Spaces:
Sleeping
Sleeping
muzammil-eds
commited on
Commit
β’
a72c9c1
1
Parent(s):
c586572
Upload 2 files
Browse files- education.py +183 -0
- requirements.txt +7 -0
education.py
ADDED
@@ -0,0 +1,183 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
import os
|
3 |
+
import pickle
|
4 |
+
import time
|
5 |
+
import g4f
|
6 |
+
import tempfile
|
7 |
+
import PyPDF2
|
8 |
+
from pdf2image import convert_from_path
|
9 |
+
import pytesseract
|
10 |
+
|
11 |
+
st.set_page_config(page_title="EDUCATIONAL ASSISTANT")
|
12 |
+
|
13 |
+
st.markdown(
|
14 |
+
"""
|
15 |
+
<style>
|
16 |
+
.title {
|
17 |
+
text-align: center;
|
18 |
+
font-size: 2em;
|
19 |
+
font-weight: bold;
|
20 |
+
}
|
21 |
+
</style>
|
22 |
+
<div class="title"> π EDUCATIONAL ASSISTANT π</div>
|
23 |
+
""",
|
24 |
+
unsafe_allow_html=True
|
25 |
+
)
|
26 |
+
# Load and Save Conversations
|
27 |
+
conversations_file = "conversations.pkl"
|
28 |
+
|
29 |
+
|
30 |
+
@st.cache_data
|
31 |
+
def load_conversations():
|
32 |
+
try:
|
33 |
+
with open(conversations_file, "rb") as f:
|
34 |
+
return pickle.load(f)
|
35 |
+
except (FileNotFoundError, EOFError):
|
36 |
+
return []
|
37 |
+
|
38 |
+
|
39 |
+
def save_conversations(conversations):
|
40 |
+
temp_conversations_file = conversations_file
|
41 |
+
with open(temp_conversations_file, "wb") as f:
|
42 |
+
pickle.dump(conversations, f)
|
43 |
+
os.replace(temp_conversations_file, conversations_file)
|
44 |
+
|
45 |
+
|
46 |
+
if 'conversations' not in st.session_state:
|
47 |
+
st.session_state.conversations = load_conversations()
|
48 |
+
|
49 |
+
if 'current_conversation' not in st.session_state:
|
50 |
+
st.session_state.current_conversation = [{"role": "assistant", "content": "How may I assist you today?"}]
|
51 |
+
|
52 |
+
|
53 |
+
def truncate_string(s, length=30):
|
54 |
+
return s[:length].rstrip() + "..." if len(s) > length else s
|
55 |
+
|
56 |
+
|
57 |
+
def display_chats_sidebar():
|
58 |
+
with st.sidebar.container():
|
59 |
+
st.header('Settings')
|
60 |
+
col1, col2 = st.columns([1, 1])
|
61 |
+
|
62 |
+
with col1:
|
63 |
+
if col1.button('Start New Chat', key="new_chat"):
|
64 |
+
st.session_state.current_conversation = []
|
65 |
+
st.session_state.conversations.append(st.session_state.current_conversation)
|
66 |
+
|
67 |
+
with col2:
|
68 |
+
if col2.button('Clear All Chats', key="clear_all"):
|
69 |
+
st.session_state.conversations = []
|
70 |
+
st.session_state.current_conversation = []
|
71 |
+
|
72 |
+
if st.sidebar.button('Solve Assignment', key="summarize_bills", use_container_width=True):
|
73 |
+
st.session_state.page = "summarize_bills"
|
74 |
+
|
75 |
+
with st.sidebar.container():
|
76 |
+
st.header('Conversations')
|
77 |
+
for idx, conversation in enumerate(st.session_state.conversations):
|
78 |
+
if conversation:
|
79 |
+
chat_title_raw = next((msg["content"] for msg in conversation if msg["role"] == "user"), "New Chat")
|
80 |
+
chat_title = truncate_string(chat_title_raw)
|
81 |
+
if st.sidebar.button(f"{chat_title}", key=f"chat_button_{idx}"):
|
82 |
+
st.session_state.current_conversation = st.session_state.conversations[idx]
|
83 |
+
|
84 |
+
|
85 |
+
def summarize_bill():
|
86 |
+
st.header("π Solve PDF Assignments π")
|
87 |
+
|
88 |
+
if st.button("Back to Chat"):
|
89 |
+
st.session_state.page = "chat"
|
90 |
+
|
91 |
+
uploaded_file = st.file_uploader("Upload an Agreement", type=['pdf'])
|
92 |
+
if uploaded_file is not None:
|
93 |
+
with tempfile.NamedTemporaryFile(delete=False) as tmp_file:
|
94 |
+
tmp_file.write(uploaded_file.read())
|
95 |
+
extracted_text = extract_text_from_pdf(tmp_file.name)
|
96 |
+
|
97 |
+
if st.button('Solve'):
|
98 |
+
# Assuming g4f.ChatCompletion can be used for summarization
|
99 |
+
# Replace with appropriate summarization logic if needed
|
100 |
+
summary = g4f.ChatCompletion.create(
|
101 |
+
