Update app.py

app.py

@@ -28,11 +28,13 @@ huggingface_token = os.environ.get("HUGGINGFACE_TOKEN")
 
 # Download necessary NLTK data
 nltk.download('punkt')
+nltk.download('averaged_perceptron_tagger')
 
 class Agent1:
     def __init__(self):
         self.question_words = set(["what", "when", "where", "who", "whom", "which", "whose", "why", "how"])
         self.conjunctions = set(["and", "or"])
+        self.pronouns = set(["it", "its", "they", "their", "them", "he", "his", "him", "she", "her", "hers"])
 
     def is_question(self, text: str) -> bool:
         words = word_tokenize(text.lower())
@@ -40,6 +42,30 @@ class Agent1:
                 text.strip().endswith('?') or 
                 any(word in self.question_words for word in words))
 
+    def replace_pronoun(self, questions: List[str]) -> List[str]:
+        if len(questions) < 2:
+            return questions
+
+        # Simple NLP to identify potential nouns in the first question
+        tokens = nltk.pos_tag(word_tokenize(questions[0]))
+        nouns = [word for word, pos in tokens if pos.startswith('NN')]
+
+        if not nouns:
+            return questions
+
+        # Use the last noun as the antecedent
+        antecedent = nouns[-1]
+
+        # Replace pronouns in subsequent questions
+        for i in range(1, len(questions)):
+            words = word_tokenize(questions[i])
+            for j, word in enumerate(words):
+                if word.lower() in self.pronouns:
+                    words[j] = antecedent
+            questions[i] = ' '.join(words)
+
+        return questions
+
     def rephrase_and_split(self, user_input: str) -> List[str]:
         words = word_tokenize(user_input)
         questions = []
@@ -61,6 +87,9 @@ class Agent1:
         if not questions:
             return [user_input]
 
+        # Handle pronoun replacement
+        questions = self.replace_pronoun(questions)
+
         return questions
 
     def process(self, user_input: str) -> tuple[List[str], Dict[str, List[Dict[str, str]]]]: