welcomecenterbot/nlp/stopwords_detector.py

28 lines
888 B
Python
Raw Normal View History

2024-09-29 06:47:49 +00:00
stopwords = []
2024-09-29 07:05:14 +00:00
with open('nlp/stop_words.txt', 'r', encoding='utf-8') as file:
2024-09-29 07:13:08 +00:00
stopwords = file.readlines()
2024-09-29 06:47:49 +00:00
# Convert stopwords to a set for faster lookup
stopword_set = set(stopwords)
2024-09-29 07:08:25 +00:00
def check_stopwords(text):
2024-09-29 06:47:49 +00:00
"""
Check if any words from the stopwords list are present in the given text.
Args:
text (str): The input text to check.
stopwords (list): A list of stopwords.
Returns:
bool: True if any stopword is found in the text, False otherwise.
"""
# Normalize the text by converting it to lower case and splitting into words
2024-09-29 06:55:58 +00:00
words = text.replace(' ', '').lower().split()
2024-09-29 06:47:49 +00:00
# Iterate through each word and check for stopwords
for word in words:
if word in stopword_set:
return True # Stop iteration and return True if a stopword is found
return False # Return False if no stopwords are found