gen_engines.py

#!/usr/bin/python
# Copyright (c) 2010 The Chromium OS Authors. All rights reserved.
# Use of this source code is governed by a BSD-style license that can be
# found in the LICENSE file.

"""Extract ibus engine information from component XML files and generate a C++
header.

This program generate a C++ header file containing the information on available
ibus engines.  It parses the XML files specified on the command line, and then
generates a static array definition from the information extracted.

Run it like:
  gen_engines.py [files]

It will produce output that looks like:

// This file is automatically generated by gen_engines.py
#ifndef IBUS_INPUT_METHODS_H
#define IBUS_INPUT_METHODS_H

namespace chromeos {

struct IBusEngineInfo {
  const char* name;
  const char* longname;
  const char* layout;
  const char* language;
};
const IBusEngineInfo ibus_engines[] = {
{"chewing", "Chewing", "us", "zh_TW"},
{"xkb:us::eng", "USA", "us", "eng"},
{"xkb:us:dvorak:eng", "USA - Dvorak", "us(dvorak)", "eng"},
{"xkb:be::fra", "Belgium", "be", "fra"},
{"xkb:br::por", "Brazil", "br", "por"},
};

}  // namespace chromeos

#endif  // IBUS_INPUT_METHODS_H

"""

__author__ = 'zork (Zachary Kuznia)'

import re
import sys
from xml.etree.ElementTree import ElementTree

OUTPUT_HEADER = """
// This file is automatically generated by gen_engines.py
#ifndef IBUS_INPUT_METHODS_H
#define IBUS_INPUT_METHODS_H

namespace chromeos {

struct IBusEngineInfo {
  const char* name;
  const char* longname;
  const char* layout;
  const char* language;
};
const IBusEngineInfo ibus_engines[] = {
"""

ENGINE_FORMAT = """{"%(name)s", "%(longname)s", "%(layout)s", "%(language)s"},
"""

OUTPUT_FOOTER = """
};

}  // namespace chromeos

#endif  // IBUS_INPUT_METHODS_H
"""

def ExtractEngines(filename):
  """Parse the specified file and extract the engine definitions.

  Arguments:
    filename: Full path of the file to parse
  Returns:
    List of engines extracted from the file
  """
  engines = []

  # Parse the file as XML
  tree = ElementTree()
  tree.parse(filename)

  for engine_list in tree.findall('engines'):
    for engine in engine_list.findall('engine'):
      cur_engine = {'name': "", 'longname': "", 'layout': "", 'language': ""}
      name = engine.findall('name')
      longname = engine.findall('longname')
      layout = engine.findall('layout')
      language = engine.findall('language')

      # If there's no name, we can't use this entry
      if not name:
        continue;

      cur_engine['name'] = name[0].text
      if longname:
        cur_engine['longname'] = longname[0].text
      if layout:
        cur_engine['layout'] = layout[0].text
      if language:
        cur_engine['language'] = language[0].text

      engines.append(cur_engine)

  return engines


def GetWhitelist(list_filename):
  """Get whitelisted IDs.

  Arguments:
    list_filename: file name of whitelist.
  Returns:
    list of whitelisted input method IDs.
  """
  whitelist = []
  whitelist_file = open(list_filename, 'r')
  for line in whitelist_file:
    lang = re.sub(r'#.*$', '', line)
    lang = lang.split()
    if not lang: continue
    whitelist.append(lang[0])
  whitelist_file.close()
  return whitelist


def SortEnginesWithWhitelist(engines, whitelist):
  """Sort engines list in the order of appearance in whitelist.

  Note that any engine in whitelist is supposed to exist in engines list.

  Arguments:
    engines: List of engines.
    whitelist: List of whitelisted input method IDs.
  Returns:
    Sorted list of whitelisted input method IDs.
  """
  sorted_engines = []
  engine_names_map = {}
  for engine in engines:
    engine_names_map[engine['name']] = engine
  for lang in whitelist:
    if lang not in engine_names_map:
      print >> sys.stderr, (
          "%s not listed in supported engine, but in whitelist.txt" % lang)
      sys.exit(1)
    sorted_engines.append(engine_names_map[lang])
  return sorted_engines


def CreateEngineHeader(engines):
  """Create the header file from a list of engines.

  Arguments:
    engines: list of ibus engine objects
  Returns:
    The text of a C++ header file containing the engine data.
  """
  output = []
  output.append(OUTPUT_HEADER)
  for engine in engines:
    output.append(ENGINE_FORMAT % engine)
  output.append(OUTPUT_FOOTER)

  return "".join(output)


def main(argv):
  if len(argv) < 3:
    print >> sys.stderr, 'Usage: gen_engines.py [whitelist] [files]'
    sys.exit(1)
  engines = []
  whitelist = GetWhitelist(argv[1])
  for filename in argv[2:]:
    new_engines = ExtractEngines(filename)
    if len(new_engines) == 0:
      print >> sys.stderr, "%s had no engines" % filename
      sys.exit(2)
    engines.extend(new_engines)
  engines = SortEnginesWithWhitelist(engines, whitelist)
  output = CreateEngineHeader(engines)
  print output


if __name__ == '__main__':
  main(sys.argv)