#!/usr/bin/python

import sys,os,re,glob
import codecs

def conv_encoding(data, to_enc='utf_8'):
    lookup = ('utf_8', 'euc_jp', 'euc_jis_2004', 'euc_jisx0213', 'shift_jis')
    for enc in lookup:
        try:
            data = data.decode(enc)
            break
        except:
            pass
    
    if isinstance(data, unicode):
        return data.encode(to_enc)
    else:
        return data

def parse_julius_result(julius_result):
    id_list = [] # list for recognized ids
    dir_list = [] # list for the directions
    utter_list = [] # list for the content of utterances

    re_id = re.compile("source_id = [0-9]+")
    re_dir = re.compile(", azimuth = -?[0-9.]+")
    #re_utter = re.compile("pass1_best: .*")
    re_utter = re.compile("sentence1: .*")
    for line in open(julius_result, 'r'):
        if re_id.search(line):
            # append id
            m = re_id.search(line)
            id_list.append(int(m.group()[12:]))
        if re_dir.search(line):
            # append direction
            m = re_dir.search(line)
            dir_list.append(float(m.group()[12:]))
                
        if re_utter.search(line):
            # append utterance
            m = re_utter.search(line)
            if(re.compile("</s>").search(line)):
                utter_list.append(conv_encoding(m.group()[15:-5])) # <s> ###### </s>
            else:
                utter_list.append(conv_encoding(m.group()[15:]))

    return id_list, dir_list, utter_list
   
if __name__ == '__main__':
	if len(sys.argv) < 2:
		print 'this result.txt'
		exit()
	
	deg = 0
	margin = 180
	if len(sys.argv) >= 4:
		deg = float(sys.argv[2])
		margin = float(sys.argv[3])
	
	(id_list, dir_list, utter_list) = parse_julius_result(sys.argv[1])
	
	print 'id\tdir\trecognition'
	for (i,d,u) in filter(lambda x: abs(x[1]-deg)<margin, sorted(zip(id_list, dir_list, utter_list), key=lambda x: x[0])):
		s = '%02d\t%02d\t%s' % (int(i),int(d),u)
		print conv_encoding(s, to_enc='shift_jis')