Maintaining CSV rules

Associated directory: 06-maintaining-lots-of-csv-rules

Lots and lots of rules

Eventually you will reach a point where your rulefile is so big that every time you add a new rule, you are faced with a choice: either search for an existing rule to add new pattern to, or create a new rule, creating boilerplate of:

if
PATTERN
  account2 ...

, which is rather wasteful and quite a lot to type.

My solution to this was to write rules in a simple pipe-separated file, and have a rule that preprocesses them into format suitable for hledger.

File would have the following format:

PATTERN|account2|comment

, where comment could be empty and tailing separator could be omitted. Lines that begin with # are ignored. Pipe was chosen as a separator because it is very unlikely to appear in either pattern, account name or comment.

For example:

EMPLOYER INC|imcome:employer
TRANSFER|assets:Lloyds:transfers
WAITROSE|expenses:groceries|weekly shopping

Account could also be left blank, in which case generated rule will just add a comment. This could be used to augment more general rule, for example you can annotate a transaction on particular date:

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

File like that could be easily edited or manipulated in Excel/OpenOffice.

If you adopt a convention of calling this file import/<source>/rules.psv, and calling generated file import/<source>/generated.rules then it is possible to write a build rule which would automatically conver this file to hledger syntax whenever necessary and redo CSV-to-journal conversion when you add or modify rules.

See the changes in 06-maintaining-lots-of-csv-rules.

Next steps

To be continued