The Beast

A flexible, declarative toolkit for transforming machine-readable data into FollowTheMoney (FTM) entities.

The Beast is currently in beta and is battle-tested in production on hundreds of data sources. While the mapping format may evolve for better flexibility, changes are introduced cautiously.

Installation

pip install thebeast

Quick Start

1. Write a YAML mapping that describes how to read your source data and transform it into FTM entities.
2. Run the mapping:

beast mapping.yaml

3. Or sample a small fraction first:

beast-sample mapping.yaml --fraction 0.01

Features

• Declarative YAML mappings - define data transformations without writing code
• Multiple input formats - CSV, TSV, JSON, JSONL, with support for compressed and remote files (via smart_open)
• Rich property pipelines - column extraction, literals, Jinja2 templates, regex operations, transformers, augmentors
• Nested collections - handle hierarchical data with JMESPath traversal
• Statement metadata - attach provenance at dataset, collection, or property level
• Multiprocessing - parallel digest for CPU-bound workloads
• Built-in transformers - date parsing, phone/email normalization, transliteration, and more
• FTM schema validation - entities are validated against FollowTheMoney schemas
• Custom FTM ontologies - extend or replace the standard FTM model with your own schemas

Mapping Example

id: my_dataset

ingest:
  cls: thebeast.ingest.CSVDictReader
  params:
    input_uri: ./people.csv

digest:
  cls: thebeast.digest.SingleProcessDigestor
  meta:
    dataset: { literal: MY_DATASET }
  collections:
    persons:
      path: "[@]"
      entities:
        person:
          schema: Person
          keys:
            - record.id
          properties:
            name:
              template: "{{ record.first }} {{ record.last }}"
            birthDate:
              column: birth
            email:
              column: emails
              regex_split: "[;,]"

dump:
  cls: thebeast.dump.StatementsCSVWriter
  params:
    output_uri: ./output.csv
    error_uri: ./errors.csv

Documentation

Full documentation is available in docs/README.md, covering:

• Mapping format and all property operations
• Ingestors, digestors, and dumpers
• Statement metadata and provenance
• Nested collections and entity references
• Record and property transformers
• Sampling and testing workflows

Running Tests

pip install thebeast[dev]
python -m pytest thebeast/tests/ -v

License

MIT