spark-logical-plan-capture-v2 0.1.1

PySpark package for Catalyst logical plan capture with Base64 logging marker SPARK_LOGICAL_PLAN_CAPTURE_V2.

spark-logical-plan-capture-v2

PyPI-пакет для PySpark 3.5.x, который подключает Spark SQL extension:

• перехватывает логические планы Catalyst
• сериализует payload в Java binary form
• кодирует payload в Base64
• пишет маркер в лог: SPARK_LOGICAL_PLAN_CAPTURE_V2:<base64>

Installation

pip install spark-logical-plan-capture-v2

Usage

from pyspark.sql import SparkSession
from spark_logical_plan_capture import configure_spark_builder

builder = SparkSession.builder.appName("capture-demo").master("local[*]")
spark = configure_spark_builder(builder).getOrCreate()

spark.sql("select 1 as value").show()

Decode SQL from log line

from spark_logical_plan_capture import decode_captured_sql_from_logline

log_line = "SPARK_LOGICAL_PLAN_CAPTURE_V2:<base64>"
sql = decode_captured_sql_from_logline(spark, log_line)
print(sql)

Build and publish

1. Соберите JVM jar:

sbt clean test package

2. Скопируйте jar в Python package data:

python scripts/prepare_python_package.py

3. Соберите wheel/sdist:

python -m build

4. Опубликуйте:

python -m twine upload dist/*