deltalake_pandas_dataframe_write_read_version_history.py

1import pandas as pd
2from deltalake import DeltaTable, write_deltalake
3
4# 1. Create a sample Pandas DataFrame
5df = pd.DataFrame({"id": [1, 2], "value": ["foo", "bar"]})
6
7# 2. Write the DataFrame to a Delta table
8# This will create a directory named 'tmp/delta-table' with the Delta log and data files
9write_deltalake("tmp/delta-table", df)
10
11# 3. Read the Delta table
12dt = DeltaTable("tmp/delta-table")
13
14# 4. Convert the Delta table back to a Pandas DataFrame to verify
15result_df = dt.to_pandas()
16print(result_df)
17
18# 5. Check the current version of the table
19print(f"Current table version: {dt.version()}")
20
21# 6. List the files in the table
22print(f"Files in table: {dt.files()}")