New Feature: extract from string

[doncat99]

# Initialize the extractor
extractor = Extractor()
extractor.load_document_loader(DocumentLoaderPyPdf())
extractor.load_llm("gpt-4o-mini")  # or any other supported model

# Extract data from the document
result = extractor.extract(test_file_path, InvoiceContract)

above is the standard usage of ExtractThinker.

What if I already have custom processing for the PDF document, such as removing headers and footers and filtering out the target string from the PDF document, and I want the extractor to continue based on my pdf_string?

[enoch3712]

Hello @doncat99!

Yes, sounds great!

There are two ways to do this:

You just do an implementation of documentLoader, that will just return this content. The only problem here is that the structure needs to be defined in a specific way, in you just return will not work.

Another way to do this, is just in the extract you can pass the raw content to enrich the extraction:

We can also make a DcoumenLoader just for injection, that checks that everything is correct.

I can do that

[doncat99]

@enoch3712, thanks for the hint!

The code works like below.

extractor.set_skip_loading(skip=True)
extractor.extract("string", Contract, instruction)
extractor.set_skip_loading(skip=False)

[enoch3712]

Hello @doncat99!

yes, sounds good!

[enoch3712]

Gonna take a look at this for the next release