2020-11-04 18:02:15 +00:00
|
|
|
# temporary directory
|
|
|
|
|
TMP_DIRECTORY = '../tmp/structure_assignment'
|
|
|
|
|
|
|
|
|
|
# scripts
|
2020-12-02 22:51:28 +00:00
|
|
|
CONLLU_TWEAK_SCRIPT_NAME = 'tweak_conllu.py'
|
2020-11-04 18:02:15 +00:00
|
|
|
CONLLU_TEI_SCRIPT_NAME = 'conllu_to_xml.py'
|
|
|
|
|
MWE_EXTRACTION_SCRIPT_NAME = 'wani.py'
|
|
|
|
|
STRUCTURE_ASSIGNMENT_SCRIPT_NAME = 'assign_structures.py'
|
|
|
|
|
STRUCTURE_CREATION_SCRIPT_NAME = 'create_structures.py'
|
|
|
|
|
STRUCTURE_CREATION_SCRIPT_NAME = 'create_structures.py'
|
|
|
|
|
TEI_DICTIONARY_SCRIPT_NAME = 'tei_to_dictionary.py'
|
|
|
|
|
|
|
|
|
|
# resources
|
|
|
|
|
OBELIKS_JAR_FILE_NAME = '../resources/obeliks.jar'
|
|
|
|
|
TRANSLATION_FILE_NAME = '../resources/dict.xml'
|
|
|
|
|
CLASSLA_MODELS_DIRECTORY = '../resources/classla'
|
|
|
|
|
STRUCTURE_CURRENT_FILE_NAME = '../resources/structures.xml'
|
|
|
|
|
STRUCTURE_SCHEMA_FILE_NAME = '../resources/structures.xsd'
|
|
|
|
|
DICTIONARY_SCHEMA_FILE_NAME = '../resources/monolingual_dictionaries.xsd'
|
|
|
|
|
|
|
|
|
|
# temporary outputs
|
|
|
|
|
STRING_LIST_FILE_NAME = TMP_DIRECTORY + '/strings.txt'
|
|
|
|
|
OBELIKS_RAW_FILE_NAME = TMP_DIRECTORY + '/obeliks_raw.conllu'
|
|
|
|
|
OBELIKS_TWEAKED_FILE_NAME = TMP_DIRECTORY + '/obeliks_tweaked.conllu'
|
|
|
|
|
CLASSLA_FILE_NAME = TMP_DIRECTORY + '/classla.conllu'
|
|
|
|
|
TEI_INIT_FILE_NAME = TMP_DIRECTORY + '/tei_initial.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_2_FILE_NAME = TMP_DIRECTORY + '/mwes2.csv'
|
|
|
|
|
STRUCTURE_NEW_FILE_NAME = TMP_DIRECTORY + '/structures_new.xml'
|
|
|
|
|
DICTIONARY_XML_FILE_NAME = TMP_DIRECTORY + '/dictionary.xml'
|
