https://github.com/google/mozc/issues/462

--- /src/build_tools/redirect.py
+++ /src/build_tools/redirect.py
@@ -58,14 +58,15 @@
     process = subprocess.Popen(sys.argv, stdout=subprocess.PIPE,
                                universal_newlines=True)
   except:
-    print '=========='
-    print ' ERROR: %s' % ' '.join(sys.argv)
-    print '=========='
+    print('==========')
+    print(' ERROR: %s' % ' '.join(sys.argv))
+    print('==========')
     raise
   (stdout_content, _) = process.communicate()
   # Write the stdout content to the output file.
   output_file = open(output_file_name, 'w')
   output_file.write(stdout_content)
+  output_file.close()
   return process.wait()
 
 if __name__ == '__main__':
--- /src/build_tools/run_after_chdir.py
+++ /src/build_tools/run_after_chdir.py
@@ -57,7 +57,7 @@
     sys.argv.insert(0, sys.executable)  # Inject the python interpreter path.
   # We don't capture stdout and stderr from Popen. The output will just
   # be emitted to a terminal or console.
-  print sys.argv
+  print(sys.argv)
   sys.exit(subprocess.call(sys.argv))
 
 if __name__ == '__main__':
--- /src/build_tools/serialized_string_array_builder.py
+++ /src/build_tools/serialized_string_array_builder.py
@@ -58,11 +58,11 @@
     f.write(struct.pack('<I', array_size))
 
     # Offset and length array of (4 + 4) * array_size bytes.
-    for i in xrange(array_size):
+    for i in range(array_size):
       f.write(struct.pack('<I', offsets[i]))
       f.write(struct.pack('<I', lengths[i]))
 
     # Strings chunk.
-    for i in xrange(array_size):
+    for i in range(array_size):
       f.write(strings[i])
-      f.write('\0')
+      f.write(b'\0')
--- /src/build_tools/test_tools/gtest_report.py
+++ /src/build_tools/test_tools/gtest_report.py
@@ -36,9 +36,9 @@
 
 __author__ = "nona"
 
-import cStringIO as StringIO
+import io
 import logging
-from xml.etree import cElementTree as ElementTree
+from xml.etree import ElementTree
 
 
 class Failure(object):
@@ -87,13 +87,13 @@
     """Returns summarized error report text."""
     if self.fail_num == 0:
       return ''
-    output = StringIO.StringIO()
+    output = io.StringIO()
     for testcase in self.testcases:
       if not testcase.failures:
         continue
-      print >>output, '%s.%s:' % (self.name, testcase.name)
+      print('%s.%s:' % (self.name, testcase.name), file=output)
       for failure in testcase.failures:
-        print >>output, failure.contents.encode('utf-8')
+        print(failure.contents.encode('utf-8'), file=output)
     return output.getvalue()
 
   @classmethod
--- /src/build_tools/test_tools/test_launcher.py
+++ /src/build_tools/test_tools/test_launcher.py
@@ -101,11 +101,11 @@
       time.sleep(1)
       try:
         shutil.rmtree(self._path)
-      except OSError, e:
+      except OSError as e:
         logging.error('Failed to remove %s. error: %s', self._path, e)
 
 
-def _ExecuteTest((command, gtest_report_dir)):
+def _ExecuteTest(args):
   """Executes tests with specified Test command.
 
   Args:
@@ -122,6 +122,7 @@
   module, which is used in multiprocessing module.
   (http://docs.python.org/library/pickle.html)
   """
+  (command, gtest_report_dir) = args
   binary = command[0]
   binary_filename = os.path.basename(binary)
   tmp_dir = tempfile.mkdtemp()
--- /src/build_tools/tweak_data.py
+++ /src/build_tools/tweak_data.py
@@ -55,7 +55,7 @@
       The value for the variable if the variable is defined in the
       environment.  Otherwise original string is returned.
     """
-    if environment.has_key(matchobj.group(1)):
+    if matchobj.group(1) in environment:
       return environment[matchobj.group(1)]
     return matchobj.group(0)
 
--- /src/build_tools/tweak_info_plist.py
+++ /src/build_tools/tweak_info_plist.py
@@ -42,8 +42,8 @@
 import logging
 import optparse
 import sys
-import mozc_version
-import tweak_data
+from . import mozc_version
+from . import tweak_data
 
