The Manager doesn't need to know that it was called from the CLI, so it
now is initialized with the database and performs an action, based on
the operation it receives and its parameters.
The work isn't finished, some Manager actions are still based on the CLI
arguments.
The CLI logic and creation of parameters to pass to the manager have
been moved to the __main__.py file, which brings it to line to the
program being called as a package from the command line.
Implements the argument parser, the manager logic and the DB client
methods.
Encapsulates the DbClient connection under the _db attribute on the
manager.
Adds verbose option to enable ORM increased logging.
Move all BL to the manager. The ArgParser now only parses the CLI
arguments and creates a command which contains which command was run. In
turn, this information is passed to the manager, which will run the
appropriate business logic.
This will make it easier to add new options, separating the parsing of
the CLI options from the implementation of the logic. It also simplifies
any future effort in adding a different input (e.g. GUI).
Warning: some function were commented out, this is only a tracer bullet.
Categorizer will work directly on ORM classes, which will cleanup the
code, since changes will automatically be persisted when change the
objects.
Adds wrapper session class inside the DbClient for the manager to use.
The manager will have to have some DB session knowledge, which adds some
unfortunate coupling.
Removes some unnecessary relations between tables that were added by
mistake.
category CLI option now uses the manager.
Can now download from all banks registered on the banks/nordigen table
and write to the PostgresSQL DB (or any DB, since we're now agnostic).
Commented out most of the managers functions until the integration with
the new DB client is complete.
Set Optional relationships. Remove the DB types as dataclasses, it only
increased the verbosity of the types w/o much going for it.
Change the name Original to Transaction, since the type is the
placeholder for the rest of the transaction information.
When the credit card was correct but the bank was wrong, it was
difficul to accertain which one was wrong. Not when one fails, it will
also log what are the available options.
A bank had an offset on the date of the transactions...
It also passes the interal Bank type to the converter, to access it's
additional options before converting.
The __conform__ can´t generate a tuple from the class, but it is still
worth to use the DB intermediate types for cleaner code. So add tuple()
method the the DBTransaction and use it when writing to the DB.
The period will only apply after the transactions have been received,
since if sent on the request, credit cards seem to use the value date
instead of the booking date, rendering the feature useless.
Also adds some robustness to the download of the transactions, by adding
a retry in case of timeout.
There is now the possibility to download the transactions for all banks
in the banks table in the DB.
`NordigenInput` parse method fully functional, and entire chain from
downloading to parsing (simple w/ converter) to writing to DB.
DbTransaction type added __conform__ to simplify writes to DB.
Get bank methods added to both `Manager` and `DatabaseClient`.
Warning: csv parser most likely not working at this point.
Issues #16 and #17
Python's single dispatch can´t use type aliases and lists of classes.
All converters now work on a class, and lists need to be generated
inplace and call the converter for each item.
Adds the bank class both to the DB schema and the common types.
This patch saves the bank information in the DB, in a new table. It also
adds two new CLI commands, register/unregister, so enter the banking
information. (This should later be done internally).
It also adds new types alias for the DB transaction type and new
converters.
Input `transactions` method renamed to `parse`.
Issue #18
Adds a new abstract `Input` interface for the manager to use different
input handlers with the same methods. The Nordigen class and a new JSON
parser descend from it. The previous csv parser will also eventually.
New converter for list input also added.
Issue #19
Move from a direct access to DB by the parsers/categorizers to a middle
layer, which will bring an easier way to have two input alternatives.
This patch starts by instantiating the manager on the cli runnable and
using it for the parser function.
This patch also moves the queries to a different file, so that
introducing new functions on the DB client becomes more manageable and
clearer.
Finally, the new manager will need converters to move from the code type
Transaction to the DB types. This will eventually simplify the code data
model by removing more of its method and leaving it a simple dataclass.
Issue #14
To improve the code organization, additional subfolders were added into
the pfbudget module.
The barebones of renewing the token and downloading transactions
directly from the banks through the Nordigen API were also added.
Issues #13, #15
This will provide differentiation between income that's regular and
stable (long-term contract) and that which is not, such as gigs,
presents, etc.
It presents the information separated in both graphs and reports.
To have cleared information on manually added transactions to the
database, this patch adds the original column so those added after
the parsing can be marked as such.
Also adds a column to add comments to transaction, when they are not
explicit from the transaction message.
Now possible to configure each group in the categories.yaml. Only color
added at this moment. The order in which the groups are listed is used
by the graph module and will later be used by the report.
Fixes search for vacations. The negative match is now done by groups,
and the key changes from negative_regex to not_in_groups.
Only updates transactions when any travel is assigned. Was previously
updating all uncategorized transactions that reached the travel search.
Regex match now correctly done lower letters to lower letters.
Transaction __repr__ method added and updated along with __str__, to
provide a more clearer output when manually categorizing.
- Remove unnecessary category null check after db select
- Adds multiple try-catch block for misbehaving parsing
- matplotlib.plt tight_layout moved to figure method
- Fixes debit/credit index with different dates
- Check for existing banks transformed to lower chars, to avoid
capitilization problem
