Update app.py
Browse files
app.py
CHANGED
@@ -1,260 +1,193 @@
|
|
1 |
-
import PyPDF2
|
2 |
-
import re
|
3 |
-
from sentence_transformers import SentenceTransformer
|
4 |
-
import faiss
|
5 |
-
from langchain.agents import initialize_agent, AgentType,Tool
|
6 |
-
from langchain.schema import HumanMessage
|
7 |
-
from langchain_google_genai import ChatGoogleGenerativeAI
|
8 |
-
import gradio as gr
|
9 |
-
import os
|
10 |
-
import pytesseract
|
11 |
-
from PIL import Image
|
12 |
-
pytesseract.pytesseract.tesseract_cmd = r"tesseract.exe"
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
-
|
17 |
-
|
18 |
-
|
19 |
-
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
|
25 |
-
|
26 |
-
|
27 |
-
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
|
32 |
-
|
33 |
-
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
|
44 |
-
|
45 |
-
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
-
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
|
61 |
-
for
|
62 |
-
|
63 |
-
|
64 |
-
|
65 |
-
|
66 |
-
|
67 |
-
|
68 |
-
|
69 |
-
|
70 |
-
|
71 |
-
|
72 |
-
|
73 |
-
|
74 |
-
|
75 |
-
|
76 |
-
|
77 |
-
|
78 |
-
|
79 |
-
|
80 |
-
|
81 |
-
|
82 |
-
|
83 |
-
|
84 |
-
|
85 |
-
|
86 |
-
|
87 |
-
|
88 |
-
|
89 |
-
|
90 |
-
|
91 |
-
|
92 |
-
|
93 |
-
|
94 |
-
|
95 |
-
|
96 |
-
|
97 |
-
|
98 |
-
|
99 |
-
|
100 |
-
|
101 |
-
|
102 |
-
|
103 |
-
|
104 |
-
|
105 |
-
|
106 |
-
|
107 |
-
|
108 |
-
|
109 |
-
|
110 |
-
|
111 |
-
|
112 |
-
|
113 |
-
|
114 |
-
|
115 |
-
|
116 |
-
|
117 |
-
|
118 |
-
|
119 |
-
|
120 |
-
|
121 |
-
|
122 |
-
|
123 |
-
|
124 |
-
|
125 |
-
|
126 |
-
|
127 |
-
|
128 |
-
|
129 |
-
|
130 |
-
|
131 |
-
|
132 |
-
|
133 |
-
|
134 |
-
|
135 |
-
|
136 |
-
|
137 |
-
|
138 |
-
|
139 |
-
|
140 |
-
|
141 |
-
|
142 |
-
|
143 |
-
|
144 |
-
|
145 |
-
|
146 |
-
)
|
147 |
-
|
148 |
-
|
149 |
-
|
150 |
-
|
151 |
-
|
152 |
-
|
153 |
-
|
154 |
-
|
155 |
-
|
156 |
-
|
157 |
-
|
158 |
-
|
159 |
-
|
160 |
-
|
161 |
-
|
162 |
-
|
163 |
-
|
164 |
-
|
165 |
-
|
166 |
-
|
167 |
-
)
|
168 |
-
|
169 |
-
|
170 |
-
|
171 |
-
|
172 |
-
|
173 |
-
|
174 |
-
|
175 |
-
|
176 |
-
|
177 |
-
|
178 |
-
)
|
179 |
-
|
180 |
-
|
181 |
-
|
182 |
-
|
183 |
-
|
184 |
-
|
185 |
-
|
186 |
-
|
187 |
-
|
188 |
-
|
189 |
-
|
190 |
-
|
191 |
-
|
192 |
-
|
193 |
-
]
|
194 |
-
)
|
195 |
-
else:
|
196 |
-
# If no image, only pass the text
|
197 |
-
message = HumanMessage(content=[{"type": "text", "text": query}])
|
198 |
-
|
199 |
-
# Invoke the model with the multimodal message
|
200 |
-
result = agent.invoke([message])
|
201 |
-
response = result['output']
|
202 |
-
intermediate_steps = result.get('intermediate_steps', [])
|
203 |
-
|
204 |
-
thought_process = ""
|
205 |
-
for action, observation in intermediate_steps:
|
206 |
-
thought_process += f"**Thought:** {action.log}\n"
|
207 |
-
thought_process += f"**Action:** {action.tool}\n"
|
208 |
-
thought_process += f"**Observation:** {observation}\n\n"
|
209 |
-
|
210 |
-
return response, thought_process.strip()
|
211 |
-
# Step 5: Gradio Interface
|
212 |
-
from gradio import ChatMessage
|
213 |
-
