Spaces:
Sleeping
Sleeping
Update src/streamlit_app.py
Browse files- src/streamlit_app.py +556 -38
src/streamlit_app.py
CHANGED
@@ -1,40 +1,558 @@
|
|
1 |
-
import altair as alt
|
2 |
-
import numpy as np
|
3 |
-
import pandas as pd
|
4 |
import streamlit as st
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5 |
|
6 |
-
"""
|
7 |
-
# Welcome to Streamlit!
|
8 |
-
|
9 |
-
Edit `/streamlit_app.py` to customize this app to your heart's desire :heart:.
|
10 |
-
If you have any questions, checkout our [documentation](https://docs.streamlit.io) and [community
|
11 |
-
forums](https://discuss.streamlit.io).
|
12 |
-
|
13 |
-
In the meantime, below is an example of what you can do with just a few lines of code:
|
14 |
-
"""
|
15 |
-
|
16 |
-
num_points = st.slider("Number of points in spiral", 1, 10000, 1100)
|
17 |
-
num_turns = st.slider("Number of turns in spiral", 1, 300, 31)
|
18 |
-
|
19 |
-
indices = np.linspace(0, 1, num_points)
|
20 |
-
theta = 2 * np.pi * num_turns * indices
|
21 |
-
radius = indices
|
22 |
-
|
23 |
-
x = radius * np.cos(theta)
|
24 |
-
y = radius * np.sin(theta)
|
25 |
-
|
26 |
-
df = pd.DataFrame({
|
27 |
-
"x": x,
|
28 |
-
"y": y,
|
29 |
-
"idx": indices,
|
30 |
-
"rand": np.random.randn(num_points),
|
31 |
-
})
|
32 |
-
|
33 |
-
st.altair_chart(alt.Chart(df, height=700, width=700)
|
34 |
-
.mark_point(filled=True)
|
35 |
-
.encode(
|
36 |
-
x=alt.X("x", axis=None),
|
37 |
-
y=alt.Y("y", axis=None),
|
38 |
-
color=alt.Color("idx", legend=None, scale=alt.Scale()),
|
39 |
-
size=alt.Size("rand", legend=None, scale=alt.Scale(range=[1, 150])),
|
40 |
-
))
|
|
|
|
|
|
|
|
|
1 |
import streamlit as st
|
2 |
+
import torch
|
3 |
+
import os
|
4 |
+
from dotenv import load_dotenv
|
5 |
+
from together import Together
|
6 |
+
from transformers import AutoTokenizer, AutoModelForSequenceClassification, BertTokenizer,DistilBertTokenizer, BertForSequenceClassification, DistilBertForSequenceClassification
|
7 |
+
from datetime import datetime, timedelta
|
8 |
+
import pandas as pd
|
9 |
+
from task_css import get_custom_css # Import the custom CSS function
|
10 |
+
import gdown
|
11 |
+
|
12 |
+
# Set environment variable for offline mode
|
13 |
+
os.environ["TRANSFORMERS_OFFLINE"] = "1"
|
14 |
+
|
15 |
+
# Load environment variables
|
16 |
+
load_dotenv()
|
17 |
+
|
18 |
+
# Together AI Client with API key from environment variable
|
19 |
+
client = Together(api_key=os.getenv("TOGETHER_API_KEY", ""))
|
20 |
+
|
21 |
+
# Set device
|
22 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
23 |
+
|
24 |
+
# Load Intent Model
|
25 |
+
intent_model_path = "intent_classifier.pth"
|
26 |
+
# Extract file ID from Google Drive URL
|
27 |
+
file_id = "1_GDGvV3MVvBguIsjMyDLg3RxUV_gnFAY"
|
28 |
+
num_intent_labels = 151 # Moved this up before model creation
|
29 |
+
|
30 |
+
# Load Emotion Model
|
31 |
+
emotions_model_path = "./saved_model"
|
32 |
+
emotions_folder_id = "1gYWkbC_XBw_GZjsfwXvubHFil4BCq_gH"
|
33 |
+
|
34 |
+
# Add new pretrained model ID
|
35 |
+
pretrained_folder_id = "13t_EB2LFhRIwb3dkKDtA0O5NXXZBoG-j"
|
36 |
+
|
37 |
+
# Initialize Session State
|
38 |
+
if "is_ready" not in st.session_state:
|
39 |
+
st.session_state.is_ready = False
|
40 |
+
st.session_state.models = {} # Initialize models dict immediately
|
41 |
+
st.session_state.tasks = []
|
42 |
+
st.session_state.task_counter = 0
|
43 |
+
st.session_state.overall_emotion = None
|
44 |
+
