# -*- encoding: iso-8859-1 -*-

import re, random, sys


def appendEntrys(wholeEntry, wholeLex):
	wholeLex.append(wholeEntry)
	return wholeLex
	
	
def randomize(wholeLex):
	randEntrys = random.sample(wholeLex, 100)
	printRand(randEntrys)
	
	
def printRand(randEntrys):
#	f = open("01randomEntries.xml", "w")
	f = open("02randomEntries.xml", "a")
	f.write('<?xml version="1.0" encoding="iso-8859-1"?>' + "\n" + "\n")
	f.write("<dictionary>" + "\n")
	for entry in randEntrys:
		for line in entry:
			f.write(line + "\n")
	f.write("</dictionary>" + "\n")		


if __name__ == "__main__":
#	lexicon = open("../res/10WAT.xml")
	lexicon = open("../res/11WAT.xml")
	wholeEntry = []
	wholeLex = []
	i = 0
	countEntry = 1
	line = lexicon.readline()
	while line:
		line = line.strip()
		print countEntry
		print "\n"
		if re.search("<entry", line):
			if wholeEntry and wholeEntry[0] is not "":
				countEntry += 1
				wholeLex = appendEntrys(wholeEntry, wholeLex)
			wholeEntry = []
		wholeEntry.append(line)
		line = lexicon.readline()
	wholeLex = appendEntrys(wholeEntry, wholeLex)
	randomize(wholeLex)
	print "Anzahl der Einträge: ", countEntry

		

