When a Power Automate document approval is enabled for Business Central we can leverage the approvals connector or another platform. In either of those scenarios then comments an approver makes will be between the requester and the approver. Now in most scenarios this is fine. However, if we compare this to standard OOTB document approvals for BC then we get all the comment history….so why should we settle for less just because we’ve adopted a different platform for approvals? Plus, the fact the approver for Power Automate doesn’t have to be a BC user means we have no real grasp on who approved. Couple this with the fact the user assigned to the Power Automate BC connector shows as the “approver” in BC
Now to cater for many we will need something to enable the logging of comments across all possibilities. So, I’m talking Sales, Purchase, Customers, Vendors and Items – all of these areas can have approvals in Power Automate for BC. All of them have a standard “Comments” table which can be used as well. Looking to avoid the standard approvals comments because the page action doesn’t exist when you use Power Automate BC Document approvals:
Instead I will use the regular “Comments” area for each piece of data:
Simply a case of publishing three pages as web services:
//Customer, Vendor or Item can use a minimum JSON body of the below. Just swap out the "Table_Name" and "No"
//Purchase Documents just swap out the "Document_Type" and "No"
"Comment": "Test Comment"
//Sales Documents just swap out the "Document_Type" and "No"
"Comment": "Test Comment"
So, that’s BC organised, what about Power Automate? I have opted for a OOTB BC template Purchase Invoice approval in Power Automate. The posting of comments, for me at least, can go in two possible places. One is right after the “Start an approval” action
Here is the result:
Minor draw back of using the standard tables/pages here is that the comments table(s) could well be editable to the user. It is reasonable that permissions can control this as the user who posts the comment is the one assigned to the BC Power Automate connector. Of course, permission sets aren’t the only standard choice. You could use change log or field monitoring too. If that doesn’t sit well with you then I’d suggest adding a boolean field to control whether the comment record can be altered or not. You can add that to the JSON body you send once it’s available. Final thing being the “Comment” field in the standard tables only houses 80 characters – but that’s the same as the standard BC approvals comment field – so no essay long approval comments folks 😜
Have you noticed when using Power Automate for BC document approvals you have the same approval buttons? One for sending and one for cancelling. The sending one works fine. Cancel on the other hand isn’t doing what we need. The approvers will still have an outstanding approval and the flow will keep waiting to run. Eventually when it does run it will fail as the request has been cancelled BC side. If no one approves and the rest of the flow doesn’t run then you have a flow run ongoing until it expires…These things are ok. I just think it makes the administration more difficult.
My goal with this post is to have the standard cancel action to actually cancel my Power Automate approval and my Power Automate flow run. Full housekeeping 😄🏡. Let’s cover some basics first:
With Power Automate you might be using different methods to get the approval dealt with – MS Approvals, Teams, Outlook, other. I will look to solve the MS Approvals scenario as it’s the most common.
So what do we need? Well first thing is a record of information from Power Automate back in BC. So for starters a table extension to the “Workflow Webhook Entry” table. This is where the data from the “Flow Entries” page (above) lives. We have 3 fields as I want to capture details on numerous attributes so I can cancel all the relevant parts.
More AL DEV follows but let’s tackle the Power Automate Document approval next. Start off with the standard template for whichever document you need this for. I’ve chosen Purchase Orders.
The compose action “Workflow ID” contains the expression workflow()?[‘name’]so that I can record the GUID of the workflow. The “Workflow Run ID” contains the expression workflow()?[‘run’][‘name’] so that I can record the text string name of the flow run. The template for a BC document approval comes with a “Start and Wait” approval. I’ve removed this and added the “Create an approval” so that I can find out and record the ID of the created approval. You then need the “Wait for an approval” so the flow has the response. The final change/addition is the HTTP step.
So all that remains is the cancellation. For the data in BC to end up back in Power Automate a codeunit is required:
codeunit 50100 PA_CancelApproval
[EventSubscriber(ObjectType::Codeunit, Codeunit::"Workflow Webhook Management", 'OnCancelWorkflow', '', false, false)]
local procedure CancelPowerAutomateApprovalEntry(WorkflowWebhookEntry: Record "Workflow Webhook Entry")
if IsNullGuid(WorkflowWebhookEntry."Power Automate Approval ID") then
SendToFlow(WorkflowWebhookEntry."Power Automate Approval ID", WorkflowWebhookEntry."Power Automate Flow ID", WorkflowWebhookEntry."Power Automate Run_ID");
local procedure SendToFlow(ApprovalID: Guid; FlowID: Guid; RunID: Text)
client.Post('<Add your URL here', cont, response);
The idea here is that another Power Automate flow will be setup to receive a POST. Note that you will need to add the URL to your Power Automate flow for this to work.