 _COPYRIGHT_YEAR = datetime.date.today().year
 
@@ -81,7 +81,7 @@
 
   version = mozc_version.MozcVersion(options.version_file)
 
-  copyright_message = (u'© %d Google Inc.' % _COPYRIGHT_YEAR).encode('utf-8')
+  copyright_message = ('© %d Google Inc.' % _COPYRIGHT_YEAR).encode('utf-8')
   long_version = version.GetVersionString()
   short_version = version.GetVersionInFormat('@MAJOR@.@MINOR@.@BUILD@')
 
--- /src/build_tools/tweak_info_plist_strings.py
+++ /src/build_tools/tweak_info_plist_strings.py
@@ -40,7 +40,7 @@
 import logging
 import optparse
 import sys
-import tweak_data
+from . import tweak_data
 
 _COPYRIGHT_YEAR = datetime.date.today().year
 
@@ -77,7 +77,7 @@
   if options.branding == 'GoogleJapaneseInput':
     variables = {
         'CF_BUNDLE_NAME_EN': 'Google Japanese Input',
-        'CF_BUNDLE_NAME_JA': u'Google 日本語入力'.encode('utf-8'),
+        'CF_BUNDLE_NAME_JA': 'Google 日本語入力'.encode('utf-8'),
         'NS_HUMAN_READABLE_COPYRIGHT': copyright_message,
         'INPUT_MODE_ANNOTATION': 'Google',
         }
--- /src/build_tools/tweak_macinstaller_script.py
+++ /src/build_tools/tweak_macinstaller_script.py
@@ -39,7 +39,7 @@
 import logging
 import optparse
 
-import mozc_version
+from . import mozc_version
 
 
 def _ReplaceVariables(data, environment):
--- /src/build_tools/tweak_pkgproj.py
+++ /src/build_tools/tweak_pkgproj.py
@@ -45,7 +45,7 @@
 import os
 import plistlib
 import re
-import mozc_version
+from . import mozc_version
 
 from os import path
 
@@ -71,7 +71,7 @@
       The value for the variable if the variable is defined in the
       environment.  Otherwise original string is returned.
     """
-    if environment.has_key(matchobj.group(1)):
+    if matchobj.group(1) in environment:
       return environment[matchobj.group(1)]
     return matchobj.group(0)
 
--- /src/build_tools/util.py
+++ /src/build_tools/util.py
@@ -73,11 +73,11 @@
   return 1
 
 
-class RunOrDieError(StandardError):
+class RunOrDieError(Exception):
   """The exception class for RunOrDie."""
 
   def __init__(self, message):
-    StandardError.__init__(self, message)
+    Exception.__init__(self, message)
 
 
 def RunOrDie(argv):
@@ -105,7 +105,7 @@
     return  # Do nothing if not exist.
   if IsWindows():
     # Read-only files cannot be deleted on Windows.
-    os.chmod(file_name, 0700)
+    os.chmod(file_name, 0o700)
   logging.debug('Removing file: %s', file_name)
   os.unlink(file_name)
 
--- /src/build_tools/zlib_util.py
+++ /src/build_tools/zlib_util.py
@@ -58,7 +58,7 @@
 
 def main():
   if len(sys.argv) != 4:
-    print >>sys.stderr, 'Invalid arguments'
+    print('Invalid arguments', file=sys.stderr)
     return
   if sys.argv[1] == 'compress':
     Compress(sys.argv[2], sys.argv[3])
@@ -66,7 +66,7 @@
   if sys.argv[1] == 'decompress':
     Decompress(sys.argv[2], sys.argv[3])
     return
-  print >>sys.stderr, 'Unknown command:', sys.argv[1]
+  print('Unknown command:', sys.argv[1], file=sys.stderr)
 
 
 if __name__ == '__main__':
--- /src/composer/internal/gen_typing_model.py
+++ /src/composer/internal/gen_typing_model.py
@@ -54,14 +54,13 @@
 __author__ = "noriyukit"
 
