DoclingConverter

Converts PDFs and other supported files into Markdown documents using Docling, an open source document processing library.

GPU Acceleration

DoclingConverter runs faster with GPU acceleration. Enable GPU acceleration in the pipeline settings to improve performance:

1. Go to Pipelines and click the pipeline that contains the DoclingConverter component. You're redirected to the Pipeline Details page.
2. Go to Settings and click the GPU Acceleration toggle to turn it on.

For details, see GPU Acceleration.

Key Features

• Converts PDFs and other supported file types to Markdown content.
• Runs OCR and preserves table structure during conversion.
• Inserts page break placeholders to maintain document structure.
• Extracts and flattens Docling metadata so downstream components can filter by Docling-specific fields.
• Accepts file paths, Path objects, or ByteStream inputs.
• For more information about Docling, see the Docling website.

Configuration

1. Drag the DoclingConverter component onto the canvas from the Component Library.
2. Click the component to open the configuration panel.
3. Configure the parameters as needed.

Connections

DoclingConverter accepts file paths, Path objects, or ByteStream inputs as sources, and optional metadata as meta. It outputs a list of Document objects (documents).

Connect the pipeline's file input to the sources input. Connect the documents output to DocumentSplitter, DocumentJoiner, or DocumentWriter for further processing.

Usage Example

Using the Component in an Index

In this example, DoclingConverter receives files from the Input component, then sends the processed documents to DocumentSplitter.

components:
  docling_converter:
    type: deepset_cloud_custom_nodes.converters.docling_converter.DoclingConverter
    init_parameters:
      do_picture_classification: false
  splitter:
    type: haystack.components.preprocessors.document_splitter.DocumentSplitter
    init_parameters:
      split_by: word
      split_length: 250
      split_overlap: 30
  writer:
    type: haystack.components.writers.document_writer.DocumentWriter
    init_parameters:
      document_store:
        type: haystack_integrations.document_stores.opensearch.document_store.OpenSearchDocumentStore
        init_parameters:
          hosts:
          - ${OPENSEARCH_HOST}
          index: docling-demo
          max_chunk_bytes: 104857600
          embedding_dim: 768
          return_embedding: false
          method:
          mappings:
          settings:
          create_index: true
          http_auth:
          use_ssl:
          verify_certs:
          timeout:
      policy: SKIP

connections:
- sender: docling_converter.documents
  receiver: splitter.documents
- sender: splitter.documents
  receiver: writer.documents

inputs:
  files:
  - docling_converter.sources

max_runs_per_component: 100

metadata: {}

Parameters

Inputs

Parameter	Type	Default	Description
sources	List[Union[str, Path, ByteStream]]		File paths, Path objects, or ByteStreams to convert.
meta	Optional[Union[Dict[str, Any], List[Dict[str, Any]]]]	None	Metadata applied to every generated document or aligned per source.

Outputs

Parameter	Type	Default	Description
documents	List[Document]		Documents created from the Docling Markdown export.

Init Parameters

These are the parameters you can configure in Pipeline Builder:

Parameter	Type	Default	Description
do_ocr	bool	True	Enable OCR when Docling processes PDF pages.
do_table_structure	bool	True	Preserve table structure information during conversion.
do_code_enrichment	bool	False	Enable Docling code OCR to capture code blocks.
do_formula_enrichment	bool	False	Enable formula OCR to keep math expressions readable.
do_picture_classification	bool	False	Attach Docling picture classification metadata.
do_picture_description	bool	False	Add Docling generated picture descriptions to metadata.
page_break_placeholder	str	"\f"	Token inserted whenever Docling encounters a page break.
filter_binary_hash	bool	True	Remove Docling binary_hash fields from metadata.
pipeline_options	PdfPipelineOptions | None	None	Advanced Docling pipeline options for Haystack users. Overrides the individual boolean flags.
convert_kwargs	Dict[str, Any] | None	None	Additional arguments forwarded to DocumentConverter.convert(), such as headers or max_num_pages.
md_export_kwargs	Dict[str, Any] | None	None	Overrides for Docling Markdown export, for example image_placeholder or custom formatting flags.

Run Method Parameters

These are the parameters you can configure for the component's run() method. This means you can pass these parameters at query time through the API, in Playground, or when running a job. For details, see Modify Pipeline Parameters at Query Time.

Parameter	Type	Default	Description
sources	List[Union[str, Path, ByteStream]]		Paths or ByteStreams for the files you want to convert.
meta	Optional[Union[Dict[str, Any], List[Dict[str, Any]]]]	None	Metadata applied to every generated document or aligned per source list.