Source Code for Module viterbi.read_probs

 1  #!/usr/bin/python 
 2   
 3 -def read_probs_file(filename, debug=False): 
 4     """ 
 5     Read in either, case(a), a file representing 2 parameter probs in format:: 
 6        p1 p2 prob 
 7     Or, case(b), a file containing a pronounciation dict with lines of the form:: 
 8        word: 
 9          phon1 phon2 prob 
10              ... 
11          phoni phonj prob 
12     In Case (a) return 2-param prob dictionary:: 
13        D[w1][2] = probability of w2 given w1 
14     In Case (b) return 3-param prob dictionary:: 
15        D[w][phon1][phon2] = probability of phon2 given phon1 when 
16                             pronouncing word w 
17     These dictionaries are printed to files with the same base 
18     as filename, but a ".py" extension. Default: extension 
19     of input files will be ".txt". 
20     """ 
21     ## Input a text file 
22     bi_dict = {} 
23     pron_dict = {} 
24     # Use bi_dict as the default dict 
25     dispatcher = DictDispatcher(bi_dict) 
26     ifh = open(filename,'r') 
27     for line in ifh: 
28         if line[0] == '%':   # Comment char 
29            continue 
30         line = line.rstrip() 
31         dispatcher.add_line_to_dict(line,pron_dict,debug) 
32     ifh.close() 
33     ## Output a python file 
34     (base,ext) = os.path.splitext(filename) 
35     ofh = open(base+'.py','w') 
36     print >> ofh, '# Automatically generated from %s' % filename 
37     if bi_dict: 
38         print >> ofh, base+'_dict'+'=', bi_dict 
39     else: 
40         print >> ofh, base+'_dict'+'=', pron_dict 
41     ofh.close() 
42     return (bi_dict,pron_dict, dispatcher) 
43   
44  import re 
45  word_re = re.compile(r'(\w+):$') 
46  bigram_re = re.compile(r'(\#|\w+)\s+(\w+)\s+(\d?\.\d+)$') 
47  blank_re = re.compile(r'\w*$') 
48  start_re = re.compile(r'(\#)') 
49   
50 -class DictDispatcher(object): 
51 -    def __init__(self,active_dictionary={}): 
52          object.__init__(self) 
53          self.active_dictionary=active_dictionary 
54   
55 -    def add_line_to_dict(self,line, pron_dict, debug): 
56          if debug: 
57              print line 
58          word_match = re.match(word_re,line) 
59          bigram_match = re.match(bigram_re,line) 
60          if word_match: 
61              'Word dict detected!' 
62              self.active_dictionary=pron_dict.setdefault(word_match.groups()[0],{}) 
63          elif bigram_match: 
64              w1 = bigram_match.groups()[0] 
65              if re.match(start_re, w1): 
66                  w1 = 'start' 
67              w1_dict = self.active_dictionary.setdefault(w1, {}) 
68              print bigram_match.groups()[2] 
69              w1_dict[bigram_match.groups()[1]] = float(bigram_match.groups()[2]) 
70          else: 
71              if not re.match(blank_re,line): 
72                  print 'Unprocessed line %s' % line 
73   
74  if __name__ == '__main__': 
75      import sys, os 
76      files = sys.argv[1:] 
77      for file_str in files: 
78          print 'Processing %s' % file_str 
79          print 
80          read_probs_file (file_str,True) 
81