Word Doc to Clean HTML

docx2html.py

#!/usr/bin/env python
# Setup: install pandoc
# 	https://github.com/jgm/pandoc/releases
# To run script:
# 	python python docx2html.py [filepath]
# Platypus app setup:
# App Name: docx2html
# Script Type: Python
# Path: /usr/bin/python
# Script Path: Edit: [Paste In] 
# Args: none
# Output: Progress Bar
# Accpets Dropped Items: True
# Remain Running After Initial Execution: True


import string
import sys
import os
import subprocess

print "This app will make clean HTML from a DOCX file."

if os.path.exists('/usr/local/bin/pandoc'):
	# System has pandoc
	if len(sys.argv) > 1:
		filepath = sys.argv[1]
		if os.path.isfile(filepath):
			filename, filesuffix = os.path.splitext(filepath)
			if filesuffix == ".docx":
				htmlfile = string.replace(filename + ".html", " ", "-")
				subprocess.call(['/usr/local/bin/pandoc -s "' + filepath + '" -t markdown | /usr/local/bin/pandoc -t html -o ' + htmlfile], shell=True)
				print "File processed: " + htmlfile
			else:
				print "Wrong file type. Please supply a .DOCX file."
		else:
			print "File path is invalid. Please supply a valid file path."
	else:
		print "Please supply file path as an argument."
else:
	# Need to install pandoc
	print "This script requires pandoc. Please install it from https://github.com/jgm/pandoc/releases"