Detect Schema Changes Across Datasets (Python)Python

import duckdb

def compare_schemas(file1, file2):
    """
    Compare schemas of two datasets and find differences.
    Args:
        file1 (str): Path to the first dataset (CSV/Parquet).
        file2 (str): Path to the second dataset (CSV/Parquet).
    Returns:
        list: Schema differences.
    """
    con = duckdb.connect()
    schema1 = con.execute(f"DESCRIBE SELECT * FROM read_csv_auto('{file1}')").fetchall()
    schema2 = con.execute(f"DESCRIBE SELECT * FROM read_csv_auto('{file2}')").fetchall()
    return {"file1_schema": schema1, "file2_schema": schema2}

# Example Usage
differences = compare_schemas("data1.csv", "data2.csv")
print(differences)
Copy code