update on topic_for_story #1

Agent_Chat_Classification.py

@@ -1,5 +1,5 @@
 from Agent_Chat_Classification_Helper import call_classification_llm
-from Classification_parameters import TRACKED_FIELDS
+from Classification_parameters import TRACKED_FIELDS, ALLOWED_STORY_TOPICS
 from supabase_ie import load_user_info, save_user_info, update_country
 from db_5_process_session import _load_history
 
@@ -28,6 +28,9 @@ def analyze_message(user_id: str, query: str):
     relevant_missing = analysis.get("relevant_missing_fields", [])
     topic = analysis.get("topic", "personal")
     response_mode = analysis.get("response_mode", "dialogic")
+    topic_for_story = analysis.get("topic_for_story", "none")
+    if topic_for_story not in ALLOWED_STORY_TOPICS:
+        topic_for_story = "none"
     
     print(f"[DEBUG][CLASSIFICATION_RAW] analysis={analysis}")
     # --- Update user_info in Supabase if new info found ---
@@ -56,6 +59,7 @@ def analyze_message(user_id: str, query: str):
         "response_mode": response_mode,
         "user_info": user_info,
         "relevant_missing": relevant_missing,
+        "topic_for_story": topic_for_story, 
         "chunks": [],  # retrieval results can be appended later
         "needs_news_fetch": analysis.get("needs_news_fetch", False),
         "news_topic": analysis.get("news_topic", ""),

Agent_Chat_Classification_Helper.py

@@ -4,7 +4,7 @@ from config import OPENAI_CHAT_MODEL
 from openai import OpenAI
 import json
 from supabase_ie import load_user_info
-from Classification_parameters import TOPIC_DESCRIPTIONS, RESPONSE_MODE_DESCRIPTIONS, TRACKED_FIELDS, FIELD_DESCRIPTIONS
+from Classification_parameters import TOPIC_DESCRIPTIONS, RESPONSE_MODE_DESCRIPTIONS, TRACKED_FIELDS, FIELD_DESCRIPTIONS, ALLOWED_STORY_TOPICS
 client = OpenAI()
 
 
@@ -35,6 +35,7 @@ def call_classification_llm(query: str, recent_history: list, user_info: dict, m
         "topic": "",
         "response_mode": "",
         "emotion": "",
+        "topic_for_story": "",
         "needs_news_fetch": False,
         "news_topic": [],
         "news_question": ""
@@ -63,6 +64,12 @@ def call_classification_llm(query: str, recent_history: list, user_info: dict, m
       * topic: one of {list(TOPIC_DESCRIPTIONS.keys())}
       * response_mode: one of {list(RESPONSE_MODE_DESCRIPTIONS.keys())}
       * emotion: one of ["curious","happy","sad","angry","neutral"]
+    
+    - topic_for_story:
+      * Choose exactly one label from this list (closed set):
+        [{allowed_str}]
+      * Return "none" if no fitting label is clearly supported.
+      * Never invent new labels.
 
     - needs_news_fetch:
       * true if the user is clearly asking for news, current events, or external information lookup.
@@ -106,71 +113,10 @@ def call_classification_llm(query: str, recent_history: list, user_info: dict, m
             "topic": "",
             "response_mode": "",
             "emotion": "",
+            "topic_for_story": "none",
             "needs_news_fetch": False,
             "news_topic": [],
             "news_question": ""
         }
 
