Spaces:
Sleeping
Sleeping
File size: 28,224 Bytes
f84c6e7 3326b9e |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 |
import gradio as gr
import os
from llama_index.core import Settings
from documents_prep import load_json_documents, load_table_documents, load_image_documents
from utils import get_llm_model, get_embedding_model, get_reranker_model, answer_question
from my_logging import log_message
from index_retriever import create_vector_index, create_query_engine
import sys
from config import (
HF_REPO_ID, HF_TOKEN, DOWNLOAD_DIR, CHUNKS_FILENAME,
JSON_FILES_DIR, TABLE_DATA_DIR, IMAGE_DATA_DIR, DEFAULT_MODEL, AVAILABLE_MODELS
)
def merge_table_chunks(chunk_info):
merged = {}
for chunk in chunk_info:
doc_type = chunk.get('type', 'text')
doc_id = chunk.get('document_id', 'unknown')
if doc_type == 'table' or doc_type == 'table_row':
table_num = chunk.get('table_number', '')
key = f"{doc_id}_{table_num}"
if key not in merged:
merged[key] = {
'document_id': doc_id,
'type': 'table',
'table_number': table_num,
'section_id': chunk.get('section_id', 'unknown'),
'chunk_text': chunk.get('chunk_text', '')
}
else:
merged[key]['chunk_text'] += '\n' + chunk.get('chunk_text', '')
else:
unique_key = f"{doc_id}_{chunk.get('section_id', '')}_{chunk.get('chunk_id', 0)}"
merged[unique_key] = chunk
return list(merged.values())
def create_chunks_display_html(chunk_info):
if not chunk_info:
return "<div style='padding: 20px; text-align: center; color: black;'>Нет данных о чанках</div>"
merged_chunks = merge_table_chunks(chunk_info)
html = "<div style='max-height: 500px; overflow-y: auto; padding: 10px; color: black;'>"
html += f"<h4 style='color: black;'>Найдено релевантных чанков: {len(merged_chunks)}</h4>"
for i, chunk in enumerate(merged_chunks):
bg_color = "#f8f9fa" if i % 2 == 0 else "#e9ecef"
section_display = get_section_display(chunk)
formatted_content = get_formatted_content(chunk)
html += f"""
<div style='background-color: {bg_color}; padding: 10px; margin: 5px 0; border-radius: 5px; border-left: 4px solid #007bff; color: black;'>
<strong style='color: black;'>Документ:</strong> <span style='color: black;'>{chunk['document_id']}</span><br>
<strong style='color: black;'>Раздел:</strong> <span style='color: black;'>{section_display}</span><br>
<strong style='color: black;'>Содержание:</strong><br>
<div style='background-color: white; padding: 8px; margin-top: 5px; border-radius: 3px; font-family: monospace; font-size: 12px; color: black; max-height: 200px; overflow-y: auto;'>
{formatted_content}
</div>
</div>
"""
html += "</div>"
return html
def get_section_display(chunk):
section_path = chunk.get('section_path', '')
section_id = chunk.get('section_id', 'unknown')
doc_type = chunk.get('type', 'text')
if doc_type == 'table' and chunk.get('table_number'):
table_num = chunk.get('table_number')
if not str(table_num).startswith('№'):
table_num = f"№{table_num}"
return f"таблица {table_num}"
if doc_type == 'image' and chunk.get('image_number'):
image_num = chunk.get('image_number')
if not str(image_num).startswith('№'):
image_num = f"№{image_num}"
return f"рисунок {image_num}"
if section_path:
return section_path
elif section_id and section_id != 'unknown':
return section_id
return section_id
def get_formatted_content(chunk):
document_id = chunk.get('document_id', 'unknown')
section_path = chunk.get('section_path', '')
section_id = chunk.get('section_id', 'unknown')
section_text = chunk.get('section_text', '')
parent_section = chunk.get('parent_section', '')
parent_title = chunk.get('parent_title', '')
