Spaces:
Sleeping
Sleeping
NLTK path modification
Browse files- components/config.py +1 -1
- components/utils.py +1 -1
- static/ntlk/corpora/stopwords.zip +0 -0
- static/{nltk β ntlk}/corpora/stopwords/README +0 -0
- static/{nltk β ntlk}/corpora/stopwords/arabic +0 -0
- static/{nltk β ntlk}/corpora/stopwords/azerbaijani +0 -0
- static/{nltk β ntlk}/corpora/stopwords/basque +0 -0
- static/{nltk β ntlk}/corpora/stopwords/bengali +0 -0
- static/{nltk β ntlk}/corpora/stopwords/catalan +0 -0
- static/{nltk β ntlk}/corpora/stopwords/chinese +0 -0
- static/{nltk β ntlk}/corpora/stopwords/danish +0 -0
- static/{nltk β ntlk}/corpora/stopwords/dutch +0 -0
- static/{nltk β ntlk}/corpora/stopwords/english +0 -0
- static/{nltk β ntlk}/corpora/stopwords/finnish +0 -0
- static/{nltk β ntlk}/corpora/stopwords/french +0 -0
- static/{nltk β ntlk}/corpora/stopwords/german +0 -0
- static/{nltk β ntlk}/corpora/stopwords/greek +0 -0
- static/{nltk β ntlk}/corpora/stopwords/hebrew +0 -0
- static/{nltk β ntlk}/corpora/stopwords/hinglish +0 -0
- static/{nltk β ntlk}/corpora/stopwords/hungarian +0 -0
- static/{nltk β ntlk}/corpora/stopwords/indonesian +0 -0
- static/{nltk β ntlk}/corpora/stopwords/italian +0 -0
- static/{nltk β ntlk}/corpora/stopwords/kazakh +0 -0
- static/{nltk β ntlk}/corpora/stopwords/nepali +0 -0
- static/{nltk β ntlk}/corpora/stopwords/norwegian +0 -0
- static/{nltk β ntlk}/corpora/stopwords/portuguese +0 -0
- static/{nltk β ntlk}/corpora/stopwords/romanian +0 -0
- static/{nltk β ntlk}/corpora/stopwords/russian +0 -0
- static/{nltk β ntlk}/corpora/stopwords/slovene +0 -0
- static/{nltk β ntlk}/corpora/stopwords/spanish +0 -0
- static/{nltk β ntlk}/corpora/stopwords/swedish +0 -0
- static/{nltk β ntlk}/corpora/stopwords/tajik +0 -0
- static/{nltk β ntlk}/corpora/stopwords/turkish +0 -0
components/config.py
CHANGED
@@ -12,7 +12,7 @@ class Settings:
|
|
12 |
0: "NGEN - Non Misogynistic Content",
|
13 |
1: "GEN - Misogynistic Content",
|
14 |
}
|
15 |
-
NLTK_DATA_PATH = "
|
16 |
|
17 |
|
18 |
app_config = Settings()
|
|
|
12 |
0: "NGEN - Non Misogynistic Content",
|
13 |
1: "GEN - Misogynistic Content",
|
14 |
}
|
15 |
+
NLTK_DATA_PATH = "static/ntlk"
|
16 |
|
17 |
|
18 |
app_config = Settings()
|
components/utils.py
CHANGED
@@ -3,7 +3,7 @@ import nltk
|
|
3 |
import re
|
4 |
from components.config import app_config
|
5 |
|
6 |
-
nltk.
|
7 |
|
8 |
|
9 |
# Cleans one text
|
|
|
3 |
import re
|
4 |
from components.config import app_config
|
5 |
|
6 |
+
nltk.download("stopwords", download_dir=app_config.NLTK_DATA_PATH)
|
7 |
|
8 |
|
9 |
# Cleans one text
|
static/ntlk/corpora/stopwords.zip
ADDED
Binary file (34.3 kB). View file
|
|
static/{nltk β ntlk}/corpora/stopwords/README
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/arabic
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/azerbaijani
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/basque
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/bengali
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/catalan
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/chinese
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/danish
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/dutch
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/english
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/finnish
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/french
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/german
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/greek
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/hebrew
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/hinglish
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/hungarian
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/indonesian
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/italian
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/kazakh
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/nepali
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/norwegian
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/portuguese
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/romanian
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/russian
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/slovene
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/spanish
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/swedish
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/tajik
RENAMED
File without changes
|
static/{nltk β ntlk}/corpora/stopwords/turkish
RENAMED
File without changes
|