 
-# def summarize_user_info(user_id: str) -> str:
-#     """Fetch user profile from Supabase, create a natural language summary, and return it as a string."""
-
-#     data = load_user_info(user_id)
-#     profile = data.get("user_profile", {})
-
-#     parts = []
-
-#     # Location
-#     if profile.get("living_country"):
-#         parts.append(f"lives in {profile['living_country']}")
-#     if profile.get("origin_country"):
-#         parts.append(f"was born in {profile['origin_country']}")
-
-#     # Kids
-#     kids = profile.get("kids")
-#     if kids:
-#         try:
-#             if isinstance(kids, str):  # sometimes JSONB comes back as string
-#                 kids = json.loads(kids)
-#             kids_list = [f"{k.get('name','?')} ({k.get('age','?')} yrs)" for k in kids if isinstance(k, dict)]
-#             if kids_list:
-#                 parts.append("has kids: " + ", ".join(kids_list))
-#         except Exception as e:
-#             print("⚠️ Could not parse kids:", kids, e)
-
-#     # Arrays: hobbies, interests, languages, sports
-#     for field, label in [
-#         ("hobbies", "enjoys"),
-#         ("interests", "is interested in"),
-#         ("sports_played", "plays"),
-#         ("sports_watched", "watches"),
-#         ("language_spoken", "speaks")
-#     ]:
-#         values = profile.get(field)
-#         if values:
-#             if isinstance(values, str):
-#                 values = [values]
-#             if isinstance(values, list) and values:
-#                 parts.append(f"{label} " + ", ".join(values))
-
-#     # Misc single fields
-#     if profile.get("profession"):
-#         parts.append(f"works as {profile['profession']}")
-#     if profile.get("favorite_team"):
-#         parts.append(f"supports {profile['favorite_team']}")
-#     if profile.get("favorite_food"):
-#         parts.append(f"likes {profile['favorite_food']}")
-#     if profile.get("favorite_book"):
-#         parts.append(f"favorite book is {profile['favorite_book']}")
-#     if profile.get("favorite_movie"):
-#         parts.append(f"favorite movie is {profile['favorite_movie']}")
-#     if profile.get("pets"):
-#         parts.append(f"has pets: {profile['pets']}")
-#     if profile.get("education"):
-#         parts.append(f"studied {profile['education']}")
-
-#     # Final summary
-#     summary = "The user " + "; ".join(parts) + "." if parts else "No profile information available."
-#     print(f"✅ User profile summary generated:\n{summary}")
-
-#     return summary

Classification_parameters.py

@@ -1,6 +1,11 @@
 # ==============================
 # Descriptions for Classification
 # ==============================
+ALLOWED_STORY_TOPICS = [
+    "fortune","confidence","mental_strength","bullying","fear_phobias",
+    "friendship","love","sex","meaning_of_life","justice_fairness","none"
+]
+
 TRACKED_FIELDS = [
     "name", "living_country", "origin_country","living_place",
     "origin_place","number_of_brothers_and_sisters","kids",

Prompt_building.py

@@ -180,6 +180,7 @@ def build_socratic_prompt(
     style_examples_path: str = "style_examples.json",
     relevant_missing: list = None,
     topic: str = None,
+    topic_for_story: str = None,
     response_mode: str = None,
     emotion: dict = None,
 ) -> str:

Retrieve.py

@@ -3,14 +3,15 @@
 import os
 import json
 import numpy as np
-from typing import List, Dict
+from typing import List, Dict, Optional, Any
+import requests
 from langchain_community.vectorstores import FAISS
 import streamlit as st
 import numpy as np
 #from db_paths import (PERSONAL_INFO_CHUNKS_PATH,CHAT_HISTORY_CHUNKS_PATH)
 import json
 from supabase_ie import load_user_info, load_history_for_display, download_faiss_from_supabase
-
+from config import SUPABASE_URL, SUPABASE_HEADERS
 
 #used in embed_query
 def normalize(v):
@@ -193,3 +194,41 @@ def compute_metadata_boost(metadata: Dict, query: str, filter_keywords: List[str
                         boost += weight_filter
     return boost
 
