Generate seeds for HABTM associations

[jfoo1984]

This PR addresses #34. The change add seeds for join tables for an HABTM association by iterating over all HABTM associations in a model.

[jfoo1984]

I haven't had a chance to look at updating the specs. I did try to run the migrator spec, but it seemed to be broken due to other reasons.

[jfoo1984]

So even though this allows us to generate a seed with that implicit model, it seems like we can't actually run that generated seed successfully. I'm getting the following error

private constant Citation::HABTM_PathogenListVersions referenced

NameError: private constant Citation::HABTM_PathogenListVersions referenced

[jfoo1984]

@pboling This branch is now working for me. Some points to consider though

• The generated seed generates the HABTM association in both direction, though each statement has a check to make sure the association is not added if it already exists. We could probably improve this by creating a list of HABTM associations and pruning off duplicates
• Are there cases where we might want duplicated HABTM associations?
• Should we add a option / var to control whether we check for HABTM associations?

Update: I'm planning to look into consolidating adding associations into a single statement, and also using sets to avoid generating associations both ways

[jfoo1984]

Addressed issues mentioned above

* The generated seed generates the HABTM association in both direction, though each statement has a check to make sure the association is not added if it already exists.  We could probably improve this by creating a list of HABTM associations and pruning off duplicates

https://github.com/pboling/seed_migration/pull/80/files#diff-6652e1848792ce36ff90d5730fcc4de2252024499451356ce8764d918c9e4614R283-R299 adds a step to go through all registered models, and use a Set to store only unique associations.

Should we add a option / var to control whether we check for HABTM associations?

This seems unnecessary, as these associations are only added if both models are registered. But it would not be too difficult to add this config option if needed.

I'm planning to look into consolidating adding associations into a single statement, and also using sets to avoid generating associations both ways.

This has been added, the seed generates statements to add necessary HABTM association ids in a single statement.

[pboling]

If you have CI working in your fork, please send a PR with those fixes too! 🗡️