diff options
| -rwxr-xr-x | realtime/mkconfig.py | 27 | ||||
| -rw-r--r-- | realtime/rt/__init__.py | 5 | ||||
| -rw-r--r-- | realtime/rt/util.py | 54 | ||||
| -rwxr-xr-x | word-aligner/force_align.py | 2 |
4 files changed, 57 insertions, 31 deletions
diff --git a/realtime/mkconfig.py b/realtime/mkconfig.py index 86c165d6..954283ce 100755 --- a/realtime/mkconfig.py +++ b/realtime/mkconfig.py @@ -6,13 +6,7 @@ import sys from cdec.configobj import ConfigObj -SA_INI_FILES = set(( - 'f_sa_file', - 'e_file', - 'a_file', - 'lex_file', - 'precompute_file', - )) +import rt def main(): @@ -39,9 +33,7 @@ def main(): 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])) + rt.util.sa_ini_basename(config) config.write() # language models @@ -50,21 +42,10 @@ def main(): 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)] + config = [[f.strip() for f in line.split('=')] for line in open(cdec_ini)] + rt.util.cdec_ini_basename(config) 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 diff --git a/realtime/rt/__init__.py b/realtime/rt/__init__.py index d52157c1..738777f3 100644 --- a/realtime/rt/__init__.py +++ b/realtime/rt/__init__.py @@ -1,3 +1,4 @@ -from aligner import * -from decoder import * from rt import * +import aligner +import decoder +import util diff --git a/realtime/rt/util.py b/realtime/rt/util.py index 263e33fb..885298e6 100644 --- a/realtime/rt/util.py +++ b/realtime/rt/util.py @@ -1,7 +1,24 @@ +import os import subprocess import sys import threading +from cdec.configobj import ConfigObj + +SA_INI_FILES = set(( + 'f_sa_file', + 'e_file', + 'a_file', + 'lex_file', + 'precompute_file', + )) + +def consume_stream(stream): + def consume(s): + for _ in s: + pass + threading.Thread(target=consume, args=(stream,)).start() + def popen_io(cmd): p = subprocess.Popen(cmd, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE) consume_stream(p.stderr) @@ -12,8 +29,35 @@ def popen_io_v(cmd): p = subprocess.Popen(cmd, stdin=subprocess.PIPE, stdout=subprocess.PIPE) return p -def consume_stream(stream): - def consume(s): - for _ in s: - pass - threading.Thread(target=consume, args=(stream,)).start() +def sa_ini_addpath(config, path): + for key in config: + if key in SA_INI_FILES: + config[key] = os.path.join(path, config[key]) + +def sa_ini_basename(config): + for key in config: + if key in SA_INI_FILES: + config[key] = os.path.join('sa', os.path.basename(config[key])) + +def cdec_ini_addpath(config, path): + cdec_ini_fn(config, lambda x: os.path.join(path, x)) + +def cdec_ini_basename(config): + cdec_ini_fn(config, os.path.basename) + +def cdec_ini_fn(config, fn): + # This is a list of (k, v), not a ConfigObj or dict + for i in range(len(config)): + if config[i][0] == 'feature_function': + if config[i][1].startswith('KLanguageModel'): + f = config[i][1].split() + f[-1] = fn(f[-1]) + config[i][1] = ' '.join(f) + elif config[i][1].startswith('External'): + f = config[i][1].split() + if f[1].endswith('libcdec_ff_hpyplm.so'): + for j in range(1, len(f)): + if not f[j].startswith('-'): + f[j] = fn(f[j]) + config[i][1] = ' '.join(f) + diff --git a/word-aligner/force_align.py b/word-aligner/force_align.py index ad6d95fa..6721dacb 100755 --- a/word-aligner/force_align.py +++ b/word-aligner/force_align.py @@ -21,7 +21,7 @@ def main(): aligner = ForceAligner(*sys.argv[1:]) for line in sys.stdin: - sys.stdout.write('{}\n'.format(aligner.align(line.strip()))) + sys.stdout.write('{}\n'.format(aligner.align_formatted(line.strip()))) aligner.close() |