fastrag.ingestion.steps#

Step base classes and the policy-aware evaluator.

Re-exports PipelineStep, ExecutionContext, SourceStep, DocumentStep, and RoutedStep from fastrag.ingestion.steps.base, their canonical module. Concrete steps (fetchers, parsers, chunkers, embedders) live in sibling packages: fetching, parsing, chunking, embedding. The pipeline’s storage backend, VectorStore, lives outside this package entirely, at fastrag.vector_store — it is queried at read time as well as written at ingest time, so it is not an ingestion-only concern.

Modules

base

PipelineStep, ExecutionContext, and the policy-aware evaluator.

chunking

Chunking: parsed documents in, embeddable chunks out.

embedding

Embedding: documents in, the same documents with vectors attached out.

fetching

Fetching: locations in, Source objects out.

parsing

Source parsing: raw bytes in, documents out.