st.session_state.overall_emotion_label = "Neutral"
|
45 |
+
|
46 |
+
# Page Configuration first
|
47 |
+
st.set_page_config(
|
48 |
+
page_title="π AI Productivity Assistant",
|
49 |
+
layout="wide",
|
50 |
+
page_icon="π―"
|
51 |
+
)
|
52 |
+
|
53 |
+
# Custom CSS for enhanced styling
|
54 |
+
st.markdown(get_custom_css(), unsafe_allow_html=True)
|
55 |
+
|
56 |
+
# Show loading screen if models aren't ready
|
57 |
+
if not st.session_state.is_ready:
|
58 |
+
st.markdown(
|
59 |
+
"""
|
60 |
+
<div class="loading-container" style="text-align: center; padding: 50px;">
|
61 |
+
<div class="loading-spinner"></div>
|
62 |
+
<h2>Setting up your AI assistant...</h2>
|
63 |
+
<p>This may take a minute. We're downloading the required models.</p>
|
64 |
+
</div>
|
65 |
+
""",
|
66 |
+
unsafe_allow_html=True
|
67 |
+
)
|
68 |
+
|
69 |
+
# Load models here
|
70 |
+
try:
|
71 |
+
# First download pretrained models
|
72 |
+
if not os.path.exists("pretrained_models"):
|
73 |
+
with st.status("Downloading base models...", expanded=True) as status:
|
74 |
+
os.makedirs("pretrained_models", exist_ok=True)
|
75 |
+
gdown.download_folder(
|
76 |
+
f"https://drive.google.com/drive/folders/{pretrained_folder_id}",
|
77 |
+
output="pretrained_models",
|
78 |
+
quiet=False
|
79 |
+
)
|
80 |
+
status.update(label="Base models downloaded!", state="complete")
|
81 |
+
|
82 |
+
# Intent Model Loading
|
83 |
+
if not os.path.exists(intent_model_path):
|
84 |
+
with st.status("Downloading intent model...", expanded=True) as status:
|
85 |
+
output = gdown.download(
|
86 |
+
f"https://drive.google.com/uc?id={file_id}",
|
87 |
+
intent_model_path,
|
88 |
+
quiet=False
|
89 |
+
)
|
90 |
+
status.update(label="Intent model downloaded!", state="complete")
|
91 |
+
|
92 |
+
# Emotion Model Loading
|
93 |
+
if not os.path.exists(emotions_model_path):
|
94 |
+
with st.status("Downloading emotion model...", expanded=True) as status:
|
95 |
+
os.makedirs(emotions_model_path, exist_ok=True)
|
96 |
+
gdown.download_folder(
|
97 |
+
f"https://drive.google.com/drive/folders/{emotions_folder_id}",
|
98 |
+
output=emotions_model_path,
|
99 |
+
quiet=False
|
100 |
+
)
|
101 |
+
status.update(label="Emotion model downloaded!", state="complete")
|
102 |
+
|
103 |
+
# Load and store intent model
|
104 |
+
intent_model = AutoModelForSequenceClassification.from_pretrained(
|
105 |
+
"pretrained_models/bert-base-uncased",
|
106 |
+
num_labels=num_intent_labels,
|
107 |
+
ignore_mismatched_sizes=True, # Add this parameter
|
108 |
+
local_files_only=True
|
109 |
+
)
|
110 |
+
intent_model.load_state_dict(
|
111 |
+
torch.load(intent_model_path, map_location=device, weights_only=True)
|
112 |
+
)
|
113 |
+
st.session_state.models["intent_model"] = intent_model.to(device).eval()
|
114 |
+
st.session_state.models["intent_tokenizer"] = AutoTokenizer.from_pretrained(
|
115 |
+
"pretrained_models/bert-base-uncased",
|
116 |
+
local_files_only=True
|
117 |
+
)
|
118 |
+
|
119 |
+
# Load and store emotion model
|
120 |
+
emotions_model = AutoModelForSequenceClassification.from_pretrained(
|
121 |
+
emotions_model_path,
|
122 |
+
ignore_mismatched_sizes=True, # Add this parameter
|
123 |
+
local_files_only=True
|
124 |
+
)
|
125 |
+
st.session_state.models["emotions_model"] = emotions_model.to(device).eval()
|
126 |
+
st.session_state.models["emotions_tokenizer"] = AutoTokenizer.from_pretrained(
|
127 |
+
emotions_model_path,
|
128 |
+
local_files_only=True
|
129 |
+
)
|
130 |
