|
|
|
@ -31,8 +31,9 @@ def run_pipeline(input_file_name, output_file_name):
|
|
|
|
|
shutil.copyfile(input_file_name, STRING_LIST_FILE_NAME)
|
|
|
|
|
run_obeliks(STRING_LIST_FILE_NAME, OBELIKS_RAW_FILE_NAME)
|
|
|
|
|
tweak_conllu(OBELIKS_RAW_FILE_NAME, OBELIKS_TWEAKED_FILE_NAME)
|
|
|
|
|
run_classla(OBELIKS_TWEAKED_FILE_NAME, CLASSLA_FILE_NAME)
|
|
|
|
|
run_tei_conversion(CLASSLA_FILE_NAME, TEI_INIT_FILE_NAME)
|
|
|
|
|
run_classla(OBELIKS_TWEAKED_FILE_NAME, CLASSLA_OUTPUT_FILE_NAME)
|
|
|
|
|
run_jos_translation(CLASSLA_OUTPUT_FILE_NAME, CLASSLA_TRANSLATED_FILE_NAME)
|
|
|
|
|
run_tei_conversion(CLASSLA_TRANSLATED_FILE_NAME, TEI_INIT_FILE_NAME)
|
|
|
|
|
shutil.copyfile(TEI_INIT_FILE_NAME, output_file_name)
|
|
|
|
|
|
|
|
|
|
def run_obeliks(list_file_name, conllu_file_name):
|
|
|
|
@ -54,6 +55,12 @@ def run_classla(obeliks_file_name, classla_file_name):
|
|
|
|
|
result = nlp(doc)
|
|
|
|
|
result.conll_file.write_conll(classla_file_name)
|
|
|
|
|
|
|
|
|
|
def run_jos_translation(input_file_name, output_file_name):
|
|
|
|
|
print('Translating JOS ...')
|
|
|
|
|
translate_command = ' '.join(['python', TRANSLATION_SCRIPT_NAME, '-infile', input_file_name, '-dict', TRANSLATION_FILE_NAME, '-outfile', output_file_name])
|
|
|
|
|
print(translate_command)
|
|
|
|
|
os.system(translate_command)
|
|
|
|
|
|
|
|
|
|
def run_tei_conversion(classla_file_name, tei_file_name):
|
|
|
|
|
print('Converting to tei ...')
|
|
|
|
|
convert_command = ' '.join(['python', CONLLU_TEI_SCRIPT_NAME, '-o', tei_file_name, classla_file_name])
|
|
|
|
|