#!/usr/bin/env python

import os
import xml.parsers.expat
from xml.sax.saxutils import escape


# How much data we process at a time
CHUNK_SIZE = 1024 * 1024

# The sequence of element leading us to the current one
path = []

# How far we are in the current file
cur_size = 0
# From how much should we start another file
MAX_SIZE = 1024*1024 # 1Mb

# The current index
cur_idx = 0
# The current file handle we are writing to
cur_file = None

# The filename we are playing with
root = None
ext = None

# What was the signature of the start element
start = None

# if we are currently in the process of changing file
ending = False

def attrs_s(attrs):
    l = ['']
    for i in range(0,len(attrs), 2):
        l.append('%s="%s"' % (attrs[i], escape(attrs[i+1])))
    return ' '.join(l)

def next_file():
    global cur_size, ending
    if (not ending) and (cur_size > MAX_SIZE):
        global cur_file, cur_idx
        print "part %d Done" % cur_idx
        ending = True
        for elem in reversed(path):
            end_element(elem[0])
        cur_file.close()
        cur_size = 0
        cur_idx += 1
        cur_file = open(root + '.%d' % cur_idx + ext, 'wt')
        for elem in path:
            start_element(*elem)
        ending = False

def start_element(name, attrs):
    global cur_size, start
    if start is not None:
        cur_file.write('<%s%s>' % (start[0], attrs_s(start[1])))
    start = (name, attrs)
    if ending:
        return
    cur_size += len(name) + sum(len(k) for k in attrs)
    path.append((name, attrs))
    next_file()
def end_element(name):
    global cur_size
    global start
    if start is not None:
        cur_file.write('<%s%s/>' % (start[0],attrs_s(start[1])))
    else:
        cur_file.write('</%s>' % name)
    start = None
    if ending:
        return
    elem = path.pop()[0]
    assert elem == name
    cur_size += len(name)
    next_file()
def char_data(data):
    global cur_size, start
    if start is not None:
        cur_file.write('<%s%s>' % (start[0], attrs_s(start[1])))
        start = None
    cur_file.write(escape(data))
    cur_size += len(data)
    next_file()

def main(filename):

    p = xml.parsers.expat.ParserCreate()
    p.ordered_attributes = 1

    p.StartElementHandler = start_element
    p.EndElementHandler = end_element
    p.CharacterDataHandler = char_data

    global cur_file, cur_idx
    global root, ext

    root, ext = os.path.splitext(filename)

    cur_file = open(root + '.%d' % cur_idx + ext, 'wt')

    with open(filename, 'rt') as xml_file:
        while True:
            chunk = xml_file.read(1024*1024)
            if chunk == '':
                break
            p.Parse(chunk)
    cur_file.close()
    print "part %d Done" % cur_idx

if __name__ == "__main__":
    import sys
    if len(sys.argv) == 1:
        print "usage: %s FILENAME [MAX_SIZE]" % sys.argv[0]
        sys.exit()
    if len(sys.argv) > 2:
        MAX_SIZE = int(sys.argv[2]) * 1024
    main(sys.argv[1])