Elasticsearch Python Client

The Elasticsearch Python client is Elastic’s official SDK for talking to an Elasticsearch cluster from Python. It wraps the full REST API surface in both synchronous and asyncio-based clients, adds connection pooling, retries, and sniffing, and layers a Django-ORM-like query DSL and an ES|QL helper on top of the raw request/response API so teams can index documents, run searches, and manage cluster state without hand-building HTTP calls.

What You Get

• Full REST API coverage generated from Elasticsearch’s API specification, kept in lockstep with server versions
• Both synchronous (Elasticsearch) and asyncio-based (AsyncElasticsearch) client classes
• A Django-style query DSL module (elasticsearch.dsl) for building complex queries and mappings as Python objects
• Built-in connection pooling, retry logic, node sniffing, and OpenTelemetry instrumentation
• Bulk and streaming helpers for high-throughput indexing and scrolling through large result sets

Common Use Cases

• Building full-text search features backed by an Elasticsearch cluster in a Python web application
• Bulk-indexing large document sets with the client’s streaming bulk helpers
• Running aggregations and analytics queries via the query DSL instead of raw JSON query bodies
• Integrating async Elasticsearch access into asyncio-based services (FastAPI, aiohttp)

Under The Hood

Architecture - The client separates transport concerns (_sync/_async connection and transport modules handling pooling, retries, and node sniffing) from the generated API surface in client.py and the dsl subpackage, which compiles high-level query objects down to the same JSON request bodies the raw client sends, giving both a low-level and ORM-like access pattern against one transport layer. Tech Stack - Pure Python supporting 3.10 through 3.14 plus PyPy, with elastic-transport as its HTTP layer, built with hatchling, and OpenTelemetry hooks (_otel.py) baked in for tracing outgoing requests. Code Quality - test_elasticsearch/ mirrors both sync and async clients with dedicated test suites, CI runs against real Elasticsearch versions via noxfile.py, and the generated API bindings carry consistent type stubs (py.typed) across the whole package. API Design - The client mirrors Elasticsearch’s REST API naming almost one-to-one (e.g. client.search(), client.index()), making it easy to map official Elasticsearch documentation directly onto client calls, while the DSL module trades some of that directness for a more Pythonic, composable query-building experience.