nubix-ai 1.1.0

AI library used for document processing and LLM calls.

nubix-ai

AI library used for document processing and LLM calls.

Installation

pip install nubix-ai

Usage

from pydantic import BaseModel
from nubix_ai import NubixAI

# Initialize the client
client = NubixAI(
    docling_api_key="your-docling-api-key",
    openai_api_key="your-openai-api-key"
)

# Process a document
markdown, metadata = client.call_docling_process_file("document.pdf")

# Or send a PDF through the full-PDF LLM markdown endpoint
markdown, metadata = client.call_docling_process_pdf_llm("document.pdf")

# Extract structured data with LLM
class MySchema(BaseModel):
    field1: str
    field2: int

result = client.extract_with_llm(
    markdown_text=markdown,
    prompt_input="Extract the following information:",
    PydanticInput=MySchema
)

Release

Create a PyPI API token for the nubix-ai project and export it as UV_PUBLISH_TOKEN.

export UV_PUBLISH_TOKEN=pypi-...

Publish the current version from pyproject.toml:

uv build
uv publish --index pypi

Or use the helper script from the project root:

scripts/publish.sh

To set the version first and publish in one step:

scripts/publish.sh 1.1.1

To test the upload flow without touching production PyPI:

uv build
uv publish --index testpypi

License

MIT