SoSa123456 commited on
Commit
b77e4f5
1 Parent(s): 8de8d38

Create Utils.py

Browse files
Files changed (1) hide show
  1. Utils.py +21 -0
Utils.py ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ '''
2
+ Utility functions for the Climate Change Radio Script Generator
3
+ '''
4
+ import nltk
5
+ import string
6
+ # Download the necessary NLTK data
7
+ nltk.download('punkt')
8
+ # Function to clean the generated paragraph
9
+ def clean_paragraph(entry):
10
+ paragraphs = entry.split('\n')
11
+ for i in range(len(paragraphs)):
12
+ split_sentences = nltk.tokenize.sent_tokenize(paragraphs[i], language='english')
13
+ if i == len(paragraphs) - 1 and split_sentences[:1][-1] not in string.punctuation:
14
+ paragraphs[i] = " ".join(split_sentences[:-1])
15
+ return capitalize_first_char("\n".join(paragraphs))
16
+ # Function to capitalize the first character of a string
17
+ def capitalize_first_char(entry):
18
+ for i in range(len(entry)):
19
+ if entry[i].isalpha():
20
+ return entry[:i] + entry[i].upper() + entry[i + 1:]
21
+ return entry