def chatbot_interface(messages,prompt):
|
214 |
-
response, thought_process = chatbot_response(prompt)
|
215 |
-
#messages.append(ChatMessage(role="user", content=prompt))
|
216 |
-
|
217 |
-
for x in prompt["files"]:
|
218 |
-
messages.append(ChatMessage(role="user", content={"path": x, "mime_type": "image/png"}))
|
219 |
-
if prompt["text"] is not None:
|
220 |
-
messages.append(ChatMessage(role="user", content=prompt['text']))
|
221 |
-
if thought_process:
|
222 |
-
messages.append(ChatMessage(role="assistant", content=thought_process,metadata={"title": "🧠 Thought Process"}))
|
223 |
-
messages.append(ChatMessage(role="assistant", content=response))
|
224 |
-
|
225 |
-
return messages, gr.MultimodalTextbox(value=None, interactive=True)
|
226 |
-
|
227 |
-
|
228 |
-
def vote(data: gr.LikeData):
|
229 |
-
if data.liked:
|
230 |
-
print("You upvoted this response: " + data.value)
|
231 |
-
else:
|
232 |
-
print("You downvoted this response: " + data.value)
|
233 |
-
|
234 |
-
with gr.Blocks(theme=gr.themes.Soft()) as iface:
|
235 |
-
|
236 |
-
gr.Markdown(
|
237 |
-
"""
|
238 |
-
<div style="font-size: 24px; font-weight: bold; color: #333;">
|
239 |
-
DoJ Chatbot
|
240 |
-
</div>
|
241 |
-
<div style="font-size: 16px; color: #555;">
|
242 |
-
Ask questions related to the Department of Justice.
|
243 |
-
</div>
|
244 |
-
"""
|
245 |
-
)
|
246 |
-
chatbot = gr.Chatbot(type="messages",avatar_images=("user.jpeg", "logo.jpeg"), bubble_full_width=True) # Chatbot component to display conversation history
|
247 |
-
query_input = gr.MultimodalTextbox(interactive=True,
|
248 |
-
placeholder="Enter message or upload file...", show_label=False)
|
249 |
-
submit_button = gr.Button("Send")
|
250 |
-
|
251 |
-
submit_button.click(chatbot_interface, [chatbot, query_input], [chatbot, query_input])
|
252 |
-
query_input.submit(chatbot_interface, [chatbot, query_input], [chatbot,query_input])
|
253 |
-
|
254 |
-
chatbot.like(vote, None, None) # Adding like/dislike functionality to the chatbot
|
255 |
-
|
256 |
-
|
257 |
-
iface.launch(
|
258 |
-
show_error=True,
|
259 |
-
prevent_thread_lock=True
|
260 |
)
|
|
|
1 |
+
import PyPDF2
|
2 |
+
import re
|
3 |
+
from sentence_transformers import SentenceTransformer
|
4 |
+
import faiss
|
5 |
+
from langchain.agents import initialize_agent, AgentType,Tool
|
6 |
+
from langchain.schema import HumanMessage
|
7 |
+
from langchain_google_genai import ChatGoogleGenerativeAI
|
8 |
+
import gradio as gr
|
9 |
+
import os
|
10 |
+
import pytesseract
|
11 |
+
from PIL import Image
|
12 |
+
pytesseract.pytesseract.tesseract_cmd = r"tesseract.exe"
|
13 |
+
|
14 |
+
model = SentenceTransformer('sentence-transformers/all-MiniLM-L6-v2')
|
15 |
+
|
16 |
+
embeddings = model.encode(flattened_data)
|
17 |
+
index = faiss.read_index('IPC_index.faiss')
|
18 |
+
index2 = faiss.read_index('CrPC_index.faiss')
|
19 |
+
|
20 |
+
|
21 |
+
# Step 3: Retrieval with Citations using PDF filename
|
22 |
+
def retrieve_info_with_citation(query, top_k=5):
|
23 |
+
query_embedding = model.encode([query])
|
24 |
+