+
|
131 |
+
# Set ready state
|
132 |
+
st.session_state.is_ready = True
|
133 |
+
st.rerun()
|
134 |
+
|
135 |
+
except Exception as e:
|
136 |
+
st.error(f"Error loading models: {str(e)}")
|
137 |
+
st.stop()
|
138 |
+
|
139 |
+
# Only show main app if models are ready
|
140 |
+
if st.session_state.is_ready:
|
141 |
+
# Title with custom styling
|
142 |
+
st.markdown('<div class="main-header">π― MoodifyTask: AI Task Prioritization & Wellness Assistant</div>', unsafe_allow_html=True)
|
143 |
+
|
144 |
+
# Emotion Labels
|
145 |
+
emotion_label_names = [
|
146 |
+
"admiration", "amusement", "anger", "annoyance", "approval",
|
147 |
+
"caring", "confusion", "curiosity", "desire", "disappointment",
|
148 |
+
"disapproval", "disgust", "embarrassment", "excitement", "fear",
|
149 |
+
"gratitude", "grief", "joy", "love", "nervousness",
|
150 |
+
"optimism", "pride", "realization", "relief", "remorse",
|
151 |
+
"sadness", "surprise", "neutral"
|
152 |
+
]
|
153 |
+
|
154 |
+
# Emotion Categories
|
155 |
+
positive_emotions = ["admiration", "amusement", "approval", "caring", "curiosity", "excitement", "gratitude", "joy", "love", "optimism", "pride", "relief", "surprise"]
|
156 |
+
negative_emotions = ["anger", "annoyance", "disappointment", "disapproval", "disgust", "embarrassment", "fear", "grief", "nervousness", "remorse", "sadness"]
|
157 |
+
neutral_emotions = ["realization", "neutral"]
|
158 |
+
|
159 |
+
# Predict Intent
|
160 |
+
def predict_intent(sentence):
|
161 |
+
inputs = st.session_state.models["intent_tokenizer"](
|
162 |
+
sentence, return_tensors="pt", padding="max_length", truncation=True, max_length=128
|
163 |
+
)
|
164 |
+
inputs = {key: val.to(device) for key, val in inputs.items()}
|
165 |
+
with torch.no_grad():
|
166 |
+
outputs = st.session_state.models["intent_model"](**inputs)
|
167 |
+
predicted_class = torch.argmax(outputs.logits, dim=1).cpu().numpy()[0]
|
168 |
+
|
169 |
+
# Mapping Intent IDs to Priorities (0-150)
|
170 |
+
PRIORITY_MAPPING = {
|
171 |
+
5: [8, 35, 42, 74, 97, 110, 118, 120, 124, 136], # freeze_account, report_lost_card, flight_status, report_fraud, credit_limit, lost_luggage, dispute_charge, overdraft, cancel_reservation, emergency
|
172 |
+
4: [14, 15, 19, 20, 39, 47, 48, 49, 50, 69, 70, 71, 72], # bill_balance, bill_due, exchange_rate, credit_score, interest_rate, insurance, medical_expenses, appointment_schedule, meeting_schedule, dentist_appointment, doctor_appointment, prescription_refill, pharmacy_hours
|
173 |
+
3: [33, 34, 41, 51, 56, 57, 62, 66, 77, 78, 85], # hotel_reservation, car_rental, restaurant_reservation, tracking_package, check_in, check_out, traffic_update, directions, smart_home_on, smart_home_off, weather_forecast
|
174 |
+
2: [0, 1, 3, 6, 9, 13, 16, 17, 21, 25, 27, 28, 36, 40, 45, 52, 61], # restaurant_reviews, shopping_list, what_song, schedule_meeting, translate, play_music, book_hotel, book_flight, gas_prices, exchange_rate, movie_showtimes, recipe, cancel_flight, book_reservation, order_food, car_services, joke
|
175 |
+
1: [2, 4, 5, 7, 10, 11, 12, 18, 22, 23, 24, 26, 30, 31, 32, 37, 38, 43, 44, 46, 53, 54, 55, 58, 59, 60, 63, 64, 65, 67, 68, 73]
|
176 |
+
# tell_joke, fun_fact, trivia, horoscope, dog_fact, cat_fact, define_word, stock_price, sports_update, lottery_results, currency_conversion, holiday_list, language_learning, random_fact, poem, quote, daily_horoscope, joke_request, music_recommendation, podcast_recommendation, celebrity_gossip, movie_recommendation, TV_show_recommendation, book_recommendation, game_recommendation, radio_recommendation, trivia_game, riddle, name_meaning, birthday_reminder, anniversary_reminder, affirmations
|
