/python/engine/PinYin/tools/check_pinyin.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 check_pinyin (phrase, pinyin, hanzi):
	pys = pinyin.split ("'")
	_len = len (phrase)
	if _len != len (pys):
		return False
	for i in range (0, _len):
		c = phrase[i]
		py = pys[i]
		if c not in hanzi:
			return False
		if py not in hanzi[c]:
			return False
	return True

def main ():
	inv = False
	if len(sys.argv) >= 2 and sys.argv[1] == "-v":
		inv = True
	hanzi = load_pinyin_table ()
	for line in sys.stdin:
		line = line.strip ()
		tmp = unicode (line, "utf8").split()
		phrase = tmp[0]
		pinyin = tmp[1]
		if check_pinyin (phrase, pinyin, hanzi) != inv:
			print line
		else:
			continue

if __name__ == "__main__":
	main ()