[Bug]: No transactions detected in bank CSV export #2696

New Issue

2026-02-28T20:24:47-06:00

GiteaMirror commented

2026-02-28 20:24:47 -06:00

Originally created by @mpeter50 on GitHub (Dec 7, 2025).

Verified issue does not already exist?

I have searched and found no existing issue

What happened?

When I tried to import the CSV exported from my bank's site, AB says there are no transactions in it:

But if I look into the file, I see that there are 5 transactions in it, and a header row, fields separated by commas.
If I set the "skip start lines" to 1, AB says I cant skip more lines than there is, so maybe it cannot handle the line endings used?

According to the linux "file" command, the is a "Little-endian UTF-16 Unicode text, with very long lines".
I dont know how to tell what line endings it has though, because suprisingly its much harder than on windows.. no text editor I know can tell that to me.

Another problem could be that AB does not allow me to select the standard date format used in my country (YYYY.MM.DD), but that would be on top of the line detection issue. If it turns out to be an issue I'll open a separate issue for that.

How can we reproduce the issue?

You would need a sample export file from my bank
Open AB's webapp in Firefox
Select an account on the left
Click Import at the top
Select the sample CSV file

Where are you hosting Actual?

Docker

What browsers are you seeing the problem on?

Firefox

Operating System

Linux

Originally created by @mpeter50 on GitHub (Dec 7, 2025). ### Verified issue does not already exist? - [x] I have searched and found no existing issue ### What happened? When I tried to import the CSV exported from my bank's site, AB says there are no transactions in it: <img width="800" height="655" alt="Image" src="https://github.com/user-attachments/assets/32c7ee9b-97fa-4e52-8547-4bfd2e516070" /> But if I look into the file, I see that there are 5 transactions in it, and a header row, fields separated by commas. If I set the "skip start lines" to 1, AB says I cant skip more lines than there is, so maybe it cannot handle the line endings used? <img width="800" height="425" alt="Image" src="https://github.com/user-attachments/assets/132823e9-e247-4d95-8771-e497ca124284" /> According to the linux "file" command, the is a "Little-endian UTF-16 Unicode text, with very long lines". I dont know how to tell what line endings it has though, because suprisingly its much harder than on windows.. no text editor I know can tell that to me. Another problem could be that AB does not allow me to select the standard date format used in my country (YYYY.MM.DD), but that would be on top of the line detection issue. If it turns out to be an issue I'll open a separate issue for that. ### How can we reproduce the issue? 0. You would need a sample export file from my bank 1. Open AB's webapp in Firefox 2. Select an account on the left 3. Click Import at the top 4. Select the sample CSV file ### Where are you hosting Actual? Docker ### What browsers are you seeing the problem on? Firefox ### Operating System Linux

GiteaMirror added the transaction import bug help wanted labels 2026-02-28 20:24:47 -06:00

GiteaMirror commented

2026-02-28 20:24:48 -06:00

@mpeter50 commented on GitHub (Dec 7, 2025):

Forgot to mention: I run version 25.12.0, but before I updated from 25.7.1 it did not work either.

@mpeter50 commented on GitHub (Dec 7, 2025): Forgot to mention: I run version 25.12.0, but before I updated from 25.7.1 it did not work either.

GiteaMirror commented

2026-02-28 20:24:48 -06:00

@mpeter50 commented on GitHub (Dec 7, 2025):

If I open it with KDE Kate and use "File > Save as with encoding" to save it as UTF-8, then AB will see all the transactions in the file.

@mpeter50 commented on GitHub (Dec 7, 2025): If I open it with KDE Kate and use "File > Save as with encoding" to save it as UTF-8, then AB will see all the transactions in the file.

GiteaMirror commented

2026-02-28 20:24:49 -06:00

@mpeter50 commented on GitHub (Dec 7, 2025):

But it seems if I select the YYYY-MM-DD date format, it recognizes the dotted version I have too, so thats nice.

@mpeter50 commented on GitHub (Dec 7, 2025): But it seems if I select the YYYY-MM-DD date format, it recognizes the dotted version I have too, so thats nice.

