python.md

sidebar_position	sidebar_label
3	Python

Python

Zed includes preliminary support for Python-based interaction with a Zed lake. The Zed Python package supports loading data into a Zed lake as well as querying and retrieving results in the ZJSON format. The Python client interacts with the Zed lake via the REST API served by super db serve.

This approach works adequately when high data throughput is not required. We plan to introduce native Super Binary support for Python that should increase performance substantially for more data intensive workloads.

Installation

Install the latest version like this:

pip3 install "git+https://github.com/brimdata/super#subdirectory=python/zed"

Install the version compatible with a local zed like this:

pip3 install "git+https://github.com/brimdata/super@$(zed -version |> cut -d ' ' -f 2)#subdirectory=python/zed"

Example

To run this example, first start a Zed lake service from your shell:

zed init -lake scratch
zed serve -lake scratch

Or you can launch the Zui app and it will run a Zed lake service on the default port at http://localhost:9867.

Then, in another shell, use Python to create a pool, load some data, and run a query:

python3 <<EOF
import zed

# Connect to the default lake at http://localhost:9867.  To use a
# different lake, supply its URL via the ZED_LAKE environment variable
# or as an argument here.
client = zed.Client()

client.create_pool('TestPool')

# Load some ZSON records from a string.  A file-like object also works.
# Data format is detected automatically and can be CSV, JSON, Zeek TSV,
# ZJSON, ZNG, or ZSON.
client.load('TestPool', '{s:"hello"} {s:"world"}')

# Begin executing a Zed query for all values in TestPool.
# This returns an iterator, not a container.
values = client.query('from TestPool')

# Stream values from the server.
for val in values:
    print(val)
EOF

You should see this output:

{'s': 'world'}
{'s': 'hello'}