 import bisect
-import codecs
 import collections
 import optparse
 import struct
 
 UNDEFINED_COST = -1
-MAX_UINT16 = struct.unpack('H', '\xFF\xFF')[0]
-MAX_UINT8 = struct.unpack('B', '\xFF')[0]
+MAX_UINT16 = struct.unpack('H', b'\xFF\xFF')[0]
+MAX_UINT8 = struct.unpack('B', b'\xFF')[0]
 
 
 def ParseArgs():
@@ -113,7 +112,7 @@
   sorted_values = list(sorted(set(values)))
   mapping_table = sorted_values[0]
   mapping_table_size_without_special_value = mapping_table_size - 1
-  span = len(sorted_values) / (mapping_table_size_without_special_value - 1)
+  span = len(sorted_values) // (mapping_table_size_without_special_value - 1)
   mapping_table = [sorted_values[i * span]
                    for i
                    in range(0, mapping_table_size_without_special_value - 1)]
@@ -150,7 +149,7 @@
 
 def GetValueTable(unique_characters, mapping_table, dictionary):
   result = []
-  for key, value in dictionary.iteritems():
+  for key, value in dictionary.items():
     index = GetIndexFromKey(unique_characters, key)
     while len(result) <= index:
       result.append(len(mapping_table) - 1)
@@ -167,13 +166,13 @@
                              romaji_transition_cost)
   with open(output_path, 'wb') as f:
     f.write(struct.pack('<I', len(unique_characters)))
-    f.write(''.join(unique_characters))
+    f.write(''.join(unique_characters).encode('utf-8'))
     offset = 4 + len(unique_characters)
 
     # Add padding to place value list size at 4-byte boundary.
     if offset % 4:
       padding_size = 4 - offset % 4
-      f.write('\x00' * padding_size)
+      f.write(b'\x00' * padding_size)
       offset += padding_size
 
     f.write(struct.pack('<I', len(value_list)))
@@ -184,7 +183,7 @@
     # Add padding to place mapping_table at 4-byte boundary.
     if offset % 4:
       padding_size = 4 - offset % 4
-      f.write('\x00' * padding_size)
+      f.write(b'\x00' * padding_size)
       offset += padding_size
 
     for v in mapping_table:
@@ -198,7 +197,8 @@
   #   - trigram['vw']['x'] = -500 * log(P(x | 'vw'))
   unigram = {}
   trigram = collections.defaultdict(dict)
-  for line in codecs.open(options.input_path, 'r', encoding='utf-8'):
+  input_file = open(options.input_path, 'r', encoding='utf-8')
+  for line in input_file:
     line = line.rstrip()
     ngram, cost = line.split('\t')
     cost = int(cost)
@@ -206,6 +206,7 @@
       unigram[ngram] = cost
     else:
       trigram[ngram[:-1]][ngram[-1]] = cost
+  input_file.close()
 
   # Calculate ngram-related cost for each 'vw' and 'x':
   #     -500 * log( P('x' | 'vw') / P('x') )
--- /src/converter/gen_boundary_data.py
+++ /src/converter/gen_boundary_data.py
@@ -70,7 +70,8 @@
 def LoadPatterns(file):
   prefix = []
   suffix = []
-  for line in open(file, 'r'):
+  fh = open(file, 'r')
+  for line in fh:
     if len(line) <= 1 or line[0] == '#':
       continue
     fields = line.split()
@@ -84,8 +85,9 @@
     elif label == 'SUFFIX':
       suffix.append([re.compile(PatternToRegexp(feature)), cost])
     else:
-      print 'format error %s' % (line)
+      print('format error %s' % (line))
       sys.exit(0)
+  fh.close()
   return (prefix, suffix)
 
 
@@ -100,19 +102,23 @@
 
 def LoadFeatures(filename):
   features = []
-  for line in open(filename, 'r'):
+  fh = open(filename, 'r')
+  for line in fh:
     fields = line.split()
     features.append(fields[1])
+  fh.close()
   return features
 
 
 def CountSpecialPos(filename):
   count = 0
-  for line in open(filename, 'r'):
+  fh = open(filename, 'r')
+  for line in fh:
     line = line.rstrip()
     if not line or line[0] == '#':
       continue
     count += 1
+  fh.close()
   return count
 
