def validate_args(args: Namespace) -> None:
"""
Validate the command line arguments
Parameters
----------
args: :class:`~argparse.Namespace`
Parsed command line arguments
Raises
------
:class:`~montreal_forced_aligner.exceptions.ArgumentError`
If there is a problem with any arguments
"""
args.output_directory = args.output_directory.rstrip("/").rstrip("\\")
args.corpus_directory = args.corpus_directory.rstrip("/").rstrip("\\")
if not os.path.exists(args.corpus_directory):
raise ArgumentError(
f"Could not find the corpus directory {args.corpus_directory}.")
if not os.path.isdir(args.corpus_directory):
raise ArgumentError(
f"The specified corpus directory ({args.corpus_directory}) is not a directory."
)
if args.corpus_directory == args.output_directory:
raise ArgumentError(
"Corpus directory and output directory cannot be the same folder.")
def validate_args(args: Namespace) -> None: # pragma: no cover
"""
Validate the command line arguments
Parameters
----------
args: :class:`~argparse.Namespace`
Parsed command line arguments
Raises
------
:class:`~montreal_forced_aligner.exceptions.ArgumentError`
If there is a problem with any arguments
"""
args.output_directory = args.output_directory.rstrip("/").rstrip("\\")
args.corpus_directory = args.corpus_directory.rstrip("/").rstrip("\\")
if args.cluster and not args.num_speakers:
raise ArgumentError(
"If using clustering, num_speakers must be specified")
if not os.path.exists(args.corpus_directory):
raise ArgumentError(
f"Could not find the corpus directory {args.corpus_directory}.")
if not os.path.isdir(args.corpus_directory):
raise ArgumentError(
f"The specified corpus directory ({args.corpus_directory}) is not a directory."
)
if args.corpus_directory == args.output_directory:
raise ArgumentError(
"Corpus directory and output directory cannot be the same folder.")
args.ivector_extractor_path = validate_model_arg(
args.ivector_extractor_path, "ivector")
def validate_args(args: Namespace) -> None:
"""
Validate the command line arguments
Parameters
----------
args: :class:`~argparse.Namespace`
Parsed command line arguments
Raises
------
:class:`~montreal_forced_aligner.exceptions.ArgumentError`
If there is a problem with any arguments
"""
try:
args.speaker_characters = int(args.speaker_characters)
except ValueError:
pass
if args.test_transcriptions and args.ignore_acoustics:
raise ArgumentError(
"Cannot test transcriptions without acoustic feature generation.")
if not os.path.exists(args.corpus_directory):
raise (ArgumentError(
f"Could not find the corpus directory {args.corpus_directory}."))
if not os.path.isdir(args.corpus_directory):
raise (ArgumentError(
f"The specified corpus directory ({args.corpus_directory}) is not a directory."
))
args.dictionary_path = validate_model_arg(args.dictionary_path,
"dictionary")
if args.acoustic_model_path:
args.acoustic_model_path = validate_model_arg(args.acoustic_model_path,
"acoustic")
def validate_args(args: Namespace) -> None:
"""
Validate the command line arguments
Parameters
----------
args: :class:`~argparse.Namespace`
Parsed command line arguments
Raises
------
:class:`~montreal_forced_aligner.exceptions.ArgumentError`
If there is a problem with any arguments
"""
try:
args.speaker_characters = int(args.speaker_characters)
except ValueError:
pass
args.output_directory = args.output_directory.rstrip("/").rstrip("\\")
args.corpus_directory = args.corpus_directory.rstrip("/").rstrip("\\")
if not os.path.exists(args.corpus_directory):
raise ArgumentError(
f"Could not find the corpus directory {args.corpus_directory}.")
if not os.path.isdir(args.corpus_directory):
raise ArgumentError(
f"The specified corpus directory ({args.corpus_directory}) is not a directory."
)
args.dictionary_path = validate_model_arg(args.dictionary_path,
"dictionary")
args.acoustic_model_path = validate_model_arg(args.acoustic_model_path,
"acoustic")
def validate(args):
if not os.path.exists(args.dictionary_path):
raise (ArgumentError('Could not find the dictionary file {}'.format(
args.dictionary_path)))
if not os.path.isfile(args.dictionary_path):
raise (ArgumentError(
'The specified dictionary path ({}) is not a text file.'.format(
args.dictionary_path)))
def validate(args, download_dictionaries=None):
if args.dictionary_path.lower() in download_dictionaries:
args.dictionary_path = get_dictionary_path(
args.dictionary_path.lower())
if not os.path.exists(args.dictionary_path):
raise (ArgumentError('Could not find the dictionary file {}'.format(
args.dictionary_path)))
if not os.path.isfile(args.dictionary_path):
raise (ArgumentError(
'The specified dictionary path ({}) is not a text file.'.format(
args.dictionary_path)))
def validate_args(args):
if not os.path.exists(args.corpus_directory):
raise (ArgumentError('Could not find the corpus directory {}.'.format(
args.corpus_directory)))
if not os.path.isdir(args.corpus_directory):
raise (ArgumentError(
'The specified corpus directory ({}) is not a directory.'.format(
args.corpus_directory)))
if args.config_path and not os.path.exists(args.config_path):
raise (ArgumentError('Could not find the config file {}.'.format(
args.config_path)))
def validate_args(args):
if args.test_transcriptions and args.ignore_acoustics:
raise ArgumentError('Cannot test transcriptions without acoustic feature generation.')