Use of a webhook could be done here as well. I opted not to use that as you need something in place to renew the webhook and it would have to POST the information to flow I chose anyway. Hardcoding the endpoint is a bit 🤮 but it is simple
A CSV export with Business Central (or NAV previously) is nothing new or exciting. Why blog about it then? Well let’s tackle the topic using a different solution. This is a CSV export using Power Automate with Business Central data. It could be argued that providing you have the relevant licensing this method would stop use needing 3rd party tools.
My scenario will be basic but the application is broad. I’ll use a web service (API page or query is fine too) to expose item data showing some fields. I’ll filter the web service to keep my results condensed and only select the fields I need. The goal: retrieve the required data from BC and create a CSV file on a recurring basis. The file should be added to a chosen storage location.
Power Automate has a recurring type flow which is suitable for this scenario. To ensure better error trapping and subsequent actions I am making use of the “Scope” feature in my flow. I will then add the actions I need. Two more scopes take place after and will only run if the 1st scope succeeds or fails – 1 scope for each possible outcome.
I also liked using a header in the HTTP request of “Accept” which has this as the value: application/json OData.metadata=none. It’s not terribly exciting but when you intension is to read data only (I don’t want to insert or modify) it removes some of the data you won’t be using like the etag (typically used for tracking modifications). I’d say the majority of the work in this entire process is with the BC data endpoint. You might find a custom api page is needed. For example, to get all the fields you want from each of the tables you need them from.
Most of what I’ve covered thus far is fairly normal in Power Automate. The worthwhile part of this post comes in explaining the use of a compose action – which from the picture is titled as “Array for CSV”. Why do we need it? Well the action that follows for creating a CSV table (standard feature in Power Automate) expects an array. Now in the case of BC it will typically output a JSON object. Without going into extensive detail on json objects or arrays just think about the fact you need the part in between [ ] (square brackets). A simple expression takes care of converting the JSON object you get from BC to an array, which can be converted to a CSV. Create a compose action and add expression body([‘Parse_JSON’]?[‘value’]). This will extract just the array part which is typically titled as “value” when we retrieve data from a BC web service. Example of this can be seen below:
The remainder of the flow is super simple. I’ve used another scope block and adjusted the settings of it to run only if the first BC data block was successful. Use a built in “Create CSV table” action and take the output of the earlier compose action with the array you need – this function only accepts an array. Then you can choose where the file should output to. I chose to create a SharePoint file – ensure you put .CSV at the end of the file name
This is a flexible solution where you can quickly have automated or on demand CSV exports working for your solution. It is much more rapid than the full BC method which would involved job queue tasks. You can of course layer in better notifications as well. This is why I opted for the Scope feature where you can alter the “Configure run after” setting for a previous action.
What am I getting at here? Well, think about a document approval scenario where you might need to have let’s say 1 shortcut dimension that you want to use as a driver for who approves the line. This could be a cost centre or a specific project, maybe even a department. Standard BC requires you to create multiple workflow records to handle different filtering. Personally found that “Header” focused fields are the main one’s which get checked. For creating line level approvals this is a challenge on performance as all workflows will be checked. Is there another way? Power Automate provides a fresh view on the topic. You are able to ignore the rigid nature of BC workflow engine and come up with something new. Couple that with the fact you have more notifications on offer it’s an intriguing proposition.
Let’s work with the business process being you need approval from the department head on a line if a specific department value appears. In my example I will use a global dimension but in theory you can adapt this to any line field exposed by an API or Odata page. There this a set of document approval templates available with Power Automate for the BC connector but they won’t be used here.
The initial event triggers of “when a document approval is requested” is the starting position of the flow. Naturally followed by GETting the details of the record. From this point onwards it’s all about the line data and working it so we can cycle through accordingly and get approvals. In my scenario I have a multi line document which has repeating “Department” values. I want to create a single approval per department not a single approval per line. The approver is approving the total for their department.