 
@@ -141,7 +147,7 @@
       f.write(struct.pack('<H', GetCost(prefix, feature)))
       f.write(struct.pack('<H', GetCost(suffix, feature)))
 
-    for _ in xrange(num_special_pos):
+    for _ in range(num_special_pos):
       f.write(struct.pack('<H', 0))
       f.write(struct.pack('<H', 0))
 
--- /src/converter/gen_quality_regression_test_data.py
+++ /src/converter/gen_quality_regression_test_data.py
@@ -84,7 +84,7 @@
                else _ENABLED)
     id = issue.attributes['id'].value
     target = GetText(issue.getElementsByTagName('target'))
-    for detail in issue.getElementsByTagName(u'detail'):
+    for detail in issue.getElementsByTagName('detail'):
       fields = []
       fields.append('mozcsu_%s' % id)
       for key in ('reading', 'output', 'actionStatus', 'rank', 'accuracy'):
@@ -104,19 +104,19 @@
 
 def GenerateHeader(files):
   try:
-    print 'namespace mozc{'
-    print 'struct TestCase {'
-    print '  const bool enabled;'
-    print '  const char *tsv;'
-    print '} kTestData[] = {'
+    print('namespace mozc{')
+    print('struct TestCase {')
+    print('  const bool enabled;')
+    print('  const char *tsv;')
+    print('} kTestData[] = {')
     for file in files:
       for enabled, line in ParseFile(file):
-        print ' {%s, "%s"},' % (enabled, EscapeString(line))
-    print '  {false, nullptr},'
-    print '};'
-    print '}  // namespace mozc'
+        print(' {%s, "%s"},' % (enabled, EscapeString(line)))
+    print('  {false, nullptr},')
+    print('};')
+    print('}  // namespace mozc')
   except:
-    print 'cannot open %s' % (file)
+    print('cannot open %s' % (file))
     sys.exit(1)
 
 
--- /src/converter/gen_segmenter_code.py
+++ /src/converter/gen_segmenter_code.py
@@ -54,18 +54,22 @@
   pos = {}
   max_id = 0
 
-  for line in open(id_file, "r"):
+  fh = open(id_file, "r")
+  for line in fh:
     fields = line.split()
     pos[fields[1]] = fields[0]
     max_id = max(int(fields[0]), max_id)
+  fh.close()
 
   max_id = max_id + 1
-  for line in open(special_pos_file, "r"):
+  fh = open(special_pos_file, "r")
+  for line in fh:
     if len(line) <= 1 or line[0] == '#':
       continue
     fields = line.split()
     pos[fields[0]] = ("%d" % max_id)
     max_id = max_id + 1
+  fh.close()
 
   return pos
 
@@ -79,8 +83,7 @@
   pat = re.compile(PatternToRegexp(pattern))
   min = -1;
   max = -1;
-  keys = pos.keys()
-  keys.sort()
+  keys = sorted(pos.keys())
 
   range = []
 
@@ -107,7 +110,7 @@
       tmp.append("(%s >= %s && %s <= %s)" % (name, r[0], name, r[1]))
 
   if len(tmp) == 0:
-    print "FATAL: No rule fiind %s" % (pattern)
+    print("FATAL: No rule fiind %s" % (pattern))
     sys.exit(-1)
 
   return " || ".join(tmp)
@@ -115,19 +118,21 @@
 def main():
   pos = ReadPOSID(sys.argv[1], sys.argv[2])
 
-  print HEADER % (len(pos.keys()), len(pos.keys()))
+  print(HEADER % (len(pos.keys()), len(pos.keys())))
 
-  for line in open(sys.argv[3], "r"):
+  fh = open(sys.argv[3], "r")
+  for line in fh:
     if len(line) <= 1 or line[0] == '#':
       continue
     (l, r, result) = line.split()
     result = result.lower()
     lcond = GetRange(pos, l, "rid") or "true";
     rcond = GetRange(pos, r, "lid") or "true";
-    print "  // %s %s %s" % (l, r, result)
-    print "  if ((%s) && (%s)) { return %s; }" % (lcond, rcond, result)
+    print("  // %s %s %s" % (l, r, result))
+    print("  if ((%s) && (%s)) { return %s; }" % (lcond, rcond, result))
+  fh.close()
 
