Add testing framework (mocks)

[tombola]

Should support the following testing

• unit test utility functions
• unit test cli commands
• test api endpoints
• integration test against local fm/wc?

[tombola]

Will need to mock external data sources:

• WooCommerce API (preferably by intercepting requests - see example)*
• FileMaker odbc connection (patch pypyodbc?)

Hopefully can provide a set of pytest marks (eg). that will run the corresponding recordings of live requests against local database. This would keep test more maintainable, as can re-run against new copy of database and commit the responses.

[tombola]

I have added mocking for WooCommerce, which will make testing that aspect simple and fast.

Next is testing the odbc connection* against an SQLite database in place of filemaker, and dumping some test data in there from the relevant filemaker tables, eg acquisitions, seed_lots, batches. Should include some non test data to catch edge cases.

Resources re odbc mock:

• https://github.com/vitalseeds/vs-data-api/pull/9
• https://stackoverflow.com/questions/31164610/connect-to-sqlite3-server-using-pyodbc-python
• http://www.ch-werner.de/sqliteodbc/
• https://web.synametrics.com/SQLite.htm
• https://formulae.brew.sh/formula/sqliteodbc
• https://realpython.com/python-sqlite-sqlalchemy/
• https://realpython.com/python-sql-libraries/#sqlite_1
• https://duckduckgo.com/?t=ffab&q=sql+insert&atb=v354-1&ia=web

* pypyodbc unless I can build pyodbc against iodbc

[tombola]

Testing stock database (ideally against sqlite)

Is SQLite a viable drop in replacement

First must check if SQL syntax used is equivalent between sqlite and FileMaker (🤞). Do this by constructing two simple tables in SQLite with as columns as per vs_tables.py then take some generated SQL (eg from batch_upload.py) and perform the (joined) query against sqlite instead.

Options for mocking:

• pyodbc driver
• monkeypatch sqlite to query/fetch/commit from sqlite instead
• test against a local fmdb (slow), with orders/batches/lineitems partially truncated

To recreate fm in sqlite, could:

• parse database design report to construct schema
• pyodbc query for each table
• write to sqlite