michael@0: #!/usr/bin/env python michael@0: # Copyright (c) 2011 The Chromium Authors. All rights reserved. michael@0: # Use of this source code is governed by a BSD-style license that can be michael@0: # found in the LICENSE file. michael@0: michael@0: """Convert any unicode characters found in the input file to C literals.""" michael@0: michael@0: import codecs michael@0: import optparse michael@0: import os michael@0: import sys michael@0: michael@0: michael@0: def main(argv): michael@0: parser = optparse.OptionParser() michael@0: usage = 'Usage: %prog -o ' michael@0: parser.set_usage(usage) michael@0: parser.add_option('-o', dest='output_dir') michael@0: michael@0: options, arglist = parser.parse_args(argv) michael@0: michael@0: if not options.output_dir: michael@0: print "output_dir required" michael@0: return 1 michael@0: michael@0: if len(arglist) != 2: michael@0: print "input_file required" michael@0: return 1 michael@0: michael@0: in_filename = arglist[1] michael@0: michael@0: if not in_filename.endswith('.utf8'): michael@0: print "input_file should end in .utf8" michael@0: return 1 michael@0: michael@0: out_filename = os.path.join(options.output_dir, os.path.basename( michael@0: os.path.splitext(in_filename)[0])) michael@0: michael@0: WriteEscapedFile(in_filename, out_filename) michael@0: return 0 michael@0: michael@0: michael@0: def WriteEscapedFile(in_filename, out_filename): michael@0: input_data = codecs.open(in_filename, 'r', 'utf8').read() michael@0: with codecs.open(out_filename, 'w', 'ascii') as out_file: michael@0: for i, char in enumerate(input_data): michael@0: if ord(char) > 127: michael@0: out_file.write(repr(char.encode('utf8'))[1:-1]) michael@0: if input_data[i + 1:i + 2] in '0123456789abcdefABCDEF': michael@0: out_file.write('""') michael@0: else: michael@0: out_file.write(char.encode('ascii')) michael@0: michael@0: michael@0: if __name__ == '__main__': michael@0: sys.exit(main(sys.argv))