Concaterate Mhtml Files

concat.mhtml.py

import sys

fout = open(sys.argv[0] + '.mht', 'w')
files = sorted(sys.argv[1:])
first = True
html_content = ''
attachment_contents = {}

for f in files:
    data = open(f, 'r').read()
    boundary_start = data.find('boundary="') + len('boundary="')
    boundary_end = data.find('"', boundary_start)
    boundary = '--' + data[boundary_start : boundary_end]
    contents = data.split(boundary)[:-1]
    html = contents[1].lower()
    if first:
        first = False
        output_boundary = boundary
        fout.write(contents[0])
        fout.write(output_boundary)
    else:
        html = html[html.find('\n\n') + 2:]

    html_content += html
    for content in contents[2:]:
        header = content[:content.find('\n\n')]
        attachment_contents.setdefault(header, content)

fout.write(html_content)
fout.write(output_boundary)
for attachment_content in attachment_contents.itervalues():
    fout.write(attachment_content)
    fout.write(output_boundary)
fout.write('--\n')