tomwalters@381: #!/usr/bin/env python
tomwalters@381: # encoding: utf-8
tomwalters@381: """
tomwalters@381: gen_results.py
tomwalters@381: 
tomwalters@381: Created by Thomas Walters on 2010-09-12.
tomwalters@381: """
tomwalters@381: 
tomwalters@381: import sys
tomwalters@381: import getopt
tomwalters@381: import re
tomwalters@381: 
tomwalters@381: 
tomwalters@381: help_message = '''
tomwalters@388: Write out all the results for a run of the
tomwalters@381: syllable recognition experiment. Expected input is a 'misclassificaions'
tomwalters@381: file of the type generated by run_test_instance.sh, along with the locations
tomwalters@381: of files containing the training talkers and all the talkers that the system
tomwalters@381: was tested on.
tomwalters@381: 
tomwalters@381: Arguments:
tomwalters@381: -i --input_file
tomwalters@381: -t --train_talkers
tomwalters@381: -s --test_talkers
tomwalters@381: -c --value_count
tomwalters@381: -p --spoke_pattern
tomwalters@381: '''
tomwalters@381: 
tomwalters@381: class Usage(Exception):
tomwalters@381:   def __init__(self, msg):
tomwalters@381:     self.msg = msg
tomwalters@381: 
tomwalters@381: 
tomwalters@381: def main(argv=None):
tomwalters@381:   if argv is None:
tomwalters@381:     argv = sys.argv
tomwalters@381:   try:
tomwalters@381:     try:
tomwalters@381:       opts, args = getopt.getopt(argv[1:], "hi:t:s:o:c:p:v",
tomwalters@381:                                  ["help", "input_file=", "train_talkers=",
tomwalters@388:                                   "test_talkers=",
tomwalters@381:                                   "value_count=", "spoke_pattern="])
tomwalters@381:     except getopt.error, msg:
tomwalters@381:       raise Usage(msg)
tomwalters@381:   
tomwalters@381:     # defaults
tomwalters@381:     input_file = "misclassified_syllables_iteration_15"
tomwalters@381:     train_talkers = "training_talkers"
tomwalters@381:     test_talkers = "testing_talkers"
tomwalters@381:     output_filename = "results.txt"
tomwalters@381:     total_value_count = 185
tomwalters@381:     spoke_pattern_file = "spoke_pattern.txt"
tomwalters@381:     
tomwalters@381:     # option processing
tomwalters@381:     for option, value in opts:
tomwalters@381:       if option == "-v":
tomwalters@381:         verbose = True
tomwalters@381:       if option in ("-h", "--help"):
tomwalters@381:         raise Usage(help_message)
tomwalters@381:       if option in ("-i", "--input_file"):
tomwalters@381:         input_file = value
tomwalters@381:       if option in ("-t", "--train_talkers"):
tomwalters@381:         train_talkers = value
tomwalters@381:       if option in ("-s", "--test_talkers"):
tomwalters@381:         test_talkers = value
tomwalters@381:       if option in ("-c", "--value_count"):
tomwalters@381:         total_value_count = int(value)
tomwalters@386:       if option in ("-p", "--spoke_pattern"):
tomwalters@381:         spoke_pattern_file = value
tomwalters@381:   
tomwalters@381:   except Usage, err:
tomwalters@381:     print >> sys.stderr, sys.argv[0].split("/")[-1] + ": " + str(err.msg)
tomwalters@381:     print >> sys.stderr, "\t for help use --help"
tomwalters@381:     return 2
tomwalters@381:     
tomwalters@381:   results = dict()
tomwalters@381:   f = open(input_file, 'r')
tomwalters@381:   for line in f:
tomwalters@381:     values = line.strip().split(' ')
tomwalters@381:     results[values[1]]=100*(1-float(values[0])/total_value_count)
tomwalters@381:  	
tomwalters@381:   f = open(test_talkers, 'r')
tomwalters@381:   test_talkers_list = f.readlines()
tomwalters@381:   f.close()
tomwalters@381: 
tomwalters@381:   f = open(train_talkers, 'r')
tomwalters@381:   train_talkers_list = f.readlines()
tomwalters@381:   f.close()
tomwalters@381: 
tomwalters@381:   spoke_pattern = []
tomwalters@381:   f = open(spoke_pattern_file, 'r')
tomwalters@381:   for line in f:
tomwalters@381: 	spoke_pattern.append(line.strip().split(' '))
tomwalters@381: 
tomwalters@381:   all_talkers_list = []
tomwalters@381:   all_talkers_list.extend(train_talkers_list)
tomwalters@381:   all_talkers_list.extend(test_talkers_list)
tomwalters@381: 
tomwalters@381:   # Here I make the rather rash assumption that the model was tested on all talkers
tomwalters@381:   # this should be true if the training and testing was done using my scripts.
tomwalters@381:   for t in all_talkers_list:
tomwalters@381: 	results.setdefault(t.strip(), 100.0)
tomwalters@381: 
tomwalters@381:   total_score = 0.0
tomwalters@381:   element_count = 0
tomwalters@381:   for t in train_talkers_list:
tomwalters@381: 	total_score += results[t.strip()]
tomwalters@381: 	element_count += 1
tomwalters@381:   score = total_score / element_count
tomwalters@381:   print ("# Score on training talkers: %.1f%%" % score)
tomwalters@381: 
tomwalters@381:   total_score = 0.0
tomwalters@381:   element_count = 0
tomwalters@381:   for t in all_talkers_list:
tomwalters@381: 	total_score += results[t.strip()]
tomwalters@381: 	element_count += 1
tomwalters@381:   score = total_score / element_count
tomwalters@381:   print ("# Score on all talkers: %.1f%%" % score)
tomwalters@381: 
tomwalters@381:   total_score = 0.0
tomwalters@381:   element_count = 0
tomwalters@381:   for t in test_talkers_list:
tomwalters@381: 	total_score += results[t.strip()]
tomwalters@381: 	element_count += 1
tomwalters@381:   score = total_score / element_count
tomwalters@381:   print ("# Score on test talkers: %.1f%%" % score)
tomwalters@381: 
tomwalters@381:   score = results[spoke_pattern[0][0]]
tomwalters@381:   print ("# Score on central talker: %.1f" % score)
tomwalters@381: 
tomwalters@381:   for s in xrange(1,9):
tomwalters@381: 	print ("# Results for spoke %d" % s)
tomwalters@381: 	for p in xrange(0, 7):
tomwalters@381: 	  score = results[spoke_pattern[s][p]]
tomwalters@381: 	  m = re.match('(.*)p(.*)s', spoke_pattern[s][p])
tomwalters@381: 	  gpr = m.group(1)
tomwalters@381: 	  vtl=m.group(2)
tomwalters@381: 	  print ("%s %s %s" % (gpr, vtl, score))
tomwalters@381: 
tomwalters@381: if __name__ == "__main__":
tomwalters@381:   sys.exit(main())