25 lines
770 B
Python
25 lines
770 B
Python
import sys
|
|
import os
|
|
sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__),"..", "src")))
|
|
|
|
from lib.custom_logger import get_logger
|
|
logger = get_logger(level=10)
|
|
|
|
from providers.duck_db_provider import DuckDBProvider
|
|
|
|
def example_duckdb():
|
|
logger.info("Starting example_duckdb function.")
|
|
file_path = "./data_science/data/y=*/news/news_*.csv"
|
|
provider = DuckDBProvider()
|
|
sql = f"""
|
|
SELECT *
|
|
FROM read_csv_auto('{file_path}' , HEADER=TRUE, IGNORE_ERRORS=TRUE)
|
|
"""
|
|
result = provider.query_df(sql)
|
|
print("latest published_parsed:", result)
|
|
|
|
# forで1件ずつ表示
|
|
for idx, row in result.iterrows():
|
|
logger.info(f"title:{row['title']}")
|
|
|
|
example_duckdb() |