def script_find_where_character_occurs():
"""
>>> script_find_where_character_occurs()
:return:
"""
d = defaultdict(list)
annotated_proper_nouns = utils.get_annotated_proper_nouns_txt(
constants.LEMMATISED_DLH_CLEAN_PERSON_NAMES)
dlh_books = get_parsed_dlh_books()
for i, book in enumerate(dlh_books):
for j, para in enumerate(book):
for k, sentence in enumerate(para):
for lemma in annotated_proper_nouns:
for form in annotated_proper_nouns[lemma]:
if form in sentence:
d[form].append(f"{i + 1}-{j + 1}-{k + 1}")
filename = os.path.join(PACKDIR, Work.DLH.get_main_directory(),
"dlh_clean_lemma_occurrences.txt")
with codecs.open(filename, "w", encoding="utf-8") as f:
l = []
for item in d:
line = f"{item}: ({len(d[item])}) " + " ".join(d[item])
l.append(line)
f.write("\n".join(l))
def get_lemmatised_dlh_place_names():
"""
>>> get_lemmatised_dlh_place_names()['Ligeris']
['Ligeris', 'Ligerem']
"""
file_path = os.path.join(PACKDIR, Work.DLH.get_main_directory(),
constants.LEMMATISED_DLH_CLEAN_PLACES)
return utils.get_annotated_proper_nouns_txt(file_path)
def get_lemmatised_dlh_person_names():
"""
>>> get_lemmatised_dlh_person_names()['Chlodovechus']
['Chlodovechus', 'Chlodovechi', 'Chlodovechum', 'Chlodovecho']
"""
file_path = os.path.join(PACKDIR, Work.DLH.get_main_directory(),
constants.LEMMATISED_DLH_CLEAN_PERSON_NAMES)
return utils.get_annotated_proper_nouns_txt(file_path)
def get_lemmatised_dlh_proper_nouns():
"""
>>> get_lemmatised_dlh_proper_nouns()['Burgundia']
['Burgundias', 'Burgundiam', 'Burgundiae', 'Burgundia', 'Burgundio']
"""
file_path = os.path.join(PACKDIR, Work.DLH.get_main_directory(),
constants.LEMMATISED_DLH_PROPER_NOUNS)
return utils.get_annotated_proper_nouns_txt(file_path)
def get_volsunga_annotated_names():
"""
>>> annotated_names = get_volsunga_annotated_names()
>>> annotated_names["Sigmundr"]
['Sigmundr', 'Sigmund', 'Sigmundi', 'Sigmundar']
:return:
"""
volsunga_names_proper_nouns = os.path.join(
PACKDIR, Work.VOL.get_main_directory(),
constants.VOLSUNGA_CLEAN_NAMES_PROPER_NOUNS)
return utils.get_annotated_proper_nouns_txt(volsunga_names_proper_nouns)