complete_xml_file

#!/usr/bin/env python

# Given a partial XML file (a prefix of one), this tells you which tags
# would be needed to close all open tags and writes them to the end of the
# XML file.  If there are problems while reading the file (non-XML junk),
# it will point you to an editor so you can manually clean the file.

# This program was written to help recover XML files which had been
# corrupted in a hard drive failure.  You should backup any XML files that
# you run it on unless you fully understand what this code does.

# Author: David McClosky
# Homepage: http://zorglish.org
# Code: http://github.com/dmcc

import xml.sax, sys

class XMLTagStack(xml.sax.ContentHandler):
    def __init__(self):
        self.stack = []
    def startElement(self, name, attrs):
        self.stack.append(name)
    def endElement(self, name):
        assert name == self.stack[-1]
        self.stack.pop(-1)
    def closing_tags(self):
        s = ''
        for tag in self.stack[::-1]:
            s += "</%s>\n" % tag
        return s

def check_or_repair(filename):
    tag_stack = XMLTagStack()
    parser = xml.sax.make_parser()
    parser.setContentHandler(tag_stack)

    try:
        parser.parse(open(filename))
    except xml.sax._exceptions.SAXParseException as e:
        if e.getMessage() != 'no element found':
            print("Error:", e)
            print('There is probably junk at', e.getLineNumber(), e.getColumnNumber(), end=' ')
            print("Remove it manually and run this again to complete the file.")
            print(f"vi {filename} +{e.getLineNumber()}")
            return

    if tag_stack.stack:
        print('stack:', tag_stack.stack)
        f = open(filename, 'a')
        f.write(tag_stack.closing_tags())
        f.close()
    else:
        print("already complete")

if __name__ == "__main__":
    for filename in sys.argv[1:]:
        print("Filename:", filename)
        check_or_repair(filename)
        print()