justmars / corpus-base Goto Github PK

The various utils functions can be absored by toolkit. Also, it includes the pdf extraction functionality for non-elibrary content.

Each segment can become a subject of FTS. Consider the discussion in the sqlite forum re: ebooks being fit for FTS.

Since the determination of opinion content is a long-running process, can consider adding segments to the data file for inclusions.

Because of deletion by prefix, we have a regression error in the form of auto-generated m2m table names.

Example:

1. pax_tbl_individuals_sc_tbl_decisions is created in corpus-base
2. setup_pax deletes all tables prefixed with pax
3. Deletion will include pax_tbl_individuals_sc_tbl_decisions because when it is created, the naming is auto-generated.

Because of this side-effect it might be more prudent to explicitly declare m2m table names when they are prefixed differently.