llama-index-readers-fastocr 0.1.0

A LlamaIndex reader that reads scanned PDFs and images with FastOCR and returns their text.

Website · Docs · Get an API key

Quickstart

pip install llama-index-readers-fastocr
export FASTOCR_API_KEY="your_api_key"

from llama_index.readers.fastocr import FastOCRReader

documents = FastOCRReader().load_data("scanned_contract.pdf")
print(documents[0].text)

Python 3.10 or newer. The distribution is llama-index-readers-fastocr and the import is llama_index.readers.fastocr.

What you get back

FastOCRReader.load_data returns one Document per file.

The FastOCR keys (fastocr_document_id, pages_billed, status, searchable_pdf_path) are left out of the text that embedding models and LLMs see. Pass extra_info={...} to load_data to add your own metadata.

If the account runs out of API pages partway through a file, you still get the text of the pages that were processed, with status set to partial and a warning in the log. A document that fails raises the error from fastocr-sdk with its message and code.

Languages and files

• 100+ languages, detected automatically. No language setting is needed.
• PDF files up to 1 GiB, and PNG, JPG, JPEG, GIF, WebP and BMP images up to 20 MiB. An image counts as one page.

Get an API key and pages

Any signed-in FastOCR account can create an API key and start right away.

1. Sign in and create an API key in the developer console. You accept the API terms with your first key.
2. Set the key as FASTOCR_API_KEY, or pass api_key= to the reader.
3. When you need more than the free pages, buy a prepaid API page pack by card in the developer console.

Each account gets 50 free document pages per month.

Examples

RAG over scanned PDFs

from llama_index.core import VectorStoreIndex
from llama_index.readers.fastocr import FastOCRReader

documents = FastOCRReader().load_data("scanned_contract.pdf")
index = VectorStoreIndex.from_documents(documents)
print(index.as_query_engine().query("When can the contract be terminated?"))

This example uses LlamaIndex's default OpenAI models, so it needs OPENAI_API_KEY. Set another embedding model and LLM in Settings to use a different provider.

A folder of scans

from pathlib import Path

from llama_index.readers.fastocr import FastOCRReader

scans = sorted(p for p in Path("scans").iterdir() if p.suffix.lower() in {".pdf", ".png", ".jpg", ".jpeg"})
for document in FastOCRReader().load_data(scans):
    print(document.metadata["file_name"], document.metadata["pages_billed"])

The reader runs one file at a time and returns when all files are done.

Use it with SimpleDirectoryReader

from llama_index.core import SimpleDirectoryReader
from llama_index.readers.fastocr import FastOCRReader

documents = SimpleDirectoryReader(
    "scans",
    file_extractor={".pdf": FastOCRReader(), ".png": FastOCRReader(), ".jpg": FastOCRReader()},
).load_data()

Searchable PDFs for an archive

from llama_index.readers.fastocr import FastOCRReader

documents = FastOCRReader(output="pdf").load_data(["invoice-001.pdf", "invoice-002.pdf"])
for document in documents:
    print(document.metadata["searchable_pdf_path"])

Each searchable PDF is saved beside its input, as invoice-001.searchable.pdf. An existing file with that name is replaced. Index the text and keep the PDF.

Options

load_data(file_path, extra_info=None) takes one path or a list of paths.

Support and license

Questions: support@fastocr.org. Bugs and requests: GitHub issues. MIT license; see LICENSE.