177 |
+
}
|
178 |
+
|
179 |
+
# Find the priority based on predicted_class
|
180 |
+
predicted_intent_score = next((priority for priority, ids in PRIORITY_MAPPING.items() if predicted_class in ids), 1) # Default to 1 if not found
|
181 |
+
|
182 |
+
return predicted_intent_score
|
183 |
+
|
184 |
+
# Emotion to Numeric Score Mapping
|
185 |
+
EMOTION_MAPPING = {
|
186 |
+
"admiration": 4, "amusement": 3, "anger": 5, "annoyance": 4, "approval": 3,
|
187 |
+
"caring": 4, "confusion": 3, "curiosity": 3, "desire": 4, "disappointment": 4,
|
188 |
+
"disapproval": 4, "disgust": 5, "embarrassment": 4, "excitement": 5, "fear": 5,
|
189 |
+
"gratitude": 3, "grief": 5, "joy": 5, "love": 5, "nervousness": 4,
|
190 |
+
"optimism": 4, "pride": 4, "realization": 3, "relief": 3, "remorse": 4,
|
191 |
+
"sadness": 5, "surprise": 3, "neutral": 3
|
192 |
+
}
|
193 |
+
|
194 |
+
# Function to get numeric emotion score
|
195 |
+
def get_emotion_score(emotion):
|
196 |
+
return EMOTION_MAPPING.get(emotion.lower(), 3) # Default to 3 if not found
|
197 |
+
# Predict Emotion
|
198 |
+
def predict_emotion(sentence):
|
199 |
+
if not sentence.strip():
|
200 |
+
return 3, "neutral"
|
201 |
+
# Ensure the input is a full sentence
|
202 |
+
if len(sentence.split()) == 1:
|
203 |
+
sentence = f"I feel {sentence}"
|
204 |
+
inputs = st.session_state.models["emotions_tokenizer"](
|
205 |
+
sentence, return_tensors="pt", padding="max_length", truncation=True, max_length=128
|
206 |
+
)
|
207 |
+
inputs = {key: val.to(device) for key, val in inputs.items() if key != "token_type_ids"}
|
208 |
+
|
209 |
+
with torch.no_grad():
|
210 |
+
outputs = st.session_state.models["emotions_model"](**inputs)
|
211 |
+
predicted_class = torch.argmax(outputs.logits, dim=1).cpu().numpy()[0]
|
212 |
+
|
213 |
+
detected_emotion = emotion_label_names[predicted_class]
|
214 |
+
|
215 |
+
# Manually adjust for stress/pressure-related words
|
216 |
+
stress_keywords = ["stress", "stressed", "overwhelmed", "pressure", "tense", "burnout"]
|
217 |
+
if any(word in sentence.lower() for word in stress_keywords):
|
218 |
+
if detected_emotion not in ["sadness", "nervousness"]:
|
219 |
+
detected_emotion = "nervousness" # Change to "sadness" if you prefer
|
220 |
+
|
221 |
+
emotion_score = get_emotion_score(detected_emotion)
|
222 |
+
if emotion_score is None:
|
223 |
+
emotion_score = 3 # Default neutral score
|
224 |
+
|
225 |
+
return emotion_score, detected_emotion
|
226 |
+
|
227 |
+
|
228 |
+
# Get Emotion Category
|
229 |
+
def get_emotion_category(emotion):
|
230 |
+
if emotion in positive_emotions:
|
231 |
+
return "positive"
|
232 |
+
elif emotion in negative_emotions:
|
233 |
+
return "negative"
|
234 |
+
else:
|
235 |
+
return "neutral"
|
236 |
+
|
237 |
+
|
238 |
+
def normalize_priority(priority, min_value=0, max_value=10):
|
239 |
+
return (priority - min_value) / (max_value - min_value) # Normalize between 0-1
|
240 |
+
|
241 |
+
# Calculate Task Priority
|
242 |
+
def calculate_priority_score(predicted_intent_score,emotion_score, emotion, time_remaining, complexity, emotion_category):
|
243 |
+
"""
|
244 |
+
Calculate an adaptive priority score for tasks based on intent, emotion, time urgency, and complexity.
