Spaces:
Sleeping
Sleeping
File size: 8,897 Bytes
1580a2d cd201bf 1580a2d | 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 | import os
import fitz
import pandas as pd
from pathlib import Path
from llama_index.core import Document, VectorStoreIndex
from llama_index.embeddings.huggingface import HuggingFaceEmbedding
from llama_index.core.query_engine import RetrieverQueryEngine
from llama_index.core.retrievers import VectorIndexRetriever
from llama_index.core.response_synthesizers import get_response_synthesizer, ResponseMode
from llama_index.core.prompts import PromptTemplate
from config import *
def log_message(message):
print(message, flush=True)
def extract_text_from_pdf(file_path):
doc = fitz.open(file_path)
text = ""
for page in doc:
text += page.get_text()
doc.close()
return text
def extract_text_from_txt(file_path):
with open(file_path, 'r', encoding='utf-8') as file:
return file.read()
def chunk_text(text, chunk_size=CHUNK_SIZE, chunk_overlap=CHUNK_OVERLAP):
log_message(f"📄 Chunking text into pieces of {chunk_size} characters...")
chunks = []
start = 0
while start < len(text):
end = start + chunk_size
chunk = text[start:end]
chunks.append(chunk)
start = end - chunk_overlap
log_message(f"✅ Created {len(chunks)} chunks")
return chunks
def process_uploaded_file(file_path, file_name, doc_name, doc_link):
try:
log_message(f"🔄 Processing file: {file_name}")
file_extension = Path(file_path).suffix.lower()
if file_extension == '.pdf':
log_message("📖 Extracting text from PDF...")
text = extract_text_from_pdf(file_path)
elif file_extension == '.txt':
log_message("📝 Reading text file...")
text = extract_text_from_txt(file_path)
else:
return None, "Unsupported file type"
word_count = len(text.split())
log_message(f"📊 Extracted {word_count} words from document")
chunks = chunk_text(text)
return {
'document': doc_name,
'file_name': file_name,
'doc_link': doc_link,
'total_words': word_count,
'extracted_text': text,
'chunks': chunks
}, None
except Exception as e:
log_message(f"❌ Error processing file: {str(e)}")
return None, str(e)
def get_existing_documents():
try:
upload_dir = "UPLOADED_DOCUMENTS"
if not os.path.exists(upload_dir):
return []
documents = []
for file_name in os.listdir(upload_dir):
if file_name.endswith(('.txt', '.pdf')):
doc_name = os.path.splitext(file_name)[0]
documents.append(doc_name)
return sorted(documents)
except Exception as e:
log_message(f"❌ Error reading documents: {str(e)}")
return []
def add_to_vector_index(new_chunks, file_info, existing_chunks_df=None):
try:
log_message("🔧 Setting up embedding model...")
embed_model = HuggingFaceEmbedding(model_name=EMBEDDING_MODEL)
log_message("📝 Creating document objects...")
new_documents = []
new_chunk_data = []
for i, chunk in enumerate(new_chunks):
doc_id = f"{file_info['file_name']}_{i}"
new_documents.append(Document(
text=chunk,
metadata={
"chunk_id": doc_id,
"document_id": file_info['file_name'],
"document_name": file_info['document'],
"document_link": file_info['doc_link']
}
))
new_chunk_data.append({
'chunk_id': doc_id,
'document_id': file_info['file_name'],
'document_name': file_info['document'],
'document_link': file_info['doc_link'],
'chunk_text': chunk
})
if existing_chunks_df is not None:
log_message("🔄 Merging with existing chunks...")
new_chunks_df = pd.DataFrame(new_chunk_data)
chunks_df = pd.concat([existing_chunks_df, new_chunks_df], ignore_index=True)
else:
chunks_df = pd.DataFrame(new_chunk_data)
log_message("🏗️ Building vector index...")
all_documents = [Document(text=str(row['chunk_text']),
metadata={
"chunk_id": row['chunk_id'],
"document_id": row['document_id'],
"document_name": row['document_name'],
"document_link": row['document_link']
})
for _, row in chunks_df.iterrows()]
vector_index = VectorStoreIndex.from_documents(all_documents, embed_model=embed_model)
log_message("🔍 Setting up retriever...")
retriever = VectorIndexRetriever(
index=vector_index,
similarity_top_k=RETRIEVER_TOP_K,
similarity_cutoff=SIMILARITY_THRESHOLD
)
log_message("🎯 Configuring response synthesizer...")
custom_prompt_template = PromptTemplate(CUSTOM_PROMPT_NEW)
response_synthesizer = get_response_synthesizer(
response_mode=ResponseMode.TREE_SUMMARIZE,
text_qa_template=custom_prompt_template
)
query_engine = RetrieverQueryEngine(
retriever=retriever,
response_synthesizer=response_synthesizer
)
log_message("💾 Saving chunks to file...")
os.makedirs(download_dir, exist_ok=True)
chunks_df.to_csv(os.path.join(download_dir, chunks_filename), index=False)
log_message("✅ Successfully added document to vector index")
return query_engine, chunks_df, None
except Exception as e:
log_message(f"❌ Error adding to vector index: {str(e)}")
return None, existing_chunks_df, str(e)
def initialize_system():
try:
log_message("🔄 Initializing AI Expert system...")
os.makedirs(download_dir, exist_ok=True)
chunks_df = None
try:
chunks_csv_path = os.path.join(download_dir, chunks_filename)
if os.path.exists(chunks_csv_path):
log_message("📂 Loading existing chunks...")
chunks_df = pd.read_csv(chunks_csv_path)
else:
log_message("📝 Creating empty chunks database...")
chunks_df = pd.DataFrame(columns=['chunk_id', 'document_id', 'document_name', 'document_link', 'chunk_text'])
except Exception as e:
log_message(f"⚠️ Could not load chunks: {str(e)}")
chunks_df = pd.DataFrame(columns=['chunk_id', 'document_id', 'document_name', 'document_link', 'chunk_text'])
query_engine = None
if not chunks_df.empty:
log_message("🤖 Setting up AI models...")
embed_model = HuggingFaceEmbedding(model_name=EMBEDDING_MODEL)
log_message("📝 Creating document objects from existing chunks...")
documents = [Document(text=str(row['chunk_text']),
metadata={
"chunk_id": row.get('chunk_id', i),
"document_id": row.get('document_id', 'unknown'),
"document_name": row.get('document_name', 'unknown'),
"document_link": row.get('document_link', '')
})
for i, (_, row) in enumerate(chunks_df.iterrows())]
log_message("🔍 Building vector index...")
vector_index = VectorStoreIndex.from_documents(documents, embed_model=embed_model)
retriever = VectorIndexRetriever(
index=vector_index,
similarity_top_k=RETRIEVER_TOP_K,
similarity_cutoff=SIMILARITY_THRESHOLD
)
custom_prompt_template = PromptTemplate(CUSTOM_PROMPT)
response_synthesizer = get_response_synthesizer(
response_mode=ResponseMode.TREE_SUMMARIZE,
text_qa_template=custom_prompt_template
)
query_engine = RetrieverQueryEngine(
retriever=retriever,
response_synthesizer=response_synthesizer
)
log_message("✅ System successfully initialized!")
return query_engine, chunks_df, True
except Exception as e:
log_message(f"❌ Initialization error: {str(e)}")
return None, None, False |