if not os.path.exists(args.corpus_directory):
raise (ArgumentError('Could not find the corpus directory {}.'.format(args.corpus_directory)))
if not os.path.isdir(args.corpus_directory):
raise (ArgumentError('The specified corpus directory ({}) is not a directory.'.format(args.corpus_directory)))
if not os.path.exists(args.dictionary_path):
raise (ArgumentError('Could not find the dictionary file {}'.format(args.dictionary_path)))
if not os.path.isfile(args.dictionary_path):
raise (ArgumentError('The specified dictionary path ({}) is not a text file.'.format(args.dictionary_path)))
def validate_args(args):
if not os.path.exists(args.corpus_directory):
raise ArgumentError('Could not find the corpus directory {}.'.format(
args.corpus_directory))
if not os.path.isdir(args.corpus_directory):
raise ArgumentError(
'The specified corpus directory ({}) is not a directory.'.format(
args.corpus_directory))
if args.corpus_directory == args.output_directory:
raise ArgumentError(
'Corpus directory and output directory cannot be the same folder.')
def validate_args(args):
args.model_type = args.model_type.lower()
if args.model_type not in ['acoustic', 'g2p', 'dictionary', 'ivector']:
raise ArgumentError(
"model_type must be one of 'acoustic', 'g2p', 'dictionary', or 'ivector"
)
if args.language is not None:
available_languages = list_available_languages(args.model_type)
if args.language not in available_languages:
possible = ', '.join(available_languages)
raise ArgumentError(
'Could not find {}, '
'possible languages for download are: {}'.format(
args.language, possible))
def validate_args(args, download_dictionaries=None):
if args.dictionary_path is not None and args.dictionary_path.lower() in download_dictionaries:
args.dictionary_path = get_dictionary_path(args.dictionary_path.lower())
if not args.source_path.endswith('.arpa'):
if not os.path.exists(args.source_path):
raise (ArgumentError('Could not find the corpus directory {}.'.format(args.source_path)))
if not os.path.isdir(args.source_path):
raise (ArgumentError('The specified corpus directory ({}) is not a directory.'.format(args.source_path)))
else:
if not os.path.exists(args.source_path):
raise (ArgumentError('Could not find the source file {}.'.format(args.source_path)))
if args.config_path and not os.path.exists(args.config_path):
raise (ArgumentError('Could not find the config file {}.'.format(args.config_path)))
if args.model_path and not os.path.exists(args.model_path):
raise (ArgumentError('Could not find the model file {}.'.format(args.model_path)))
def validate_args(args, downloaded_acoustic_models, download_dictionaries,
downloaded_language_models):
if not os.path.exists(args.corpus_directory):
raise ArgumentError('Could not find the corpus directory {}.'.format(
args.corpus_directory))
if not os.path.isdir(args.corpus_directory):
raise ArgumentError(
'The specified corpus directory ({}) is not a directory.'.format(
args.corpus_directory))
if args.dictionary_path.lower() in download_dictionaries:
args.dictionary_path = get_dictionary_path(
args.dictionary_path.lower())
if not os.path.exists(args.dictionary_path):
raise ArgumentError('Could not find the dictionary file {}'.format(
args.dictionary_path))
if not os.path.isfile(args.dictionary_path):
raise ArgumentError(
'The specified dictionary path ({}) is not a text file.'.format(
args.dictionary_path))
if args.corpus_directory == args.output_directory:
raise ArgumentError(
'Corpus directory and output directory cannot be the same folder.')
if args.acoustic_model_path.lower() in downloaded_acoustic_models:
args.acoustic_model_path = get_pretrained_acoustic_path(
args.acoustic_model_path.lower())
elif args.acoustic_model_path.lower().endswith(AcousticModel.extension):
if not os.path.exists(args.acoustic_model_path):
raise ArgumentError('The specified model path does not exist: ' +
args.acoustic_model_path)
else:
raise ArgumentError(
'The acoustic model \'{}\' is not currently downloaded, '
'please download a pretrained model, align via training or specify one of the following language names: {}.'
