Quickstart

from mixpeek import Mixpeek
import time

# Initialize the Mixpeek client
mp = Mixpeek(api_key="YOUR_API_KEY")

# Create a namespace
namespace = mp.namespaces.create(
    name="quickstart",
    description="My first Mixpeek project"
)
namespace_id = namespace["namespace_id"]

# Create a bucket
bucket = mp.buckets.create(
    namespace_id=namespace_id,
    name="content-bucket",
    description="Storage for our content",
    schema={
      "my_img": {
        "type" : "image"
      }
    }
)
bucket_id = bucket["bucket_id"]

# Create a collection
collection = mp.collections.create(
    namespace_id=namespace_id,
    name="content-collection",
    description="Processed content documents",
    source={
      "bucket": bucket_id
    }
    feature_extractors=[
      {"feature_extractor_name": "scene_splitter"}
    ]
)
collection_id = collection["collection_id"]

# Upload an object
object = mp.objects.create(
    bucket_id=bucket_id,
    name="content-bucket",
    files=[
        {
            "my_img": "https://example.com/sample-image.jpg"
        }
    ]
)
object_id = object["object_id"]

# Wait for processing to complete
print("Processing content, please wait...")
time.sleep(30)  # Adjust based on content size

# Create a retriever
retriever = mp.retrievers.create(
    namespace_id=namespace_id,
    name="content-retriever",
    description="Search across processed content",
    stages=[
        {
            "name": "embedding_search",
            "type": "vector",
            "collection_id": collection_id,
            "index": "multimodal",
            "limit": 20
        }
    ]
)
retriever_id = retriever["retriever_id"]

# Search for content
results = mp.retrievers.execute(
    retriever_id=retriever_id,
    query={
        "text": "show me images with people"
    }
)

# Display results
print("\nSearch Results:")
for result in results["results"]:
    print(f"Document ID: {result['document_id']}")
    print(f"Score: {result['score']}")
    print(f"Content: {result['content']}")
    print("---")