Dataset Versioning

import requests

API_URL = "https://api.mixpeek.com"
headers = {"Authorization": "Bearer YOUR_API_KEY", "X-Namespace": "your-namespace"}

# Create collection linked to versioned object storage
collection = requests.post(f"{API_URL}/v1/collections", headers=headers, json={
    "collection_name": "training_data_v2",
    "source": {"type": "bucket", "bucket_id": "versioned-training-data"},
    "feature_extractor": {
        "feature_extractor_name": "multimodal_extractor",
        "version": "v1"
    }
}).json()

# Index from versioned object storage (e.g., Tigris, S3)
requests.post(f"{API_URL}/v1/buckets/versioned-training-data/objects", headers=headers, json={
    "blobs": [{"property": "content", "url": "s3://bucket/training/"}],
    "metadata": {"version": "v2", "snapshot_date": "2024-10-01"}
})

# Create cluster snapshot for this version
cluster = requests.post(f"{API_URL}/v1/clusters", headers=headers, json={
    "cluster_name": "training_v2_snapshot",
    "source_collection_ids": [collection["collection_id"]],
    "feature_addresses": ["mixpeek://multimodal_extractor@v1/embedding"],
    "algorithm": "hdbscan"
}).json()

# Execute to create snapshot
execution = requests.post(
    f"{API_URL}/v1/clusters/{cluster['cluster_id']}/execute",
    headers=headers
).json()
print(f"Snapshot created: {execution['run_id']}")

# Query historical dataset state by filtering on metadata
results = requests.post(
    f"{API_URL}/v1/retrievers/versioned-search/execute",
    headers=headers,
    json={"query": {"text": "product demos"}}
).json()

print(f"Found {len(results['documents'])} documents")