colivara-py 1.5.1

The official Python SDK for the ColiVara API

colivara-py

The official Python SDK for the ColiVara API. ColiVara is a document search and retrieval API that uses advanced machine learning techniques to index and search documents. This SDK allows you to interact with the API to create collections, upload documents, search for documents, and generate embeddings.

Installation

Install colivara-py using pip:

pip install colivara-py

---

Usage

Refer to the ColiVara API documentation for detailed guidance on how to use this library.

Requirements

• You need access to the ColiVara API, which you can self-host (see ColiVara API repo) or use the hosted version at colivara.com.
• Obtain an API key by signing up at ColiVara or from your self-hosted API.

Example Code

import os
from colivara_py import ColiVara

rag_client = ColiVara(
    api_key=os.environ.get("COLIVARA_API_KEY"),  # Default is `None`
    base_url="https://api.colivara.com"  # Default is `https://api.colivara.com`
)

# Create a new collection (optional)
new_collection = rag_client.create_collection(name="my_collection", metadata={"description": "A sample collection"})
print(f"Created collection: {new_collection.name}")

# Upload a document to the collection
document = rag_client.upsert_document(
    name="sample_document",
    collection_name="my_collection",  # Defaults to "default_collection"
    url="https://example.com/sample.pdf",
    metadata={"author": "John Doe"}
)
print(f"Uploaded document: {document.name}")

# Search for documents
search_results = rag_client.search(
    query="machine learning",
    collection_name="my_collection",
    top_k=3
)
for result in search_results.results:
    print(f"Page {result.page_number} of {result.document_name}: Score {result.normalized_score}")

# List documents in a collection
documents = rag_client.list_documents(collection_name="my_collection")
for doc in documents:
    print(f"Document: {doc.name}, Pages: {doc.num_pages}")

# Generate embeddings
embeddings = rag_client.create_embedding(
    input_data=["This is a sample text for embedding"],
    task="query"
)
print(f"Generated {len(embeddings.data)} embeddings")

# Delete a document
rag_client.delete_document("sample_document", collection_name="my_collection")
print("Document deleted")

---

Development

Setting up the Development Environment

1. Clone the repository and navigate to the project directory:

   cd colivara-py
   

2. Create a virtual environment:

   uv venv
   

3. Activate the virtual environment:

   macOS/Linux:

   source .venv/bin/activate
   

   Windows:

   .venv\Scripts\activate
   

4. Install the development dependencies:

   uv sync --extra dev-dependencies
   

5. Run tests:

   pytest
   

Regenerating the SDK

If the OpenAPI specification is updated, regenerate the SDK as follows:

1. Install the OpenAPI generator (on macOS, use Homebrew):

   brew install openapi-generator
   

2. Verify the installation:

   openapi-generator version
   

3. Run the OpenAPI generator from the project directory:

   openapi-generator generate -i https://api.colivara.com/v1/openapi.json -g python -c config.yaml --ignore-file-override .openapi-generator-ignore --template-dir ./templates
   

---

Updating the SDK and Documentation

Follow these steps for major changes to the OpenAPI spec:

1. Regenerate the SDK using the OpenAPI generator.
2. Update the client interface in colivara_py/client.py. if needed
3. Modify tests in the tests directory to reflect the changes. if needed.
4. Run tests to ensure functionality.

---

Building Documentation Locally

Generate and view the SDK documentation:

1. To serve the documentation locally:

   pdocs server colivara_py
   

2. To generate documentation as HTML:

   pdocs as_html colivara_py --overwrite
   

3. To generate documentation as Markdown:

   pdocs as_markdown colivara_py
   

---

License

This SDK is licensed under the Apache License, Version 2.0. The ColiVara API is licensed under the Functional Source License, Version 1.1, Apache 2.0 Future License. See LICENSE.md for details.

For commercial licensing, contact us via tjmlabs.com. We’re happy to work with you to provide a license tailored to your needs.