.format(args.acoustic_model_path.lower(),
', '.join(downloaded_acoustic_models)))
if args.language_model_path.lower() in downloaded_language_models:
args.language_model_path = get_pretrained_language_model_path(
args.language_model_path.lower())
elif args.language_model_path.lower().endswith(LanguageModel.extension) or \
args.language_model_path.lower().endswith(FORMAT):
if not os.path.exists(args.language_model_path):
raise ArgumentError('The specified model path does not exist: ' +
args.language_model_path)
else:
raise ArgumentError(
'The language model \'{}\' is not currently downloaded, '
'please download, train a new model, or specify one of the following language names: {}.'
.format(args.language_model_path.lower(),
', '.join(downloaded_language_models)))
def validate(args, pretrained_languages):
if not args.g2p_model_path:
args.g2p_model_path = None
elif args.g2p_model_path in pretrained_languages:
args.g2p_model_path = get_pretrained_g2p_path(args.g2p_model_path)
if args.g2p_model_path and not os.path.exists(args.g2p_model_path):
raise (ArgumentError('Could not find the G2P model file {}.'.format(
args.g2p_model_path)))
if args.g2p_model_path and (not os.path.isfile(args.g2p_model_path)
or not args.g2p_model_path.endswith('.zip')):
raise (ArgumentError(
'The specified G2P model path ({}) is not a zip file.'.format(
args.g2p_model_path)))
if not os.path.exists(args.input_path):
raise (ArgumentError('Could not find the input path {}.'.format(
args.input_path)))
def validate_args(args, downloaded_acoustic_models, download_dictionaries):
if args.config_path and not os.path.exists(args.config_path):
raise (ArgumentError('Could not find the config file {}.'.format(args.config_path)))
if not os.path.exists(args.corpus_directory):
raise (ArgumentError('Could not find the corpus directory {}.'.format(args.corpus_directory)))
if not os.path.isdir(args.corpus_directory):
raise (ArgumentError('The specified corpus directory ({}) is not a directory.'.format(args.corpus_directory)))
if args.dictionary_path.lower() in download_dictionaries:
args.dictionary_path = get_dictionary_path(args.dictionary_path.lower())
if not os.path.exists(args.dictionary_path):
raise ArgumentError('Could not find the dictionary file {}'.format(args.dictionary_path))
if not os.path.isfile(args.dictionary_path):
raise ArgumentError('The specified dictionary path ({}) is not a text file.'.format(args.dictionary_path))
if args.acoustic_model_path.lower() in downloaded_acoustic_models:
args.acoustic_model_path = get_pretrained_acoustic_path(args.acoustic_model_path.lower())
elif args.acoustic_model_path.lower().endswith(AcousticModel.extension):
if not os.path.exists(args.acoustic_model_path):
raise ArgumentError('The specified model path does not exist: ' + args.acoustic_model_path)
else:
raise ArgumentError(
'The language \'{}\' is not currently included in the distribution, '
'please align via training or specify one of the following language names: {}.'.format(
args.acoustic_model_path.lower(), ', '.join(downloaded_acoustic_models)))
def validate_args(args):
if args.corpus_directory == args.output_directory:
raise Exception(
'Corpus directory and output directory cannot be the same folder.')
if not os.path.exists(args.corpus_directory):
raise (ArgumentError('Could not find the corpus directory {}.'.format(
args.corpus_directory)))
if not os.path.isdir(args.corpus_directory):
raise (ArgumentError(
'The specified corpus directory ({}) is not a directory.'.format(
args.corpus_directory)))
if not os.path.exists(args.dictionary_path):
raise (ArgumentError('Could not find the dictionary file {}'.format(
args.dictionary_path)))
if not os.path.isfile(args.dictionary_path):
raise (ArgumentError(
'The specified dictionary path ({}) is not a text file.'.format(
args.dictionary_path)))
def validate_args(args: Namespace) -> None:
"""
Validate the command line arguments
Parameters
----------
args: :class:`~argparse.Namespace`
Parsed command line arguments
Raises
------
:class:`~montreal_forced_aligner.exceptions.ArgumentError`
If there is a problem with any arguments
"""
try:
args.speaker_characters = int(args.speaker_characters)
except ValueError:
pass
args.output_directory = None
if not args.output_model_path:
args.output_model_path = None
output_paths = args.output_paths
if len(output_paths) > 2:
raise ArgumentError(f"Got more arguments for output_paths than 2: {output_paths}")
for path in output_paths:
if path.endswith(".zip"):
args.output_model_path = path
else:
args.output_directory = path.rstrip("/").rstrip("\\")
args.corpus_directory = args.corpus_directory.rstrip("/").rstrip("\\")
if args.corpus_directory == args.output_directory:
raise ArgumentError("Corpus directory and output directory cannot be the same folder.")
if not os.path.exists(args.corpus_directory):
raise (ArgumentError(f'Could not find the corpus directory "{args.corpus_directory}".'))
if not os.path.isdir(args.corpus_directory):
raise (
ArgumentError(
f'The specified corpus directory "{args.corpus_directory}" is not a directory.'
