cgiacofei/pyledgertools

Tools for managing banking data with Ledger-cli.

pyledgertools's Issues

Not reading existing transactions to skip dupicates.

Pick classification probablity to accept match without user input.

Merge budget scripts into this repo.

Maintaining shell scripts separate from this repo is annoying.
It would be easier to merge them together and not bother making this an installable package.

Build parser for paycheck json data.

Pulling data from ADP using a ruby gem (adp-downloader).

Paycheck is formatted as json.

Rule parse crashes the script if rule location is not valid.

Naive bayes trainer can't handle postings with calculated values.

Need to either ignore calculated values or teach the script how to dot the calculation.

example transaction that breaks the system:

YYYY-MM-DD * Payee name
    Account:Name                 ($ 50.00 + $ 45.00)

Pull check images from suntrust.

As long as we're in there might as well try to grab check images.

Use editor instead of taking input from the command line directly.

This will simplify manual account entry by allowing for autocomplete of account names.

Example code from stackoverflow

import sys, tempfile, os
from subprocess import call

EDITOR = os.environ.get('EDITOR','vim') 

initial_message = '' 

with tempfile.NamedTemporaryFile(suffix='.tmp') as tf:
  tf.write(initial_message)
  tf.flush()
  call([EDITOR, tf.name])

  # Read new file contents.
  tf.seek(0)
  edited_message = tf.read()

Something to consider:

I ran into the same problem of reading back the old contents of the file, and its true that in many cases, vim moves the old file aside and writes the new file, leaving our tf file descriptor still open on the now-backup file. The command line option "+set backupcopy=yes" added before the file name prevents this problem, and I'm now happily able to read the new contents of the file. call([EDITOR, '+set backupcopy=yes', tf.name]) – Chuck Buche Nov 22 '16 at 6:56

Suntrust scraper only pulls what's on the front page.

Need to be able to expand the transaction list and select more transsactions.

Implement bayesian classifier.

Package is already available at https://github.com/muatik/naive-bayes-classifier.

Mortgage parser calculates incorrect values if escrow is not processed last.

Current Config

Process Mortgage:
  conditions:
    - AND:
      - PAYEE CONTAINS Mortgage
  process:
    mortgage_process:
      file: "mortgage_schedule.csv"
      currency: $
      principal_pmt: "Liabilities:Mortgage:Principle"
      interest_pmt: "Expenses:Interest:Mortgage"
      escrow_pmt: "Assets:Escrow:Mortgage"

Fixed Config

Put accounts in a list so the order is preserved, ensuring escrow is always processed last.

Process Mortgage:
  conditions:
    - AND:
      - PAYEE CONTAINS Mortgage
  process:
    mortgage_process:
      file: "mortgage_schedule.csv"
      currency: $
      accounts:
        - principal_pmt: "Liabilities:Mortgage:Principle"
        - interest_pmt: "Expenses:Interest:Mortgage"
        - escrow_pmt: "Assets:Escrow:Mortgage"

Web scraper not working when run in a cron job.

I suspect the issue is the cron environment not finding phantomjs or its libraries...

Use this stackoverflow answer to check.

Need module for parsing ledger text into data structure.

Ledger comes with python bindings that may work out of the box. Need to try and get them to work...

Otherwise, may need to write something from scratch.

Combine Selenium with Beautifulsoup for page processing.

This will make it easier to provide a raw html file for processing without having to load the page automatically.

CapitalOne 360 scraper.

Redo ofx download to use new ofxtools interface.

See README for a usage example.

Need actual unit tests.

Need csv parser.