将 Pandas 数据框导出到 Excel 多表文件的正确方法是什么？

Question

I need to output two cleaned and recalculated dataframes to Excel file as separate sheets. This code works, but opening resulting file in Excel produces "file corrupted" - it gets repaired and opens fine afterwards, but this is annoying.

The code is on Azure Jupiter Notebook, Python 3.6, I download Excel file and open in Excel 365, Win 10.

# Create a Pandas Excel writer using XlsxWriter as the engine.
writer = pd.ExcelWriter('PR_weatherGDDid.xlsx', engine='xlsxwriter') 

# Write each dataframe to a different worksheet.
df.to_excel(writer, sheet_name='Daily', index=False)     
doystats.to_excel(writer, sheet_name='stats')    

# Close the Pandas Excel writer and output the Excel file.
writer.save()

So: Excel file gets created but has a problem to be opened in Excel.

Answer 1

Here is the correct way.

>>> with pd.ExcelWriter('PR_weatherGDDid.xlsx') as writer: 
...     df.to_excel(writer, sheet_name='Daily')
...     doystats.to_excel(writer, sheet_name='stats')

Answer 2

This is my code and I can open the Excell file allright:

# Create a Pandas Excel writer using XlsxWriter as the engine.
writer = pd.ExcelWriter('PR_weatherGDDid.xlsx') 

data = [['AMN987','Ok'],['AMN987','Ok'],['AMN987','Error'], ['BBB987','Ok'],['BBB987','Ok'],['CCC','Error']]
df = pd.DataFrame(data, columns=['Serial', 'Status'])

days_to = [['02/08/19',4],['02/08/19',8],['02/08/19',3], ['02/08/19',6],['02/08/19',0],['02/08/19',9]]
doystats = pd.DataFrame(days_to, columns=['Date', 'Day'])

# Write each dataframe to a different worksheet.
df.to_excel(writer, sheet_name='Daily', index=False)     
doystats.to_excel(writer, sheet_name='stats')    

# Close the Pandas Excel writer and output the Excel file.
writer.save()
writer.close()

The output looks like this:

Answer 3

The problem with Excel only opening the created file after "repairs" seems to stem from the fact that file was created in Azure Jupiter notebook online. All 3 code variants (mine and suggested by @atlas and @sharif) produced file needing "repairs" in the online environment, but made normal Excel file when I run it through local-installed Jupiter Notebooks (Anaconda).

Answer 4

As Larisa Golovko noted, this appears to be an issue only with XlsxWriter on Azure Notebooks. It doesn't happen with XlsxWriter, Pandas or Jupyter in offline environments.

I dug into it a bit more here and it looks like it there is a zipfile compression error on the .rels files in the xlsx archive. Currently I don't know what is causing that but it appears to be related to the standard Python zipfile library on that environment. I'll try to put together a simpler test case without XlsxWriter.

A workaround is to use the XlsxWriter in_memory constructor option:

workbook = xlsxwriter.Workbook('hello_world.xlsx', {'in_memory': True})

# Or:

writer = pd.ExcelWriter('pandas_example.xlsx',
                        engine='xlsxwriter',
                        options={'in_memory': True})