Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -9,12 +9,16 @@ from q_learning_chatbot import QLearningChatbot
|
|
9 |
|
10 |
from gtts import gTTS
|
11 |
from io import BytesIO
|
|
|
|
|
|
|
|
|
|
|
|
|
12 |
|
13 |
mdl = ModelPipeLine()
|
14 |
final_chain = mdl.create_final_chain()
|
15 |
|
16 |
-
st.set_page_config(page_title="PeacePal")
|
17 |
-
|
18 |
# Define states and actions
|
19 |
states = [
|
20 |
"Negative",
|
@@ -24,15 +28,35 @@ states = [
|
|
24 |
"Positive",
|
25 |
]
|
26 |
|
27 |
-
#
|
28 |
-
|
29 |
-
#st.sidebar.image(logo_path, use_column_width=True)
|
30 |
|
31 |
-
#
|
32 |
-
|
33 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
34 |
|
35 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
36 |
|
37 |
## generated stores AI generated responses
|
38 |
if 'generated' not in st.session_state:
|
@@ -41,6 +65,24 @@ if 'generated' not in st.session_state:
|
|
41 |
if 'past' not in st.session_state:
|
42 |
st.session_state['past'] = ['Hi!']
|
43 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
44 |
# Layout of input/response containers
|
45 |
|
46 |
colored_header(label='', description='', color_name='blue-30')
|
@@ -56,28 +98,7 @@ def get_text():
|
|
56 |
def generate_response(prompt):
|
57 |
response = mdl.call_conversational_rag(prompt,final_chain)
|
58 |
return response['answer']
|
59 |
-
|
60 |
-
def text_to_speech(text):
|
61 |
-
# Use gTTS to convert text to speech
|
62 |
-
tts = gTTS(text=text, lang='en')
|
63 |
-
# Save the speech as bytes in memory
|
64 |
-
fp = BytesIO()
|
65 |
-
tts.write_to_fp(fp)
|
66 |
-
return fp
|
67 |
-
|
68 |
-
def speech_recognition_callback():
|
69 |
-
# Ensure that speech output is available
|
70 |
-
if st.session_state.my_stt_output is None:
|
71 |
-
st.session_state.p01_error_message = "Please record your response again."
|
72 |
-
return
|
73 |
|
74 |
-
# Clear any previous error messages
|
75 |
-
st.session_state.p01_error_message = None
|
76 |
-
|
77 |
-
# Store the speech output in the session state
|
78 |
-
st.session_state.speech_input = st.session_state.my_stt_output
|
79 |
-
|
80 |
-
|
81 |
## Applying the user input box
|
82 |
with input_container:
|
83 |
# Add a radio button to choose input mode
|
@@ -101,11 +122,60 @@ with input_container:
|
|
101 |
response = generate_response(query)
|
102 |
st.session_state.past.append(query)
|
103 |
st.session_state.generated.append(response)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
104 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
105 |
# Convert the response to speech
|
106 |
speech_fp = text_to_speech(response)
|
107 |
# Play the speech
|
108 |
st.audio(speech_fp, format='audio/mp3')
|
|
|
109 |
else:
|
110 |
# Add a text input field for query
|
111 |
query = st.text_input("Query: ", key="input")
|
@@ -116,7 +186,59 @@ with input_container:
|
|
116 |
response = generate_response(query)
|
117 |
st.session_state.past.append(query)
|
118 |
st.session_state.generated.append(response)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
119 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
120 |
# Convert the response to speech
|
121 |
speech_fp = text_to_speech(response)
|
122 |
# Play the speech
|
|
|
9 |
|
10 |
from gtts import gTTS
|
11 |
from io import BytesIO
|
12 |
+
st.set_page_config(page_title="PeacePal")
|
13 |
+
#image to the sidebar
|
14 |
+
image_path = os.path.join('images', 'sidebar.jpg')
|
15 |
+
st.sidebar.image(image_path, use_column_width=True)
|
16 |
+
|
17 |
+
st.title('PeacePal 🌱')
|
18 |
|
19 |
mdl = ModelPipeLine()
|
20 |
final_chain = mdl.create_final_chain()
|
21 |
|
|
|
|
|
22 |
# Define states and actions
|
23 |
states = [
|
24 |
"Negative",
|
|
|
28 |
"Positive",
|
29 |
]
|
30 |
|
31 |
+
# Initialize Q-learning chatbot and mental health classifier
|
32 |
+
chatbot = QLearningChatbot(states)
|
|
|
33 |
|
34 |
+
# Function to display Q-table
|
35 |
+
def display_q_table(q_values, states):
|
36 |
+
q_table_dict = {"State": states}
|
37 |
+
q_table_df = pd.DataFrame(q_table_dict)
|
38 |
+
return q_table_df
|
39 |
+
|
40 |
+
def text_to_speech(text):
|
41 |
+
# Use gTTS to convert text to speech
|
42 |
+
tts = gTTS(text=text, lang="en")
|
43 |
+
# Save the speech as bytes in memory
|
44 |
+
fp = BytesIO()
|
45 |
+
tts.write_to_fp(fp)
|
46 |
+
return fp
|
47 |
|
48 |
+
|
49 |
+
def speech_recognition_callback():
|
50 |
+
# Ensure that speech output is available
|
51 |
+
if st.session_state.my_stt_output is None:
|
52 |
+
st.session_state.p01_error_message = "Please record your response again."
