mirror of
https://github.com/rspeer/wordfreq.git
synced 2024-12-24 18:01:38 +00:00
fixes to tokenization
This commit is contained in:
parent
bb5df3b074
commit
d376f4e2e2
@ -149,10 +149,15 @@ def turkish_tokenize(text, include_punctuation=False):
|
||||
in Turkish, and modifies 'comma-below' characters to use cedillas.
|
||||
"""
|
||||
text = unicodedata.normalize('NFC', text).replace('İ', 'i').replace('I', 'ı')
|
||||
token_expr = TOKEN_RE_WITH_PUNCTUATION if include_punctuation else TOKEN_RE
|
||||
if include_punctuation:
|
||||
return [
|
||||
smash_numbers(commas_to_cedillas(token.casefold()))
|
||||
for token in TOKEN_RE_WITH_PUNCTUATION.findall(text)
|
||||
]
|
||||
else:
|
||||
return [
|
||||
smash_numbers(commas_to_cedillas(token.strip("'").casefold()))
|
||||
for token in token_expr.findall(text)
|
||||
for token in TOKEN_RE.findall(text)
|
||||
]
|
||||
|
||||
|
||||
@ -161,10 +166,15 @@ def romanian_tokenize(text, include_punctuation=False):
|
||||
Like `simple_tokenize`, but modifies the letters ş and ţ (with cedillas)
|
||||
to use commas-below instead.
|
||||
"""
|
||||
token_expr = TOKEN_RE_WITH_PUNCTUATION if include_punctuation else TOKEN_RE
|
||||
if include_punctuation:
|
||||
return [
|
||||
smash_numbers(cedillas_to_commas(token.casefold()))
|
||||
for token in TOKEN_RE_WITH_PUNCTUATION.findall(text)
|
||||
]
|
||||
else:
|
||||
return [
|
||||
smash_numbers(cedillas_to_commas(token.strip("'").casefold()))
|
||||
for token in token_expr.findall(text)
|
||||
for token in TOKEN_RE.findall(text)
|
||||
]
|
||||
|
||||
|
||||
@ -353,6 +363,9 @@ def tokenize(text, lang, include_punctuation=False, external_wordlist=False):
|
||||
does not support these languages yet. It will split on spaces and
|
||||
punctuation, giving tokens that are far too long.
|
||||
"""
|
||||
# A really simple way to handle language codes with more than just the
|
||||
# language
|
||||
lang = lang.split('-')[0]
|
||||
if lang == 'ja' or lang == 'ko':
|
||||
return tokenize_mecab_language(text, lang, include_punctuation)
|
||||
elif lang == 'zh':
|
||||
|
Loading…
Reference in New Issue
Block a user