+PERSONAL_BUCKET = {"personal", "advice"}
+
+def get_story_from_supabase(
+    user_id: str,
+    username: str,
+    conversation_type: str,
+    topic_for_story: str | None,
+) -> dict | None:
+    """
+    If conversation_type is personal/advice and topic_for_story is set,
+    call a Supabase RPC that: 
+      - selects a story with your rules (unseen first; else seen<=1 and >90d ago),
+      - logs usage,
+      - returns a compact JSON payload for the prompt builder.
+
+    Returns None if no suitable story.
+    """
+    if conversation_type not in PERSONAL_BUCKET:
+        return None
+    if not topic_for_story or topic_for_story == "none":
+        return None
+
+    fn = "pick_and_log_story_with_history_rpc"
+    payload = {
+        "p_user_id": user_id,
+        "p_topic": topic_for_story,
+    }
+
+    url = f"{SUPABASE_URL}/rest/v1/rpc/{fn}"
+    r = requests.post(url, headers=SUPABASE_HEADERS, json=payload, timeout=20)
+
+    if r.status_code == 404 or not r.text or r.text == "null":
+        return None
+    r.raise_for_status()
+    story = r.json()
+    return story
+
+    #return r.json()

app_nn.py

@@ -28,7 +28,7 @@ from Agent_Emotional_Evaluation import analyze_emotions_from_history
 from Agent_Chat_Classification import analyze_message
 #from Agent_Chat_Classification_Helper import load_user_info
 
-from Retrieve import retrieve_all_chunks
+from Retrieve import retrieve_all_chunks, get_story_from_supabase
 from Prompt_building import TOPIC_TO_DBS
 from ask_llm_final_prompt import ask_socrates
 
@@ -207,12 +207,20 @@ def run_chat_app(user_id: str, username:str, profile: Dict[str, Any],ui_lang:str
         topic_to_dbs=TOPIC_TO_DBS,
     )
 
+    socratic_story = get_story_from_supabase(
+        user_id=user_id, 
+        username=username, 
+        topic_for_story=analysis.get("topic_for_story"), 
+        conversation_type=analysis.get("topic"),
+    )
+
     # ---- Compose final answer ----
     final_reply_en = ask_socrates(
         user_input=user_msg_en,
         retrieved_chunks=chunks,
         relevant_missing=analysis.get("relevant_missing"),
         topic=analysis.get("topic"),
+        topic_for_story=analysis.get("topic_for_story"),
         response_mode=analysis.get("response_mode"),
         emotion=emotion_result,
         user_id=user_id,

ask_llm_final_prompt.py

@@ -19,6 +19,7 @@ def ask_socrates(
     model=OPENAI_CHAT_MODEL,
     relevant_missing=None,
     topic=None,
+    topic_for_story=None,
     response_mode=None,
     emotion=None     
 ):
@@ -29,6 +30,7 @@ def ask_socrates(
         user_id=user_id,   
         relevant_missing=relevant_missing,
         topic=topic,
+        topic_for_story=topic_for_story,
         response_mode=response_mode,
         emotion=emotion 
     )

config.py

@@ -28,6 +28,7 @@ HF_EMBEDDING_MODEL       = _get("HF_EMBEDDING_MODEL", required=False, default="s
 GNEWS_KEY                = _get("GNEWS_KEY", required=False, default="")
 
 
+
 # ── Models ─────────────────────────────────────────────────────────────────────
 OPENAI_CHAT_MODEL       = os.getenv("OPENAI_CHAT_MODEL", "gpt-4.1-mini")
 OPENAI_CLASSIFIER_MODEL = os.getenv("OPENAI_CLASSIFIER_MODEL", "gpt-4.1-mini")
@@ -77,3 +78,10 @@ def get_username() -> str | None:
         st.session_state.get("username")
         or username_from_email(st.session_state.get("email"))
     )
+
+# Reusable headers for PostgREST / RPC calls
+SUPABASE_HEADERS = {
+    "apikey": SUPABASE_SERVICE_KEY,
+    "Authorization": f"Bearer {SUPABASE_SERVICE_KEY}",
+    "Content-Type": "application/json",
+}