import socket
import sys,re
# -*- coding: utf-8 -*-

def tounicode(data):
	f = lambda d, enc: d.decode(enc);
	codecs = ['utf-8','shift_jis','euc_jp','cp932',
			'euc_jis_2004','euc_jisx0213','iso2022_jp','iso2022_jp_1',
			'iso2022_jp_2','iso2022_jp_2004','iso2022_jp_3','iso2022_jp_ext',
			'shift_jis_2004','shift_jisx0213','utf_16','utf_16_be',
			'utf_16_le','utf_7','utf_8_sig'];

	for codec in codecs:
		try: return f(data, codec);
		except: continue;
	return None;

def convert_uni2enc(data,enc):
	data1 = tounicode(data);
	data2 = '';
	for char in data1[:]:
		try:
			data2 += char.encode(enc);
		except:
			pass;
	return data2;


if __name__ == '__main__':

	host = "localhost"
	port = 10500

	sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
	sock.connect((host, port))
	print "Connection to Julius succeeded"
	sys.stdout.flush()

	# clear the log file
	#handle = open("julius_out.txt", "w")
	#handle.close()

	word = {}
	angle = {}
	parsed = []
	currentID = None

	line = ""
	while True:
	    # build a valid line
	    tmpline = sock.recv(1024).replace(".\n", "").rstrip()
	    line += tmpline
	    if not tmpline[-1] == ">":
	        continue

		# make sure line is unicode
		line = tounicode(line)
		
	    # parse it.
	    res = re.search(
	        '<SOURCEINFO SOURCEID="([^"]*)".*AZIMUTH="([^"]*)".*', line)
	    if not res is None:
	        angle[int(res.group(1))] = float(res.group(2))

	    res = re.search(
	        '<RECOG(OUT|FAIL) SOURCEID="([0-9]*)".*', line)
	    if not res is None:
	        currentID = int(res.group(2))

	    if not currentID is None and "<WHYPO" in line:
	        res = re.search('<WHYPO.*WORD="([^<][^"]+)".*', line)
	        if not res is None:
	            word[currentID] = res.group(1)
	            parsed.append((currentID, angle[currentID], res.group(1)))

	            out = "%d, %f, %s\n" % (currentID, angle[currentID], res.group(1))
	            #handle = open("julius_out.txt", "a")
	            #handle.write(out)
	            #handle.close()
	            #print out
	            print 'source_id = %d, azimuth = %f' % (currentID, angle[currentID])
	            s = 'sentence1: <s> %s </s>' % res.group(1)
	            print convert_uni2enc(s,'shift_jis')
	            sys.stdout.flush()
	    line = ""

	"""
	for p in parsed:
	    if abs(p[1] - target) < tolerance:
	        print p
	"""