# for 1.0 use "wiki."+embedding_code+".vec"
# for 1.1 use "wiki.dummy.vec"
metadata.embeddings_file_name = "wiki.dummy.vec"
# token delimiter. Must be either space (default) or "" (for languages like Japanese, Chinese, etc.)
if language_code in "zh ja":
metadata.token_delimiter = ""
else:
metadata.token_delimiter = " "
# OPTIONAL: model build date: string
metadata.model_build_date = str(datetime.now())
# OPTIONAL: model build source: what corpus was it built from. Ex: UD-Romanian-RRT v2.2
metadata.model_build_source = local_model
# OPTIONAL: other notes, string value
metadata.notes = "Source: ud-treebanks-v2.2"
metadata.info()
try:
input_folder = os.path.join(input_models_root_folder, local_model)
model_store.package_model(
input_folder,
output_models_root_folder,
metadata,
should_contain_tokenizer=True,
should_contain_compound_word_expander=False,
should_contain_lemmatizer=True,
should_contain_tagger=True,
should_contain_parser=True)
except:
print("Error processing model")
#break # test just one package
print("\n\tModel folder: " + model_folder_path)
print("\tUse tokenizer: {}".format(_tokenizer))
print("\tUse compound word expander: {}".format(_compound_word_expander))
print("\tUse lemmatizer: {}".format(_lemmatizer))
print("\tUse tagger: {}".format(_tagger))
print("\tUse parser: {}\n".format(_parser))
# check if path exists
if not os.path.exists(model_folder_path):
raise Exception("Model folder not found!")
# check if metadata exists
if not os.path.exists(os.path.join(model_folder_path, "metadata.json")):
raise Exception("metadata.json not found in model folder!")
# check if metadata is valid
metadata = ModelMetadata()
metadata.read(os.path.join(model_folder_path, "metadata.json"))
output_folder_path = os.path.dirname(model_folder_path)
model_store_object = ModelStore(disk_path=output_folder_path)
model_store_object.package_model(
model_folder_path,
output_folder_path,
metadata,
should_contain_tokenizer=_tokenizer,
should_contain_compound_word_expander=_compound_word_expander,
should_contain_lemmatizer=_lemmatizer,
should_contain_tagger=_tagger,
should_contain_parser=_parser)