Qdrant.jl

Julia wrapper to Qdrant Vector Database. The code is mostly generated using their OpenAPI spec.

WIP Notice: The code is very basic and does not include any safety checks. At best it's work in progress.

Environment

This code is being worked against a locally running Qdrant instance in a container using the following command docker run -p 6333:6333 qdrant/qdrant

This package is not registered and would need to be installed using the url to this repository using Pkg; Pkg.add("https://github.com/asbisen/Qdrant.jl.git")

Example Code

Create a Connection

using Qdrant

conn = QdrantConnection("http://localhost:6333")

Sample Workflow

# Get Existing Collections
existing_collections = get_collections(conn)

# Create a new collection
collection_name = "custom_collection"
vector_params = QdrantVectorParams(size=128, distance=QdrantDistance("Cosine"))
hnsw_conf = QdrantHnswConfig(m=32, ef_construct=200, on_disk=true)
response = create_collection(conn, collection_name;
                vectors_config=vector_params,
                hnsw_config=hnsw_conf,
                shard_number=2,
                replication_factor=2,
                on_disk_payload=true
            )

# Get Collection Info
collection_config = get_collection(conn, "custom_collection")

# Check if a collection exists
result = collection_exists(conn, collection_name)
println("Collection $collection_name exists: $result")

# Insert Vector
id      = UInt(110)
emb     = rand(Float32, 128)
payload = Dict("Name" => "John Doe", "Age" => 20)
point = Qdrant.QdrantPointStruct(id, emb, payload)

res = upsert_points(conn, collection_name, [point])

# Search for a vector
query = Qdrant.QdrantSearchRequest(rand(128), 25; score_threshold=0.2, with_vector=false)
r = search_points(conn, collection_name, query)


# Search for vectors with filter
filter = QdrantFilter(
    should = [
        Dict("key" => "color",   "match" => Dict("value" => "blue")),
        Dict("key" => "country", "match" => Dict("value" => "Canada"))
    ],
    must = [
        Dict("key" => "bool", "match" => Dict("value" => true))
    ],
    must_not = [
        Dict("key" => "age",  "range" => Dict("lt" => 50))
        ]
)

query = Qdrant.QdrantSearchRequest(rand(128), 5; with_vector=false, filter=filter)
r = search_points(conn, collection_name, query)