BaseTools/Scripts/ConvertUni.py

   1 ## @file
   2 #  Check a patch for various format issues
   3 #
   4 #  Copyright (c) 2015, Intel Corporation. All rights reserved.<BR>
   5 #
   6 #  SPDX-License-Identifier: BSD-2-Clause-Patent
   7 #
   8
   9 from __future__ import print_function
  10
  11 VersionNumber = '0.1'
  12 __copyright__ = "Copyright (c) 2015, Intel Corporation  All rights reserved."
  13
  14 import argparse
  15 import codecs
  16 import os
  17 import sys
  18
  19 class ConvertOneArg:
  20     """Converts utf-16 to utf-8 for one command line argument.
  21
  22        This could be a single file, or a directory.
  23     """
  24
  25     def __init__(self, utf8, source):
  26         self.utf8 = utf8
  27         self.source = source
  28
  29         self.ok = True
  30
  31         if not os.path.exists(source):
  32             self.ok = False
  33         elif os.path.isdir(source):
  34             for (root, dirs, files) in os.walk(source):
  35                 files = filter(lambda a: a.endswith('.uni'), files)
  36                 for filename in files:
  37                     path = os.path.join(root, filename)
  38                     self.ok &= self.convert_one_file(path)
  39                     if not self.ok:
  40                         break
  41
  42                 if not self.ok:
  43                     break
  44         else:
  45             self.ok &= self.convert_one_file(source)
  46
  47     def convert_one_file(self, source):
  48         if self.utf8:
  49             new_enc, old_enc = 'utf-8', 'utf-16'
  50         else:
  51             new_enc, old_enc = 'utf-16', 'utf-8'
  52         #
  53         # Read file
  54         #
  55         f = open(source, mode='rb')
  56         file_content = f.read()
  57         f.close()
  58
  59         #
  60         # Detect UTF-16 Byte Order Mark at beginning of file.
  61         #
  62         bom = (file_content.startswith(codecs.BOM_UTF16_BE) or
  63                file_content.startswith(codecs.BOM_UTF16_LE))
  64         if bom != self.utf8:
  65             print("%s: already %s" % (source, new_enc))
  66             return True
  67
  68         #
  69         # Decode old string data
  70         #
  71         str_content = file_content.decode(old_enc, 'ignore')
  72
  73         #
  74         # Encode new string data
  75         #
  76         new_content = str_content.encode(new_enc, 'ignore')
  77
  78         #
  79         # Write converted data back to file
  80         #
  81         f = open(source, mode='wb')
  82         f.write(new_content)
  83         f.close()
  84
  85         print(source + ": converted, size", len(file_content), '=>', len(new_content))
  86         return True
  87
  88
  89 class ConvertUniApp:
  90     """Converts .uni files between utf-16 and utf-8."""
  91
  92     def __init__(self):
  93         self.parse_options()
  94         sources = self.args.source
  95
  96         self.ok = True
  97         for patch in sources:
  98             self.process_one_arg(patch)
  99
 100         if self.ok:
 101             self.retval = 0
 102         else:
 103             self.retval = -1
 104
 105     def process_one_arg(self, arg):
 106         self.ok &= ConvertOneArg(self.utf8, arg).ok
 107
 108     def parse_options(self):
 109         parser = argparse.ArgumentParser(description=__copyright__)
 110         parser.add_argument('--version', action='version',
 111                             version='%(prog)s ' + VersionNumber)
 112         parser.add_argument('source', nargs='+',
 113                             help='[uni file | directory]')
 114         group = parser.add_mutually_exclusive_group()
 115         group.add_argument("--utf-8",
 116                            action="store_true",
 117                            help="Convert from utf-16 to utf-8 [default]")
 118         group.add_argument("--utf-16",
 119                            action="store_true",
 120                            help="Convert from utf-8 to utf-16")
 121         self.args = parser.parse_args()
 122         self.utf8 = not self.args.utf_16
 123
 124 if __name__ == "__main__":
 125     sys.exit(ConvertUniApp().retval)