Skip to main content

Deephaven Python Client

Deephaven Python Client is a Python package created by Deephaven Data Labs. It is a client API that allows Python applications to remotely access Deephaven data servers.

Building on Windows 10/11.

To build on Windows, please see the file cpp-client/README-windows.md in this repository.

venv

It's recommended to install in a Python virtual environment (venv). Use a command like the below to create a venv. Then, activate the venv.

python3 -m venv ~/py/dhenv
source ~/py/dhenv/bin/activate

Source Directory

From the deephaven-core repository root

(clone from https://github.com/deephaven/deephaven-core)

It is assumed that you have the repository checked out at the location specified by ${DHROOT}

$ cd ~/dhsrc  # or another directory you choose
$ git clone https://github.com/deephaven/deephaven-core.git
$ cd deephaven-core
$ export DHROOT=`pwd`

Change to the py/client directory inside the deephaven-core repository

$ cd $DHROOT/py/client

Dev environment setup

$ pip3 install -r requirements-dev.txt

Build

$ DEEPHAVEN_VERSION=$(../../gradlew :printVersion -q) python3 setup.py bdist_wheel

Run tests

$ python3 -m unittest discover tests

Run examples

$ python3 -m examples.demo_table_ops
$ python3 -m examples.demo_query
$ python3 -m examples.demo_run_script
$ python3 -m examples.demo_merge_tables
$ python3 -m examples.demo_asof_join

Install

Note the actual name of the .whl file may be different depending on system details.

$ pip3 install dist/pydeephaven-<x.y>-py3-none-any.whl

Quick start

    >>> from pydeephaven import Session
    >>> session = Session() # assuming Deephaven Community Edition is running locally with the default configuration
    >>> table1 = session.time_table(period=1000000000).update(formulas=["Col1 = i % 2"])
    >>> df = table1.to_arrow().to_pandas()
    >>> print(df)
                        Timestamp  Col1
    0     1629681525690000000     0
    1     1629681525700000000     1
    2     1629681525710000000     0
    3     1629681525720000000     1
    4     1629681525730000000     0
    ...                   ...   ...
    1498  1629681540670000000     0
    1499  1629681540680000000     1
    1500  1629681540690000000     0
    1501  1629681540700000000     1
    1502  1629681540710000000     0

    >>> session.close()

Initialize

The Session class is your connection to Deephaven. This is what allows your Python code to interact with a Deephaven server.

from pydeephaven import Session

session = Session()

Ticking table

The Session class has many methods that create tables. This example creates a ticking time table and binds it to Deephaven.

from pydeephaven import Session

session = Session()

table = session.time_table(period=1000000000).update(formulas=["Col1 = i % 2"])
session.bind_table(name="my_table", table=table)

This is the general flow of how the Python client interacts with Deephaven. You create a table (new or existing), execute some operations on it, and then bind it to Deephaven. Binding the table gives it a named reference on the Deephaven server, so that it can be used from the Web API or other Sessions.

Execute a query on a table

table.update() can be used to execute an update on a table. This updates a table with a query string.

from pydeephaven import Session

session = Session()

# Create a table with no columns and 3 rows
table = session.empty_table(3)
# Create derived table having a new column MyColumn populated with the row index "i"
table = table.update(["MyColumn = i"])
# Update the Deephaven Web Console with this new table
session.bind_table(name="my_table", table=table)

Sort a table

table.sort() can be used to sort a table. This example sorts a table by one of its columns.

from pydeephaven import Session

session = Session()

table = session.empty_table(5)
table = table.update(["SortColumn = 4-i"])

table = table.sort(["SortColumn"])
session.bind_table(name="my_table", table=table)

Filter a table

table.where() can be used to filter a table. This example filters a table using a filter string.

from pydeephaven import Session

session = Session()

table = session.empty_table(5)
table = table.update(["Values = i"])

table = table.where(["Values % 2 == 1"])
session.bind_table(name="my_table", table=table)

Query objects

Query objects are a way to create and manage a sequence of Deephaven query operations as a single unit. Query objects have the potential to perform better than the corresponding individual queries, because the query object can be transmitted to the server in one request rather than several, and because the system can perform certain optimizations when it is able to see the whole sequence of queries at once. They are similar in spirit to prepared statements in SQL.

The general flow of using a query object is to construct a query with a table, call the table operations (sort, filter, update, etc.) on the query object, and then assign your table to the return value of query.exec().

Any operation that can be executed on a table can also be executed on a query object. This example shows two operations that compute the same result, with the first one using the table updates and the second one using a query object.

from pydeephaven import Session

session = Session()

table = session.empty_table(10)

# executed immediately
table1= table.update(["MyColumn = i"]).sort(["MyColumn"]).where(["MyColumn > 5"]);

# create Query Object (execution is deferred until the "exec" statement)
query_obj = session.query(table).update(["MyColumn = i"]).sort(["MyColumn"]).where(["MyColumn > 5"]);
# Transmit the QueryObject to the server and execute it
table2 = query_obj.exec();

session.bind_table(name="my_table1", table=table1)
session.bind_table(name="my_table2", table=table2)

Join two tables

table.join() is one of many operations that can join two tables, as shown below.

from pydeephaven import Session

session = Session()

table1 = session.empty_table(5)
table1 = table1.update(["Values1 = i", "Group = i"])
table2 = session.empty_table(5)
table2 = table2.update(["Values2 = i + 10", "Group = i"])

table = table1.join(table2, on=["Group"])
session.bind_table(name="my_table", table=table)

