Spaces:
Build error
Build error
Update utils/legal_notebook_interface.py
Browse files- utils/legal_notebook_interface.py +105 -156
utils/legal_notebook_interface.py
CHANGED
|
@@ -4,6 +4,7 @@ import datetime
|
|
| 4 |
|
| 5 |
class LegalNotebookInterface:
|
| 6 |
def __init__(self, case_manager, vector_store, document_processor):
|
|
|
|
| 7 |
self.case_manager = case_manager
|
| 8 |
self.vector_store = vector_store
|
| 9 |
self.document_processor = document_processor
|
|
@@ -13,35 +14,19 @@ class LegalNotebookInterface:
|
|
| 13 |
st.session_state.active_documents = []
|
| 14 |
if "chat_history" not in st.session_state:
|
| 15 |
st.session_state.chat_history = []
|
| 16 |
-
if "
|
| 17 |
-
st.session_state.
|
| 18 |
-
if "suggestions" not in st.session_state:
|
| 19 |
-
st.session_state.suggestions = []
|
| 20 |
|
| 21 |
def render(self):
|
| 22 |
"""Render the main notebook interface."""
|
| 23 |
-
#
|
| 24 |
st.markdown("""
|
| 25 |
<style>
|
| 26 |
-
.main-container {
|
| 27 |
-
max-width: 1200px;
|
| 28 |
-
margin: auto;
|
| 29 |
-
}
|
| 30 |
.document-panel {
|
| 31 |
background-color: #f8f9fa;
|
| 32 |
border-radius: 10px;
|
| 33 |
padding: 20px;
|
| 34 |
margin: 10px;
|
| 35 |
-
height: calc(100vh - 100px);
|
| 36 |
-
overflow-y: auto;
|
| 37 |
-
}
|
| 38 |
-
.chat-panel {
|
| 39 |
-
background-color: white;
|
| 40 |
-
border-radius: 10px;
|
| 41 |
-
padding: 20px;
|
| 42 |
-
margin: 10px;
|
| 43 |
-
height: calc(100vh - 100px);
|
| 44 |
-
overflow-y: auto;
|
| 45 |
}
|
| 46 |
.message {
|
| 47 |
padding: 15px;
|
|
@@ -64,13 +49,6 @@ class LegalNotebookInterface:
|
|
| 64 |
border-radius: 20px;
|
| 65 |
cursor: pointer;
|
| 66 |
}
|
| 67 |
-
.document-snippet {
|
| 68 |
-
padding: 10px;
|
| 69 |
-
background-color: #fff;
|
| 70 |
-
border-left: 3px solid #ffd700;
|
| 71 |
-
margin: 10px 0;
|
| 72 |
-
border-radius: 4px;
|
| 73 |
-
}
|
| 74 |
</style>
|
| 75 |
""", unsafe_allow_html=True)
|
| 76 |
|
|
@@ -90,153 +68,124 @@ class LegalNotebookInterface:
|
|
| 90 |
st.markdown("### 📚 Documents")
|
| 91 |
|
| 92 |
# Document selection
|
| 93 |
-
|
| 94 |
-
if
|
| 95 |
-
|
| 96 |
-
|
| 97 |
-
|
| 98 |
-
|
| 99 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 100 |
|
| 101 |
-
if
|
| 102 |
-
|
| 103 |
-
|
| 104 |
-
"
|
| 105 |
-
|
| 106 |
-
|
| 107 |
-
|
| 108 |
-
|
| 109 |
-
|
| 110 |
-
|
| 111 |
-
|
| 112 |
-
# Show document previews
|
| 113 |
-
for doc in selected_docs:
|
| 114 |
-
with st.expander(f"📄 {doc['title']}", expanded=False):
|
| 115 |
-
st.markdown("#### Quick Preview")
|
| 116 |
-
content = doc.get('content', '')[:500]
|
| 117 |
-
st.markdown(f"{content}...")