model="gpt-3.5-turbo",
|
102 |
+
messages=[{"role": "user", "content": "Please solve this Agreement: \n" + extracted_text}],
|
103 |
+
temperature=0.5, # You can adjust parameters as needed
|
104 |
+
max_tokens=150 # Adjust the token limit as needed
|
105 |
+
)
|
106 |
+
st.text_area("Summary", summary, height=400)
|
107 |
+
|
108 |
+
|
109 |
+
def extract_text_from_pdf(file_path: str) -> str:
|
110 |
+
try:
|
111 |
+
with open(file_path, 'rb') as file:
|
112 |
+
reader = PyPDF2.PdfReader(file)
|
113 |
+
text = ''
|
114 |
+
for page_number in range(len(reader.pages)):
|
115 |
+
page = reader.pages[page_number]
|
116 |
+
text += page.extract_text()
|
117 |
+
return text
|
118 |
+
except Exception as e:
|
119 |
+
try:
|
120 |
+
images = convert_from_path(file_path)
|
121 |
+
extracted_texts = [pytesseract.image_to_string(image) for image in images]
|
122 |
+
return "\n".join(extracted_texts)
|
123 |
+
except Exception as e:
|
124 |
+
raise ValueError(f"Failed to process {file_path} using PDF Reader and OCR. Error: {e}")
|
125 |
+
|
126 |
+
|
127 |
+
def main_app():
|
128 |
+
for message in st.session_state.current_conversation:
|
129 |
+
with st.chat_message(message["role"]):
|
130 |
+
st.write(message["content"])
|
131 |
+
|
132 |
+
def generate_response(prompt_input):
|
133 |
+
string_dialogue = '''
|
134 |
+
You are an educational assistant chatbot, designed to provide insightful and accurate answers in the educational domain. Your responses should be engaging and emulate a human educator to create a comfortable learning environment. Instead of simply presenting facts, aim to inspire curiosity and deeper understanding.
|
135 |
+
|
136 |
+
Context:
|
137 |
+
|
138 |
+
Understand the essence of the user's educational query.
|
139 |
+
Consider the academic level and subject matter of the question.
|
140 |
+
Access a broad knowledge base to provide well-informed responses.
|
141 |
+
Organize the response clearly and logically.
|
142 |
+
Deliver the answer in a manner that is both educational and relatable to human interaction.
|
143 |
+
|
144 |
+
Human:
|
145 |
+
'''
|
146 |
+
for dict_message in st.session_state.current_conversation:
|
147 |
+
string_dialogue += dict_message["role"].capitalize() + ": " + dict_message["content"] + "\\n\\n"
|
148 |
+
|
149 |
+
prompt = f"{string_dialogue}\n {prompt_input} Assistant: "
|
150 |
+
response_generator = g4f.ChatCompletion.create(
|
151 |
+
model="gpt-3.5-turbo",
|
152 |
+
messages=[{"role": "user", "content": prompt}],
|
153 |
+
stream=True,
|
154 |
+
)
|
155 |
+
return response_generator
|
156 |
+
|
157 |
+
if prompt := st.chat_input('Send a Message'):
|
158 |
+
st.session_state.current_conversation.append({"role": "user", "content": prompt})
|
159 |
+
with st.chat_message("user"):
|
160 |
+
st.write(prompt)
|
161 |
+
|
162 |
+
with st.chat_message("assistant"):
|
163 |
+
with st.spinner("Thinking..."):
|
164 |
+
response = generate_response(prompt)
|
165 |
+
placeholder = st.empty()
|
166 |
+
full_response = ''
|
167 |
+
for item in response:
|
168 |
+
full_response += item
|
169 |
+
time.sleep(0.003)
|
170 |
+
placeholder.markdown(full_response)
|
171 |
+
placeholder.markdown(full_response)
|
172 |
+
st.session_state.current_conversation.append({"role": "assistant", "content": full_response})
|
173 |
+
save_conversations(st.session_state.conversations)
|
174 |
+
|
175 |
+
|
176 |
+
display_chats_sidebar()
|
177 |
+
if st.session_state.get('page') == "summarize_bills":
|
178 |
+
summarize_bill()
|
179 |
+
elif st.session_state.get('page') == "chat":
|
180 |
+
main_app()
|
181 |
+
else:
|
182 |
+
# Default page when the app starts or when the state is not set
|
183 |
+
main_app()
|
requirements.txt
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
gpt4all
|
2 |
+
streamlit
|
3 |
+
g4f
|
4 |
+
docx2txt
|
5 |
+
pypdf2
|
6 |
+
pytesseract
|
7 |
+
pdf2image
|