The principal mechanism that makes this flow produce the desired results is counting the approvals result. If we have a unique total of values, so departments in my case, then I need to match that with approval results to get an approved document. So 3 unique values means 3 approval results to get an approved document.
I draw some comparisons with the standard workflow in terms of maintenance with a “Switch” statement. This is where I check for results on my chosen line level field value. If I get a match then I send an approval to the relevant person. 📝 Note that we could lookup the approver recipients from a setup table if needed. Feel like how it is currently setup resembles some of the standard BC workflow setup.
There are apps that handle this and I’ve seen a colleagues version of it recently which is great. However, why code when you don’t have to? Limiting factor with no code will be that there is no restriction on other companies i.e. other companies can create data. Some solutions look to reduce this possibility. This is a basic version that doesn’t cover that 🤪
So here’s the goal – have one BC company create master data and pass it onto other companies. Have one companies master data modifications passed to other companies. Have one companies master data deletions result in deletions in other companies.
Note that each type of data action will be in an individual Power Automate Flow – i.e. one for Insert, one for modify and one for delete.
Code when you don’t have to isn’t quite what I’ve gone for but it is certainly low code. My reasons are totally justified as you’ll see. This solution could be used on G/L Accounts, Customers, Vendors, Items, Dimensions, Fixed Assets, Posting Groups etc. I have chosen to demo it with G/L Accounts as you need a little extra there. If you check the standard API for accounts (g/l accounts) it only has a GET command. Which is fine for one part of the process but no good after that. I’ve knocked up a page with the idea of using it as a web service, could be done with an API page too. Either way that gives us an endpoint to work with in Power Automate. Whether you are dealing with insert, modify or delete the start of the flow will use the regular BC connector. Only after that will you look at the HTTP connector which enables you to work with endpoint URL, like that of our web service page. Once you have that page you are able to POST, PATCH or DELETE. The start of the flow can use an event based trigger from the BC connector for insert, modified or deleted. Insert and modify follow the same concept so I’ll cover just one. For DELETE something else is needed. Code can be found here: https://github.com/JAng13sea/Blogs/tree/master/Master%20Data%20Sync
In my example with G/L Accounts not much information can be taken from the standard BC connector with a GET command. That is why I have the supplementary page in addition to it allowing me to perform more CRUD actions.
Once the HTTP connector is dealt with for pulling the right information the in built Parse JSON action can be used. This will make each part of the response data from the GET command be accessible dynamically so we can produce a new JSON for patching or posting data into one of the other companies 👍 The parse JSON allows you to paste a schema. If you have tested the URI from the GET command you will have something to copy and paste
An “Apply to each” starts each block that handles the companies being updated – this is due to the JSON we parse – it is just a loop but it will only run once per record. Place the “value” into the “Apply to each” so it can iterate through the record. Add a new HTTP action and choose the relevant command (PATCH for modify, POST for insert, DELETE for delete). The URI here selects the “Number” of the record to be worked with and that has come directly from the JSON that was parsed. Once your headers are entered you just need to build up a body. If you access the URI from a browser or with a HTTP test tool grab a sample body. Remove the static elements and place in the dynamic JSON. As mentioned the copy and paste feature speeds this process up for the other companies where a change on the company name or ID is the only added step.
What has to happen for the delete option though? If we delete the data from the “master” company the reference information is lost and of course per company the ID value will be different as that is a unique GUID. To aid this I have opted for an event subscriber in a codeunit for the OnDelete trigger for each table I intend to use the feature for. There is a supporting table which then holds the deleted information so I have enough to then use a primary key value in one of the other companies to do a delete.
A month end procedure is a regular topic to be covered when implementing finance with D365 BC. Some finance systems throw a user into a defined routine but with D365 BC it’s much more simplistic. To remove the need for users to be experts wouldn’t a defined routine be useful?
Here is the goal: Create a Power Automate flow that will handle the month end procedure in BC with minimal user input. Principal thing here is that month end is routine so why not have something to cater for it. Caveat here being that job knowledge makes up for a big portion of this finance process so not all of it can be catered for. The scope is laid out in the next paragraph.
