Spaces:
Sleeping
Sleeping
| import streamlit as st | |
| from google import genai | |
| from google.genai import types | |
| from transformers import pipeline | |
| from huggingface_hub import login | |
| from streamlit_elements import elements, nivo, mui | |
| import requests | |
| import os | |
| from collections import Counter | |
| hf_token = os.getenv("HF_TOKEN") | |
| genai_token = os.getenv("G_TOKEN") | |
| maps_token = os.getenv("MAP_TOKEN") | |
| login(token=hf_token) | |
| client = genai.Client(api_key=genai_token) | |
| logo_path = "https://huggingface.co/spaces/Elite13/mental-health/resolve/main/logo.png" | |
| st.logo(logo_path, size="large") | |
| def find_psychiatrists(location, token): | |
| endpoint = "https://maps.googleapis.com/maps/api/place/textsearch/json" | |
| params = { | |
| "query": f"psychiatrist near {location}", | |
| "key": token | |
| } | |
| try: | |
| res = requests.get(endpoint, params=params) | |
| res.raise_for_status() | |
| data = res.json() | |
| return data.get("results", [])[:5] | |
| except Exception as e: | |
| st.error(f"Failed to fetch psychiatrists: {e}") | |
| return [] | |
| label_map = { | |
| "LABEL_0": 'anxiety', | |
| "LABEL_1": 'depression', | |
| "LABEL_2": 'bipolar', | |
| "LABEL_3": 'normal', | |
| "LABEL_4": 'personality disorder', | |
| "LABEL_5": 'stress', | |
| "LABEL_6": 'suicidal' | |
| } | |
| model_name = "Elite13/bert-finetuned-mental-health" | |
| classifier = pipeline("text-classification", model=model_name, tokenizer=model_name) | |
| if "messages" not in st.session_state: | |
| st.session_state.messages = [] | |
| if "total_messages" not in st.session_state: | |
| st.session_state.total_messages = 0 | |
| if "emotion_log" not in st.session_state: | |
| st.session_state.emotion_log = [] | |
| if "accepted" not in st.session_state: | |
| st.session_state.accepted = False | |
| if "analysis_done" not in st.session_state: | |
| st.session_state.analysis_done = False | |
| if "psychiatrist_results" not in st.session_state: | |
| st.session_state.psychiatrist_results = [] | |
| if "last_location" not in st.session_state: | |
| st.session_state.last_location = "" | |
| st.markdown("<h1 style='text-align: center;'>Mental Health Chatbot</h1>", unsafe_allow_html=True) | |
| st.markdown("---") | |
| if not st.session_state.accepted: | |
| st.markdown("<h2 style ='text-align:center;'>Terms and Conditions</h2>", unsafe_allow_html=True) | |
| st.markdown(""" | |
| **This is a University Project.** | |
| - The application is part of academic coursework or research. | |
| - Data entered here may be used for project evaluation or improvement. | |
| - No personal data will be stored or shared beyond academic purposes. | |
| """) | |
| if st.button("✅ I Agree"): | |
| st.session_state.accepted = True | |
| st.rerun() | |
| if st.session_state.accepted: | |
| # Chat History | |
| for msg in st.session_state.messages: | |
| with st.chat_message(msg["role"]): | |
| st.markdown(msg["content"]) | |
| # Chat Input | |
| if prompt := st.chat_input("Say something..."): | |
| st.session_state.messages.append({"role": "user", "content": prompt}) | |
| with st.chat_message("user"): | |
| st.markdown(prompt) | |
| st.session_state.total_messages += 1 | |
| chat_texts = [ | |
| types.Content(role=msg["role"], parts=[types.Part(text=msg["content"])]) | |
| for msg in st.session_state.messages | |
| ] | |
| # Gemini Response | |
| response = client.models.generate_content( | |
| model="gemini-2.0-flash", | |
| config=types.GenerateContentConfig(system_instruction=( | |
| "You are a kind and supportive friend. Speak casually like a real human, Give your answer in less than 2 sentences. " | |
| "offering empathy, emotional comfort, and helpful suggestions in a non-judgmental tone." | |
| )), | |
| contents=chat_texts | |
| ) | |
| reply = response.text | |
| st.session_state.messages.append({"role": "assistant", "content": reply}) | |
| with st.chat_message("assistant"): | |