level = chunk.get('level', '')
chunk_text = chunk.get('chunk_text', '')
doc_type = chunk.get('type', 'text')
# For text documents
if level in ['subsection', 'sub_subsection', 'sub_sub_subsection'] and parent_section:
current_section = section_path if section_path else section_id
parent_info = f"{parent_section} ({parent_title})" if parent_title else parent_section
return f"В разделе {parent_info} в документе {document_id}, пункт {current_section}: {chunk_text}"
else:
current_section = section_path if section_path else section_id
clean_text = chunk_text
if section_text and chunk_text.startswith(section_text):
section_title = section_text
elif chunk_text.startswith(f"{current_section} "):
clean_text = chunk_text[len(f"{current_section} "):].strip()
section_title = section_text if section_text else f"{current_section} {clean_text.split('.')[0] if '.' in clean_text else clean_text[:50]}"
else:
section_title = section_text if section_text else current_section
return f"В разделе {current_section} в документе {document_id}, пункт {section_title}: {clean_text}"
def initialize_system(repo_id, hf_token, download_dir, chunks_filename=None,
json_files_dir=None, table_data_dir=None, image_data_dir=None,
use_json_instead_csv=False):
try:
log_message("Инициализация системы")
os.makedirs(download_dir, exist_ok=True)
from config import CHUNK_SIZE, CHUNK_OVERLAP
from llama_index.core.text_splitter import TokenTextSplitter
embed_model = get_embedding_model()
llm = get_llm_model(DEFAULT_MODEL)
reranker = get_reranker_model()
Settings.embed_model = embed_model
Settings.llm = llm
Settings.text_splitter = TokenTextSplitter(
chunk_size=CHUNK_SIZE,
chunk_overlap=CHUNK_OVERLAP,
separator=" ",
backup_separators=["\n", ".", "!", "?"]
)
log_message(f"Configured chunk size: {CHUNK_SIZE} tokens")
log_message(f"Configured chunk overlap: {CHUNK_OVERLAP} tokens")
all_documents = []
chunks_df = None
# CHANGED: Use load_all_documents instead of loading separately
if use_json_instead_csv and json_files_dir:
log_message("Используем JSON файлы вместо CSV")
from documents_prep import load_all_documents
# This will handle text, tables, and images all together with proper logging
all_documents = load_all_documents(
repo_id=repo_id,
hf_token=hf_token,
json_dir=json_files_dir,
table_dir=table_data_dir if table_data_dir else "",
image_dir=image_data_dir if image_data_dir else ""
)
else:
# OLD PATH: Loading separately (fallback)
if chunks_filename:
log_message("Загружаем данные из CSV")
if table_data_dir:
log_message("Добавляю табличные данные")
from documents_prep import load_table_documents
table_chunks = load_table_documents(repo_id, hf_token, table_data_dir)
log_message(f"Загружено {len(table_chunks)} табличных чанков")
all_documents.extend(table_chunks)
if image_data_dir:
log_message("Добавляю данные изображений")
from documents_prep import load_image_documents
image_documents = load_image_documents(repo_id, hf_token, image_data_dir)
log_message(f"Загружено {len(image_documents)} документов изображений")
all_documents.extend(image_documents)
log_message(f"Всего документов после всей обработки: {len(all_documents)}")
vector_index = create_vector_index(all_documents)
query_engine = create_query_engine(vector_index)
# Create chunk_info for display (extract from documents metadata)
chunk_info = []
for doc in all_documents:
chunk_info.append({
'document_id': doc.metadata.get('document_id', 'unknown'),
'section_id': doc.metadata.get('section_id', 'unknown'),
'type': doc.metadata.get('type', 'text'),
