Documentation on workflow about hledger-flow HOT 8 CLOSED

OK, I've got what I needed by actually reading the Full-fledged Hledger.
When I read the documentation in my head "Full-fledged Hledger" was just referring to running Hledger manually.

After reading it I see that I really should be adding rules at the level of matching a single transaction and use the table format for rules to keep this manageable:

if|account2|comment
...
2017-10-11.*EMPLOYER INC||per diem for the trip to New York

I was also worried about my rules files becoming huge an unmanageable, but this kind of single transaction rule only needs to run for a specific account.

from hledger-flow.

I just added this nice dateutils package with the following example preprocess script:

#!/usr/bin/bash

cat $1 | dateconv -S -i "%d-%m-%Y" -f "%F" > $2

This was particularly useful for some pesky files where they had 3/5/2023 as a date, and I didn't want to think about adding those leading zeros back in for ISO date formatting.

<3

from hledger-flow.

@TC72, I personally make sure there's a rule for everything. That can get a bit awkward, but I like that there's nothing precious about my 2-processed and 3-journal directories, I can always recreate them from my csv and rule files.

I have no idea if this is a good long term plan (for example, it means my rules files are going to grow without bound). I'd be quite curious to hear how other people handle this.

from hledger-flow.

Thanks for the reply.

There are some things I can't do with rules.
For example if I go on holiday I tag every transaction related to it with holiday:summer-2023.
That way I can tag flights, car rental, restaurants and have a total cost for the holiday.

But I go to the same places in different years and a rules file can't handle this without me adding rules to match the exact date and month of each transaction which is a mess.

For this kind of thing manually editing the report is much simpler and it's how hledger was designed to work with the .latest files and only appending to existing reports. To lose manual edits you have to go out of your way to delete both the .latest file and the journal file.

I know I can make this work, even if it involves some git magic to deal with it. Just figuring it out and seeing if I'm missing the right way to do it.

from hledger-flow.

One thing that kept confusing me about this though, is that every time I had to enter a specific rule like this I had to look up how that particular CSV file was formatted. I can't just match against the YYYY-MM-DD format because some banks have very different date formats, or put the payee before/after the description etc. This made me have to open the raw CSV, double check the entry, then write the regex to match that line. Am I missing something here?

from hledger-flow.

@japhir, I found that quite frustrating as well. I eventually got fed up and opted to write preprocess scripts for each of my financial institutions to coerce them all to the same ISO date format.

from hledger-flow.

Ah yeah sounds like a nice solution. Only downside I can think of is the extra stupid work + the fact that you can no longer share your bank parsing .rules files easily with others who may want to use them (I was thinking about committing all my new banks parsing rules files to https://github.com/simonmichael/hledger/tree/master/examples/csv soon)

from hledger-flow.

from hledger-flow.