Perform aggregations on a table

Aggregations can be applied on tables in the Python client. This example creates an aggregation that averages the Count column of a table, and aggregates it by the Group column.

from pydeephaven import Session, agg

session = Session()

table = session.empty_table(10)
table = table.update(["Count = i", "Group = i % 2"])

my_agg = agg.avg(["Count"])

table = table.agg_by(aggs=[my_agg], by=["Group"])
session.bind_table(name="my_table", table=table)

Convert a PyArrow table to a Deephaven table

Deephaven natively supports PyArrow tables. This example converts between a PyArrow table and a Deephaven table.

import pyarrow as pa
from pydeephaven import Session

session = Session()

arr = pa.array([4,5,6], type=pa.int32())
pa_table = pa.Table.from_arrays([arr], names=["Integers"])

table = session.import_table(pa_table)
session.bind_table(name="my_table", table=table)

#Convert the Deephaven table back to a pyarrow table
pa_table = table.to_arrow()

Execute a script server side

session.run_script() can be used to execute code on the Deephaven server. This is useful when operations cannot be done on the client-side, such as creating a dynamic table writer. This example shows how to execute a script server-side and retrieve a table generated from the script.

from pydeephaven import Session

session = Session()

script = """
from deephaven import empty_table

table = empty_table(8).update(["Index = i"])
"""

session.run_script(script)

table = session.open_table("table")
print(table.to_arrow())

Error handling

The DHError is thrown whenever the client package encounters an error. This example shows how to catch a DHError.

from pydeephaven import Session, DHError

try:
    session = Session(host="invalid_host")
except DHError as e:
    print("Deephaven error when connecting to session")
    print(e)
except Exception as e:
    print("Unknown error")
    print(e)

Related documentation

API Reference

[start here] https://deephaven.io/core/client-api/python/

Release files for pydeephaven 42.5

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Built distribution (wheel)

Table of built distributions (wheels) for pydeephaven 42.5
File Interpreter ABI Platform
pydeephaven-42.5-py3-none-any.whl Python 3 none any Details

Release files / pydeephaven-42.5-py3-none-any.whl

Download URL pydeephaven-42.5-py3-none-any.whl
Size 121.3 kB
Tags Python 3
SHA-256 checksum
How to use checksums
69dc9133ed9ee82c80b6c4ebbea80d5a5e73042505fef0a5b2444f4288b28877
BLAKE2b-256 checksum
How to use checksums
bb27cd401e29d46b757c8cae22f971a08b4375ce898ea6c67e022397ac4f530c
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Sep 10, 2026.

Transparency log

Release history Release notifications | RSS feed

This release

42.5 This release

1 release file

42.4

1 release file

42.3

1 release file

42.2

1 release file

42.1

1 release file

42.0

1 release file

41.10

1 release file

41.9

1 release file

41.8

1 release file

41.7

1 release file

41.6

1 release file

41.5

1 release file

41.4

1 release file

41.3

1 release file

41.2

1 release file

41.1

1 release file

41.0

1 release file

0.40.9

1 release file

0.40.8

1 release file

0.40.7

1 release file

0.40.6

1 release file

0.40.5

1 release file

0.40.4

1 release file

0.40.3

1 release file

0.40.2

1 release file

0.40.1

1 release file

0.40.0

1 release file

0.39.8

1 release file

0.39.7

1 release file

0.39.6

1 release file

0.39.5

1 release file

0.39.4

1 release file

0.39.3

1 release file

0.39.2

1 release file

0.39.1

1 release file

0.39.0

1 release file

0.38.0

1 release file

0.37.6

1 release file

0.37.5

1 release file

0.37.4

1 release file

0.37.3

1 release file

0.37.2

1 release file

0.37.1

1 release file

0.37.0

1 release file

0.36.2

1 release file

0.36.1

1 release file

0.36.0

1 release file

0.35.3

1 release file

0.35.2

1 release file

0.35.1

1 release file

0.35.0

1 release file

0.34.4

1 release file

0.34.3

1 release file

0.34.2

1 release file

0.34.1

1 release file

0.34.0

1 release file

0.33.6

1 release file

0.33.5

1 release file

0.33.4

1 release file

0.33.3

1 release file

0.33.2

1 release file

0.33.1

1 release file

0.33.0

1 release file

0.32.1

1 release file

0.32.0

1 release file

0.31.0

1 release file

0.30.4

1 release file

0.30.3

1 release file

0.30.2

1 release file

0.30.1

1 release file

0.30.0

1 release file

0.29.1

1 release file

0.29.0

1 release file

0.28.1

1 release file

0.28.0

1 release file

0.27.1

1 release file

0.27.0

1 release file

0.26.1

1 release file

0.26.0

1 release file

0.25.3

1 release file

0.25.2

1 release file

0.25.0

1 release file

0.24.1

1 release file

0.24.0

1 release file

0.23.0

1 release file

0.22.0

1 release file

0.21.1

1 release file

0.21.0

1 release file

0.20.0

1 release file

0.19.0

1 release file

0.18.0

1 release file

0.16.0

1 release file

0.15.1

1 release file

0.14.0

1 release file

0.13.0

1 release file

0.12.0

1 release file

0.11.0

1 release file

0.10.0

1 release file

0.9.0

1 release file

0.8.0

1 release file

0.7.0

1 release file

0.6.0

1 release file

0.5.2

1 release file

0.5.1

1 release file

0.5.0

1 release file

0.4.1

1 release file

0.4.0

1 release file

0.1.1

1 release file

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page