|
245 |
+
"""
|
246 |
+
emotion_score = emotion_score if emotion_score is not None else 3
|
247 |
+
# Normalize time urgency (scale 0 to 1 based on 7 days)
|
248 |
+
time_score = max(0, min(1, 1 - (time_remaining.total_seconds() / (7 * 24 * 3600))))
|
249 |
+
|
250 |
+
# Set emotion-based adjustments
|
251 |
+
stress_emotions = ["nervousness", "sadness", "fear"]
|
252 |
+
frustration_emotions = ["anger", "frustration","disappointment","annoyance"]
|
253 |
+
anxiety_emotions = ["anxiety", "uncertainty"]
|
254 |
+
|
255 |
+
|
256 |
+
if emotion_category == "negative":
|
257 |
+
if emotion in stress_emotions:
|
258 |
+
# Prioritize **easy, quick** tasks to reduce cognitive load
|
259 |
+
priority = (predicted_intent_score * 0.15) + (emotion_score * 0.1) + (time_score * 0.3) + ((10 - complexity) * 0.45)
|
260 |
+
|
261 |
+
elif emotion in frustration_emotions:
|
262 |
+
# Prioritize **engaging** tasks (not too easy) but keep urgency in mind
|
263 |
+
priority = (predicted_intent_score * 0.2) + (emotion_score * 0.15) + (time_score * 0.25) + (complexity * 0.4)
|
264 |
+
|
265 |
+
elif emotion in anxiety_emotions:
|
266 |
+
# Prioritize **urgent, low-complexity** tasks
|
267 |
+
priority = (predicted_intent_score * 0.2) + (emotion_score * 0.1) + (time_score * 0.4) + ((10 - complexity) * 0.3)
|
268 |
+
|
269 |
+
else:
|
270 |
+
# Default for negative emotions: balance urgency and ease
|
271 |
+
priority = (predicted_intent_score * 0.2) + (emotion_score * 0.1) + (time_score * 0.3) + ((10 - complexity) * 0.4)
|
272 |
+
|
273 |
+
elif emotion_category == "positive":
|
274 |
+
# If the user is in a **good mood**, favor challenging, high-impact tasks
|
275 |
+
priority = (predicted_intent_score * 0.35) + (emotion_score * 0.2) + (time_score * 0.25) + (complexity * 0.2)
|
276 |
+
|
277 |
+
else: # Neutral emotion
|
278 |
+
# Keep a balance between difficulty and urgency
|
279 |
+
priority = (predicted_intent_score * 0.3) + (emotion_score * 0.2) + (time_score * 0.2) + (complexity * 0.3)
|
280 |
+
|
281 |
+
return normalize_priority(priority) # Ensure no negative priority values
|
282 |
+
|
283 |
+
|
284 |
+
|
285 |
+
|
286 |
+
# AI-Generated Plan Based on Start Time
|
287 |
+
from datetime import datetime
|
288 |
+
|
289 |
+
def get_llama_suggestion(emotion, tasks, selected_datetime):
|
290 |
+
"""Generate AI plan based on full datetime instead of just time"""
|
291 |
+
# Sort tasks by priority (higher priority first)
|
292 |
+
sorted_tasks = sorted(tasks, key=lambda x: x["priority_score"], reverse=True)
|
293 |
+
|
294 |
+
# Filter tasks based on selected datetime
|
295 |
+
filtered_tasks = [
|
296 |
+
task for task in sorted_tasks
|
297 |
+
if task["due_date_time"] >= selected_datetime
|
298 |
+
]
|
299 |
+
|
300 |
+
if not filtered_tasks:
|
301 |
+
well_being_prompts = {
|
302 |
+
"nervousness": "Suggest mindfulness exercises and short relaxation techniques.",
|
303 |
+
"sadness": "Suggest comforting activities like journaling or light exercise.",
|
304 |
+
"anger": "Suggest ways to channel frustration productively.",
|
305 |
+
"joy": "Suggest ways to maintain productivity while feeling good.",
|
306 |
+
"neutral": "Suggest general relaxation activities like listening to music."
|
307 |
+
}
|
308 |
+
well_being_prompt = f"""
|
309 |
+
The user is feeling {emotion}.
|
310 |
+
They have no tasks scheduled after {selected_datetime.strftime('%B %d, %I:%M %p')}.
