---

name: opensearch-client description: OpenSearch Python client library for hybrid search. Use when writing OpenSearch queries, creating indexes, or implementing text/vector/hybrid search with Korean support.

OpenSearch Client

Python client library for OpenSearch with hybrid search support.

Installation

uv add opensearch-client           # Basic
uv add opensearch-client[openai]   # OpenAI embeddings
uv add opensearch-client[local]    # FastEmbed (local)
uv add opensearch-client[all]      # All features

Quick Start

from opensearch_client import OpenSearchClient

client = OpenSearchClient(host="localhost", port=9200, use_ssl=False)
client.ping()

Text Search

from opensearch_client import OpenSearchClient, TextQueryBuilder, IndexManager

client = OpenSearchClient(host="localhost", port=9200, use_ssl=False)

body = IndexManager.create_text_index_body(text_field="content", use_korean_analyzer=True)
client.create_index("docs", body)

client.bulk_index("docs", [{"content": "OpenSearch is a search engine."}])
client.refresh("docs")

query = TextQueryBuilder.multi_match(query="search", fields=["content"])
results = client.search("docs", TextQueryBuilder.build_search_body(query))

Hybrid Search

from opensearch_client import OpenSearchClient, IndexManager
from opensearch_client.semantic_search.embeddings import OpenAIEmbedding

client = OpenSearchClient(host="localhost", port=9200, use_ssl=False)
embedder = OpenAIEmbedding()

body = IndexManager.create_hybrid_index_body(
    vector_dimension=embedder.dimension,
    vector_field="embedding",
    use_korean_analyzer=True
)
client.create_index("hybrid", body)
client.setup_hybrid_pipeline("pipeline", text_weight=0.3, vector_weight=0.7)

text = "OpenSearch supports hybrid search"
client.index_document("hybrid", {"content": text, "embedding": embedder.embed(text)})
client.refresh("hybrid")

results = client.hybrid_search(
    index_name="hybrid", query="search", query_vector=embedder.embed("search"),
    pipeline="pipeline", text_fields=["content"], vector_field="embedding", k=10
)

VectorStore

from opensearch_client import OpenSearchClient, VectorStore
from opensearch_client.semantic_search.embeddings import FastEmbedEmbedding

client = OpenSearchClient(host="localhost", port=9200, use_ssl=False)
store = VectorStore("store", FastEmbedEmbedding(), client)
store.add(["Doc 1", "Doc 2"])
results = store.search("query", k=5)
for r in results:
    print(f"{r.text} (score: {r.score:.4f})")

API Reference

IndexManager

Method	Description
create_text_index_body(text_field, use_korean_analyzer)	Text search index
create_vector_index_body(vector_field, vector_dimension, space_type)	Vector search index
create_hybrid_index_body(vector_dimension, vector_field, use_korean_analyzer)	Hybrid search index

VectorStore

Method	Returns
add(texts, metadata?, ids?)	list[str] (doc IDs)
add_one(text, metadata?, doc_id?)	str (doc ID)
search(query, k, filter?)	list[SearchResult]
delete(ids)	None
clear()	None
count()	int

SearchResult (dataclass)

Attribute	Type
text	str
score	float
metadata	dict
id	str

Links

• PyPI: https://pypi.org/project/opensearch-client/
• GitHub: https://github.com/namyoungkim/opensearch-client