|
53 |
+
return
|
54 |
+
|
55 |
+
# Clear any previous error messages
|
56 |
+
st.session_state.p01_error_message = None
|
57 |
+
|
58 |
+
# Store the speech output in the session state
|
59 |
+
st.session_state.speech_input = st.session_state.my_stt_output
|
60 |
|
61 |
## generated stores AI generated responses
|
62 |
if 'generated' not in st.session_state:
|
|
|
65 |
if 'past' not in st.session_state:
|
66 |
st.session_state['past'] = ['Hi!']
|
67 |
|
68 |
+
# Initialize memory
|
69 |
+
if "entered_text" not in st.session_state:
|
70 |
+
st.session_state.entered_text = []
|
71 |
+
if "entered_mood" not in st.session_state:
|
72 |
+
st.session_state.entered_mood = []
|
73 |
+
if "messages" not in st.session_state:
|
74 |
+
st.session_state.messages = []
|
75 |
+
if "user_sentiment" not in st.session_state:
|
76 |
+
st.session_state.user_sentiment = "Neutral"
|
77 |
+
if "mood_trend" not in st.session_state:
|
78 |
+
st.session_state.mood_trend = "Unchanged"
|
79 |
+
if "mood_trend_symbol" not in st.session_state:
|
80 |
+
st.session_state.mood_trend_symbol = ""
|
81 |
+
if "show_question" not in st.session_state:
|
82 |
+
st.session_state.show_question = False
|
83 |
+
if "asked_questions" not in st.session_state:
|
84 |
+
st.session_state.asked_questions = []
|
85 |
+
|
86 |
# Layout of input/response containers
|
87 |
|
88 |
colored_header(label='', description='', color_name='blue-30')
|
|
|
98 |
def generate_response(prompt):
|
99 |
response = mdl.call_conversational_rag(prompt,final_chain)
|
100 |
return response['answer']
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
101 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
102 |
## Applying the user input box
|
103 |
with input_container:
|
104 |
# Add a radio button to choose input mode
|
|
|
122 |
response = generate_response(query)
|
123 |
st.session_state.past.append(query)
|
124 |
st.session_state.generated.append(response)
|
125 |
+
# Detect sentiment
|
126 |
+
user_sentiment = chatbot.detect_sentiment(user_message)
|
127 |
+
|
128 |
+
# Retrieve question
|
129 |
+
if user_sentiment in ["Negative", "Moderately Negative", "Neutral"]:
|
130 |
+
question = retriever.get_response(
|
131 |
+
user_message, predicted_mental_category
|
132 |
+
)
|
133 |
+
st.session_state.asked_questions.append(question)
|
134 |
+
show_question = True
|
135 |
+
else:
|
136 |
+
show_question = False
|
137 |
+
question = ""
|
138 |
|
139 |
+
# Update mood history / mood_trend
|
140 |
+
chatbot.update_mood_history()
|
141 |
+
mood_trend = chatbot.check_mood_trend()
|
142 |
+
|
143 |
+
# Define rewards
|
144 |
+
if user_sentiment in ["Positive", "Moderately Positive"]:
|
145 |
+
if mood_trend == "increased":
|
146 |
+
reward = +1
|
147 |
+
mood_trend_symbol = " ⬆️"
|
148 |
+
elif mood_trend == "unchanged":
|
149 |
+
reward = +0.8
|
150 |
+
mood_trend_symbol = ""
|
151 |
+
else: # decreased
|
152 |
+
reward = -0.2
