fix(falkordb): skip group_id tokenized away by RedisSearch

[bsolomon1124]

Summary

Fixes #1319; fixes FalkorDB default group ID and handling of the ID in query construction.

uv run examples/quickstart/quickstart_falkordb.py now succeeds.

Type of Change

• Bug fix

Objective

Fix a bug related to the default group ID for FalkorDB/RedisSearch.

Testing

• Unit tests added/updated
• Integration tests added/updated
• All existing tests pass

Breaking Changes

• This PR contains breaking changes

Luckily, this PR fixes other breaking changes that broke functionality of FalkorDB driver when used with the default group_id.

PR #914 attempted to fix RedisSearch syntax. However, in attempting to deal with the fact that _ is a separator/tokenizer character in RedisSearch (resulting in @group_id:"_" matching nothing), the PR changed get_default_group_id for FalkorDB from '_' to '\\_' while keeping validate_group_id regex as is.

Fix:

• Revert the default group_id for FalkorDB back to '_'
• Skip the @group_id: filter when the group_id would be tokenized away by RedisSearch

Checklist

• Code follows project style guidelines (make lint passes)
• Self-review completed
• Documentation updated where necessary
• No secrets or sensitive information committed

Related Issues

Closes #1319

[danielchalef]

All contributors have signed the CLA ✍️ ✅
_{Posted by the CLA Assistant Lite bot.}

[bsolomon1124]

I have read the CLA Document and I hereby sign the CLA

[bsolomon1124]

One thing: if a group_id is something like "foo_bar", sanitize would replace the underscore with a space, producing "foo bar" — which would match nodes with group_id "foo" OR "bar" instead of the exact "foo_bar". The quoting handles this for the fulltext query, but worth checking that the Cypher fallback still filters correctly in that case.

Thanks for the review @themavik . I'm not sure if I'm following here, can you elaborate?

if a group_id is something like "foo_bar", sanitize would replace the underscore with a space, producing "foo bar" — which would match nodes with group_id "foo" OR "bar" instead of the exact "foo_bar".

The sanitized value is only used for a truthiness check; it's not used to construct the fulltext query.

https://github.com/getzep/graphiti/pull/1330/changes#diff-99e252c05fd20e7c1e63fdaaac51f39bb8e6cce92c97c1b9e7c4ac8de469f37eR411

In the case of group_ids=["foo", "bar"], the resulting Cypher is WHERE n.group_id IN ["foo_bar"] since the sanitized value is discarded after the check.