#!/usr/bin/env python

# Convert a doc or docx file to html, and call a new firefox window on it.

import sys, os

import subprocess

import magic

import time

magic_parser = None

def view_word_files(file_list):

outfiles = []

for filename in file_list:

pathparts = os.path.split(filename)

fileparts = os.path.splitext(pathparts[1])

outfilename = os.path.join("/tmp", fileparts[0] + ".html")

# Extra credit: could use tempfile.mkstemp

# wvHtml does a pretty good job with simple .doc files.

if filename.lower().endswith(".doc"):

subprocess.call(["wvHtml", filename, outfilename])

outfiles.append(outfilename)

# docx is harder to convert to HTML. Use libreoffice's unoconv.

elif filename.lower().endswith(".docx"):

subprocess.call(["unoconv", "-f", "html", "-o", outfilename,

filename])

outfiles.append(outfilename)

else:

global magic_parser

if not magic_parser:

magic_parser = magic.open(magic.MAGIC_MIME)

magic_parser.load()

mimetype = magic_parser.file(filename)

print "mime type is", mimetype

if mimetype.startswith("application/msword"):

subprocess.call(["wvHtml", filename, outfilename])

outfiles.append(outfilename)

elif mimetype.startswith("application/vnd.openxmlformats-officedocument.wordprocessingml.document"):

subprocess.call(["unoconv", "-f", "html", "-o", outfilename,

filename])

outfiles.append(outfilename)

if outfiles:

subprocess.call(["firefox", "-new-window", outfiles[0]])

for f in outfiles[1:]:

# If we don't wait for the new window to pop up before

# calling new-tab, bad things will happen: the document

# may load in a new tab in the old window and THEN pop up

# an unwanted third window. Go firefox.

time.sleep(1)

subprocess.call(["firefox", "-new-tab", f])

if __name__ == "__main__":

view_word_files(sys.argv[1:])