Commit 480c3901 authored by Serafino's avatar Serafino
Browse files

fixed inconsistencies

parent 802f8b27

html_to_docx.py

+1 −2
Original line number Diff line number Diff line 
import re

import sys

from os import path

from dataclasses import dataclass



import bs4
@@ -393,7 +392,7 @@ def handle_tag(HTML_tag : bs4.element.Tag): 
            # if we are inside a table we use the cell width instead, but it need to be adjusted because the python docx image width works in a strange way

            scale_factor = .8 #80% of the width looks better then 100%

            width = cell.width*scale_factor if cell else Emu(5486400*scale_factor) 

            run.add_picture(path.join(path.dirname(path.abspath(__file__)),HTML_tag.attrs["src"]), width = width) #use .png pictures that will later be changed with .emf

            run.add_picture(HTML_tag.attrs["src"], width = width) #use .png pictures that will later be changed with .emf

            run = None # this makes sure we are not adding text to the same run that contains the picture

        elif HTML_tag.name == "span": #this is a run with a different style

            if "class" in HTML_tag.attrs:

postprocessing.py

+2 −2
Original line number Diff line number Diff line
@@ -27,7 +27,7 @@ foldername = filename[:-5] # same as filename but without extension .docx 
media_folder = sys.argv[2]



#EXTRACT docx

os.system(f"7zz x {filename} -o{foldername}")

os.system(f"7z x {filename} -o{foldername}")



#PARSE rels

path = f"{foldername}/word/_rels/document.xml.rels"
@@ -59,7 +59,7 @@ with open(path,"w") as file: 
    doc.writexml(file, encoding="UTF-8")



#rebuild docx

os.system(f"7zz a {foldername+'_fixed'}.docx ./{foldername}/*")

os.system(f"7z a {foldername+'_fixed'}.docx ./{foldername}/*")



#delete temp folder

os.system(f"rm -r ./{foldername}")