foreign spelling in uppercase

[eadmaster]

is it possible to get foreign spellings in uppercase?

import cutlet
katsu = cutlet.Cutlet()
katsu.use_foreign_spelling_uppercase = True
katsu.romaji("カツカレーは美味しい")
# => 'CUTLET CURRY wa oishii'

[polm]

It's not currently possible to do that. I don't think it would make sense as a hard-coded feature, but it would be easy to add information to output tokens about whether they use foreign spelling or not, for example.

Out of curiosity, what situation would you use this in?

[eadmaster]

Out of curiosity, what situation would you use this in?

with this lrctransliterate script, to better differentiate between native and foreign words.

[polm]

OK, I added a feature for this in #39, you can try that out and see if it fulfills your needs before I merge it.

[eadmaster]

ok, thanks, can you post me an example usage?

[polm]

import cutlet

katsu = cutlet.Cutlet()

text = "カツカレーを食べた"

tokens = katsu.romaji_tokens(katsu.tagger(text))

def upper_foreign(tok):
    if tok.foreign:
        return tok.surface.upper()
    else:
        return tok.surface

res = [upper_foreign(tok) for tok in tokens]
print(*res)

[eadmaster]

thanks, it works fine.

is it also possible to detect words that are in foreign spelling already in the input text?
e.g.

text = "First kissから始まる" 
# => 'FIRST KISS kara hajimaru'

[polm]

You can use char_type in the MeCab token features to tell if a token is alphabetic (5).

import cutlet
katsu = cutlet.Cutlet()

for tok in katsu.tagger("First kissから始まる"):
  print(tok.char_type, tok.surface)

[eadmaster]

is there a way to obtain the original untransliterated token from the same loop?

	tokens = trans_obj.romaji_tokens(trans_obj.tagger(line_text), capitalize=False)
	for tok in tokens:	
		if tok.foreign:  # --> or tok.original_surface.isascii() 
			s += tok.surface.upper()
		else:
			s += tok.surface

EDIT: nevermind, i've found a way using 2 separate loops over the tokens.

[eadmaster]

i did some extra testing with my lyrics collection and found a line that is throwing an exception:

Traceback (most recent call last):
  File "/home/andy/bin/lrctransliterate.py", line 109, in <module>
    line_text = jap_trans(trans_obj, line_text)
  File "/home/andy/bin/lrctransliterate.py", line 48, in jap_trans
    tokens = trans_obj.romaji_tokens(trans_obj.tagger(line_text), capitalize=False)
  File "/home/andy/.local/lib/python3.10/site-packages/cutlet/cutlet.py", line 224, in romaji_tokens
    roma = self.romaji_word(word)
  File "/home/andy/.local/lib/python3.10/site-packages/cutlet/cutlet.py", line 325, in romaji_word
    return self.map_kana(kana)
  File "/home/andy/.local/lib/python3.10/site-packages/cutlet/cutlet.py", line 370, in map_kana
    out += self.get_single_mapping(pk, char, nk)
  File "/home/andy/.local/lib/python3.10/site-packages/cutlet/cutlet.py", line 423, in get_single_mapping
    return self.table[kk]
KeyError: 'ﾞ'

the line was
(What!)誰かの声かﾞ

(the updated script is here)

[polm]

You need to clean the text by calling cutlet.normalize_text(text) before passing it to the tagger to avoid edge cases like this. That's mentioned in the docstring, I omitted it from the example for brevity.