Can Pandas read and modify a single Excel file worksheet (tab) without modifying the rest of the file?

Solution 1:

I had a similar question regarding the interaction between excel and python (in particular, pandas), and I was referred to this question.

Thanks to some pointers by stackoverflow community, I found a package called xlwings that seems to cover a lot of the functionalities HaPsantran required.

To use the OP's example:

Working with an existing excel file, you can drop an anchor in the data block (Sheet3) you want to import to pandas by naming it in excel and do:

# opened an existing excel file

wb = Workbook(Existing_file)

# Find in the excel file a named cell and reach the boundary of the cell block (boundary defined by empty column / row) and read the cell 

df = Range(Anchor).table.value

# import pandas and manipulate the data block
df = pd.DataFrame(df) # into Pandas DataFrame
df['sum'] = df.sum(axis= 1)

# write back to Sheet3
Range(Anchor).value = df.values

tested that this implementation didn't temper existing formula in the excel file

Let me know if this solves your problem and if there's anything I can help.

Big kudos to the developer of xlwings, they made this possible.

Below is an update to my earlier answer after further question from @jamzsabb, and to reflect a changed API after xlwings updated to >= 0.9.0.

import xlwings as xw
import pandas as pd
target_df = xw.Range('A7').options(pd.DataFrame, expand='table').value # only do this if the 'A7' cell (the cell within area of interest) is in active worksheet
#otherwise do:
#sht = xw.Book(r'path to your xlxs file\name_of_file.xlsx`).sheets['name of sheet']
#target_df = sht.Range('A7').options(pd.DataFrame, expand='table').value # you can also change 'A7' to any name that you've given to a cell like 'interest_table`

Solution 2:

I'm 90% confident the answer to "can pandas do this" is no. Posting a negative is tough, because there always might be something clever that I've missed, but here's a case:

Possible interface engines are xlrd/xlwt/xlutils, openpyxl, and xlsxwriter. None will work for your purposes, as xlrd/wt don't support all formulae, xlsxwriter can't modify existing xlsx files, and openpyxl loses images and charts.

Since I often need to do this, I've taken to only writing simple output to a separate file and then calling the win32api directly to copy the data between the workbooks while preserving all of my colleague's shiny figures. It's annoying, because it means I have to do it under Windows instead of *nix, but it works.

If you're working under Windows, you could do something similar. (I wonder if it makes sense to add a native insert option using this approach to help people in this situation, or if we should simply post a recipe.)

P.S.: This very problem has annoyed me enough from time to time that I've thought of learning enough of the modern Excel format to add support for this to one of the libraries.

P.P.S.: But since ignoring things you're not handling and returning them unmodified seems easy enough, the fact that no one seems to support it makes me think there are some headaches, and where Redmond's involved I'm willing to believe it. @john-machin would know the details, if he's about..

Solution 3:

I'm adding an answer that uses openpyxl. As of version 2.5, you can preserve charts in existing files (further details on the issue are available here).

For demonstration purposes, I create an xlsx file using pandas following the OPs guidelines. The tab named 'Sheet2' has formulas that reference 'Sheet3' and contains a chart.

import pandas as pd

df = pd.DataFrame({'col_a': [1,2,3],
                  'col_b': [4,5,6]})

writer = pd.ExcelWriter('test.xlsx', engine='xlsxwriter')
df.to_excel(writer, sheet_name='Sheet1', index=False)
workbook=writer.book
worksheet = writer.sheets['Sheet1']
df.head(0).to_excel(writer, sheet_name='Sheet2', index=False)
workbook=writer.book
worksheet = writer.sheets['Sheet2']
for i in range(2, len(df) + 2):
    worksheet.write_formula('A%d' % (i), "=Sheet3!A%d" % (i))
    worksheet.write_formula('B%d' % (i), "=Sheet3!B%d" % (i))
chart = workbook.add_chart({'type': 'column'})

chart.add_series({'values': '=Sheet2!$A$2:$A$4'})
chart.add_series({'values': '=Sheet2!$B$2:$B$4'})

worksheet.insert_chart('A7', chart)

df.to_excel(writer, sheet_name='Sheet3', index=False)
df.to_excel(writer, sheet_name='Sheet4', index=False)

writer.save()

Expected test.xlsx after running the code above:

Then if we run the code below, using openpyxl, we can modify the data in 'Sheet3' while preserving formulas and chart in 'Sheet2' and the updated data is now in this file.

from openpyxl import load_workbook

wb = load_workbook('test.xlsx')
ws = wb['Sheet3']
ws['B2'] = 7
ws['B3'] = 8
ws['B4'] = 9
wb.save('test.xlsx')

Expected test.xlsx after running the second block of code: