Drastically improved CSV imports
May 8th, 2008 by ShashankUntil now, we had a very restrictive format in which we accepted CSV files for import. This was a big pain not only because of the strict format, but because one would need to manually convert (into the format we required) a bank-downloaded CSV every single time one wanted to import.
Not anymore.
We now accept almost all formats for CSVs as long as each transaction line contains at least the following fields:
date, description and amount
Each line may also contain an optional tag field.
These fields may occur in any order which we will try our best to identify. We will also try our best to get rid of the garbage that most banks like spewing in their CSV files. If we don’t succeed, we will ask you for help — only one time per account. So once you have set this up, future CSV imports into that account will all be achieved with a single click.
Enjoy!
May 8th, 2008 at 7:18 am
I’m very pleased to see this; it was one of the (very few) key things keeping me from using Buxfer in earnest.
Just a quick question: can you specify separate “debit” and “credit” transaction amount columns? One of my primary banks does this, without a sign (+/-) preceding the amount.
May 8th, 2008 at 5:58 pm
Steve- I *knew* these banks would throw some weird thing at us no matter what we do :) Thanks for the note. I will try to incorporate or detect such weirdness as well. Could you send us a sample CSV statement of this kind? Send it to support AT buxfer.
May 14th, 2008 at 6:04 am
This feature is still not working for me at all. Either it cannot recognise a date because of some cruft at the bottom of the CSV (e.g. “Total of this column:”) or it cannot recognise a simple date format (”01 May 2008″). I additionally have the separate credit and debit columns mentioned by the user above if this is still an issue (and in multiple banks in multiple countries - I don’t think it’s as strange as you believe). I do not get offered the opportunity to specify which columns are which.
Additionally, I would always have to import these statements manually as the synchronisation feature seems to fail completely. I have Firebux and it tells me that it is recording, but seems to do nothing. When I press stop recording it doesn’t give me an error, but the accounts never sync either. A lack of automatic syncing sort of defeats the point for me…
Although, that sort of brings me to my final point - how do you hope to deal with banks that actually have some security? I’ve heard American banks can be notoriously poor at web security but in Europe we have SecurID dongles and/or partially, mouse-entered passcodes as standard…
June 14th, 2008 at 4:09 am
Steve, Sam: we have updated our CSV import process now — you can specify separate credit and debit columns as well as multiple description columns.
June 20th, 2008 at 6:18 pm
how can you import with separate debit/credit columns w/o - or + sign?
it doesn’t seem to recognize it. :(
August 12th, 2008 at 4:16 pm
Ok so I made a mistake the first time Buxfer was proposing the matching collumns. Now this “only ask once” feature is nice but how can I reset this so Buxfer asks me again how to match collumns?
August 12th, 2008 at 4:25 pm
David: yah, that is annoying. We will get that fixed. In the meantime, you can just create another bank account within your Buxfer account and try uploading into that account. Would that work?
August 13th, 2008 at 2:20 am
I guess I could do that. The mistake I made was that I did not identified enough collumns as “description”. Some of my descriptions were blank because of that and Buxfer did not import transactions with empty description I think. Am I correct about this?
August 14th, 2008 at 1:33 pm
David: yep, that’s exactly right.
December 16th, 2008 at 3:29 am
Hi there, I’m glad that the CSV import has improved and the date format is selectable, but I still get hit-or-miss results with seemingly identical data.
In the following CSV, I have the fields defined as follows in Buxfer:
Date -> DATE
Time -> DESCRIPTION
Description -> DESCRIPTION
Amount -> Credit/Debit
Currency -> DESCRIPTION
Tags -> TAGS
All I can think is that for some odd reason Buxfer is trying to parse known words (like ‘Shell’ or ‘Subway’) out of the description fields and use the recognized words as the complete description (instead of aggregating all of the description fields, as usual.)
This must be Buxfer’s attempt to filter out the ‘garbage’ that many banks put into their data, but would it be possible to -disable- that feature somehow?
I’ve tried prefixing “Shell” and “Subway” with single quotes, hyphens and underscores, but to no avail (also, this would be an ugly solution…)
I’m just trying to get consistent behavior whether my descriptions contain ‘English’ or not.
An extra added bonus would be for you to support multibyte characters, but I realize that’s probably not going to happen soon ;)
December 16th, 2008 at 3:31 am
Oops, here’s the CSV:
Date, Time, Description, Amount, Currency, Tags
2008/12/15, “14:00″, “Donki, Shibuya”, “-498″, “JPY”, “sustenance coffee”
2008/12/14, “23:41″, “Shell Gas, Shinjuku”, “-650″, “JPY”, “transportation gas”
2008/12/13, “11:53″, “Subway, Kitasando”, “-1750″, “JPY”, “sustenance lunch”
And my DESCRIPTIONs become:
“14:00 Donki, Shibuya Jpy”
“Shell”
“Subway”
Instead of:
“14:00 Donki, Shibuya JPY”
“23:41 Shell Gas, Shinjuku JPY”
“11:53 Subway, Kitasando JPY”
December 16th, 2008 at 3:41 am
I tried a test, but the underscore seems to work.
All of your descriptions will end up with a ‘_’ in front, though, and Buxfer will try to CamelCase the words for you so it will look like “23:41 _shell Gas, Shinjuku”
A ‘Don’t parse’ flag of some kind would be nice.