So what makes up a month end in BC and what is the scope of this procedure?:
User setup – change the allow from and to posting dates
Our first pointer is around the User Setup and this table could be home to a number of users that are in and out of scope for the procedure. So without any code the solution here is to define a “User Group” – perfect standard system placeholder. The added benefit is this table has “Company Name” included so if you want a differing solution per company you got it! 😊
Once page 9831 is published as a web service some ODATA filtering is needed to get what we need from the exposed data. I called my web service “MonthEndUsers” so I just add the orange part onto the end of the published web service URL: MonthEndUsers?$filter=User_Group_Code%20eq%20%27MONTH-END%27
Whilst on the web services page publish page 119 – User Setup – as this is needed too.
Head over to Power Automate and create a scheduled flow:
First thing needed is the dates for the posting from and to which will be dynamic each time. Add a “Compose” action for both of these and use the following formulas to get what we need:
Start of the month = startOfMonth(utcNow(),’yyyy-MM-dd’)
End of the month = formatDateTime(subtractFromTime(startOfMonth(addToTime(utcNow(),1,’month’)),1,’day’),’yyyy-MM-dd’)
Add a HTTP action to the flow and paste in the ODATA URL with the filter and use the GET method. Add a Parse JSON to grab the body from the HTTP trigger. Paste in a schema which you can get from going to the URL in a browser and authenticating. Add an “Apply to each” control and use the “value” of the Parse JSON. Last step is to add two PATCH requests to the User Setup page. Why two you ask? Well the page as standard won’t allow you to have a “From” date ahead of the “To” date and the body of a HTTP request is not dealt with in the order we send it so to prevent errors two requests are needed. First one to the “To” date and then the second to the “From” date.
The PATCH command needs specific syntax for filtering on the user record. At the end of the published web service URL the orange text represents one user value. The flow will use the “UserName” from the parse JSON step: UserSetup(‘UserName’). In the Header of you HTTP add the below to prevent the likelihood of errors:
General Ledger setup cannot be altered with ODATA so a different solution would be needed for that page most likely with modification. I will not cover that.
Remaining activity is the recurring journals and the General Ledger Setup. At the time of writing the standard BC connector for Power Automate has a post-journal action but it does not work with recurring journals. If your scenario is without recurring journals the standard guide on journal posting will work fine. I will cover what to do about recurring journals and I will use the batch name instead of the ID for ease of use. Publishing the General Ledger Setup page as a web service does not enable you to modify the data, it errors. As a result these two areas need development to give the final result needed.
To do this I’ve opted for a single codeunit which I will publish as a web service and use as an unbound action to complete the work by passing specific parameters:
Unbound actions with codeunits published as web services is fairly new and highly useful. Simply publish your new codeunit as a web service and then use this logic within the URL (note the orange part changes depending on the function you need to use):
Within Power Automate we now have an endpoint to use and we can simply pass in the necessary parameters. Note that in this instance both will be POST actions. The body is just the name of the parameters and the required values:
Also a mention to my colleague Dan Kinsella (https://dankinsella.blog/). He helped with the codeunit element of this solution. Cheers Dan 🍻 I owe you a beer!
Why have I decided to blog about this? Well I had a meeting with a prospect recently that sparked it. Idea was that they would use SharePoint to hold website images as a repository. Naturally they wanted to see the images in BC. A modification to store more images could have been explored but I left that alone. Instead I suggested the use of the “Links” feature. User then clicks on the SharePoint URL to see the image…no need to store it twice, so better for database size 👍
Getting the data from SharePoint to BC though was another thing as the “Links” area is a system table so no chance of a direct HTTP call with Power Automate. So the goal here is: Create a codeunit that is accessed from a web service that stamps a link on an item record.
So we start off with a very basic codeunit
To go along with that we need a page as it is the page we publish as a web service. Read Stefano’s blog for the reason why, he discovered it after all.
Add the fields that you need for the circumstance. In my case it is just the item number, system ID and the description – only the item number is needed really in my scenario. A function is then needed on the page and you need to ensure you have the [ServiceEnabled] part. Again check out Stefano’s blog to understand why.
Testing it in postman the final result looks like this. Note the addition to the URL which has /NAV.AddItemURL which is the name of the page function from the last screenshot
After posting with Power Automate I get the following:
Batch posting has been in the product for some time but I am not seeing a clear way, happy to be wrong, on batch post and e-mail. Here are my options (below) as a user conducting the posting. Yes, I have post via the job queue as well but from what I’ve tested this isn’t sending my e-mail for me. Even the print option breeds no results here. Thought it would at least use the “Document Sending Profile” on the customer record.
