IssueID #1487: kept more intermediate results

This commit is contained in:
Cyprian Laskowski 2020-09-30 19:49:36 +02:00
parent 58f37ecd08
commit c9a9d9a8bc

View File

@ -28,11 +28,11 @@ OBELIKS_RAW_FILE_NAME = TMP_DIRECTORY + '/obeliks_raw.conllu'
OBELIKS_TWEAKED_FILE_NAME = TMP_DIRECTORY + '/obeliks_tweaked.conllu' OBELIKS_TWEAKED_FILE_NAME = TMP_DIRECTORY + '/obeliks_tweaked.conllu'
CLASSLA_FILE_NAME = TMP_DIRECTORY + '/classla.conllu' CLASSLA_FILE_NAME = TMP_DIRECTORY + '/classla.conllu'
TEI_INIT_FILE_NAME = TMP_DIRECTORY + '/tei_initial.xml' TEI_INIT_FILE_NAME = TMP_DIRECTORY + '/tei_initial.xml'
TEI_STRUCTURE_FILE_NAME = TMP_DIRECTORY + '/tei_with_structure_ids.xml' TEI_STRUCTURE_1_FILE_NAME = TMP_DIRECTORY + '/tei_with_structure_ids1.xml'
TEI_STRUCTURE_2_FILE_NAME = TMP_DIRECTORY + '/tei_with_structure_ids2.xml'
MWE_CSV_1_FILE_NAME = TMP_DIRECTORY + '/mwes1.csv' MWE_CSV_1_FILE_NAME = TMP_DIRECTORY + '/mwes1.csv'
MWE_CSV_2_FILE_NAME = TMP_DIRECTORY + '/mwes2.csv' MWE_CSV_2_FILE_NAME = TMP_DIRECTORY + '/mwes2.csv'
STRUCTURE_NEW_FILE_NAME = TMP_DIRECTORY + '/structures_new.xml' STRUCTURE_NEW_FILE_NAME = TMP_DIRECTORY + '/structures_new.xml'
TEI_STRUCTURE_FILE_NAME = TMP_DIRECTORY + '/tei_with_structure_ids.xml'
NLP_CONFIG_MAP = { NLP_CONFIG_MAP = {
'treebank': 'sl_ssj_jos', 'treebank': 'sl_ssj_jos',
@ -52,12 +52,12 @@ def run_pipeline(input_file_name, output_file_name):
run_classla(OBELIKS_TWEAKED_FILE_NAME, CLASSLA_FILE_NAME) run_classla(OBELIKS_TWEAKED_FILE_NAME, CLASSLA_FILE_NAME)
convert_to_tei(CLASSLA_FILE_NAME, TEI_INIT_FILE_NAME) convert_to_tei(CLASSLA_FILE_NAME, TEI_INIT_FILE_NAME)
run_mwe_extraction(STRUCTURE_CURRENT_FILE_NAME, TEI_INIT_FILE_NAME, MWE_CSV_1_FILE_NAME) run_mwe_extraction(STRUCTURE_CURRENT_FILE_NAME, TEI_INIT_FILE_NAME, MWE_CSV_1_FILE_NAME)
run_structure_assignment(STRING_LIST_FILE_NAME, TEI_INIT_FILE_NAME, MWE_CSV_1_FILE_NAME, TEI_STRUCTURE_FILE_NAME) run_structure_assignment(STRING_LIST_FILE_NAME, TEI_INIT_FILE_NAME, MWE_CSV_1_FILE_NAME, TEI_STRUCTURE_1_FILE_NAME)
run_structure_creation(STRUCTURE_CURRENT_FILE_NAME, TEI_STRUCTURE_FILE_NAME, STRUCTURE_NEW_FILE_NAME) run_structure_creation(STRUCTURE_CURRENT_FILE_NAME, TEI_STRUCTURE_1_FILE_NAME, STRUCTURE_NEW_FILE_NAME)
run_mwe_extraction(STRUCTURE_NEW_FILE_NAME, TEI_INIT_FILE_NAME, MWE_CSV_2_FILE_NAME) run_mwe_extraction(STRUCTURE_NEW_FILE_NAME, TEI_INIT_FILE_NAME, MWE_CSV_2_FILE_NAME)
run_structure_assignment(STRING_LIST_FILE_NAME, TEI_INIT_FILE_NAME, MWE_CSV_2_FILE_NAME, TEI_STRUCTURE_FILE_NAME) run_structure_assignment(STRING_LIST_FILE_NAME, TEI_INIT_FILE_NAME, MWE_CSV_2_FILE_NAME, TEI_STRUCTURE_2_FILE_NAME)
# run_format_lexicon() # TODO: implement # run_format_lexicon() # TODO: implement
shutil.copyfile(TEI_STRUCTURE_FILE_NAME, output_file_name) shutil.copyfile(TEI_STRUCTURE_2_FILE_NAME, output_file_name)
def run_obeliks4J(obeliks_file_name, classla_file_name): def run_obeliks4J(obeliks_file_name, classla_file_name):
print('Running obeliks ...') print('Running obeliks ...')