| st.markdown(reply) | |
| with st.spinner("Clarifying and Classifying Emotion..."): | |
| clarified_response = client.models.generate_content( | |
| model="gemini-2.0-flash", | |
| config=types.GenerateContentConfig( | |
| system_instruction="Paraphrase this user message in clear, emotionally direct language. " | |
| "Make it easier for a classifier to detect the emotional state, without changing the meaning." | |
| ), | |
| contents=prompt | |
| ) | |
| clarified_text = clarified_response.text.strip() | |
| results = classifier(clarified_text, return_all_scores=True) | |
| top_pred = max(results[0], key=lambda x: x['score']) | |
| emotion = label_map[top_pred['label']] | |
| confidence = top_pred['score'] | |
| st.session_state.emotion_log.append(emotion) | |
| st.markdown(f"🧠 **Detected Emotion:** `{emotion}` ({confidence:.2f} confidence)") | |
| st.session_state.analysis_done = False | |
| st.session_state.psychiatrist_results = [] | |
| # View Analysis Button | |
| if st.session_state.total_messages > 0 and len(st.session_state.emotion_log) > 0 and not st.session_state.analysis_done: | |
| st.markdown("---") | |
| if st.button("📊 View Analysis "): | |
| emotion_counts = Counter(st.session_state.emotion_log) | |
| total = st.session_state.total_messages | |
| percentages = {e: (c / total) * 100 for e, c in emotion_counts.items()} | |
| st.markdown("### Emotion Breakdown") | |
| for e, p in percentages.items(): | |
| st.markdown(f"- **{e.capitalize()}**: {p:.2f}%") | |
| # Pie Chart | |
| nivo_data = [{"id": e, "label": e.capitalize(), "value": c} for e, c in emotion_counts.items()] | |
| with elements("nivo_pie"): | |
| with mui.Box(sx={"height": 500}): | |
| nivo.Pie( | |
| data=nivo_data, | |
| margin={"top": 40, "right": 80, "bottom": 80, "left": 80}, | |
| innerRadius=0.5, padAngle=0.7, cornerRadius=3, | |
| activeOuterRadiusOffset=8, borderWidth=1, | |
| borderColor={"from": "color", "modifiers": [["darker", 0.2]]}, | |
| arcLinkLabelsSkipAngle=10, arcLabelsSkipAngle=10, | |
| arcLabelsTextColor={"from": "color", "modifiers": [["darker", 2]]}, | |
| legends=[{ | |
| "anchor": "bottom", "direction": "row", "translateY": 56, | |
| "itemWidth": 100, "itemHeight": 18, "symbolSize": 18, | |
| "symbolShape": "circle", "itemTextColor": "#999", | |
| "effects": [{"on": "hover", "style": {"itemTextColor": "#000"}}] | |
| }], | |
| theme={"background": "#fff", "textColor": "#333", | |
| "tooltip": {"container": {"background": "#fff", "color": "#333"}}} | |
| ) | |
| # Summary Report | |
| st.markdown("---") | |
| st.markdown("### Summary Report") | |
| summary = client.models.generate_content( | |
| model="gemini-2.0-flash", | |
| config=types.GenerateContentConfig(system_instruction=( | |
| "You are a mental health chatbot summarizing emotional state based on the user's chats. " | |
| "Include key emotions, percentages, and patterns in a warm, encouraging tone." | |
| )), | |
| contents=st.session_state.emotion_log | |
| ) | |
| st.markdown(summary.text) | |
| st.session_state.analysis_done = True | |
| # Psychiatrist Finder | |
| if st.session_state.analysis_done: | |
| st.markdown("---") | |
| st.markdown("### 🩺 Find Psychiatrists Near You") | |
| with st.form("find_psych_form"): | |
| location = st.text_input("Enter your City or Area:") | |
| submitted = st.form_submit_button("🔍 Find Psychiatrists") | |
| if submitted and location: | |
| with st.spinner("Searching nearby psychiatrists..."): | |
| results = find_psychiatrists(location, maps_token) | |
| st.session_state.psychiatrist_results = results | |
| st.session_state.last_location = location | |
| if st.session_state.psychiatrist_results: | |
| st.markdown(f"### 🏥 Psychiatrists near {st.session_state.last_location.title()}:") | |
| for place in st.session_state.psychiatrist_results: | |
| name = place.get("name") | |
| address = place.get("formatted_address") | |
| rating = place.get("rating", "N/A") | |
| st.markdown(f"- **{name}**, 📍 {address}, ⭐ {rating} rating") | |