Even when you use the “Post Batch” you get nothing about e-mailing? 😬 Not forgetting the workflows in BC have no option for “Post & Send” it’s just “Post”. Usually these gaps would need filling with a code modification but not in this blog.
Where there is a will there is a way and my idea here is to use Power Automate to do the heavy lifting. On this occasion I will use a manual trigger but if this was a production ready scenario I would use a scheduled type flow.
The connector for BC in Power Automate is fairly small but I’m sure it will get better over time. There is a slight restriction though as the action of posting and emailing exists but it’s on a singular level like it is in the regular UI. So the first thing I need to do is fire in some data I can use in Power Automate from BC. I will of course need the GUID ID reference to the sales invoice as that’s what the BC Power Automate connector likes. The page that fits the bill here is available as a web service already. I will use the top one in the list 2811. This page is a mixture of open and posted sales invoices so some filtering is needed:
Example of the payload from this page:
Adding a filter to the ODATA query so I only post invoices that are ready to be posted. In the case of this page the “status” values are different to regular BC. In my case I have a choice between draft aka “Open” or open aka “Released”. So I have added ?$filter=status%20eq%20%27Draft%27%20and%20totalAmountIncludingTax%20gt%200 to the end of the web service URL I got from BC. Reason being the feature in the Power Automate BC connector restricts me to draft only and I don’t want to post invoices with a 0 value. Will log something on the Power Automate forum about the connector wanting “Open” invoices rather than “Released”. Backwards logic otherwise, what if I’m using approvals for instance. I think this is possibly wrong so I have posted this to find out more: https://powerusers.microsoft.com/t5/Connecting-To-Data/Is-the-Business-Central-Post-amp-Send-Invoice-action-correct/m-p/637081#M9492
Once we add the “id” from the parse JSON step you will get an “Apply to each” wrap around the chosen BC action which is exactly what we want (automatically). Remember our goal Batch post and e-mail invoices – so we better placed than standard.
In my case a small batch of invoices have been handled and emails dished out accordingly. I added an extra step for updating the posting date as well. Just so it takes something useful from the existing “Batch Posting” feature 👍
Recently had a requirement where alerts to a vendor need sending for orders expected in the next two weeks. A reminder is needed as the lead times for the goods are so long.
The goal is therefore, have an email sent to a vendors email address for purchase order lines in the next two weeks from today’s date.
To achieve this Power Automate or Logic Apps are the most appropriate choices. For this blog I have chosen Power Automate.
Given that I will need line data and I want to minimise the need to make multiple HTTP calls so some small DEV BC side will help. I have produced a query object so that I have all the data I require for all the areas I want data from (link to my code is at the end). I’ve chosen header level here for the “Expected Receipt Date” but you could do it for the line level if needed:
Ensure the query is published as a web service and check you have output by using the web link or from an API test tool like Postman:
To get the exact data we want for the scenario some ODATA query logic is needed. Add the following to the end of the web service URL. Replace <Your Date> with the date you want to filter by. The format must be like this: 2020-07-21T00:00:00Z:
?$filter=Expected_Receipt_Date ge (<Your Date>) and Expected_Receipt_Date le (<Your Date>) and Outstanding_Quantity gt 0
The date values will be replaced with calculated fields in the flow which will be a scheduled type flow so that is our starting point along with the ability to create variables for the date/time values which are then used in our ODATA query above:
Use the HTTP connector and call the web service with a GET command and use the current time and future time variables.
The HTTP will produce JSON which we saw earlier, from the web service call, and to use the data from this we need to use the PARSE JSON feature and we’ll be able to then select the content for use in subsequent steps like sending the email. To generate the schema just paste in an example from calling the web service. Works nicely by using an API test tool like Postman as you get it formatted in a nicer structure.
Once the JSON is read by the PARSE JSON step you are able to create a “Apply to each” step which will iterate through each of the received rows of data until it has read them all. Whilst that is happening we can initiate further actions like our email:
Each reference comes from the PARSE JSON and you can use format functions on certain data like I have for the date. The result of this flow is an email much like this one: