Update app.py

app.py

@@ -29,9 +29,17 @@ def remove_page_numbers(text):
 def clean_sentence_start(sentence):
     """
     Removes numbering like "1.", "2)", "3.4" from the start of a sentence.
+    Also removes standalone footnote numbers (e.g. "1 The...")
     """
-    # Regex: Start of line, optional space, digits/letters, dot/paren, space
-    return re.sub(r'^\s*(?:\d+|[a-zA-Z])[\.\)]\s+', '', sentence)
+    # Regex 1: Start of line, optional space, digits/letters, dot/paren, space
+    # Matches "1.", "2)", "A.", "b)"
+    sentence = re.sub(r'^\s*(?:\d+|[a-zA-Z])[\.\)]\s+', '', sentence)
+    
+    # Regex 2: Standalone numbers (1-3 digits) followed by space and Capital letter
+    # Matches "1 The", "12 It", but avoids years like "1999 The" (4 digits)
+    sentence = re.sub(r'^\s*\d{1,3}\s+(?=[A-Z])', '', sentence)
+    
+    return sentence
 
 def clean_text_block(raw_text):
     """