Even though I'm sure that there's no swift way of doing this here, I thought I'd come back and show the logic that I've used to affect a similar result.
I'm now resigned to the fact that I'm going to have to process each one in an 'Apply to each' connector run, it's a shame, but thems the breaks. Also, in an effort to avoid Condition actions which would work just fine, I'm introducing Integer Traps.
Still, if I'm going down that route, I can at least ensure that I process it as simply as possible.
What I've done here, is to map out using a few flow variables what I'll be subsuming into a single (large) expression once I have got to the end.
Separately, I've also found another source of information to confirm what this end of these strings, which is a more sure method of confirming that data. However the logic below would still work here, too. The key is to assume that I have isolated the data I need to analyse.
So ... this doesn't quite answer the question, but it gets me most of the way there ... ... anyway ... forward!
----
So, each of these codes needs to:
- Be 9 characters long.
- Start with a 3 character string of letters that is a month.
- Finish with a 6 character set of numbers.
It was realising that the 6 numbers would be an integer every time that got me set right here.
I'll work with one item for this example, and without the brackets.
Integer Traps Flow
You may notice a dependence on what I call "Integer Traps" here. Essentially, these will mirror any failures to create what is needed in a full expression, but they're really useful if you are running out of layers in your flow due to many scopes, conditions, loops, etc. I'm sure I'm not the first to use them, and hopefully I'll not be the last.
My Integer Traps are basically; if a particular condition is true, you will create an integer (useful in this case!), if it is false, you try to make an integer out of the word "ERROR", which will fail. You then run a success branch for if it makes the integer, and a failure branch if it doesn't. On the failure branch you 'Configure run after' to only run on failure.
Always ensure that you 'close' an integer trap, though. Usually by making the closing action run after all possibilities from all previous branches. In this case, though, it wasn't required.
-----
On to the logic ... and up front there is the month and monthsVAR (an array of months):
substring(variables('input'), 0, 3)
![[ "JAN", "FEB", "MAR", "APR", "MAY", "JUN", "JUL", "AUG", "SEP", "OCT", NOV", "DEC" ]](/_api/msdyn_richtextfiles%28BCA75A98-148D-4066-AA1C-121BE09CF2CA%29/msdyn_imageblob/$value?size=full "init months.jpg")
[ "JAN", "FEB", "MAR", "APR", "MAY", "JUN", "JUL", "AUG", "SEP", "OCT", NOV", "DEC" ]
These are going to be used to ensure that the first three characters are the right format.
The first integer trap that's here will be counting the length of the input, and only allowing the flow to continue if the length is exactly 9 characters long.
if(equals(int(length(variables('input'))), 9), 9, 'ERROR')
You can see here that there is a true/false check on whether the length is 9, then if it is, the value is set to 9, if not, the value is set to 'ERROR'. If I get the 9 that I want, then I will just run a quick check on that 3 character month format:
and(contains(variables('monthsVAR'), substring(variables('input'), 0, 3)), equals(variables('length'), 9))
Breaking down the conditions in the 'and' logical expression (all conditions must be true) that is in the Success branch of integer trap 1:
- The 'contains' logical check will loop through the monthsVAR array to see if those first 3 characters match one of the entries there, and if it does, it presents a 'true' value.
- The 'equals' check feels redundant as it is performed in the previous step, but it feeds a boolean to the and upon whether the length is 9 or not.
Since I am setting an Integer variable to 'ERROR' upon a false statement, this would normally cause the flow to fail, so I need to have a branch to allow things to progress despite that. My Compose action does that.
Configure run after
All that is done here is is click the menu and select 'Configure run after':
Select 'Configure run after'
Then select only the option "has failed":
has failed
This branch will now only run if the integer creation fails.
Great. Now on to handling the next check, the last 6 characters being all numbers. Luckily, if they are all numbers, then they will convert in to an integer perfectly. So, I get to use an integer trap ... to trap some integers!
if(variables('allOk'), int(substring(variables('input'), 3)), 'ERROR')
However, it's very important here (as mentioned above) to close the previous integer trap. So whilst that would usually involve me ensuring that I have action run *whatever* happens in either the success or failure branches previously (check all items on each - you'd have to go in twice) ... here, I actually want THIS branch to be skipped if the failure branch runs. So here I will *only* allow the 'Initialize int' action to go if:
- It can 'run after' the allOk branch "is successful."
- It can 'run after' the failure branch "is skipped."
allOk is successful OR Compose is skipped
So, now, that will either leave me with a successful code to use wherever, or, it won't run, because the previous error branch was successful, and so this branch will be skipped.
All that leaves me to do in order to progress forward is to manage the success and failure branches of this particular action.
Here, in this example flow it's the end, and I have Success/Failure branches set.
However in a longer flow, I would probably append the successfully managed code to a new array, or on failure, do nothing, then have a nonsense Compose action to close the success/failure branches and run on all eventualities of both branches. Which (in an 'Apply to each' loop) would then mean it can move on to the next item.
------
That's it, basically. With the added bonus of with a tiny bit of tweaking, this can all be hearded into one MASSIVE expression - which I'll edit in once I've tested it! 😅
EDIT 1 - OK, here is that whole flow as one expression, assuming that there is still the 'input' variable.
if(and(contains(variables('monthsVAR'), substring(variables('input'), 0, 3)), equals(length(variables('input')), 9)), int(substring(variables('input'), 3)), 'ERROR')
Obviously that's great because it's down to 3 steps (months variable), but you could make it two by replacing the month variable with:
json('["JAN","FEB","MAR","APR","MAY","JUN","JUL","AUG","SEP","OCT","NOV","DEC"]')
The important question to ask here is whether this is too complex for someone without this knowledge to parse in the future. The reason we're all here is to have easy access to this stuff, and that's dangerously 'codey'. 😉
Either way, edit two will have it in one step. 😅
EDIT 2 - One step - Since I'll be handling the input of a 'Apply to each' action, I can use item() to refer to the current item. The current item in this case is a pre-sorted, tab separated, line of information. Of which we only need the first piece. So:
if(and(contains(json('["JAN","FEB","MAR","APR","MAY","JUN","JUL","AUG","SEP","OCT","NOV","DEC"]'), substring(trim(first(split(item(), uriComponentToString('%09')))), 0, 3)), equals(length(trim(first(split(item(), uriComponentToString('%09'))))), 9)), int(substring(trim(first(split(item(), uriComponentToString('%09')))), 3)), 'ERROR')
It's basically some excel monkey's wet dream, though now. Which is the reason we're here to avoid this stuff. 😏
It's not an answer, but it's how I'm managing this for now.
((( If I was going to manage each integer separately, I'd use a similar method to the months. )))
