|
|
|
@ -17,21 +17,41 @@ STRUCTURE_SCHEMA_FILE_NAME = '../resources/structures.xsd'
|
|
|
|
|
DICTIONARY_SCHEMA_FILE_NAME = '../resources/monolingual_dictionaries.xsd'
|
|
|
|
|
|
|
|
|
|
# temporary outputs
|
|
|
|
|
STRING_LIST_FILE_NAME = 'strings.txt'
|
|
|
|
|
OBELIKS_RAW_FILE_NAME = 'obeliks_raw.conllu'
|
|
|
|
|
OBELIKS_TWEAKED_FILE_NAME = 'obeliks_tweaked.conllu'
|
|
|
|
|
CLASSLA_OUTPUT_FILE_NAME = 'classla_raw.conllu'
|
|
|
|
|
CLASSLA_TRANSLATED_FILE_NAME = 'classla_translated.conllu'
|
|
|
|
|
TEI_INIT_FILE_NAME = 'tei_initial.xml'
|
|
|
|
|
TEI_SINGLE_FILE_NAME = 'tei_single.xml'
|
|
|
|
|
TEI_SINGLE_STRUCTURE_FILE_NAME = 'tei_single_with_ids.xml'
|
|
|
|
|
TEI_MULTIPLE_FILE_NAME = 'tei_multiple.xml'
|
|
|
|
|
TEI_MULTIPLE_STRUCTURE_1_FILE_NAME = 'tei_multiple_with_ids1.xml'
|
|
|
|
|
TEI_MULTIPLE_STRUCTURE_2_FILE_NAME = 'tei_multiple_with_ids2.xml'
|
|
|
|
|
MWE_CSV_1_FILE_NAME = 'mwes1.csv'
|
|
|
|
|
MWE_CSV_2_FILE_NAME = 'mwes2.csv'
|
|
|
|
|
STRUCTURE_OLD_FILE_NAME = 'structures_old.xml'
|
|
|
|
|
STRUCTURE_NEW_FILE_NAME = 'structures_new.xml'
|
|
|
|
|
DICTIONARY_SINGLE_FILE_NAME = 'dictionary_single.xml'
|
|
|
|
|
DICTIONARY_MULTIPLE_FILE_NAME = 'dictionary_multiple.xml'
|
|
|
|
|
DICTIONARY_FILE_NAME = 'dictionary.xml'
|
|
|
|
|
FILE_NAME_MAP = {'strings-list': 'strings.txt',
|
|
|
|
|
'obeliks-tokenised': 'obeliks_raw.conllu',
|
|
|
|
|
'obeliks-tweaked': 'obeliks_tweaked.conllu',
|
|
|
|
|
'classla-parsed': 'classla_raw.conllu',
|
|
|
|
|
'classla-translated': 'classla_translated.conllu',
|
|
|
|
|
'tei-initial': 'tei_initial.xml',
|
|
|
|
|
'tei-single': 'tei_single.xml',
|
|
|
|
|
'tei-single-ids': 'tei_single_with_ids.xml',
|
|
|
|
|
'tei-multiple': 'tei_multiple.xml',
|
|
|
|
|
'tei-multiple-ids-1': 'tei_multiple_with_ids1.xml',
|
|
|
|
|
'tei-multiple-ids-2': 'tei_multiple_with_ids2.xml',
|
|
|
|
|
'mwes-1': 'mwes1.csv',
|
|
|
|
|
'mwes-2': 'mwes2.csv',
|
|
|
|
|
'structures-old': 'structures_old.xml',
|
|
|
|
|
'structures-new': 'structures_new.xml',
|
|
|
|
|
'dictionary-single': 'dictionary_single.xml',
|
|
|
|
|
'dictionary-multiple': 'dictionary_multiple.xml',
|
|
|
|
|
'dictionary': 'dictionary.xml'
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
# STRING_LIST_FILE_NAME = 'strings.txt'
|
|
|
|
|
# OBELIKS_RAW_FILE_NAME = 'obeliks_raw.conllu'
|
|
|
|
|
# OBELIKS_TWEAKED_FILE_NAME = 'obeliks_tweaked.conllu'
|
|
|
|
|
# CLASSLA_OUTPUT_FILE_NAME = 'classla_raw.conllu'
|
|
|
|
|
# CLASSLA_TRANSLATED_FILE_NAME = 'classla_translated.conllu'
|
|
|
|
|
# TEI_INIT_FILE_NAME = 'tei_initial.xml'
|
|
|
|
|
# TEI_SINGLE_FILE_NAME = 'tei_single.xml'
|
|
|
|
|
# TEI_SINGLE_STRUCTURE_FILE_NAME = 'tei_single_with_ids.xml'
|
|
|
|
|
# TEI_MULTIPLE_FILE_NAME = 'tei_multiple.xml'
|
|
|
|
|
# TEI_MULTIPLE_STRUCTURE_1_FILE_NAME = 'tei_multiple_with_ids1.xml'
|
|
|
|
|
# TEI_MULTIPLE_STRUCTURE_2_FILE_NAME = 'tei_multiple_with_ids2.xml'
|
|
|
|
|
# MWE_CSV_1_FILE_NAME = 'mwes1.csv'
|
|
|
|
|
# MWE_CSV_2_FILE_NAME = 'mwes2.csv'
|
|
|
|
|
# STRUCTURE_OLD_FILE_NAME = 'structures_old.xml'
|
|
|
|
|
# STRUCTURE_NEW_FILE_NAME = 'structures_new.xml'
|
|
|
|
|
# DICTIONARY_SINGLE_FILE_NAME = 'dictionary_single.xml'
|
|
|
|
|
# DICTIONARY_MULTIPLE_FILE_NAME = 'dictionary_multiple.xml'
|
|
|
|
|
# DICTIONARY_FILE_NAME = 'dictionary.xml'
|
|
|
|
|