process_document()

Process a file through the full 7-stage pipeline.

Signature


def process_document(
    source: str | Path,
    *,
    config: DistillConfig | None = None,
    format: str | None = None,
    embed: bool = True,
) -> ProcessingResult:

Parameters

Parameter	Type	Default	Description
`source`	`str \| Path`	required	File path to process
`config`	`DistillConfig`	`None`	Pipeline configuration (uses defaults if None)
`format`	`str`	`None`	Override format detection (e.g., “pdf”, “docx”)
`embed`	`bool`	`True`	Whether to generate embeddings

Returns

ProcessingResult with:


class ProcessingResult:
    document: Document          # structured document with sections
    chunks: list[DocumentChunk] # chunked and optionally embedded/enriched
    validation: ValidationReport # coverage metrics at each stage

Examples

Basic usage


from distillcore import process_document
 
result = process_document("report.pdf")

With configuration


from distillcore import process_document, DistillConfig, ChunkConfig
 
result = process_document(
    "report.pdf",
    config=DistillConfig(
        chunk=ChunkConfig(target_tokens=300),
        enrich_chunks=False,
    ),
)

With custom embedding function


from distillcore import process_document, DistillConfig, EmbeddingConfig
from distillcore.embedding import ollama_embedder
 
config = DistillConfig(
    embedding=EmbeddingConfig(embed_fn=ollama_embedder()),
)
result = process_document("report.pdf", config=config)

Without embeddings


result = process_document("report.pdf", embed=False)

Async Version


from distillcore import process_document_async
 
result = await process_document_async("report.pdf")

Same parameters and return type. Extraction is offloaded to a thread to avoid blocking the event loop.