[Python] Weird IndexError when creating hdf5 file with pandas in table mode

I have a very weird error when creating a hdf5 file with pandas in table mode. I create the file with the store.put() function:

storage_file.put(some_name, current_dataset.meta_data, format="table", data_columns=True,min_itemsize={"values": 100})


The meta_data is - obviously - a pandas dataframe with the column names:

[
'index',
'xPosition',
'yPosition',
'approachID',
'segment_0_type',
'segment_0_style',
'segment_0_duration',
'segment_0_num_points',
'segment_0_z_start',
'segment_0_z_end',
'segment_0_setpoint',
'segment_1_type',
'segment_1_style',
'segment_1_duration',
'segment_1_num_points',
'segment_1_z_start',
'segment_1_z_end',
'segment_1_setpoint',
'segment_2_type',
'segment_2_style',
'segment_2_duration',
'segment_2_num_points',
'segment_2_height_limit',
'pauseBeforeFirst',
'pauseOnTipsaver',
'segmentIndex',
'heightMultiplier',
'heightOffset',
'sensitivity',
'springConstant',
'measurementID'
]


I get the following error message:

IndexError: tuple index out of range


together with

File "PATH\Lib\site-packages\pandas\io\pytables.py", line 4473, in write_data
new_shape = (nrows,) + self.dtype[names[nindexes + i]].shape


I am sure, that there is something off with the column names, cause when I change the column names to ["0" ... "30"] with "..." the respective numbers in between, the error vanishes. The length of the column names shouldn't be a problem either cause I renamed every coumn names with

for index in range(meta_data.shape[1]):
meta_data.rename({meta_data.columns[index]: "".join([str(index) for i in range(25)])}, axis=1, inplace=True)


which is incredibly ugly, but works.

Any thoughts about this?

Continue reading...