GiteaMirror commented

2026-02-28 20:24:49 -06:00

@youngcw commented on GitHub (Dec 8, 2025):

We need an example file or there is nothing we can do

@youngcw commented on GitHub (Dec 8, 2025): We need an example file or there is nothing we can do

GiteaMirror commented

2026-02-28 20:24:49 -06:00

@mpeter50 commented on GitHub (Dec 9, 2025):

Is there a way I could send it privately?

I'll also try to just redact it, and if it still reproduces the issue I'll send it here.

@mpeter50 commented on GitHub (Dec 9, 2025): Is there a way I could send it privately? I'll also try to just redact it, and if it still reproduces the issue I'll send it here.

GiteaMirror commented

2026-02-28 20:24:50 -06:00

@youngcw commented on GitHub (Dec 9, 2025):

You can redact it or also reduce it to just one or two transactions.

@youngcw commented on GitHub (Dec 9, 2025): You can redact it or also reduce it to just one or two transactions.

GiteaMirror commented

2026-02-28 20:24:50 -06:00

@mpeter50 commented on GitHub (Dec 9, 2025):

Here is a redacted version: bank.csv

It produces the same behavior in AB as with the original export.

@mpeter50 commented on GitHub (Dec 9, 2025): Here is a redacted version: [bank.csv](https://github.com/user-attachments/files/24062370/bank.csv) It produces the same behavior in AB as with the original export.

GiteaMirror commented

2026-02-28 20:24:50 -06:00

@sublettk commented on GitHub (Dec 17, 2025):

Hi @mpeter50. I took a look at your issue. Thanks for sharing an example file. I was able to replicate what you were experiencing. The good news is that from my findings, that data inside your CSV file is not the issue. I converted your file from UTF-16 LE to UTF-8 and the dates map fine (although they would be converted from dot notation to the predefined date formats on Actual (e.g. 2025.12.01 → 12/01/2025)).

I looked at the code for Actual and it seems that the application only accepts UTF-8 encoded files. To unblock you in the mean time so that you can continue using the application while this is fixed, you can convert your UTF-16 files to UTF-8. If you are on Linux, you can use the terminal to run a command like this iconv -f UTF-16 -t UTF-8 bank-utf16.csv > bank-utf8.csv. There are also websites that you can use to do this.

In terms of implementing a solution for this, I am working on proposing a feature that automatically and seamlessly converts a UTF-16 file to UTF-8 during the file import process.

@sublettk commented on GitHub (Dec 17, 2025): Hi @mpeter50. I took a look at your issue. Thanks for sharing an example file. I was able to replicate what you were experiencing. The good news is that from my findings, that data inside your CSV file is not the issue. I converted your file from UTF-16 LE to UTF-8 and the dates map fine (although they would be converted from dot notation to the predefined date formats on Actual (e.g. 2025.12.01 → 12/01/2025)). I looked at the code for Actual and it seems that the application only accepts UTF-8 encoded files. To unblock you in the mean time so that you can continue using the application while this is fixed, you can convert your UTF-16 files to UTF-8. If you are on Linux, you can use the terminal to run a command like this `iconv -f UTF-16 -t UTF-8 bank-utf16.csv > bank-utf8.csv`. There are also websites that you can use to do this. In terms of implementing a solution for this, I am working on proposing a feature that automatically and seamlessly converts a UTF-16 file to UTF-8 during the file import process.

GiteaMirror commented

2026-02-28 20:24:51 -06:00

@mpeter50 commented on GitHub (Dec 17, 2025):

Thanks!

While we are at it, its probably the topic of a different issue, but does AB have a way to recognize CSV column names in other languages than english? It would be useful if that would work automatically, or if some of the last selections were remembered.
Actually in my case, my bank's CSV contains columns such that sometimes I want to use column A, and sometimes column B, because they are similar in nature but for some transactions one or the othher does not hold meaningful information.

@mpeter50 commented on GitHub (Dec 17, 2025): Thanks! While we are at it, its probably the topic of a different issue, but does AB have a way to recognize CSV column names in other languages than english? It would be useful if that would work automatically, or if some of the last selections were remembered. Actually in my case, my bank's CSV contains columns such that sometimes I want to use column A, and sometimes column B, because they are similar in nature but for some transactions one or the othher does not hold meaningful information.