Is your feature request related to a problem? Please describe.
The OpenDataLoader PDF integration currently extracts document text but does not expose extracted images, which limits its use in multimodal RAG pipelines.
Describe the solution you'd like
Add an optional extract_images parameter that extracts images and returns them as Haystack Document objects.
Describe alternatives you've considered
N/A
Additional context
Haystack already has components such as ImageFileToImageContent, PDFToImageContent, and DocumentToImageContent which expose images as ImageContent objects for multimodal pipelines. Existing Jina and Cohere integrations also support image-based Documents using meta["file_path"]. The OpenDataLoader integration could follow a similar pattern when exposing images extracted from documents.
Is your feature request related to a problem? Please describe.
The OpenDataLoader PDF integration currently extracts document text but does not expose extracted images, which limits its use in multimodal RAG pipelines.
Describe the solution you'd like
Add an optional
extract_imagesparameter that extracts images and returns them as HaystackDocumentobjects.Describe alternatives you've considered
N/A
Additional context
Haystack already has components such as
ImageFileToImageContent,PDFToImageContent, andDocumentToImageContentwhich expose images asImageContentobjects for multimodal pipelines. Existing Jina and Cohere integrations also support image-based Documents usingmeta["file_path"]. The OpenDataLoader integration could follow a similar pattern when exposing images extracted from documents.