|
311 |
+
{well_being_prompts.get(emotion, 'Provide general well-being tips.')}
|
312 |
+
"""
|
313 |
+
try:
|
314 |
+
response = client.chat.completions.create(
|
315 |
+
messages=[{"role": "user", "content": well_being_prompt}],
|
316 |
+
model="meta-llama/Llama-3.3-70B-Instruct-Turbo",
|
317 |
+
temperature=0.7,
|
318 |
+
)
|
319 |
+
return response.choices[0].message.content
|
320 |
+
except Exception as e:
|
321 |
+
return f"Error generating well-being tips: {e}"
|
322 |
+
|
323 |
+
# Prepare the prompt with more detailed datetime information
|
324 |
+
task_details = "\n".join([
|
325 |
+
f"- {task['description']} (Priority: {task['priority_score']:.2f}, Complexity: {task['complexity']}, Due: {task['due_date_time'].strftime('%B %d, %I:%M %p')})"
|
326 |
+
for task in filtered_tasks
|
327 |
+
])
|
328 |
+
|
329 |
+
prompt = f"""
|
330 |
+
The user is feeling {emotion}.
|
331 |
+
They need a structured productivity plan starting from {selected_datetime.strftime('%B %d, %I:%M %p')}, not the current time.
|
332 |
+
|
333 |
+
Their prioritized tasks (due on or after the selected time), sorted by priority score:
|
334 |
+
{task_details}
|
335 |
+
|
336 |
+
Please provide:
|
337 |
+
1. A detailed schedule with specific times for each task
|
338 |
+
2. Strategic breaks based on task complexity and emotional state
|
339 |
+
3. Wellness activities that complement their current emotion
|
340 |
+
4. Tips for managing tasks effectively given their emotional state
|
341 |
+
5. Suggestions for handling high-priority tasks first while maintaining well-being
|
342 |
+
"""
|
343 |
+
|
344 |
+
try:
|
345 |
+
response = client.chat.completions.create(
|
346 |
+
messages=[{"role": "user", "content": prompt}],
|
347 |
+
model="meta-llama/Llama-3.3-70B-Instruct-Turbo",
|
348 |
+
temperature=0.7,
|
349 |
+
)
|
350 |
+
return response.choices[0].message.content
|
351 |
+
except Exception as e:
|
352 |
+
return f"Error generating AI plan: {e}"
|
353 |
+
|
354 |
+
|
355 |
+
# Layout with improved spacing
|
356 |
+
col1, col2 = st.columns([1, 1], gap="medium")
|
357 |
+
|
358 |
+
with col1:
|
359 |
+
# st.markdown('<div class="emotion-analysis">', unsafe_allow_html=True)
|
360 |
+
st.markdown('<h3>π Mood Analysis</h3>', unsafe_allow_html=True)
|
361 |
+
emotion_sentence = st.text_area(
|
362 |
+
"Describe how you're feeling today:",
|
363 |
+
value="",
|
364 |
+
height=150,
|
365 |
+
help="Your emotional state helps us prioritize tasks more effectively"
|
366 |
+
)
|
367 |
+
|
368 |
+
if emotion_sentence:
|
369 |
+
emotion_score, emotion_label = predict_emotion(emotion_sentence)
|
370 |
+
st.session_state.overall_emotion = emotion_score
|
371 |
+
st.session_state.overall_emotion_label = emotion_label
|
372 |
+
|
373 |
+
st.markdown(f'<div class="emotion-badge">Detected Emotion: {emotion_label}</div>', unsafe_allow_html=True)
|
374 |
+
|
375 |
+
# Emotion-based task reprioritization
|
376 |
+
for task in st.session_state.tasks:
|
377 |
+
task["priority_score"] = calculate_priority_score(
|
378 |
+
task["predicted_intent_score"],
|
379 |
+
emotion_score,
|
380 |
+
emotion_label,
|
381 |
+
task["time_remaining"],
|
382 |
+
task["complexity"],
|
383 |
+
get_emotion_category(emotion_label)
|
384 |
+
)
|
385 |
+
st.markdown('</div>', unsafe_allow_html=True)
|
386 |
+
|
387 |
+
with col2:
|
388 |
+
# st.markdown('<div class="task-input">', unsafe_allow_html=True)
|
389 |
+
st.markdown('<h3>π
Add New Task</h3>', unsafe_allow_html=True)
|
390 |
+
with st.form("task_form", clear_on_submit=True):
|
391 |
+
task_description = st.text_input("Task Description", help="Be specific about what needs to be done")
|
392 |
+
col_date, col_time = st.columns(2)
|
393 |
+
|
394 |
+
with col_date:
|
395 |
+
due_date = st.date_input("Due Date")
|
396 |
+
|
397 |
+
with col_time:
|
398 |
+
due_time = st.time_input("Due Time")
|
399 |
+
|
400 |
+
complexity = st.slider(
|
401 |
+
"Task Complexity (1-10)",
|
402 |
+
1, 10, 5,
|
403 |
+
help="Higher complexity may affect task priority"
|
404 |
+
)
|
405 |
+
|
406 |
+
submitted = st.form_submit_button("β Add Task")
|
407 |
+
|
408 |
+
