def test_old_swedish(self):
sentence = "Far man kunu oc dör han för en hun far barn. oc sigher hun oc hænnæ frændær."
tr = ut.Transcriber(old_swedish.DIPHTHONGS_IPA, old_swedish.DIPHTHONGS_IPA_class, old_swedish.IPA_class,
old_swedish.old_swedish_rules)
transcribed_sentence = tr.main(sentence)
self.assertEqual("[far man kunu ok dør han før ɛn hun far barn ok siɣɛr hun ok hɛnːɛ frɛndɛr]",
transcribed_sentence)
def test_gothic_transcriber(self):
example_sentence = "Anastodeins aiwaggeljons Iesuis Xristaus sunaus gudis."
tr = ut.Transcriber(gothic.DIPHTHONGS_IPA,
gothic.DIPHTHONGS_IPA_class, gothic.IPA_class, gothic.gothic_rules)
transcribed_sentence = tr.main(example_sentence)
target = "[anastoːðiːns ɛwaŋgeːljoːns jeːsuis kristɔs sunɔs guðis]"
self.assertEqual(target, transcribed_sentence)
def test_old_norse_transcriber(self):
example_sentence = "Almáttigr guð skapaði í upphafi himin ok jörð ok alla þá hluti, er þeim fylgja, og " \
"síðast menn tvá, er ættir eru frá komnar, Adam ok Evu, ok fjölgaðist þeira kynslóð ok " \
"dreifðist um heim allan."
tr = ut.Transcriber(ont.DIPHTHONGS_IPA, ont.DIPHTHONGS_IPA_class, ont.IPA_class, ont.old_norse_rules)
transcribed_sentence = tr.main(example_sentence)
target = "[almaːtːiɣr guð skapaði iː upːhavi himin ɔk jœrð ɔk alːa θaː hluti ɛr θɛim fylɣja ɔɣ siːðast mɛnː " \
"tvaː ɛr ɛːtːir ɛru fraː kɔmnar adam ɔk ɛvu ɔk fjœlɣaðist θɛira kynsloːð ɔk drɛivðist um hɛim alːan]"
self.assertEqual(target, transcribed_sentence)
from xml.etree import ElementTree
from xml.etree.ElementTree import XMLParser
from cltk.phonology import utils as phu
from cltk.phonology.old_norse import transcription as ont
from cltk.phonology.syllabify import Syllabifier
from cltk.tokenize.word import WordTokenizer
from cltk.corpus.old_norse.syllabifier import hierarchy, invalid_onsets
from cltk.text_reuse.levenshtein import Levenshtein
from zoegas.constants import postags, dictionary_name, pos_verbose
# phonetic transcriber
phonetic_transcriber = phu.Transcriber(ont.DIPHTHONGS_IPA,
ont.DIPHTHONGS_IPA_class, ont.IPA_class,
ont.old_norse_rules)
# Old Norse syllabifier
s = Syllabifier(language="old_norse", break_geminants=True)
s.set_invalid_onsets(invalid_onsets)
s.set_hierarchy(hierarchy)
old_norse_word_tokenizer = WordTokenizer("old_norse")
def clean(text: str) -> Optional[str]:
"""
:param text:
:return:
