xref computation for Alfred; the xref.py CLI takes a collectionID at the first
argument, e.g. ./xref.py 7.
The main.go programs reads from the xref river queue to run that CLI. This is
how Alfred runs it.
Use -dev to output text logs (rather than JSON), and -debug to enable debug logs (these work for both xref.py and main.go).
Configuration via environment; the most useful ones with with their defaults:
ALFRED_ES http://elastic:[email protected]:9200
ALFRED_DB postgresql://aleph:[email protected]/aleph
ALFRED_DB_FTM postgresql://aleph:[email protected]/aleph_ftm
FTM_COMPARE_MODEL ./data/model.pkl
FTM_COMPARE_FREQUENCIES_DIR ./data/word_frequencies
See xref.py for a full list.
Note that the river tables are expected to be on the ALFRED_DB_FTM database, rather than the regular one (as per Alfred).
Run setup.sh to download the models and stuff, or run it via the Dockerfile.