|
| 118 |
-
|
| 119 |
-
# Document metadata
|
| 120 |
-
st.markdown("#### Metadata")
|
| 121 |
-
st.json({
|
| 122 |
-
'Type': doc.get('metadata', {}).get('type', 'Unknown'),
|
| 123 |
-
'Added': doc.get('added_at', 'Unknown'),
|
| 124 |
-
'Pages': doc.get('metadata', {}).get('page_count', 'Unknown')
|
| 125 |
-
})
|
| 126 |
|
| 127 |
def _render_chat_panel(self):
|
| 128 |
-
"""Render the chat and analysis
|
| 129 |
-
st.markdown("### 💬
|
| 130 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 131 |
# Display chat history
|
| 132 |
for message in st.session_state.chat_history:
|
| 133 |
-
|
| 134 |
-
|
| 135 |
-
|
| 136 |
-
|
| 137 |
-
|
| 138 |
-
|
| 139 |
-
for idx, suggestion in enumerate(st.session_state.suggestions):
|
| 140 |
-
with cols[idx % 2]:
|
| 141 |
-
if st.button(suggestion, key=f"suggestion_{idx}"):
|
| 142 |
-
self._handle_input(suggestion)
|
| 143 |
|
| 144 |
# Chat input
|
| 145 |
-
|
| 146 |
-
|
| 147 |
-
col1, col2 = st.columns([1, 5])
|
| 148 |
-
with col1:
|
| 149 |
-
if st.button("Send", type="primary"):
|
| 150 |
-
if chat_input:
|
| 151 |
-
self._handle_input(chat_input)
|
| 152 |
-
with col2:
|
| 153 |
-
st.markdown("*Press Enter to send*")
|
| 154 |
-
|
| 155 |
-
def _render_message(self, message: Dict):
|
| 156 |
-
"""Render a single message with citations and context."""
|
| 157 |
-
message_class = "user-message" if message['role'] == 'user' else 'assistant-message'
|
| 158 |
-
|
| 159 |
-
st.markdown(f"""
|
| 160 |
-
<div class="message {message_class}">
|
| 161 |
-
<p>{message['content']}</p>
|
| 162 |
-
{self._render_citations(message.get('citations', [])) if message.get('citations') else ''}
|
| 163 |
-
</div>
|
| 164 |
-
""", unsafe_allow_html=True)
|
| 165 |
|
| 166 |
-
def
|
| 167 |
-
"""Render document citations and snippets."""
|
| 168 |
-
if not citations:
|
| 169 |
-
return ""
|
| 170 |
-
|
| 171 |
-
citations_html = "<div class='citations'>"
|
| 172 |
-
for citation in citations:
|
| 173 |
-
citations_html += f"""
|
| 174 |
-
<div class="document-snippet">
|
| 175 |
-
<small>From {citation['document']}</small>
|
| 176 |
-
<p>{citation['text']}</p>
|
| 177 |
-
</div>
|
| 178 |
-
"""
|
| 179 |
-
citations_html += "</div>"
|
| 180 |
-
return citations_html
|
| 181 |
-
|
| 182 |
-
def _handle_input(self, input_text: str):
|
| 183 |
"""Process user input and generate response."""
|
|
|
|
|
|
|
|
|
|
|
|
|
| 184 |
# Add user message
|
| 185 |
st.session_state.chat_history.append({
|
| 186 |
-
|
| 187 |
-
|
| 188 |
-
|
| 189 |
})
|
| 190 |
|
| 191 |
-
# Get
|
| 192 |
-
|
| 193 |
-
|
| 194 |
-
|
| 195 |
-
|
| 196 |
-
|
| 197 |
-
|
| 198 |
-
|
| 199 |
-
|
| 200 |
-
|
| 201 |
-
|
| 202 |
-
|
| 203 |
-
|
| 204 |
-
|
| 205 |
-
|
| 206 |
-
|
| 207 |
-
|
| 208 |
-
'text': chunk['text']
|
| 209 |
-
}
|
| 210 |
-
for chunk in context_chunks[:2]
|
| 211 |
-
],
|
| 212 |
-
'timestamp': datetime.datetime.now().isoformat()
|
| 213 |
-
}
|
| 214 |
|
| 215 |
-
|
| 216 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 217 |
|
| 218 |
-
|
| 219 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 220 |
|
| 221 |
-
def _generate_response(self, query: str, context_chunks: List[Dict]) -> str:
|
| 222 |
-
"""Generate response using the legal analysis pipeline."""
