/python/engine/PinYin/tools/duoyin_filter.py

import sys
import bz2
sys.path.append ("..")
import PYUtil

def load_pinyin_table ():
	hanzi_dict = PYUtil.load_pinyin_table (open ("../../../../data/pinyin_table.txt"))
	tmp = {}
	for key, value in hanzi_dict.items ():
		pinyins = []
		for pinyin, freq in value.items ():
			pinyins.append (pinyin)
		tmp[key] = pinyins

	return tmp

def main ():
	hanzi = load_pinyin_table ()
	duoyin = []
	for key , value in hanzi.items ():
		if len (value) > 1:
			duoyin.append (key)
	hanzi = set(duoyin)

	for l in sys.stdin:
		l = l.strip ()
		phrase = unicode (l, "utf8").split()[0]
		for c in phrase:
			if c in hanzi:
				print l
				break

if __name__ == "__main__":
	main ()