tasks/offline-compute/PySpark/pyspark_002/init/init_db.py

#!/usr/bin/env python3
"""pyspark_002 database initialization: create tables + load seed data"""
from pyspark.sql import SparkSession

spark = SparkSession.builder \
    .appName('dataclaw_eval_init_pyspark_002') \
    .enableHiveSupport() \
    .config('spark.sql.warehouse.dir', '/tmp/hive_warehouse') \
    .getOrCreate()

spark.sql('CREATE DATABASE IF NOT EXISTS internal_platform_db')

# Create input table
spark.sql('''
CREATE TABLE IF NOT EXISTS internal_platform_db.case5_ods_code_v2_note_diff_files_diff_v3 (
    `databus_imp_date` STRING COMMENT '业务日期，原为分区列',
    `id` BIGINT COMMENT '-',
    `diff` STRING COMMENT '-'
) STORED AS ORC
''')

# Load seed data
from pyspark.sql.types import StructType, StructField, StringType, LongType

schema = StructType([
    StructField("databus_imp_date", StringType(), True),
    StructField("id", LongType(), True),
    StructField("diff", StringType(), True),
])

print('Loading seed data...')
df = spark.read.csv(
    "/tmp_workspace/seed_data/case5_ods_code_v2_note_diff_files_diff_v3.csv",
    header=True, schema=schema,
)
df.write.mode("overwrite").insertInto("internal_platform_db.case5_ods_code_v2_note_diff_files_diff_v3")

print('Database initialization complete')
spark.stop()