从 Python 网络应用程序：将数据插入电子表格（例如 LibreOffice / Excel），计算并另存为 pdf

Question

I am facing the problem, that I would like to push data (one large dataframe and one image) from my python web app (running on Tornado Webserver and Ubuntu) into a spreadsheet, calculate, save as pdf and the deliver to the frontend.

I took a look at several libs like openpyxl for writing Sheets in MS Excel, but that would solve just one part. I was thinking about using LibreOffice and pyoo, but it seems that I need the same python version on my backend as shipped with LibeOffice when importing pyuno.

Does somebody has solved a similar issue and have a recommendation how to solve this?

Thanks

Answer 1

I came up to a let's say not pretty, but rare solution that works very flexible for me.

• use openpyxl to open an existing Excel workbook that includes layout (Template)
• insert the dataframe into a separate sheet in that workbook
• use openpyxl to save as temporary_file.xlsx
• call LibeOffice with --headless --convert-to pdf temporary_file.xlsx

While executing the last call, all integrated formulas are recalculated/updated and the pdf is created ( you have to configure calc so that auto calc is enabled when files are opened )

• deliver pdf to frontend or process as you like
• delete temporary_file.xlsx

---

import openpyxl
import pandas as pd
from subprocess import call

d   = {'col1': [1, 2], 'col2': [3, 4]}
df  = pd.DataFrame(data=d)
now = datetime.datetime.now().strftime("%Y%m%d_%H%M_%f")

wb_template_name = 'Template.xlsx'
wb_temp_name     = now + wb_template_name
wb               = openpyxl.load_workbook(wb_template_name)
ws               = wb['dataframe_sheet']
pdf_convert_cmd  = 'soffice --headless --convert-to pdf ' + wb_temp_name

for r in dataframe_to_rows(df, index=True, header=True):
   ws.append(r)
wb.save(wb_temp_name)
call(pdf_convert_cmd, shell=True)

---

The reason why I'm doing this, is that I would like to be able to style the layout of the pdf independently from the data. I use named ranges or lookups that are referenced to the separate dataframe-sheet in excel.

I didn't try the image insertion yet, but this should work similar. I think there could be a way to increase the performance while simply dump the dataframe into the xlsx file (which is a zipped file of xmls), so that you don't need openpyxl.