workspacex 0.1.4

workspacex is a Python library for managing AIGC (AI-Generated Content) artifacts. It provides a collaborative workspace environment for handling multiple artifacts with features like version control, update notifications, artifact management, and pluggable storage and embedding backends.

workspacex

workspacex is a Python library for managing AIGC (AI-Generated Content) artifacts. It provides a collaborative workspace environment for handling multiple artifacts with features like version control, update notifications, artifact management, and pluggable storage and embedding backends.

---

Features

• Artifact Management: Create, update, and manage different types of artifacts (text, code, etc.)
• Workspace Organization: Group related artifacts in collaborative workspaces
• Storage Backends: Local file system and S3-compatible storage (via s3fs)
• Embedding Backends: OpenAI-compatible and Ollama embedding support
• Reranking: Local reranking using Qwen3-Reranker models
• HTTP Service: FastAPI-based reranking service

---

Installation

Basic Installation

pip install workspacex

With Reranker Support

pip install "workspacex[reranker]"  # For using reranker in your code
pip install "workspacex[reranker-server]"  # For running the reranker HTTP service

Using Poetry:

poetry install --extras "reranker-server"  # Installs all features

---

Usage

Basic Example

import asyncio
import logging

from workspacex import WorkSpace, ArtifactType

if __name__ == '__main__':
    workspace = WorkSpace.from_local_storages(workspace_id="demo")
    asyncio.run(workspace.create_artifact(ArtifactType.TEXT, "artifact_001"))

Using the Reranker

from workspacex.reranker.base import RerankConfig
from workspacex.reranker.local import Qwen3RerankerRunner
from workspacex.artifact import Artifact, ArtifactType

# Initialize reranker
config = RerankConfig(
    model_name="Qwen/Qwen3-Reranker-0.6B",  # or "Qwen/Qwen3-Reranker-8B"
    api_key="not_needed",  # Local model doesn't need these
    base_url="not_needed"
)
reranker = Qwen3RerankerRunner(config)

# Create some test documents
documents = [
    Artifact(artifact_type=ArtifactType.TEXT, content="Python is a programming language."),
    Artifact(artifact_type=ArtifactType.TEXT, content="Python is a type of snake.")
]

# Rerank documents
results = reranker.run(
    query="What is Python programming?",
    documents=documents,
    top_n=2
)

# Print results
for result in results:
    print(f"Score: {result.score}, Content: {result.artifact.content}")

Running the Reranker Server

1. Install server dependencies:

pip install "workspacex[reranker-server]"

2. Start the server:

python -m workspacex.reranker.server.reranker_server

The server will start on http://localhost:8000 with the following endpoints:

• POST /rerank: Main reranking endpoint
• GET /health: Health check endpoint
• Interactive API docs at /docs and /redoc

Example API usage:

curl -X POST "http://localhost:8000/rerank" \
     -H "Content-Type: application/json" \
     -d '{
       "query": "What is Python?",
       "documents": [
         {
           "content": "Python is a programming language.",
           "metadata": {}
         },
         {
           "content": "Python is a type of snake.",
           "metadata": {}
         }
       ],
       "top_n": 2,
       "score_threshold": 0.5
     }'

Response format:

{
  "results": [
    {
      "content": "Python is a programming language.",
      "metadata": {},
      "score": 0.95
    },
    {
      "content": "Python is a type of snake.",
      "metadata": {},
      "score": 0.45
    }
  ]
}

Storage Backends

• Local: Default, stores data in the local file system.

  from workspacex.storage.local import LocalPathRepository
  repo = LocalPathRepository("data/workspaces/demo")
  

• S3: Store artifacts in S3-compatible storage.

  from workspacex.storage.s3 import S3Repository
  repo = S3Repository(storage_path="demo", bucket="your-bucket", s3_kwargs={"key": "...", "secret": "..."})
  

Embedding Backends

• OpenAI-Compatible:

  from workspacex.embedding.openai_compatible import OpenAICompatibleEmbeddings, EmbeddingsConfig
  config = EmbeddingsConfig(api_key="sk-...", base_url="https://api.openai.com/v1", model_name="text-embedding-ada-002")
  embedder = OpenAICompatibleEmbeddings(config)
  

• Ollama:

  from workspacex.embedding.ollama import OllamaEmbeddings, OllamaConfig
  config = OllamaConfig(model="nomic-embed-text", base_url="http://localhost:11434")
  embedder = OllamaEmbeddings(config)
  

Example Scripts

• See src/examples/ for ready-to-run scripts:
  • noval_example.py
  • embeddings/openai_example.py
  • embeddings/ollama_embedding_example.py
  • image_examples.py

Run an example:

export PYTHONPATH=src
python src/examples/embeddings/openai_example.py

---

Notes

• All source code is under src/.
• Make sure to activate the correct conda environment before using Poetry commands or running code.
• If you see ModuleNotFoundError: No module named 'workspacex', ensure your PYTHONPATH includes src.
• Storage and embedding backends are pluggable and extensible.
• For S3 support, install s3fs and configure credentials as needed.
• For reranking, CUDA is recommended for better performance.
• The reranker server supports both CPU and GPU inference.

---

Let me know if you want to add more details, such as advanced usage, API docs, or contribution guidelines!