|
153 |
+
mood_trend_symbol = " ⬇️"
|
154 |
+
else:
|
155 |
+
if mood_trend == "increased":
|
156 |
+
reward = +1
|
157 |
+
mood_trend_symbol = " ⬆️"
|
158 |
+
elif mood_trend == "unchanged":
|
159 |
+
reward = -0.2
|
160 |
+
mood_trend_symbol = ""
|
161 |
+
else: # decreased
|
162 |
+
reward = -1
|
163 |
+
mood_trend_symbol = " ⬇️"
|
164 |
+
|
165 |
+
print(
|
166 |
+
f"mood_trend - sentiment - reward: {mood_trend} - {user_sentiment} - 🛑{reward}🛑"
|
167 |
+
)
|
168 |
+
|
169 |
+
# Update Q-values
|
170 |
+
chatbot.update_q_values(
|
171 |
+
user_sentiment, reward, user_sentiment
|
172 |
+
)
|
173 |
+
|
174 |
# Convert the response to speech
|
175 |
speech_fp = text_to_speech(response)
|
176 |
# Play the speech
|
177 |
st.audio(speech_fp, format='audio/mp3')
|
178 |
+
|
179 |
else:
|
180 |
# Add a text input field for query
|
181 |
query = st.text_input("Query: ", key="input")
|
|
|
186 |
response = generate_response(query)
|
187 |
st.session_state.past.append(query)
|
188 |
st.session_state.generated.append(response)
|
189 |
+
# Detect sentiment
|
190 |
+
user_sentiment = chatbot.detect_sentiment(user_message)
|
191 |
+
|
192 |
+
# Retrieve question
|
193 |
+
if user_sentiment in ["Negative", "Moderately Negative", "Neutral"]:
|
194 |
+
question = retriever.get_response(
|
195 |
+
user_message, predicted_mental_category
|
196 |
+
)
|
197 |
+
st.session_state.asked_questions.append(question)
|
198 |
+
show_question = True
|
199 |
+
else:
|
200 |
+
show_question = False
|
201 |
+
question = ""
|
202 |
+
# Convert the response to speech
|
203 |
+
speech_fp = text_to_speech(response)
|
204 |
+
# Play the speech
|
205 |
+
st.audio(speech_fp, format='audio/mp3')
|
206 |
|
207 |
+
# Update mood history / mood_trend
|
208 |
+
chatbot.update_mood_history()
|
209 |
+
mood_trend = chatbot.check_mood_trend()
|
210 |
+
|
211 |
+
# Define rewards
|
212 |
+
if user_sentiment in ["Positive", "Moderately Positive"]:
|
213 |
+
if mood_trend == "increased":
|
214 |
+
reward = +1
|
215 |
+
mood_trend_symbol = " ⬆️"
|
216 |
+
elif mood_trend == "unchanged":
|
217 |
+
reward = +0.8
|
218 |
+
mood_trend_symbol = ""
|
219 |
+
else: # decreased
|
220 |
+
reward = -0.2
|
221 |
+
mood_trend_symbol = " ⬇️"
|
222 |
+
else:
|
223 |
+
if mood_trend == "increased":
|
224 |
+
reward = +1
|
225 |
+
mood_trend_symbol = " ⬆️"
|
226 |
+
elif mood_trend == "unchanged":
|
227 |
+
reward = -0.2
|
228 |
+
mood_trend_symbol = ""
|
229 |
+
else: # decreased
|
230 |
+
reward = -1
|
231 |
+
mood_trend_symbol = " ⬇️"
|
232 |
+
|
233 |
+
print(
|
234 |
+
f"mood_trend - sentiment - reward: {mood_trend} - {user_sentiment} - 🛑{reward}🛑"
|
235 |
+
)
|
236 |
+
|
237 |
+
# Update Q-values
|
238 |
+
chatbot.update_q_values(
|
239 |
+
user_sentiment, reward, user_sentiment
|
240 |
+
)
|
241 |
+
|
242 |
# Convert the response to speech
|
243 |
speech_fp = text_to_speech(response)
|
244 |
# Play the speech
|