'chunk_text': doc.text[:200] + '...' if len(doc.text) > 200 else doc.text,
'table_number': doc.metadata.get('table_number', ''),
'image_number': doc.metadata.get('image_number', ''),
'section': doc.metadata.get('section', ''),
'connection_type': doc.metadata.get('connection_type', '') # ADD THIS
})
log_message(f"Система успешно инициализирована")
return query_engine, chunks_df, reranker, vector_index, chunk_info
except Exception as e:
log_message(f"Ошибка инициализации: {str(e)}")
import traceback
log_message(traceback.format_exc())
return None, None, None, None, []
def switch_model(model_name, vector_index):
from llama_index.core import Settings
from index_retriever import create_query_engine
try:
log_message(f"Переключение на модель: {model_name}")
new_llm = get_llm_model(model_name)
Settings.llm = new_llm
if vector_index is not None:
new_query_engine = create_query_engine(vector_index)
log_message(f"Модель успешно переключена на: {model_name}")
return new_query_engine, f"✅ Модель переключена на: {model_name}"
else:
return None, "❌ Ошибка: система не инициализирована"
except Exception as e:
error_msg = f"Ошибка переключения модели: {str(e)}"
log_message(error_msg)
return None, f"❌ {error_msg}"
# Add these global variables near the top with other globals
retrieval_params = {
'vector_top_k': 50,
'bm25_top_k': 50,
'similarity_cutoff': 0.55,
'hybrid_top_k': 100,
'rerank_top_k': 20
}
# MODIFIED: Update create_query_engine call signature
def create_query_engine(vector_index, vector_top_k=50, bm25_top_k=50,
similarity_cutoff=0.55, hybrid_top_k=100):
try:
from config import CUSTOM_PROMPT
from index_retriever import create_query_engine as create_index_query_engine
# Pass parameters to the index_retriever function
query_engine = create_index_query_engine(
vector_index=vector_index,
vector_top_k=vector_top_k,
bm25_top_k=bm25_top_k,
similarity_cutoff=similarity_cutoff,
hybrid_top_k=hybrid_top_k
)
log_message(f"Query engine created with params: vector_top_k={vector_top_k}, "
f"bm25_top_k={bm25_top_k}, cutoff={similarity_cutoff}, hybrid_top_k={hybrid_top_k}")
return query_engine
except Exception as e:
log_message(f"Ошибка создания query engine: {str(e)}")
raise
# MODIFIED: Update answer_question to use global retrieval_params
def main_answer_question(question):
global query_engine, reranker, current_model, chunks_df, retrieval_params
if not question.strip():
return ("<div style='color: black;'>Пожалуйста, введите вопрос</div>",
"<div style='color: black;'>Источники появятся после обработки запроса</div>",
"<div style='color: black;'>Чанки появятся после обработки запроса</div>")
try:
answer_html, sources_html, chunks_html = answer_question(
question, query_engine, reranker, current_model, chunks_df,
rerank_top_k=retrieval_params['rerank_top_k']
)
return answer_html, sources_html, chunks_html
except Exception as e:
log_message(f"Ошибка при ответе на вопрос: {str(e)}")
return (f"<div style='color: red;'>Ошибка: {str(e)}</div>",
"<div style='color: black;'>Источники недоступны из-за ошибки</div>",
"<div style='color: black;'>Чанки недоступны из-за ошибки</div>")
# NEW: Function to update retrieval parameters and recreate query engine
def update_retrieval_params(vector_top_k, bm25_top_k, similarity_cutoff, hybrid_top_k, rerank_top_k):
global query_engine, vector_index, retrieval_params
try:
retrieval_params['vector_top_k'] = vector_top_k
retrieval_params['bm25_top_k'] = bm25_top_k
retrieval_params['similarity_cutoff'] = similarity_cutoff
retrieval_params['hybrid_top_k'] = hybrid_top_k
