SQL over frames in Python

Polars will run SQL against registered frames, planner and all.

import polars as pl

runs = pl.DataFrame({
    "lang": ["python", "rust", "python"],
    "tpm": [104, 98, 91],
})

with pl.SQLContext(runs=runs) as ctx:
    answer = ctx.execute(
        "SELECT lang, COUNT(*) AS n, AVG(tpm) AS mean_tpm "
        "FROM runs GROUP BY lang ORDER BY mean_tpm DESC",
        eager=True)
    print(answer)

print(runs.sql("SELECT * FROM self WHERE tpm > 95"))

How it works

SQLContext registers frames under names.
execute returns a LazyFrame unless you ask to collect.
DataFrame.sql is the shorthand for one frame.

Keywords and builtins used here

as
print
with

The run, in numbers

Lines: 15
Characters to type: 348
Tokens: 87
Three-star pace: 115 tpm

At the three-star pace of 115 tokens a minute, this run takes about 45 seconds.

Type this snippet

Step 3 of 4 in Lazy & SQL, step 27 of 32 in Polars.

← Previous Next →