[ARROW-2633] [Python] Parquet file not accesible to write after first read using PyArrow

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Cannot Reproduce
• Component/s: Python
• Labels: parquet
• External issue URL: https://github.com/apache/arrow/issues/19029
• Assignee: Unassigned
• Reporter: Suman
• Created: 23/May/18 18:22
• Updated: 11/Jan/23 07:22
• Resolved: 06/Feb/19 04:46

Description

 
I am trying to read a parquet file in pandas dataframe, do some manipulation and write it back in the same file, however it seems file is not accessible to write after the first read in same function.

It only works, if I don't perform STEP 1 below. Is there anyway to unlock the file as such?

#STEP 1: Read entire parquet file
pq_file = pq.ParquetFile('\dev\abc.parquet')
exp_df = pq_file.read(nthreads=1, use_pandas_metadata=True).to_pandas()
#STEP 2: Change some data in dataframe
#
#STEP 3: write merged dataframe
pyarrow_table = pa.Table.from_pandas(exp_df)
pq.write_table(pyarrow_table, '\dev\abc.parquet',compression='none',)

Error:

File "C:\Python36\lib\site-packages\pyarrow\parquet.py", line 943, in write_table
 **kwargs)
File "C:\Python36\lib\site-packages\pyarrow\parquet.py", line 286, in __init__
 **options)
File "_parquet.pyx", line 832, in pyarrow._parquet.ParquetWriter.__cinit__
File "error.pxi", line 79, in pyarrow.lib.check_status
pyarrow.lib.ArrowIOError: Failed to open local file: \dev\abc.parquet , error: Invalid argument