PDFLoader

[chrishay]

Hi,
Sorry for the stupid question, but once a PDF is loaded how is it used in the chain as part of the context?
Thanks

[Abraxas-365]

Hi you can check this example

https://github.com/Abraxas-365/langchain-rust/blob/main/examples/qa_chain.rs

Also if you want to create a custom chain you can follow the same steps as the qa_chain

[chrishay]

I have this example of loading the PDF which is working, I just don't know how to add it to the LLM query. Thanks again!

let path = "./test.pdf";

let loader = LoPdfLoader::from_path(path).expect("Failed to create PdfLoader");

    let docs = loader
        .load()
        .await
        .unwrap()
        .map(|d| d.unwrap())
        .collect::<Vec<_>>()

[Abraxas-365]

@chrishay If you pdf is big, i recommend you use some vector store to store the documents, and then using a qa retriver chain.

Lets say you pdf is a small one, you can do this

let path = "./test.pdf";

let loader = LoPdfLoader::from_path(path).expect("Failed to create PdfLoader");

    let your_docs = loader
        .load()
        .await
        .unwrap()
        .map(|d| d.unwrap())
        .collect::<Vec<_>>()
        .await;
        
         let llm = OpenAI::default();
    let chain = StuffDocument::load_stuff_qa(llm);
    let input = chain
        .qa_prompt_builder()
        //You could also get the documents form a retriver
        .documents(&your_docs)
        .question("How old is luis and whats his favorite text editor")
        .build();
        
       let ouput = chain.invoke(input).await.unwrap(); 

[Abraxas-365]

@chrishay if you want you can join the discord, and we can go to the voice channel 😄

[chrishay]

Thanks i will try and join as am getting a compile error, for .documents(&your_docs)
expected &[Document], found &Collect<Map<..., ...>, ...>

[Abraxas-365]

Thatas weird, are you using feature_utils? lets jumpt to the discord

[chrishay]

@chrishay if you want you can join the discord, and we can go to the voice channel 😄
I'm in Discord General channel Thanks :)

[Abraxas-365]

It was missing the las await to joing the streams.

    let path = "./chris.pdf";
    let loader = LoPdfLoader::from_path(path).expect("Failed to create PdfLoader");

    let your_docs = loader
        .load()
        .await
        .unwrap()
        .map(|d| d.unwrap())
        .collect::<Vec<_>>()
        .await; //this was missing

[chrishay]

thank you for all the help

[Abraxas-365]

No worries. Please, @chrishay check your Discord DM.