retrieval_params['rerank_top_k'] = rerank_top_k
# Recreate query engine with new parameters
if vector_index is not None:
query_engine = create_query_engine(
vector_index=vector_index,
vector_top_k=vector_top_k,
bm25_top_k=bm25_top_k,
similarity_cutoff=similarity_cutoff,
hybrid_top_k=hybrid_top_k
)
log_message(f"Параметры поиска обновлены: vector_top_k={vector_top_k}, "
f"bm25_top_k={bm25_top_k}, cutoff={similarity_cutoff}, "
f"hybrid_top_k={hybrid_top_k}, rerank_top_k={rerank_top_k}")
return f"✅ Параметры обновлены"
else:
return "❌ Система не инициализирована"
except Exception as e:
error_msg = f"Ошибка обновления параметров: {str(e)}"
log_message(error_msg)
return f"❌ {error_msg}"
def retrieve_chunks(question: str, top_k: int = 20) -> list:
from index_retriever import rerank_nodes
global query_engine, reranker
if query_engine is None:
return []
try:
retrieved_nodes = query_engine.retriever.retrieve(question)
log_message(f"Получено {len(retrieved_nodes)} узлов")
reranked_nodes = rerank_nodes(
question,
retrieved_nodes,
reranker,
top_k=top_k,
min_score_threshold=0.5
)
chunks_data = []
for i, node in enumerate(reranked_nodes):
metadata = node.metadata if hasattr(node, 'metadata') else {}
chunk = {
'rank': i + 1,
'document_id': metadata.get('document_id', 'unknown'),
'section_id': metadata.get('section_id', ''),
'section_path': metadata.get('section_path', ''),
'section_text': metadata.get('section_text', ''),
'type': metadata.get('type', 'text'),
'table_number': metadata.get('table_number', ''),
'image_number': metadata.get('image_number', ''),
'text': node.text
}
chunks_data.append(chunk)
log_message(f"Возвращено {len(chunks_data)} чанков")
return chunks_data
except Exception as e:
log_message(f"Ошибка получения чанков: {str(e)}")
return []
def create_demo_interface(answer_question_func, switch_model_func, current_model, chunk_info=None):
with gr.Blocks(title="AIEXP - AI Expert для нормативной документации", theme=gr.themes.Soft()) as demo:
gr.api(retrieve_chunks, api_name="retrieve_chunks")
gr.Markdown("""
# AIEXP - Artificial Intelligence Expert
## Инструмент для работы с нормативной документацией
""")
with gr.Tab("Поиск по нормативным документам"):
gr.Markdown("### Задайте вопрос по нормативной документации")
with gr.Row():
with gr.Column(scale=2):
model_dropdown = gr.Dropdown(
choices=list(AVAILABLE_MODELS.keys()),
value=current_model,
label="Выберите языковую модель",
info="Выберите модель для генерации ответов"
)
with gr.Column(scale=1):
switch_btn = gr.Button("Переключить модель", variant="secondary")
model_status = gr.Textbox(
value=f"Текущая модель: {current_model}",
label="Статус модели",
interactive=False
)
with gr.Row():
with gr.Column(scale=3):
question_input = gr.Textbox(
label="Ваш вопрос к базе знаний",
placeholder="Введите вопрос по нормативным документам...",
lines=3
)
ask_btn = gr.Button("Найти ответ", variant="primary", size="lg")
gr.Examples(
examples=[
"О чем этот рисунок: ГОСТ Р 50.04.07-2022 Приложение Л. Л.1.5 Рисунок Л.2",
"Л.9 Формула в ГОСТ Р 50.04.07 - 2022 что и о чем там?",
"Какой стандарт устанавливает порядок признания протоколов испытаний продукции в области использования атомной энергии?",
"Кто несет ответственность за организацию и проведение признания протоколов испытаний продукции?",
"В каких случаях могут быть признаны протоколы испытаний, проведенные лабораториями?",
"В какой таблице можно найти информацию о методы исследований при аттестационных испытаниях технологии термической обработки заготовок из легированных сталей? Какой документ и какой раздел?"
