Python visualizer for qbeast index, a tool to visually examine your index, get clues about tree balance, and gather information about sampling efficiency.
With a given qbeast table path and a target RevisionID, it scans the _delta_log/ to find the cubes of the target Revision and builds the index structure for visual display.
One can also provide a sampling fraction and the selected nodes will be highlighted with a different color. Details about on the sampling efficiency will be shown in the terminal.
First of all, you need to make sure you have python installed in your pc and then install poetry in order to download all the dependencies.
# Clone qbeast-spark repo
git clone git@github.com:Qbeast-io/qbeast-spark.git
cd qbeast-spark/utils/visualizer
# Pyhton version should be 3.12
python3 --version
# Install tool and required dependencies
brew install poetry
poetry install Launch a Flask serve with the following command and open the link with a browser:
# Run the tool on the test table
poetry run qviz docs/test_table/
# optionally, specify the index revision(defaulted to 1)
# qviz <table-path> --revision-id=2- Visualize cubes retrieved for a given sampling fraction.
- Sampling details: when a valid value for sampling fraction is given, a set of sampling metrics are displayed. Only the chosen revision is taken into account for computing sampling metrics.
Sampling Info:
Disclaimer:
The displayed sampling metrics are valid only for single revision indexes(excluding revision 0):
sample fraction: 0.3
number of cubes read:8/20, 40.00%
number of rows: 41858/100000, 41.86%
sample size: 0.00141/0.00324GB, 43.49%
- To visualize any table, point to the folder in which the target
_delta_log/is contained. - To visualize a remote table, download its
_delta_log/and point to the folder location.
- To run all tests:
make test