CLI Reference

The CLI is invoked via:

python3 -m alligator.cli [OPTIONS]

All options are auto-exposed from Alligator.__init__ under the --gator.* namespace using jsonargparse.

Arguments

Argument	Default	Description
--gator.input_csv	—	Required. Path to the input CSV file
--gator.entity_retrieval_endpoint	$ENTITY_RETRIEVAL_ENDPOINT	Entity lookup API URL
--gator.entity_retrieval_token	$ENTITY_RETRIEVAL_TOKEN	API auth token
--gator.mongo_uri	mongodb://gator-mongodb:27017	MongoDB connection URI
--gator.target_columns	None	JSON string with NE/LIT/IGNORED column assignments
--gator.column_types	{}	JSON map of col_idx → [QID, ...] for type-constrained search
--gator.num_workers	cpu_count // 2	Number of parallel retrieval workers
--gator.worker_batch_size	64	Number of rows per worker batch
--gator.candidate_retrieval_limit	20	Max candidates fetched per entity
--gator.max_candidates_in_result	5	Max candidates kept in final output
--gator.num_ml_workers	2	Number of ML pipeline workers
--gator.ml_worker_batch_size	256	Batch size for ML prediction
--gator.candidate_retrieval_only	False	Stop after Phase 2 (skip ML ranking)
--gator.save_output	False	Persist output to MongoDB
--gator.save_output_to_csv	False	Write results to a CSV file
--disable-logging	False	Fully suppress all logging output

Examples

Run on a CSV with defaults

python3 -m alligator.cli --gator.input_csv tables/my_table.csv

With explicit column types

python3 -m alligator.cli \
  --gator.input_csv tables/imdb_top_100.csv \
  --gator.column_types '{"0": ["Q11424"], "7": ["Q5"]}' \
  --gator.num_workers 8

Manual column type assignment

python3 -m alligator.cli \
  --gator.input_csv tables/my_table.csv \
  --gator.target_columns '{"NE": {"0": "OTHERS", "2": "LOC"}, "LIT": {"1": "NUMBER"}, "IGNORED": [3]}' \
  --gator.num_workers 4

Candidate retrieval only (skip ML)

python3 -m alligator.cli \
  --gator.input_csv tables/my_table.csv \
  --gator.candidate_retrieval_only true

Save results to CSV and suppress logging

python3 -m alligator.cli \
  --gator.input_csv tables/my_table.csv \
  --gator.save_output true \
  --gator.save_output_to_csv true \
  --disable-logging

Config File

Because the CLI uses jsonargparse, you can also pass a YAML or JSON config file:

python3 -m alligator.cli --config my_config.yaml

Where my_config.yaml might look like:

gator:
  input_csv: tables/my_table.csv
  num_workers: 8
  num_ml_workers: 4
  worker_batch_size: 64
  candidate_retrieval_limit: 20
  save_output: true
  save_output_to_csv: true