D, I = index.search(query_embedding, k=top_k)
|
25 |
+
|
26 |
+
results = []
|
27 |
+
for i in range(min(top_k, len(I[0]))):
|
28 |
+
if D[0][i] < 1.0: # Relevance threshold
|
29 |
+
chunk_index = I[0][i]
|
30 |
+
citation = f"Source: IPC"
|
31 |
+
results.append((match, citation))
|
32 |
+
else:
|
33 |
+
break
|
34 |
+
|
35 |
+
if results:
|
36 |
+
return results
|
37 |
+
else:
|
38 |
+
return [("I'm sorry, I couldn't find relevant information.", "Source: N/A")]
|
39 |
+
|
40 |
+
|
41 |
+
def retrieve_info_with_citation2(query, top_k=5):
|
42 |
+
query_embedding = model.encode([query])
|
43 |
+
D, I = index2.search(query_embedding, k=top_k)
|
44 |
+
|
45 |
+
results = []
|
46 |
+
for i in range(min(top_k, len(I[0]))):
|
47 |
+
if D[0][i] < 1.0: # Relevance threshold
|
48 |
+
chunk_index = I[0][i]
|
49 |
+
citation = f"Source: CrPC"
|
50 |
+
results.append((match, citation))
|
51 |
+
else:
|
52 |
+
break
|
53 |
+
|
54 |
+
if results:
|
55 |
+
return results
|
56 |
+
else:
|
57 |
+
return [("I'm sorry, I couldn't find relevant information.", "Source: N/A")]
|
58 |
+
|
59 |
+
def retrieve_info(query):
|
60 |
+
results = retrieve_info_with_citation(query)
|
61 |
+
formatted_results = "\n\n".join([f"{i+1}. {match}\n{citation}" for i, (match, citation) in enumerate(results)])
|
62 |
+
return formatted_results
|
63 |
+
|
64 |
+
def retrieve_info2(query):
|
65 |
+
results = retrieve_info_with_citation2(query)
|
66 |
+
formatted_results = "\n\n".join([f"{i+1}. {match}\n{citation}" for i, (match, citation) in enumerate(results)])
|
67 |
+
return formatted_results
|
68 |
+
|
69 |
+
ipc_tool = Tool(
|
70 |
+
name="IPC Information Retrieval",
|
71 |
+
func=retrieve_info,
|
72 |
+
description="Retrieve information from the Indian Penal Code Related to query keyword(s)."
|
73 |
+
)
|
74 |
+
|
75 |
+
crpc_tool=Tool(
|
76 |
+
name="CrPC Information Retrieval",
|
77 |
+
func=retrieve_info2,
|
78 |
+
description="Retrieve information from the Code of Criminal Procedure(CrPC) Related to query keyword(s)."
|
79 |
+
)
|
80 |
+
llm = ChatGoogleGenerativeAI(
|
81 |
+
model="gemini-1.5-pro",
|
82 |
+
temperature=0.25,
|
83 |
+
max_tokens=None,
|
84 |
+
timeout=None,
|
85 |
+
max_retries=2,
|
86 |
+
prompt_template="""
|
87 |
+
You are a highly specialized legal assistant with deep knowledge of the Indian Penal Code (IPC).
|
88 |
+
Your primary task is to retrieve and summarize legal information accurately from the IPC.pdf document provided to you.
|
89 |
+
Your responses should be highly specific, fact-based, and free from any speculation or hallucinations.
|
90 |
+
Always cite the exact section from the IPC when providing an answer.
|
91 |
+
If the information is not available in the document, clearly state that and do not make any assumptions.
|
92 |
+
|
93 |
+
Example task: "What is the punishment for theft according to the IPC?"
|
94 |
+
Example response: "According to Section 379 of the IPC, the punishment for theft is imprisonment of either description for a term which may extend to three years, or with fine, or with both."
