Facture is FAC-tories that generate partition-isolated fix-TURE data, ideal for elaborating scenarios.
pip install facturedata
Originally this tool was created to generate the test data for a database that had very high overhead per-insert. Using traditional data factories and individual transactions with rollback would not work; batching the inserts was necessary.
Fixtures alone could solve this problem, but it's really hard to enumerate and manage the joins in a large number of scenarios in fixtures effectively. The primary advantage that factories have is their ease of use in testing individual scenarios without needing to understand other scenarios or manage a corpus of shared data. Carefully paritioning inserted data so the data is unambiguously owned by a scenario can provide the same benefit as factories, where it becomes easy to understand the data associated with individual scenarios.
There is a fair amount of configuration that goes into setting facture up, but here is a quick inline example data group that shows how foreign keys work:
'group': 'facture_group_shawshank_redemption',
'offset': 100,
'data': [
['actors a_mf', {'attrs': {'first_name': 'Morgan', 'last_name': 'Freeman'}}],
['actors a_tr', {'attrs': {'first_name': 'Tim', 'last_name': 'Robbins'}}],
['films f', {'attrs': {'name': 'Shawshank Redemption', 'year': '1994'}}],
['roles r1', {'refs': {'actor_id': '.a_mf.id', 'film_id': '.f.id'}}],
['roles r1', {'refs': {'actor_id': '.a_tr.id', 'film_id': '.f.id'}}]
]
For a deeper dive I recommend that you look at this example: https://github.com/gmccreight/facture/tree/master/tests/examples/sql_inject_target
Copy the factureconf.py there to the directory you will be running facture
in. Look at the facture_json target section in the original.sql file
and copy it into a file of your own choosing. Update the conf_targets
section of your newly created factureconf.py file to point to the file
where you put the facture_json.
Run:
facture
Your target file should now be filled in with some generated data. You're off to the races!
- "compile"-time factory configuration consistency checks. Checks for many typos, join issues, etc.
- Materializes the output into your version-controlled files, unrolling the complexity. Because the fixtures that we generate are automatically created, we can put a lot of effort into making the generated data comprehensible (and lineage-providing). This means we annotate all generated data with the column names, and the scenario name. It's also extremely easy to see what the effects of any change you make is, since it's all laid out for you.
- Code as configuration. The configuration files are written in Python, so you have the full power of the language.
- Easy to reason about each scenario in complete isolation because of data partitioning
- Plays well with others... can be introduced into existing fixtures. Integrates with how you do things, doesn't try to take over completely. Can target files or sections within files.