Spaces:
Runtime error
Runtime error
File size: 12,621 Bytes
fe52a97 |
1 2 3 4 5 6 7 8 9 10 11 12 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 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 |
import streamlit as st
from pathlib import Path
import tempfile
import os
import time
from typing import List, Dict
from pinecone import Pinecone
from src.table_aware_chunker import TableRecursiveChunker
from src.processor import TableProcessor
from src.llm import LLMChat
from src.embedding import EmbeddingModel
from chonkie import RecursiveRules
from src.vectordb import ChunkType, process_documents, ingest_data, PineconeRetriever
# Custom CSS for better UI
st.set_page_config(
page_title="📚 Table RAG Assistant",
layout="wide",
initial_sidebar_state="expanded"
)
st.markdown("""
<style>
.stApp {
max-width: 1200px;
margin: 0 auto;
}
.chat-message {
padding: 1.5rem;
border-radius: 0.5rem;
margin-bottom: 1rem;
box-shadow: 0 2px 4px rgba(0,0,0,0.1);
}
.user-message {
background-color: #f0f2f6;
}
.assistant-message {
background-color: #e8f0fe;
}
.st-emotion-cache-1v0mbdj.e115fcil1 {
border-radius: 0.5rem;
}
</style>
""", unsafe_allow_html=True)
# Initialize session states
if "messages" not in st.session_state:
st.session_state.messages = []
if "documents_processed" not in st.session_state:
st.session_state.documents_processed = False
if "retriever" not in st.session_state:
st.session_state.retriever = None
if "llm" not in st.session_state:
st.session_state.llm = None
if "uploaded_files" not in st.session_state:
st.session_state.uploaded_files = []
# Enhanced RAG Template using LangChain's ChatPromptTemplate
RAG_TEMPLATE = [
{
"role": "system",
"content": """You are a knowledgeable assistant specialized in analyzing documents and tables.
Your responses should be:
- Accurate and based on the provided context
- Concise (three sentences maximum)
- Professional yet conversational
- Include specific references to tables when relevant
If you cannot find an answer in the context, acknowledge this clearly."""
},
{
"role": "human",
"content": "Context: {context}\n\nQuestion: {question}"
}
]
def simulate_streaming_response(text: str, delay: float = 0.02) -> str:
"""Simulate streaming response by yielding chunks of text with delay."""
words = text.split()
result = ""
for i, word in enumerate(words):
result += word + " "
time.sleep(delay)
# Add punctuation pause
if any(p in word for p in ['.', '!', '?', ',']):
time.sleep(delay * 2)
yield result
def clear_pinecone_index(pc, index_name="vector-index"):
"""Clear the Pinecone index and reset app state."""
try:
if pc.has_index(index_name):
pc.delete_index(index_name)
st.session_state.documents_processed = False
st.session_state.retriever = None
st.session_state.messages = []
st.session_state.llm = None
st.session_state.uploaded_files = []
st.success("🧹 Database cleared successfully!")
except Exception as e:
st.error(f"❌ Error clearing database: {str(e)}")
def format_context(results: List[Dict]) -> str:
"""Format retrieved results into context string."""
context_parts = []
for result in results:
if result.get("chunk_type") == ChunkType.TABLE.value:
table_text = f"Table: {result['markdown_table']}"
if result.get("table_description"):
table_text += f"\nDescription: {result['table_description']}"
context_parts.append(table_text)
else:
context_parts.append(result.get("page_content", ""))
return "\n\n".join(context_parts)
def format_chat_message(message: Dict[str, str], results: List[Dict] = None) -> str:
"""Format chat message with retrieved tables in a visually appealing way."""
content = message["content"]
if results:
for result in results:
if result.get("chunk_type") == ChunkType.TABLE.value:
content += "\n\n---\n\n📊 **Relevant Table:**\n" + result['markdown_table']
return content
def initialize_components(pinecone_api_key: str):
"""Initialize all required components with LangChain integration."""
try:
# Initialize Pinecone
pc = Pinecone(api_key=pinecone_api_key)
# Initialize LangChain LLM with custom parameters
llm = LLMChat(
model_name="mistral:7b",
temperature=0.3 # Lower temperature for more focused responses
)
st.session_state.llm = llm
# Initialize LangChain Embeddings
embedder = EmbeddingModel("nomic-embed-text")
# Initialize Chunker
chunker = TableRecursiveChunker(
tokenizer="gpt2",
chunk_size=512,
rules=RecursiveRules(),
min_characters_per_chunk=12
)
# Initialize Processor
processor = TableProcessor(
llm_model=llm,
embedding_model=embedder,
batch_size=8
)
return pc, llm, embedder, chunker, processor
except Exception as e:
st.error(f"❌ Error initializing components: {str(e)}")
return None, None, None, None, None
def process_all_documents(uploaded_files, chunker, processor, pc, embedder):
"""Process uploaded documents with enhanced progress tracking."""
