diff options
author | Michael Denkowski <mdenkows@cs.cmu.edu> | 2013-09-03 16:35:38 -0700 |
---|---|---|
committer | Michael Denkowski <mdenkows@cs.cmu.edu> | 2013-09-03 16:35:38 -0700 |
commit | 93c9b10cb84118b552a82917871978df3f2af51c (patch) | |
tree | aeb6b6596897957f16295fdcba0b999631e222ee | |
parent | c08eef6a36534ac5b9506f5b18af1765442d3fde (diff) |
Config directory assembly
-rwxr-xr-x | realtime/mkconfig.py | 74 | ||||
-rwxr-xr-x | realtime/realtime.py | 12 |
2 files changed, 82 insertions, 4 deletions
diff --git a/realtime/mkconfig.py b/realtime/mkconfig.py new file mode 100755 index 00000000..86c165d6 --- /dev/null +++ b/realtime/mkconfig.py @@ -0,0 +1,74 @@ +#!/usr/bin/env python + +import os +import shutil +import sys + +from cdec.configobj import ConfigObj + +SA_INI_FILES = set(( + 'f_sa_file', + 'e_file', + 'a_file', + 'lex_file', + 'precompute_file', + )) + +def main(): + + if len(sys.argv[1:]) != 12: + sys.stderr.write('usage: {} a.fwd_params a.fwd_err a.rev_params a.rev_err sa sa.ini mono.klm libcdec_ff_hpyplm.so corpus.hpyplm cdec.ini weights.final output.d\n'.format(sys.argv[0])) + sys.exit(2) + + (a_fwd_params, a_fwd_err, a_rev_params, a_rev_err, sa, sa_ini, mono_klm, libcdec_ff_hpyplm_so, corpus_hpyplm, cdec_ini, weights_final, output_d) = sys.argv[1:] + + if os.path.exists(output_d): + sys.stderr.write('Directory {} exists, exiting.\n'.format(output_d)) + sys.exit(1) + + # output.d + os.mkdir(output_d) + + # alignment model + shutil.copy(a_fwd_params, os.path.join(output_d, 'a.fwd_params')) + shutil.copy(a_fwd_err, os.path.join(output_d, 'a.fwd_err')) + shutil.copy(a_rev_params, os.path.join(output_d, 'a.rev_params')) + shutil.copy(a_rev_err, os.path.join(output_d, 'a.rev_err')) + + # grammar extractor + shutil.copytree(sa, os.path.join(output_d, 'sa')) + config = ConfigObj(sa_ini) + config.filename = os.path.join(output_d, 'sa.ini') + for key in config: + if key in SA_INI_FILES: + config[key] = os.path.join('sa', os.path.basename(config[key])) + config.write() + + # language models + shutil.copy(mono_klm, os.path.join(output_d, 'mono.klm')) + shutil.copy(libcdec_ff_hpyplm_so, os.path.join(output_d, 'libcdec_ff_hpyplm.so')) + shutil.copy(corpus_hpyplm, os.path.join(output_d, 'corpus.hpyplm')) + + # decoder config + config = [(f.strip() for f in line.split('=')) for line in open(cdec_ini)] + with open(os.path.join(output_d, 'cdec.ini'), 'w') as output: + for (k, v) in config: + if k == 'feature_function': + if v.startswith('KLanguageModel'): + f = v.split() + f[-1] = os.path.basename(f[-1]) + v = ' '.join(f) + elif v.startswith('External'): + f = v.split() + if f[1].endswith('libcdec_ff_hpyplm.so'): + for i in range(1, len(f)): + if not f[i].startswith('-'): + f[i] = os.path.basename(f[i]) + v = ' '.join(f) + output.write('{}={}\n'.format(k, v)) + + # weights + shutil.copy(weights_final, os.path.join(output_d, 'weights.final')) + +if __name__ == '__main__': + main() diff --git a/realtime/realtime.py b/realtime/realtime.py index c3ba6f0b..eeaca0f4 100755 --- a/realtime/realtime.py +++ b/realtime/realtime.py @@ -6,18 +6,22 @@ import sys import rt +class Parser(argparse.ArgumentParser): + + def error(self, message): + self.print_help() + sys.stderr.write('\n{}\n'.format(message)) + sys.exit(2) + def main(): - parser = argparse.ArgumentParser(description='Real-time adaptive translation with cdec.') + parser = Parser(description='Real-time adaptive translation with cdec.') parser.add_argument('-c', '--config', required=True, help='Config directory (see README.md)') parser.add_argument('-T', '--temp', help='Temp directory (default /tmp)', default='/tmp') parser.add_argument('-a', '--cache', help='Grammar cache size (default 5)', default='5') parser.add_argument('-v', '--verbose', help='Info to stderr', action='store_true') args = parser.parse_args() - if not args.config: - parser.error('specify a configuration directory') - if args.verbose: logging.basicConfig(level=logging.INFO) |