if submitted and task_description and due_date and due_time:
|
409 |
+
due_date_time = datetime.combine(due_date, due_time)
|
410 |
+
time_remaining = due_date_time - datetime.now()
|
411 |
+
predicted_intent_score = predict_intent(task_description)
|
412 |
+
|
413 |
+
task = {
|
414 |
+
"id": st.session_state.task_counter, # Add unique ID
|
415 |
+
"description": task_description,
|
416 |
+
"due_date_time": due_date_time,
|
417 |
+
"time_remaining": time_remaining,
|
418 |
+
"complexity": complexity,
|
419 |
+
"predicted_intent_score": predicted_intent_score,
|
420 |
+
"predicted_emotion": st.session_state.overall_emotion,
|
421 |
+
"predicted_label_name": st.session_state.overall_emotion_label,
|
422 |
+
"priority_score": calculate_priority_score(
|
423 |
+
predicted_intent_score,
|
424 |
+
st.session_state.overall_emotion,
|
425 |
+
st.session_state.overall_emotion_label,
|
426 |
+
time_remaining,
|
427 |
+
complexity,
|
428 |
+
get_emotion_category(st.session_state.overall_emotion_label)
|
429 |
+
),
|
430 |
+
"completed": False
|
431 |
+
}
|
432 |
+
|
433 |
+
st.session_state.tasks.append(task)
|
434 |
+
st.session_state.task_counter += 1 # Increment counter
|
435 |
+
st.success("β
Task Added Successfully!")
|
436 |
+
st.markdown('</div>', unsafe_allow_html=True)
|
437 |
+
|
438 |
+
# Task List with Improved Visualization
|
439 |
+
if st.session_state.tasks:
|
440 |
+
st.markdown('<h3>π Task Priority List</h3>', unsafe_allow_html=True)
|
441 |
+
|
442 |
+
# Sort tasks by priority
|
443 |
+
sorted_tasks = sorted(st.session_state.tasks, key=lambda x: x["priority_score"], reverse=True)
|
444 |
+
|
445 |
+
# Create task overview cards
|
446 |
+
st.markdown('<div class="task-overview">', unsafe_allow_html=True)
|
447 |
+
col1, col2 = st.columns(2)
|
448 |
+
with col1:
|
449 |
+
st.markdown(f'<div class="metric-card"><div class="metric-value">{len(sorted_tasks)}</div><div class="metric-label">Total Tasks</div></div>', unsafe_allow_html=True)
|
450 |
+
# with col2:
|
451 |
+
# high_priority = len([t for t in sorted_tasks if t["priority_score"] > 0.7])
|
452 |
+
# st.markdown(f'<div class="metric-card"><div class="metric-value">{high_priority}</div><div class="metric-label">High Priority</div></div>', unsafe_allow_html=True)
|
453 |
+
with col2:
|
454 |
+
today = datetime.now()
|
455 |
+
due_today = len([t for t in sorted_tasks if t["due_date_time"].date() == today.date()])
|
456 |
+
st.markdown(f'<div class="metric-card"><div class="metric-value">{due_today}</div><div class="metric-label">Due Today</div></div>', unsafe_allow_html=True)
|
457 |
+
st.markdown('</div>', unsafe_allow_html=True)
|
458 |
+
|
459 |
+
# Display tasks with priority-based styling
|
460 |
+
for idx, task in enumerate(sorted_tasks):
|
461 |
+
priority_class = "high-priority" if task["priority_score"] > 0.7 else "medium-priority"
|
462 |
+
|
463 |
+
# Create a single row for task and buttons
|
464 |
+
task_container = st.container()
|
465 |
+
with task_container:
|
466 |
+
cols = st.columns([0.8, 0.1, 0.1])
|
467 |
+
|
468 |
+
# Task content in first column
|
469 |
+
with cols[0]:
|
470 |
+
st.markdown(f"""
|
471 |
+
<div class="priority-task {priority_class}">
|
472 |
+
<div class="task-content">
|
473 |
+
<div class="task-header">
|
474 |
+
<span class="task-title">{task["description"]}</span>
|
475 |
+
<span class="priority-score">Priority: {task["priority_score"]:.2f}</span>
|
476 |
+
</div>
|
477 |
+
<div class="task-details">
|
478 |
+
<span class="task-stat">Due: {task["due_date_time"].strftime("%d %b, %I:%M %p")}</span>
|
479 |
+
<span class="task-stat">Complexity: {task["complexity"]}</span>
|
480 |
+
</div>
|
481 |
+
</div>
|
482 |
+
</div>
|
483 |
+
""", unsafe_allow_html=True)
|
484 |
+
st.session_state.editing_task_id = None
|
485 |
+
# Edit button
|
486 |
+
with cols[1]:
|
487 |
+
if st.button("βοΈ", key=f"edit_{idx}", help="Edit task"):
|
488 |
+
st.session_state.editing_task_id = idx
|
489 |
+
|
490 |
+
# Delete button
|
491 |
+
with cols[2]:
|
492 |
+
if st.button("ποΈ", key=f"delete_{idx}", help="Delete task"):
|
493 |
+
st.session_state.tasks.pop(idx)
|
494 |
+
st.success("Task deleted!")