|
95 |
+
|
96 |
+
Task: {{query}}
|
97 |
+
|
98 |
+
Response:
|
99 |
+
""",
|
100 |
+
)
|
101 |
+
|
102 |
+
agent_tools = [ipc_tool,crpc_tool]
|
103 |
+
|
104 |
+
agent = initialize_agent(
|
105 |
+
tools=agent_tools,
|
106 |
+
llm=llm,
|
107 |
+
agent=AgentType.STRUCTURED_CHAT_ZERO_SHOT_REACT_DESCRIPTION,
|
108 |
+
verbose=True,
|
109 |
+
return_intermediate_steps=True,
|
110 |
+
handle_parsing_errors=True,
|
111 |
+
)
|
112 |
+
def encode_image_to_base64(image_path):
|
113 |
+
return pytesseract.image_to_string(Image.open(image_path))
|
114 |
+
def chatbot_response(query):
|
115 |
+
if query.get('files'):
|
116 |
+
# Encode image to base64
|
117 |
+
image_data=""
|
118 |
+
for x in range(len(query["files"])):
|
119 |
+
image_data += f"{x}. "+encode_image_to_base64(query["files"][x]) +"\n"
|
120 |
+
|
121 |
+
# Create a multimodal message with both text and image data
|
122 |
+
message = HumanMessage(
|
123 |
+
content=[
|
124 |
+
{"type": "text", "text": query['text'] +" System :Image(s) was added to this prompt by this user. Text Extracted from this image (Some words may be misspelled ,Use your understanding ):"+image_data}, # Add text input
|
125 |
+
|
126 |
+
]
|
127 |
+
)
|
128 |
+
else:
|
129 |
+
# If no image, only pass the text
|
130 |
+
message = HumanMessage(content=[{"type": "text", "text": query}])
|
131 |
+
|
132 |
+
# Invoke the model with the multimodal message
|
133 |
+
result = agent.invoke([message])
|
134 |
+
response = result['output']
|
135 |
+
intermediate_steps = result.get('intermediate_steps', [])
|
136 |
+
|
137 |
+
thought_process = ""
|
138 |
+
for action, observation in intermediate_steps:
|
139 |
+
thought_process += f"**Thought:** {action.log}\n"
|
140 |
+
thought_process += f"**Action:** {action.tool}\n"
|
141 |
+
thought_process += f"**Observation:** {observation}\n\n"
|
142 |
+
|
143 |
+
return response, thought_process.strip()
|
144 |
+
# Step 5: Gradio Interface
|
145 |
+
from gradio import ChatMessage
|
146 |
+
def chatbot_interface(messages,prompt):
|
147 |
+
response, thought_process = chatbot_response(prompt)
|
148 |
+
#messages.append(ChatMessage(role="user", content=prompt))
|
149 |
+
|
150 |
+
for x in prompt["files"]:
|
151 |
+
messages.append(ChatMessage(role="user", content={"path": x, "mime_type": "image/png"}))
|
152 |
+
if prompt["text"] is not None:
|
153 |
+
messages.append(ChatMessage(role="user", content=prompt['text']))
|
154 |
+
if thought_process:
|
155 |
+
messages.append(ChatMessage(role="assistant", content=thought_process,metadata={"title": "🧠 Thought Process"}))
|
156 |
+
messages.append(ChatMessage(role="assistant", content=response))
|
157 |
+
|
158 |
+
return messages, gr.MultimodalTextbox(value=None, interactive=True)
|
159 |
+
|
160 |
+
|
161 |
+
def vote(data: gr.LikeData):
|
162 |
+
if data.liked:
|
163 |
+
print("You upvoted this response: " + data.value)
|
164 |
+
else:
|
165 |
+
print("You downvoted this response: " + data.value)
|
166 |
+
|
167 |
+
with gr.Blocks(theme=gr.themes.Soft()) as iface:
|
168 |
+
|
169 |
+
gr.Markdown(
|
170 |
+
"""
|
171 |
+
<div style="font-size: 24px; font-weight: bold; color: #333;">
|
172 |
+
DoJ Chatbot
|
173 |
+
</div>
|
174 |
+
<div style="font-size: 16px; color: #555;">
|
175 |
+
Ask questions related to the Department of Justice.
|
176 |
+
</div>
|
177 |
+
"""
|
178 |
+
)
|
179 |
+
chatbot = gr.Chatbot(type="messages",avatar_images=("user.jpeg", "logo.jpeg"), bubble_full_width=True) # Chatbot component to display conversation history
|
180 |
+
query_input = gr.MultimodalTextbox(interactive=True,
|
181 |
+
placeholder="Enter message or upload file...", show_label=False)
|
182 |
+
submit_button = gr.Button("Send")
|
183 |
+
|
184 |
+
submit_button.click(chatbot_interface, [chatbot, query_input], [chatbot, query_input])
|
185 |
+
query_input.submit(chatbot_interface, [chatbot, query_input], [chatbot,query_input])
|
186 |
+
|
187 |
+
chatbot.like(vote, None, None) # Adding like/dislike functionality to the chatbot
|
188 |
+
|
189 |
+
|
190 |
+
iface.launch(
|
191 |
+
show_error=True,
|
192 |
+
prevent_thread_lock=True
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
193 |
)
|