docs: update how_to docs to reflect new loaders interface and functionalities

[MedlockM]

• Description: Updates two notebooks in the how_to documentation to reflect new loader interfaces and functionalities.
• Issue: Some how_to notebooks were still using loader interfaces from previous versions of LangChain and did not demonstrate the latest loader functionalities (e.g., extracting images with ImageBlobParser, extracting tables in specific output formats, parsing documents using Vision-Language Models with ZeroxPDFLoader, and using CloudBlobLoader in the GenericLoader, etc.).
• Dependencies: py-zerox
• Twitter handle: @MarcMedlock2

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
langchain	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	May 13, 2025 9:26pm

[MedlockM]

@ccurme
Could you please justify your changes? Especially concerning the pdf guide. For example, why not keep the part explaining the new method for parsing with a vision model using Zerox? The way it's currently presented (convert pdf to an image then send the image to a chat object) is "old-fashioned", isn't it?

Likewise, the fact that you can set up an ImageBlobParser directly in a loader to extract tables or images with different strategies is a new feature that should be highlighted, don't you think?