating Write URLs

On the flip side, here's the method to create writable URLs. This can be handed off, for example to the Photoshop APIs, and used for outputs.

async function getSignedUploadUrl(name, client, containerName, duration=5) {
	let permissions = new BlobSASPermissions();
	permissions.write = true;

	let currentDateTime = new Date();
	let expiryDateTime = new Date(currentDateTime.setMinutes(currentDateTime.getMinutes()+duration));
	let blobSasModel = {
		containerName,
		permissions,
		expiresOn: expiryDateTime
	};

	let tempBlockBlobClient = client.getBlockBlobClient(name);
	return await tempBlockBlobClient.generateSasUrl(blobSasModel);
}

Using it looks like so:

let outputInvertedURL = await getSignedUploadUrl(fileName, containerClient, AZURE_CONTAINERNAME);

Uploading to Azure

Normally I didn't have to worry about uploading to Azure. If I made an upload URL and the API used it, then I didn't need to worry about it. But I was curious how it would work. My 'usual' upload code failed because Azure requires a special header. Here's the function:

async function uploadFile(url, filePath) {
	let size = fs.statSync(filePath).size;

	await fetch(url, {
		method:'PUT', 
		headers: {
			'Content-Type':'image/*',
			'Content-Length':size,
			'x-ms-blob-type':'BlockBlob'
		},
		body: fs.readFileSync(filePath)
	});

}

That x-ms-blob-type is the special header you need. Also note I've hard coded an image content-type. You could make that an argument or get the value dynamically.

Using it just requires the URL, which you get from the previous method, and a file path:

// sourceInput is something like ./cats_rules.jpg'
let fileName = sourceInput.split('/').pop();
let uploadURL = await getSignedUploadUrl(fileName, containerClient, AZURE_CONTAINERNAME);
await uploadFile(uploadURL, sourceInput);

That's it. I hope this helps because this post is the post I wish I had found when I started. ;)