if not uploaded_files:
st.warning("📤 Please upload at least one document.")
return False
try:
temp_dir = tempfile.mkdtemp()
file_paths = []
with st.status("📑 Processing Documents", expanded=True) as status:
# Save uploaded files
st.write("📁 Saving uploaded files...")
for uploaded_file in uploaded_files:
st.write(f"Saving {uploaded_file.name}...")
file_path = Path(temp_dir) / uploaded_file.name
with open(file_path, "wb") as f:
f.write(uploaded_file.getvalue())
file_paths.append(str(file_path))
# Process documents
st.write("🔄 Processing documents...")
processed_chunks = process_documents(
file_paths=file_paths,
chunker=chunker,
processor=processor,
output_path='./output.md'
)
# Ingest data
st.write("📥 Ingesting data to vector database...")
ingest_data(
processed_chunks=processed_chunks,
embedding_model=embedder,
pinecone_client=pc
)
# Setup retriever
st.write("🎯 Setting up retriever...")
st.session_state.retriever = PineconeRetriever(
pinecone_client=pc,
index_name="vector-index",
namespace="rag",
embedding_model=embedder,
llm_model=st.session_state.llm
)
st.session_state.documents_processed = True
status.update(label="✅ Processing complete!", state="complete", expanded=False)
return True
except Exception as e:
st.error(f"❌ Error processing documents: {str(e)}")
return False
finally:
# Cleanup
for file_path in file_paths:
try:
os.remove(file_path)
except Exception:
pass
try:
os.rmdir(temp_dir)
except Exception:
pass
def main():
st.title("📚 Table RAG Assistant")
st.markdown("---")
pc = None
# Sidebar Configuration with improved styling
with st.sidebar:
st.title("⚙️ Configuration")
pinecone_api_key = st.text_input("🔑 Enter Pinecone API Key:", type="password")
st.markdown("---")
col1, col2 = st.columns(2)
with col1:
if st.button("🧹 Clear DB", use_container_width=True):
clear_pinecone_index(pc)
with col2:
if st.button("🗑️ Clear Chat", use_container_width=True):
st.session_state.messages = []
st.session_state.llm.clear_history()
st.rerun()
# Display uploaded files
if st.session_state.uploaded_files:
st.markdown("---")
st.subheader("📁 Uploaded Files")
for file in st.session_state.uploaded_files:
st.write(f"- {file.name}")
pc = None
if not pinecone_api_key:
st.sidebar.warning("⚠️ Please enter Pinecone API key to continue.")
st.stop()
# Initialize components if not already done
if st.session_state.retriever is None:
pc, llm, embedder, chunker, processor = initialize_components(pinecone_api_key)
clear_pinecone_index(pc)
if None in (pc, llm, embedder, chunker, processor):
st.stop()
# Document Upload Section with improved UI
if not st.session_state.documents_processed:
st.header("📄 Document Upload")
st.markdown("Upload your documents to get started. Supported formats: PDF, DOCX, TXT, CSV, XLSX")
uploaded_files = st.file_uploader(
"Drop your files here",
accept_multiple_files=True,
type=["pdf", "docx", "txt", "csv", "xlsx"]
)
if uploaded_files:
st.session_state.uploaded_files = uploaded_files
if st.button("🚀 Process Documents", use_container_width=True):
if process_all_documents(uploaded_files, chunker, processor, pc, embedder):
st.success("✨ Documents processed successfully!")
# Enhanced Chat Interface with Simulated Streaming
if st.session_state.documents_processed:
st.header("💬 Chat Interface")
st.markdown("Ask questions about your documents and tables")
# Display chat history with improved styling
for message in st.session_state.messages:
with st.chat_message(message["role"]):
st.markdown(format_chat_message(message, message.get("results")))
# Chat input with simulated streaming
if prompt := st.chat_input("Ask a question..."):
# Display user message
with st.chat_message("user"):
st.markdown(prompt)
st.session_state.messages.append({"role": "user", "content": prompt})
# Generate response with simulated streaming
with st.chat_message("assistant"):
response_placeholder = st.empty()
with st.spinner("🤔 Thinking..."):
# Retrieve relevant content
results = st.session_state.retriever.invoke(
question=prompt,
top_k=3
)
# Format context and get response from LLM
context = format_context(results)
chat = st.session_state.llm
input_vars = {
"question": prompt,
"context": context
}
# Get full response first
full_response = chat.chat_with_template(RAG_TEMPLATE, input_vars)
# Simulate streaming of the response
for partial_response in simulate_streaming_response(full_response):
response_placeholder.markdown(partial_response + "▌")
# Display final response with tables
formatted_response = format_chat_message(
{"role": "assistant", "content": full_response},
results
)
response_placeholder.markdown(formatted_response)
# Save to chat history
st.session_state.messages.append({
"role": "assistant",
"content": full_response,
"results": results
})
if __name__ == "__main__":
main() |