Requested feature

Background

• Docling's DocumentConverter supports several formats like PDF, HTML, or .docx, which allows the conversion of those files into a DoclingDocument object. Each supported format is enumerated as an InputFormat instance.
• The DocumentConverter is leveraged in many integrations. For instance, in LlamaIndex the DoclingReader can be leveraged in SimpleDirectoryReader to convert PDF files

  from llama_index.core import SimpleDirectoryReader
  from llama_index.readers.docling import DoclingReader
  
  reader = DoclingReader(export_type=DoclingReader.ExportType.JSON)
  dir_reader = SimpleDirectoryReader(
      input_dir="/my/temp/dir",
      file_extractor={".pdf": reader}
  )

• Since the conversion may be computationally costly, users may want to persist the converted documents as .json files and use them later in other data processing pipelines

Request

• Create a new conversion backend that simply reads the content of a JSON file that contains a DoclingDocumentexport.
• In this way, the pattern example above could be reused

  from llama_index.core import SimpleDirectoryReader
  from llama_index.readers.docling import DoclingReader
  
  reader = DoclingReader(export_type=DoclingReader.ExportType.JSON)
  dir_reader = SimpleDirectoryReader(
      input_dir=/my/temp/dir",
      file_extractor={".json": reader}
  )

Alternatives

• Delegate the reading of JSON DoclingDocument files to integration frameworks (e.g., create another DoclingReader for JSON in LlamaIndex)
• Delegate the reading of JSON to docling-core