juniper-data-client 0.4.0

HTTP client for the JuniperData dataset generation service

juniper-data-client

Python client library for the JuniperData REST API.

Overview

juniper-data-client provides a simple, robust client for interacting with the JuniperData dataset generation service. It is the official client library used by both JuniperCascor (neural network backend) and juniper-canopy (web dashboard).

Ecosystem Compatibility

This package is part of the Juniper ecosystem. Compatible with:

juniper-data	juniper-cascor	juniper-canopy
0.4.x	0.3.x	0.2.x

Installation

pip install juniper-data-client

Or install from source:

cd juniper-data-client
pip install -e .

Quick Start

from juniper_data_client import JuniperDataClient

# Create client (default: localhost:8100)
client = JuniperDataClient("http://localhost:8100")

# Check service health
health = client.health_check()
print(f"Service status: {health['status']}")

# Create a spiral dataset
result = client.create_spiral_dataset(
    n_spirals=2,
    n_points_per_spiral=100,
    noise=0.1,
    seed=42,
)
dataset_id = result["dataset_id"]
print(f"Created dataset: {dataset_id}")

# Download as numpy arrays
arrays = client.download_artifact_npz(dataset_id)
X_train = arrays["X_train"]  # (160, 2) float32
y_train = arrays["y_train"]  # (160, 2) float32 one-hot
X_test = arrays["X_test"]    # (40, 2) float32
y_test = arrays["y_test"]    # (40, 2) float32 one-hot

print(f"Training samples: {len(X_train)}")
print(f"Test samples: {len(X_test)}")

Features

• Simple API: Easy-to-use methods for all JuniperData endpoints
• Automatic Retries: Built-in retry logic for transient failures (429, 5xx)
• Connection Pooling: Efficient HTTP connection reuse
• Type Hints: Full type annotations for IDE support
• Context Manager: Resource cleanup with with statement
• Custom Exceptions: Granular error handling

Usage Examples

Context Manager

with JuniperDataClient("http://localhost:8100") as client:
    result = client.create_spiral_dataset(seed=42)
    arrays = client.download_artifact_npz(result["dataset_id"])
# Session automatically closed

Wait for Service

client = JuniperDataClient("http://localhost:8100")

# Wait up to 30 seconds for service to be ready
if client.wait_for_ready(timeout=30):
    result = client.create_spiral_dataset(seed=42)
else:
    print("Service not available")

Custom Parameters

# Using the general create_dataset method
result = client.create_dataset(
    generator="spiral",
    params={
        "n_spirals": 3,
        "n_points_per_spiral": 200,
        "noise": 0.2,
        "seed": 12345,
        "algorithm": "legacy_cascor",
        "radius": 10.0,
    }
)

Error Handling

from juniper_data_client import (
    JuniperDataClient,
    JuniperDataConnectionError,
    JuniperDataNotFoundError,
    JuniperDataValidationError,
)

client = JuniperDataClient()

try:
    result = client.create_dataset("spiral", {"n_spirals": -1})
except JuniperDataValidationError as e:
    print(f"Invalid parameters: {e}")
except JuniperDataConnectionError as e:
    print(f"Service unreachable: {e}")

PyTorch Integration

import torch
from juniper_data_client import JuniperDataClient

client = JuniperDataClient()
result = client.create_spiral_dataset(seed=42)
arrays = client.download_artifact_npz(result["dataset_id"])

# Convert to PyTorch tensors
X_train = torch.from_numpy(arrays["X_train"])  # torch.float32
y_train = torch.from_numpy(arrays["y_train"])  # torch.float32

API Reference

JuniperDataClient

Method	Description
health_check()	Get service health status
is_ready()	Check if service is ready (boolean)
wait_for_ready(timeout)	Wait for service to become ready
list_generators()	List available generators
get_generator_schema(name)	Get parameter schema for generator
create_dataset(generator, params)	Create dataset with generator
create_spiral_dataset(**kwargs)	Convenience method for spiral datasets
list_datasets(limit, offset)	List dataset IDs
get_dataset_metadata(id)	Get dataset metadata
download_artifact_npz(id)	Download NPZ as dict of arrays
download_artifact_bytes(id)	Download raw NPZ bytes
get_preview(id, n)	Get JSON preview of samples
delete_dataset(id)	Delete a dataset
list_versions(name)	List version history for a dataset
get_latest(name)	Get latest version by name
batch_delete(dataset_ids)	Delete multiple datasets
batch_create(datasets)	Create multiple datasets
batch_update_tags(dataset_ids, add_tags, remove_tags)	Update tags on multiple datasets
batch_export(dataset_ids)	Export multiple datasets as ZIP
close()	Close the client session

Exceptions

Exception	Description
JuniperDataClientError	Base exception for all errors
JuniperDataConnectionError	Connection to service failed
JuniperDataTimeoutError	Request timed out
JuniperDataNotFoundError	Resource not found (404)
JuniperDataValidationError	Invalid parameters (400/422)

NPZ Artifact Schema

Downloaded artifacts contain the following numpy arrays (all float32):

Key	Shape	Description
X_train	(n_train, 2)	Training features
y_train	(n_train, n_classes)	Training labels (one-hot)
X_test	(n_test, 2)	Test features
y_test	(n_test, n_classes)	Test labels (one-hot)
X_full	(n_total, 2)	Full dataset features
y_full	(n_total, n_classes)	Full dataset labels (one-hot)

Configuration

Parameter	Default	Description
base_url	http://localhost:8100	JuniperData service URL
timeout	30	Request timeout in seconds
retries	3	Number of retry attempts
backoff_factor	0.5	Backoff multiplier between retries

Requirements

• Python >=3.12
• numpy >=1.24.0
• requests >=2.28.0
• urllib3 >=2.0.0

License

MIT License - Copyright (c) 2024-2026 Paul Calnon

Juniper Ecosystem

This package is part of the Juniper Cascade Correlation Neural Network Research Platform.

Package	Description	Install
juniper-data-client	Dataset service client (this package)	pip install juniper-data-client
juniper-cascor-client	Neural network service client	pip install juniper-cascor-client
juniper-cascor-worker	Distributed training worker	pip install juniper-cascor-worker