structure_assignment/scripts/constants.py

42 lines
2.0 KiB
Python

# temporary directory
TMP_DIRECTORY = '../tmp/structure_assignment'
# scripts
TEI_SPLIT_SCRIPT_NAME = 'split_tei.py'
CONLLU_TWEAK_SCRIPT_NAME = 'tweak_conllu.py'
TRANSLATION_SCRIPT_NAME = 'translate_jos.py'
CONLLU_TEI_SCRIPT_NAME = 'conllu_to_xml.py'
MWE_EXTRACTION_SCRIPT_NAME = 'wani.py'
STRUCTURE_SINGLE_ASSIGNMENT_SCRIPT_NAME = 'assign_single_structures.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'
DICTIONARY_MERGE_SCRIPT_NAME = 'merge_dictionaries.py'
# resources
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_OUTPUT_FILE_NAME = TMP_DIRECTORY + '/classla_raw.conllu'
CLASSLA_TRANSLATED_FILE_NAME = TMP_DIRECTORY + '/classla_translated.conllu'
TEI_INIT_FILE_NAME = TMP_DIRECTORY + '/tei_initial.xml'
TEI_SINGLE_FILE_NAME = TMP_DIRECTORY + '/tei_single.xml'
TEI_SINGLE_STRUCTURE_FILE_NAME = TMP_DIRECTORY + '/tei_single_with_ids.xml'
TEI_MULTIPLE_FILE_NAME = TMP_DIRECTORY + '/tei_multiple.xml'
TEI_MULTIPLE_STRUCTURE_1_FILE_NAME = TMP_DIRECTORY + '/tei_multiple_with_ids1.xml'
TEI_MULTIPLE_STRUCTURE_2_FILE_NAME = TMP_DIRECTORY + '/tei_multiple_with_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_SINGLE_FILE_NAME = TMP_DIRECTORY + '/dictionary_single.xml'
DICTIONARY_MULTIPLE_FILE_NAME = TMP_DIRECTORY + '/dictionary_multiple.xml'
DICTIONARY_FILE_NAME = TMP_DIRECTORY + '/dictionary.xml'