|
| 223 |
# Combine context from chunks
|
| 224 |
-
context = "\n\n".join([chunk['text'] for chunk in
|
| 225 |
|
| 226 |
-
#
|
| 227 |
-
|
| 228 |
-
|
| 229 |
-
|
| 230 |
-
response_content += "3. Recommendations...\n"
|
| 231 |
|
| 232 |
-
return
|
| 233 |
-
|
| 234 |
-
def
|
| 235 |
-
"""
|
| 236 |
-
|
| 237 |
-
st.
|
| 238 |
-
|
| 239 |
-
|
| 240 |
-
|
| 241 |
-
|
| 242 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4 |
|
| 5 |
class LegalNotebookInterface:
|
| 6 |
def __init__(self, case_manager, vector_store, document_processor):
|
| 7 |
+
"""Initialize the notebook interface with required components."""
|
| 8 |
self.case_manager = case_manager
|
| 9 |
self.vector_store = vector_store
|
| 10 |
self.document_processor = document_processor
|
|
|
|
| 14 |
st.session_state.active_documents = []
|
| 15 |
if "chat_history" not in st.session_state:
|
| 16 |
st.session_state.chat_history = []
|
| 17 |
+
if "current_case" not in st.session_state:
|
| 18 |
+
st.session_state.current_case = None
|
|
|
|
|
|
|
| 19 |
|
| 20 |
def render(self):
|
| 21 |
"""Render the main notebook interface."""
|
| 22 |
+
# Add custom styling
|
| 23 |
st.markdown("""
|
| 24 |
<style>
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
.document-panel {
|
| 26 |
background-color: #f8f9fa;
|
| 27 |
border-radius: 10px;
|
| 28 |
padding: 20px;
|
| 29 |
margin: 10px;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 30 |
}
|
| 31 |
.message {
|
| 32 |
padding: 15px;
|
|
|
|
| 49 |
border-radius: 20px;
|
| 50 |
cursor: pointer;
|
| 51 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 52 |
</style>
|
| 53 |
""", unsafe_allow_html=True)
|
| 54 |
|
|
|
|
| 68 |
st.markdown("### 📚 Documents")
|
| 69 |
|
| 70 |
# Document selection
|
| 71 |
+
cases = self.case_manager.get_all_cases()
|
| 72 |
+
if not cases:
|
| 73 |
+
st.info("No cases found. Please create a case and add documents first.")
|
| 74 |
+
return
|
| 75 |
+
|
| 76 |
+
selected_case = st.selectbox(
|
| 77 |
+
"Select Case",
|
| 78 |
+
cases,
|
| 79 |
+
format_func=lambda x: x['title']
|
| 80 |
+
)
|
| 81 |
+
|
| 82 |
+
if selected_case:
|
| 83 |
+
st.session_state.current_case = selected_case['id']
|
| 84 |
+
documents = self.case_manager.list_documents(selected_case['id'])
|
| 85 |
|
| 86 |
+
if documents:
|
| 87 |
+
st.write("#### Available Documents")
|
| 88 |
+
for doc in documents:
|
| 89 |
+
with st.expander(f"📄 {doc['title']}", expanded=False):
|
| 90 |
+
st.write("**Added:** ", doc.get('added_at', 'Unknown'))
|
| 91 |
+
st.write("**Type:** ", doc.get('metadata', {}).get('document_type', 'Unknown'))
|
| 92 |
+
if st.button("View Details", key=f"view_{doc['id']}"):
|
| 93 |
+
self._show_document_details(doc)
|
| 94 |
+
else:
|
| 95 |
+
st.info("No documents in this case. Please add documents first.")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 96 |
|
| 97 |
def _render_chat_panel(self):
|
| 98 |
+
"""Render the chat and analysis interface."""
|
| 99 |
+
st.markdown("### 💬 Document Analysis")
|
| 100 |
|
| 101 |
+
if not st.session_state.current_case:
|
| 102 |
+
st.info("Please select a case from the left panel to start analysis.")
|
| 103 |
+
return
|
| 104 |
+
|
| 105 |
# Display chat history
|
| 106 |
for message in st.session_state.chat_history:
|
| 107 |
+
message_class = "user-message" if message["role"] == "user" else "assistant-message"
|
| 108 |
+
st.markdown(f"""
|
| 109 |
+
<div class="message {message_class}">
|
| 110 |
+
{message["content"]}
|
| 111 |
+
</div>
|
| 112 |
+
""", unsafe_allow_html=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
| 113 |
|
| 114 |
# Chat input
|
| 115 |
+
if prompt := st.chat_input("Ask about your documents..."):
|
| 116 |
+
self._handle_chat_input(prompt)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 117 |
|
| 118 |
+
def _handle_chat_input(self, prompt: str):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 119 |
"""Process user input and generate response."""