-  print FOOTER
+  print(FOOTER)
 
 if __name__ == "__main__":
   main()
--- /src/data_manager/gen_connection_data.py
+++ /src/data_manager/gen_connection_data.py
@@ -32,8 +32,7 @@
 
 __author__ = "hidehiko"
 
-import cStringIO as StringIO
-import itertools
+import io
 import logging
 import optparse
 import os
@@ -45,7 +44,7 @@
 INVALID_COST = 30000
 INVALID_1BYTE_COST = 255
 RESOLUTION_FOR_1BYTE = 64
-FILE_MAGIC = '\xAB\xCD'
+FILE_MAGIC = b'\xAB\xCD'
 
 FALSE_VALUES = ['f', 'false', '0']
 TRUE_VALUES = ['t', 'true', '1']
@@ -79,28 +78,28 @@
   # The result is a square matrix.
   mat_size = pos_size + special_pos_size
 
-  matrix = [[0] * mat_size for _ in xrange(mat_size)]
+  matrix = [[0] * mat_size for _ in range(mat_size)]
   with open(text_connection_file) as stream:
     stream = code_generator_util.SkipLineComment(stream)
     # The first line contains the matrix column/row size.
-    size = stream.next().rstrip()
+    size = next(stream).rstrip()
     assert (int(size) == pos_size), '%s != %d' % (size, pos_size)
 
     for array_index, cost in enumerate(stream):
       cost = int(cost.rstrip())
-      rid = array_index / pos_size
+      rid = array_index // pos_size
       lid = array_index % pos_size
       if rid == 0 and lid == 0:
         cost = 0
       matrix[rid][lid] = cost
 
   # Fill INVALID_COST in matrix elements for special POS.
-  for rid in xrange(pos_size, mat_size):
-    for lid in xrange(1, mat_size):  # Skip EOS
+  for rid in range(pos_size, mat_size):
+    for lid in range(1, mat_size):  # Skip EOS
       matrix[rid][lid] = INVALID_COST
 
-  for lid in xrange(pos_size, mat_size):
-    for rid in xrange(1, mat_size):  # Skip BOS
+  for lid in range(pos_size, mat_size):
+    for rid in range(1, mat_size):  # Skip BOS
       matrix[rid][lid] = INVALID_COST
 
   return matrix
@@ -116,7 +115,7 @@
         # Heuristically, we do not compress INVALID_COST.
         continue
       m[cost] = m.get(cost, 0) + 1
-    mode_value = max(m.iteritems(), key=lambda (_, count): count)[0]
+    mode_value = max(m.items(), key=lambda x: x[1])[0]
     result.append(mode_value)
   return result
 
@@ -126,8 +125,8 @@
   # list, and fill None into the matrix if it equals to the corresponding
   # mode value.
   assert len(matrix) == len(mode_value_list)
-  for row, mode_value in itertools.izip(matrix, mode_value_list):
-    for index in xrange(len(row)):
+  for row, mode_value in zip(matrix, mode_value_list):
+    for index in range(len(row)):
       if row[index] == mode_value:
         row[index] = None
 
@@ -179,7 +178,7 @@
     resolution = RESOLUTION_FOR_1BYTE
   else:
     resolution = 1
-  stream = StringIO.StringIO()
+  stream = io.BytesIO()
 
   # Output header.
   stream.write(FILE_MAGIC)
@@ -194,7 +193,7 @@
 
   # 4 bytes alignment.
   if len(mode_value_list) % 2:
-    stream.write('\x00\x00')
+    stream.write(b'\x00\x00')
 
   # Process each row:
   for row in matrix:
@@ -218,7 +217,7 @@
             if cost == INVALID_COST:
               cost = INVALID_1BYTE_COST
             else:
-              cost /= resolution
+              cost //= resolution
               assert cost != INVALID_1BYTE_COST
           values.append(cost)
 
@@ -237,7 +236,7 @@
       values_size = len(values) * 2
 
     # Output the bits for a row.
-    stream.write(struct.pack('<HH', len(compact_bits) / 8, values_size))
+    stream.write(struct.pack('<HH', len(compact_bits) // 8, values_size))
     OutputBitList(chunk_bits, stream)
     OutputBitList(compact_bits, stream)
     if use_1byte_cost: