Skip to main content
Version: Next

Python Library

Tenzir ships with a Python library to enable interaction with Tenzir with primitives that integrate well with the Python ecosystem. We distribute the library as PyPI package called tenzir.


The Python library is considered experimental and subject to change without notice.

Install the PyPI package

Use pip to install Tenzir:

pip install tenzir[module]

Use the Tenzir Python library


The following snippet illustrates a small script to query Tenzir.

#!/usr/bin/env python3

import asyncio
from tenzir import Tenzir, to_json_rows

async def example():
tenzir = Tenzir()

generator = tenzir.export("", limit=10)
async for row in to_json_rows(generator):


The Python library is meant to expose all the Tenzir features that are relevant in a Python environment. For now though, it is still in active development and only the following interfaces are exposed:

  • export
  • count
  • status

Many options that exist on the CLI are not mapped to the library. The idea here is to avoid overwhelming the API with options that are actually not needed when interacting with Tenzir from Python.

class Tenzir

    class Tenzir(
endpoint: Optional[str]

Create a connection to a Tenzir node that is listening at the specified endpoint. If no enpdoint is given the TENZIR_ENDPOINT environment variable is used, if that is also not present the tenzir.endpoint value from a local tenzir.yaml configuration file is used. In case that value is also not present the default connection endpoint of is used.


    coroutine export(
expression: str,
mode: ExportMode = ExportMode.HISTORICAL,
limit: int = 100
) -> AsyncIterable[TableSlice]

Evaluate an expression in a Tenzir node and receive the resulting events in an asynchronous stream of TableSlices.

The mode argument can be set to one of HISTORICAL, CONTINUOUS, or UNIFIED. A historical export evaluates the expression against data that is stored in the Tenzir database, the resulting output stream ends when all eligible data has been evaluated. A CONTINUOUS one looks at data as it flows into the node, it will continue to run until the event limit is reached, it gets discarded, or the node terminates.

The limit argument sets an upper bound on the number of events that should be produced. The special value 0 indicates that the number of results is unbounded.


    coroutine count(
expression: str
) -> int

Evaluate the sum of all events in the database that match the given expression.


    coroutine status() -> dict

Retrieve the current status from Tenzir.

>>> st = await tenzir.status()
>>> pprint.pprint(st["system"])
{'current-memory-usage': 729628672,
'database-path': '/var/lib/tenzir',
'in-memory-table-slices': 0,
'peak-memory-usage': 729628672,
'swap-space-usage': 0}

class TableSlice

    coroutine collect_pyarrow(
stream: AsyncIterable[TableSlice],
) -> dict[str, list[pyarrow.Table]]

Collect a stream of TableSlice and return a dictionary of Arrow tables indexed by schema name.

class TenzirRow

A TenzirRow is a Python native representation of an "event" from Tenzir. It consists of a name and a data dictionary.

    coroutine to_json_rows(
stream: AsyncIterable[TableSlice],
) -> AsyncIterable[TenzirRow]

Convert a stream of TableSlices to a stream of TenzirRows.