mhamza-007's picture
Upload 56 files
d840583 verified
Raw
History Blame Contribute Delete
1.23 kB
"""Text cleaning / normalization for Arabic sentiment analysis."""
from preprocessing.arabic_normalizer import Arabic_Light_Stemmer, normalizeArabic
from preprocessing.emoji_handler import (
emoji_counter,
extract_emoji,
remove_emoji,
replace_emojis_with_text,
replace_emoticon_with_emojis,
space_between_emojis,
)
from preprocessing.pipeline import (
CLEANING_STEPS,
EMOJI_STEPS,
clean_text,
preprocess_dataframe,
)
from preprocessing.text_cleaning import (
Removing_non_arabic,
Removing_numbers,
Removing_punctuations,
Removing_urls,
remove_extra_Space,
remove_hashtags_and_mentions,
remove_small_sentences,
remove_stop_words,
)
__all__ = [
"Arabic_Light_Stemmer",
"CLEANING_STEPS",
"EMOJI_STEPS",
"Removing_non_arabic",
"Removing_numbers",
"Removing_punctuations",
"Removing_urls",
"clean_text",
"emoji_counter",
"extract_emoji",
"normalizeArabic",
"preprocess_dataframe",
"remove_emoji",
"remove_extra_Space",
"remove_hashtags_and_mentions",
"remove_small_sentences",
"remove_stop_words",
"replace_emojis_with_text",
"replace_emoticon_with_emojis",
"space_between_emojis",
]