remove old tagging system

src/chat.py

@@ -1,7 +1,6 @@
 from huggingface_hub import InferenceClient
 from src.config import BASE_MODEL, MY_MODEL, HF_TOKEN
 import os
-from src.utils.tags import tag_user_input
 from src.utils.profile import load_schema, create_empty_profile, extract_profile_updates, merge_profile, profile_to_summary
 from src.utils.resources import load_resources, filter_resources, score_resources, format_resources_for_context
 
@@ -73,9 +72,6 @@ class Chatbot:
         """
         model_id = MY_MODEL if MY_MODEL else BASE_MODEL # define MY_MODEL in config.py if you create a new model in the HuggingFace Hub
         self.client = InferenceClient(model=model_id, token="HF_TOKEN")
-        # Initialize tag lists
-        self.user_tags = []
-        self.substance_tags = []
         # Initialize user profile
         current_dir = os.path.dirname(os.path.abspath(__file__))
         data_dir = os.path.join(current_dir, '..', 'data')
@@ -91,8 +87,6 @@ class Chatbot:
 
     def reset(self):
         """Reset conversation state for a new session without re-initializing the client or resources."""
-        self.user_tags = []
-        self.substance_tags = []
         self.user_profile = create_empty_profile()
 
     def update_profile(self, user_input):
@@ -109,16 +103,13 @@ class Chatbot:
     def format_prompt(self, user_input, turn_number=0):
         """
         Format the user's input into a list of chat messages with system context.
-        Also tags the input with relevant keywords and substances that appear in the text,
-        and updates the user profile with any new information detected.
+        Updates the user profile with any new information detected from the message.
 
         This method:
-        1. Loads system prompt from system_prompt.txt
-        2. Detects keywords from keywords.txt in user input (case-insensitive, partial matches)
-        3. Detects substances from substances.txt in user input (case-insensitive, partial matches)
-        4. Updates user profile from schema-based keyword matching
-        5. Injects profile summary into the system prompt so the model knows what's been gathered
-        6. Returns a list of message dicts for the chat completion API
+        1. Loads system prompt from system_prompt.md
+        2. Updates user profile from schema-based extraction
+        3. Injects profile summary into the system prompt so the model knows what's been gathered
+        4. Returns a list of message dicts for the chat completion API
 
         Args:
             user_input (str): The user's question
@@ -135,13 +126,6 @@ class Chatbot:
         with open(system_prompt_path, 'r', encoding='utf-8') as f:
             system_prompt = f.read().strip()
 
-        # Tag user input with keywords and substances
-        keywords_path = os.path.join(current_dir, '../data/keywords.txt')
-        substances_path = os.path.join(current_dir, '../data/substances.txt')
-
-        self.user_tags = tag_user_input(keywords_path, user_input)
-        self.substance_tags = tag_user_input(substances_path, user_input)
-
         # Update user profile from this message
         self.update_profile(user_input)
 
@@ -199,12 +183,12 @@ class Chatbot:
             print("[Harbor] Crisis keywords detected — returning crisis response.")
             return CRISIS_RESPONSE
 
-        # 1. Format messages (also updates profile and tags)
+        # 1. Format messages (also updates profile)
         turn_number = len(history) if history else 0
         messages = self.format_prompt(user_input, turn_number=turn_number)
 
         # 1b. After the user's first message, return a fixed follow-up instead of calling the LLM.
-        # Profile and tags have already been updated above so the first message is not lost.
+        # Profile has already been updated above so the first message is not lost.
         if history and len(history) == 1:
             return (
                 "Thank you for sharing that. Before I give you any recommendations, "