|
495 |
+
st.rerun()
|
496 |
+
|
497 |
+
# Show edit form below the task if being edited
|
498 |
+
if st.session_state.editing_task_id == idx:
|
499 |
+
with st.form(key=f"edit_form_{idx}"):
|
500 |
+
col1, col2 = st.columns(2)
|
501 |
+
with col1:
|
502 |
+
new_description = st.text_input("Description", value=task["description"])
|
503 |
+
new_complexity = st.slider("Complexity", 1, 10, value=task["complexity"])
|
504 |
+
with col2:
|
505 |
+
new_due_date = st.date_input("Due Date", value=task["due_date_time"].date())
|
506 |
+
new_due_time = st.time_input("Due Time", value=task["due_date_time"].time())
|
507 |
+
|
508 |
+
col1, col2 = st.columns(2)
|
509 |
+
with col1:
|
510 |
+
if st.form_submit_button("πΎ Save"):
|
511 |
+
# Update task
|
512 |
+
task["description"] = new_description
|
513 |
+
task["due_date_time"] = datetime.combine(new_due_date, new_due_time)
|
514 |
+
task["time_remaining"] = task["due_date_time"] - datetime.now()
|
515 |
+
task["complexity"] = new_complexity
|
516 |
+
|
517 |
+
# Recalculate priority
|
518 |
+
task["priority_score"] = calculate_priority_score(
|
519 |
+
task["predicted_intent_score"],
|
520 |
+
task["predicted_emotion"],
|
521 |
+
task["predicted_label_name"],
|
522 |
+
task["time_remaining"],
|
523 |
+
task["complexity"],
|
524 |
+
get_emotion_category(task["predicted_label_name"])
|
525 |
+
)
|
526 |
+
st.session_state.editing_task_id = None
|
527 |
+
st.success("Task updated!")
|
528 |
+
st.rerun()
|
529 |
+
|
530 |
+
with col2:
|
531 |
+
if st.form_submit_button("β Cancel"):
|
532 |
+
st.session_state.editing_task_id = None
|
533 |
+
st.rerun()
|
534 |
+
|
535 |
+
# AI Plan Section
|
536 |
+
if st.session_state.tasks:
|
537 |
+
st.markdown('<div class="custom-card">', unsafe_allow_html=True)
|
538 |
+
st.markdown('<h3>β° AI Task Planning</h3>', unsafe_allow_html=True)
|
539 |
+
|
540 |
+
col_date, col_time = st.columns(2)
|
541 |
+
|
542 |
+
with col_date:
|
543 |
+
plan_date = st.date_input("Select Plan Date", datetime.now().date())
|
544 |
+
|
545 |
+
with col_time:
|
546 |
+
plan_time = st.time_input("Select Plan Start Time", datetime.now().time())
|
547 |
+
|
548 |
+
selected_datetime = datetime.combine(plan_date, plan_time)
|
549 |
+
|
550 |
+
if st.button("π
Generate AI Plan"):
|
551 |
+
suggestion = get_llama_suggestion(
|
552 |
+
st.session_state.overall_emotion_label,
|
553 |
+
st.session_state.tasks,
|
554 |
+
selected_datetime # Pass full datetime object
|
555 |
+
)
|
556 |
+
st.markdown(f'<div class="info-box">{suggestion}</div>', unsafe_allow_html=True)
|
557 |
+
st.markdown('</div>', unsafe_allow_html=True)
|
558 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|