In the last week I’ve been asked 3 times how to handle commas in csv files.
Reading CSV files
In this post I will use the following example CSV
Name,Description,Amount First item,This is the test of the first item,1 Second Item,"The second, and last item",2
So I’ve got one header line. Then I’ve got a simple line with some data followed by a more complicated line where my description contains a comma.
Note that the csv will have a double quote around the data with a comma. If only every field could have double quotes around all data then that would be great.
In my case i want to replace the commas that separate the fields with 3 hashes (###). But I don’t want to replace the comma in my data.
Creating a flow
As so often I’m going to create a manually started flow. in this flow I will initialize 3 variables. In general I try to avoid variables as much as possible and use compose steps instead, but there will be too loops and conditions for that to work this time.
With these 3 variables I will control the manipulation of the csv file.
Reading the CSV data
Now I’ve got 3 steps to get my CSV data and turning it into an array of lines.
For the CSV content compose action i’m using the following expression:
base64ToString( body('Get_file_content')?['$content'] )
for the CSV Lines split, I’m using the following expression to split by the new lines.
Note that I added a new line in the middle of my expression!
When I run this flow I’ve now got the following array of CSV lines.
Constructing new CSV lines
In the next step I will process the CSV lines in an Apply to each step. this Apply to each will take the output from the earlier CSV Lines compose action and take it as its input.
Then for each line we want to split the data.
The expression used to split the fields is
All quite simple so far. But what does the result look like? Especially that second item in our CSV will now not look right.
[ "Second Item", "\"The second", " and last item\"", "2" ]
there are a couple of different approaches, but in my approach in this post I will go for standard flow options available. I could have called and Azure Function to do all the work, but I want to stick to standard Power Automate actions today.
The general process is shown below.
The New CSV line is set to an empty value using the concat function ( concat(”) will do this).
Then Inside the Apply to each 2 will will set the New CSV line to the required ### separated text like this:
Second Item,"The second, and last item",2
Then we will use a compose action to get each of the lines. So that we can use the Pieter’s method to merge all of the lines.
Inside the Apply to each 2
Inside the Apply to each 2 we will need to check if a field of data starsts with a double quote (“). If it does then we have found data that has commas and my csv editor (e.g. Excel) decided that the double quotes were needed.
First I created a compose action which gets the first character
We can now use this in a condition step. when the data starts with a double quote we will simply store the text in the merge Text variable. Nothing to complicated.
If we find data without a double quote then there are two options. Either we have found data to merge before or we haven’t.
If we have found a field starting with a quotes before then we set the merged text variable to merge the values found. If we haven’t found quotes before then we simply set the merged text to the value of the field.
The code for the nothing to merge is as follows:
When I need to merge two texts together I use the following 4 steps:
So now we have a New CSV Line that looks like the has separated lines that I mentioned before.
The only thing left to do is now to set the compose with the content of the variable so t that a second compose can collect all the lines.
This final Compose will now hold the following data:
This solves my problem.
I could still remove the initial 3 hashes, but that would clutter the post rather than anything else. They aren’t sitting in my way for now.
Other Power Automate posts?
Have a look at the Power Automate User Guide with many other posts.