DeepsetCSVRowsToDocumentsConverter

Read CSV files from various sources and convert each row into Haystack documents.

DeepsetCSVRowsToDocumentsConverter reads a CSV file and converts each row into a Document object, using one column as the document's main content. All other columns are added to the document's metadata.

Key Features

• Converts each CSV row into a separate document — useful for pre-chunked datasets.
• Configurable content column to select which column becomes the document content.
• All remaining columns are automatically stored in document metadata.
• Supports UTF-8 encoding by default with configurable encoding.

Configuration

1. Drag the DeepsetCSVRowsToDocumentsConverter component onto the canvas from the Component Library.
2. Click on the component to open the configuration panel.
3. Configure the component settings:
   • Set the Content Column to specify which CSV column to use as the document content. The default is content.
   • Set the Encoding for the CSV files. The default is UTF-8.

Connections

DeepsetCSVRowsToDocumentsConverter accepts a list of file paths or ByteStream objects through its sources input. It outputs a list of Document objects, one per CSV row.

Because the output documents are already one-row chunks, you can bypass a document splitter and connect directly to an embedder. It typically connects with:

• FileTypeRouter: receives CSV files routed by MIME type.
• SentenceTransformersDocumentEmbedder or other embedders: sends documents directly for embedding.

Usage Examples

Basic Configuration

  DeepsetCSVRowsToDocumentsConverter:
    type: deepset_cloud_custom_nodes.converters.csv_rows_to_documents.DeepsetCSVRowsToDocumentsConverter
    init_parameters:
      content_column: content
      encoding: utf-8

Using the Component in a Pipeline

This is an example of an index that processes multiple file types. It starts with FilesInput followed by file_classifier (FileTypeRouter) which classifies files by type and sends them to an appropriate converter.

DeepsetCSVRowsToDocumentsConverter receives CSV files from file_classifier (FileTypeRouter) and outputs a list of pre-chunked documents. Since these documents are already chunked, they bypass the splitter (DeepsetDocumentSplitter) and go directly to the document_embedder (SentenceTransformersDocumentEmbedder) and finally to the writer (DocumentWriter), which writes them into the document store.

YAML configuration:

# haystack-pipeline
components:
  file_classifier:
    type: haystack.components.routers.file_type_router.FileTypeRouter
    init_parameters:
      mime_types:
      - text/markdown
      - text/html
      - text/csv
  markdown_converter:
    type: haystack.components.converters.markdown.MarkdownToDocument
    init_parameters: {}
  html_converter:
    type: haystack.components.converters.html.HTMLToDocument
    init_parameters:
      extraction_kwargs:
        output_format: txt
        target_language:
        include_tables: true
        include_links: false
  document_embedder:
    type: haystack.components.embedders.sentence_transformers_document_embedder.SentenceTransformersDocumentEmbedder
    init_parameters:
      model: intfloat/e5-base-v2
  writer:
    type: haystack.components.writers.document_writer.DocumentWriter
    init_parameters:
      document_store:
        type: haystack_integrations.document_stores.opensearch.document_store.OpenSearchDocumentStore
        init_parameters:
          embedding_dim: 768
          similarity: cosine
          hosts:
          index: ""
          max_chunk_bytes: 104857600
          return_embedding: false
          method:
          mappings:
          settings:
            index.knn: true
          create_index: true
          http_auth:
          use_ssl:
          verify_certs:
          timeout:
      policy: OVERWRITE
  DeepsetCSVRowsToDocumentsConverter:
    type: deepset_cloud_custom_nodes.converters.csv_rows_to_documents.DeepsetCSVRowsToDocumentsConverter
    init_parameters:
      content_column: content
      encoding: utf-8
  DocumentSplitter:
    type: haystack.components.preprocessors.document_splitter.DocumentSplitter
    init_parameters:
      split_by: word
      split_length: 200
      split_overlap: 0
      split_threshold: 0
      splitting_function:
      respect_sentence_boundary: false
      language: en
      use_split_rules: true
      extend_abbreviations: true
      skip_empty_documents: true

connections:
- sender: file_classifier.text/markdown
  receiver: markdown_converter.sources
- sender: file_classifier.text/html
  receiver: html_converter.sources
- sender: DeepsetCSVRowsToDocumentsConverter.documents
  receiver: document_embedder.documents
- sender: document_embedder.documents
  receiver: writer.documents
- sender: file_classifier.text/csv
  receiver: DeepsetCSVRowsToDocumentsConverter.sources
- sender: markdown_converter.documents
  receiver: DocumentSplitter.documents
- sender: html_converter.documents
  receiver: DocumentSplitter.documents
- sender: DocumentSplitter.documents
  receiver: document_embedder.documents

metadata: {}

inputs:
  files:
  - file_classifier.sources

max_runs_per_component: 100

Parameters

Inputs

Parameter	Type	Default	Description
sources	List[Union[str, Path, ByteStream]]		List of CSV file paths (str or Path) or ByteStream objects.
meta	Optional[Union[Dict[str, Any], List[Dict[str, Any]]]]	None	Optional metadata to attach to the documents. Can be a single dict or a list of dicts.

Outputs

Parameter	Type	Description
documents	List[Document]	A list of Haystack Documents, one per CSV row.

Init Parameters

These are the parameters you can configure in Pipeline Builder:

Parameter	Type	Default	Description
content_column	str	content	Name of the column to use as content when processing the CSV file.
encoding	str	utf-8	Encoding type to use when reading the files.

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]]		List of CSV file paths (str or Path) or ByteStream objects.
meta	Optional[Union[Dict[str, Any], List[Dict[str, Any]]]]	None	Optional metadata to attach to the documents. Can be a single dict or a list of dicts.