|
| 120 |
+
if not st.session_state.current_case:
|
| 121 |
+
st.error("Please select a case first.")
|
| 122 |
+
return
|
| 123 |
+
|
| 124 |
# Add user message
|
| 125 |
st.session_state.chat_history.append({
|
| 126 |
+
"role": "user",
|
| 127 |
+
"content": prompt,
|
| 128 |
+
"timestamp": datetime.datetime.now().isoformat()
|
| 129 |
})
|
| 130 |
|
| 131 |
+
# Get documents from current case
|
| 132 |
+
documents = self.case_manager.list_documents(st.session_state.current_case)
|
| 133 |
+
if not documents:
|
| 134 |
+
st.error("No documents available for analysis.")
|
| 135 |
+
return
|
| 136 |
+
|
| 137 |
+
try:
|
| 138 |
+
# Get relevant chunks from documents
|
| 139 |
+
chunks = []
|
| 140 |
+
for doc in documents:
|
| 141 |
+
doc_chunks = self.vector_store.similarity_search(
|
| 142 |
+
prompt,
|
| 143 |
+
filter_criteria={"doc_id": doc['id']},
|
| 144 |
+
k=2
|
| 145 |
+
)
|
| 146 |
+
if doc_chunks:
|
| 147 |
+
chunks.extend(doc_chunks)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 148 |
|
| 149 |
+
# Generate response
|
| 150 |
+
response = {
|
| 151 |
+
"role": "assistant",
|
| 152 |
+
"content": self._generate_response(prompt, chunks),
|
| 153 |
+
"timestamp": datetime.datetime.now().isoformat()
|
| 154 |
+
}
|
| 155 |
|
| 156 |
+
# Add response to chat history
|
| 157 |
+
st.session_state.chat_history.append(response)
|
| 158 |
+
|
| 159 |
+
except Exception as e:
|
| 160 |
+
st.error(f"Error analyzing documents: {str(e)}")
|
| 161 |
+
|
| 162 |
+
def _generate_response(self, prompt: str, chunks: List[Dict]) -> str:
|
| 163 |
+
"""Generate response based on document chunks."""
|
| 164 |
+
if not chunks:
|
| 165 |
+
return "I couldn't find any relevant information in the documents. Please try rephrasing your question."
|
| 166 |
|
|
|
|
|
|
|
| 167 |
# Combine context from chunks
|
| 168 |
+
context = "\n\n".join([chunk['text'] for chunk in chunks])
|
| 169 |
|
| 170 |
+
# Generate a simple response (replace with your actual response generation)
|
| 171 |
+
response = f"Based on the documents, I found {len(chunks)} relevant sections:\n\n"
|
| 172 |
+
for i, chunk in enumerate(chunks, 1):
|
| 173 |
+
response += f"{i}. {chunk['text'][:200]}...\n\n"
|
|
|
|
| 174 |
|
| 175 |
+
return response
|
| 176 |
+
|
| 177 |
+
def _show_document_details(self, doc: Dict):
|
| 178 |
+
"""Show detailed document information."""
|
| 179 |
+
st.markdown("#### Document Details")
|
| 180 |
+
st.write("**Title:** ", doc['title'])
|
| 181 |
+
st.write("**Added:** ", doc.get('added_at', 'Unknown'))
|
| 182 |
+
|
| 183 |
+
metadata = doc.get('metadata', {})
|
| 184 |
+
if metadata:
|
| 185 |
+
st.write("**Document Type:** ", metadata.get('document_type', 'Unknown'))
|
| 186 |
+
st.write("**Word Count:** ", metadata.get('word_count', 'Unknown'))
|
| 187 |
+
|
| 188 |
+
if 'entities' in metadata:
|
| 189 |
+
st.write("**Named Entities:** ")
|
| 190 |
+
for entity_type, entities in metadata['entities'].items():
|
| 191 |
+
st.write(f"- {entity_type}: {', '.join(entities[:5])}")
|