Python client

pip install ./clients/python             # from a clone of the repository
pip install "./clients/python[numpy]"    # numpy arrays as vectors

from quiver import Client
 
client = Client("http://localhost:6900")   # api_key="...", timeout=30.0
print(client.health())
print(client.info())

{'status': 'ok', 'version': '0.1.0'}
{'collections': 0, 'points': 0, 'snapshots_enabled': True, 'storage': 'file', 'version': '0.1.0'}

from quiver import Flat, HNSW, KDTree
 
movies = client.get_or_create_collection(
    "movies",
    dimension=4,
    metric="cosine",  # "cosine" | "euclidean" | "dot_product" | "manhattan" | "hamming"
    index=HNSW(m=16),  # or Flat() / KDTree()
)
print(movies, movies.info())

Collection('movies') CollectionInfo(name='movies', dimension=4, metric='cosine', index={'type': 'hnsw', 'm': 16, 'ef_construction': 128, 'ef_search': 64, 'seed': 6512614579346866177}, count=0, tombstones=0, field_indexes=[], quantization='none', on_disk=False, vectors={}, sparse_vectors={})

from quiver import Point
 
print(movies.upsert([
    Point("alien", [0.9, 0.1, 0.0, 0.2], {"genre": "sci-fi", "year": 1979}),
    {"id": "arrival", "vector": [0.8, 0.3, 0.1, 0.1], "metadata": {"genre": "sci-fi", "year": 2016}},
    ("amelie", [0.0, 0.2, 0.9, 0.4], {"genre": "romance", "year": 2001}),
]))
generated = movies.insert([0.1, 0.1, 0.1, 0.9], {"genre": "drama"})
print(len(generated), len(movies))
print(movies.get("amelie", with_vectors=False), movies.get("nope"))

3
36 4
Point(id='amelie', vector=[], metadata={'genre': 'romance', 'year': 2001}) None

import numpy as np
 
query = np.array([1.0, 0.0, 0.0, 0.1], dtype=np.float32)
print([hit.id for hit in movies.search(query, k=2)])

['alien', 'arrival']

from quiver import Flat, Point, Prefetch, SparseVector, SparseVectorParams, VectorParams
 
papers = client.create_collection(
    "papers",
    dimension=4,
    vectors={"title": VectorParams(2, metric="euclidean", index=Flat())},
    sparse_vectors={"text": SparseVectorParams("idf")},  # or just ["text"]
)
papers.upsert([
    Point("p1", [0.1, 0.2, 0.3, 0.4], {"year": 2021},
          vectors={"title": [0.5, 0.5]}, sparse_vectors={"text": {17: 1.0, 4: 0.5}}),
])
papers.search([0.5, 0.5], k=5, using="title")
papers.search_sparse(SparseVector([17], [1.0]), k=5, using="text")
hits = papers.query(
    [Prefetch(vector=[0.1, 0.2, 0.3, 0.4]), Prefetch(sparse={17: 1.0}, using="text", weight=2.0)],
    k=10,
)
print([(h.id, h.score) for h in hits])

papers.update(index={"ef_search": 128}, vectors={"title": {"on_disk": True}})
papers.rename("papers-v1")            # papers.name is now "papers-v1"
client.set_alias("papers", "papers-v1")
print(client.list_aliases())          # {'papers': 'papers-v1'}

from quiver import ConflictError, InvalidRequestError, NotFoundError
 
try:
    movies.search([1.0, 0.0], k=1)
except InvalidRequestError as e:
    print(e.status_code, e.code, e.message)
try:
    movies.search_by_id("nope")
except NotFoundError as e:
    print(e.status_code, e.code)
try:
    client.create_collection("movies", dimension=4)
except ConflictError as e:
    print(e.status_code, e.code)

400 dimension_mismatch dimension mismatch: expected 4, got 2
404 point_not_found
409 already_exists

import asyncio
from quiver import AsyncClient
 
 
async def main():
    async with AsyncClient("http://localhost:6900") as client:
        docs = await client.get_or_create_collection("docs", dimension=3)
        await docs.upsert([("a", [1, 0, 0]), ("b", [0, 1, 0])])
        print([hit.id for hit in await docs.search([1, 0, 0], k=1)])
        same = client.collection("docs")  # no await
        print([p.id async for p in same.iter_points()])
 
 
asyncio.run(main())

['a']
['a', 'b']