)
)
args.dictionary_path = validate_model_arg(args.dictionary_path, "dictionary")
def validate_args(args, downloaded_ivector_extractors):
if args.cluster and not args.num_speakers:
raise ArgumentError(
'If using clustering, num_speakers must be specified')
if not os.path.exists(args.corpus_directory):
raise ArgumentError('Could not find the corpus directory {}.'.format(
args.corpus_directory))
if not os.path.isdir(args.corpus_directory):
raise ArgumentError(
'The specified corpus directory ({}) is not a directory.'.format(
args.corpus_directory))
if args.corpus_directory == args.output_directory:
raise ArgumentError(
'Corpus directory and output directory cannot be the same folder.')
if args.ivector_extractor_path.lower() in downloaded_ivector_extractors:
args.ivector_extractor_path = get_pretrained_ivector_path(
args.ivector_extractor_path.lower())
elif args.ivector_extractor_path.lower().endswith(
IvectorExtractor.extension):
if not os.path.exists(args.ivector_extractor_path):
raise ArgumentError('The specified model path does not exist: ' +
args.ivector_extractor_path)
else:
raise ArgumentError(
'The language \'{}\' is not currently included in the distribution, '
'please align via training or specify one of the following language names: {}.'
.format(args.ivector_extractor_path.lower(),
', '.join(downloaded_ivector_extractors)))
def validate_args(args, download_dictionaries=None):
if args.config_path and not os.path.exists(args.config_path):
raise (ArgumentError('Could not find the config file {}.'.format(
args.config_path)))
if not os.path.exists(args.corpus_directory):
raise (ArgumentError('Could not find the corpus directory {}.'.format(
args.corpus_directory)))
if not os.path.isdir(args.corpus_directory):
raise (ArgumentError(
'The specified corpus directory ({}) is not a directory.'.format(
args.corpus_directory)))
if args.dictionary_path.lower() in download_dictionaries:
args.dictionary_path = get_dictionary_path(
args.dictionary_path.lower())
if not os.path.exists(args.dictionary_path):
raise (ArgumentError('Could not find the dictionary file {}'.format(
args.dictionary_path)))
if not os.path.isfile(args.dictionary_path):
raise (ArgumentError(
'The specified dictionary path ({}) is not a text file.'.format(
args.dictionary_path)))
def validate_args(args: Namespace) -> None:
"""
Validate the command line arguments
Parameters
----------
args: :class:`~argparse.Namespace`
Parsed command line arguments
Raises
------
:class:`~montreal_forced_aligner.exceptions.ArgumentError`
If there is a problem with any arguments
"""
args.source_path = args.source_path.rstrip("/").rstrip("\\")
if args.dictionary_path:
args.dictionary_path = validate_model_arg(args.dictionary_path,
"dictionary")
if not args.source_path.endswith(".arpa"):
if not os.path.exists(args.source_path):
raise (ArgumentError(
f"Could not find the corpus directory {args.source_path}."))
if not os.path.isdir(args.source_path):
raise (ArgumentError(
f"The specified corpus directory ({args.source_path}) is not a directory."
))
else:
if not os.path.exists(args.source_path):
raise (ArgumentError(
f"Could not find the source file {args.source_path}."))
if args.config_path and not os.path.exists(args.config_path):
raise (ArgumentError(
f"Could not find the config file {args.config_path}."))
if args.model_path and not os.path.exists(args.model_path):
raise (
ArgumentError(f"Could not find the model file {args.model_path}."))
def validate_dictionary_args(args: Namespace) -> None:
"""
Validate the command line arguments
Parameters
----------
args: :class:`~argparse.Namespace`
Parsed command line arguments
Raises
------
:class:`~montreal_forced_aligner.exceptions.ArgumentError`
If there is a problem with any arguments
"""
if sys.platform == "win32":
raise ArgumentError(
"Cannot validate dictionaries on native Windows, please use Windows Subsystem for Linux."
)
args.dictionary_path = validate_model_arg(args.dictionary_path,
"dictionary")
if args.g2p_model_path:
args.g2p_model_path = validate_model_arg(args.g2p_model_path, "g2p")