],
inputs=question_input
)
with gr.Row():
with gr.Column(scale=2):
answer_output = gr.HTML(
label="",
value=f"<div style='background-color: #2d3748; color: white; padding: 20px; border-radius: 10px; text-align: center;'>Здесь появится ответ на ваш вопрос...<br><small>Текущая модель: {current_model}</small></div>",
)
with gr.Column(scale=1):
sources_output = gr.HTML(
label="",
value="<div style='background-color: #2d3748; color: white; padding: 20px; border-radius: 10px; text-align: center;'>Здесь появятся релевантные чанки...</div>",
)
with gr.Column(scale=1):
chunks_output = gr.HTML(
label="Релевантные чанки",
value="<div style='background-color: #2d3748; color: white; padding: 20px; border-radius: 10px; text-align: center;'>Здесь появятся релевантные чанки...</div>",
)
# NEW TAB: Retrieval Parameters
with gr.Tab("⚙️ Параметры поиска"):
gr.Markdown("### Настройка параметров векторного поиска и переранжирования")
with gr.Row():
with gr.Column():
vector_top_k = gr.Slider(
minimum=10,
maximum=200,
value=50,
step=10,
label="Vector Top K",
info="Количество результатов из векторного поиска"
)
with gr.Column():
bm25_top_k = gr.Slider(
minimum=10,
maximum=200,
value=50,
step=10,
label="BM25 Top K",
info="Количество результатов из BM25 поиска"
)
with gr.Row():
with gr.Column():
similarity_cutoff = gr.Slider(
minimum=0.0,
maximum=1.0,
value=0.55,
step=0.05,
label="Similarity Cutoff",
info="Минимальный порог схожести для векторного поиска"
)
with gr.Column():
hybrid_top_k = gr.Slider(
minimum=10,
maximum=300,
value=100,
step=10,
label="Hybrid Top K",
info="Количество результатов из гибридного поиска"
)
with gr.Row():
with gr.Column():
rerank_top_k = gr.Slider(
minimum=5,
maximum=100,
value=20,
step=5,
label="Rerank Top K",
info="Количество результатов после переранжирования"
)
with gr.Column():
update_btn = gr.Button("Применить параметры", variant="primary")
update_status = gr.Textbox(
value="Параметры готовы к применению",
label="Статус",
interactive=False
)
gr.Markdown("""
### Рекомендации:
- **Vector Top K**: Увеличьте для более полного поиска по семантике (50-100)
- **BM25 Top K**: Увеличьте для лучшего поиска по ключевым словам (30-80)
- **Similarity Cutoff**: Снизьте для более мягких критериев (0.3-0.6), повысьте для строгих (0.7-0.9)
- **Hybrid Top K**: Объединённые результаты (100-150)
- **Rerank Top K**: Финальные результаты (10-30)
""")
update_btn.click(
fn=update_retrieval_params,
inputs=[vector_top_k, bm25_top_k, similarity_cutoff, hybrid_top_k, rerank_top_k],
outputs=[update_status]
)
# Display current parameters
gr.Markdown("### Текущие параметры:")
current_params_display = gr.Textbox(
value="Vector: 50 | BM25: 50 | Cutoff: 0.55 | Hybrid: 100 | Rerank: 20",
label="",
interactive=False,
lines=2
)
def display_current_params():
return f"""Vector Top K: {retrieval_params['vector_top_k']}
BM25 Top K: {retrieval_params['bm25_top_k']}
Similarity Cutoff: {retrieval_params['similarity_cutoff']}
Hybrid Top K: {retrieval_params['hybrid_top_k']}
Rerank Top K: {retrieval_params['rerank_top_k']}"""
# Refresh params display on tab change
demo.load(
fn=display_current_params,
outputs=[current_params_display]
)
update_btn.click(
fn=display_current_params,
outputs=[current_params_display]
)
# Original tab logic
switch_btn.click(
fn=switch_model_func,
inputs=[model_dropdown],
outputs=[model_status]
)
ask_btn.click(
fn=answer_question_func,
inputs=[question_input],
outputs=[answer_output, sources_output, chunks_output]
)
question_input.submit(
fn=answer_question_func,
inputs=[question_input],
outputs=[answer_output, sources_output, chunks_output]
)
return demo
query_engine = None
chunks_df = None
reranker = None
vector_index = None
current_model = DEFAULT_MODEL
def main_switch_model(model_name):
global query_engine, vector_index, current_model
new_query_engine, status_message = switch_model(model_name, vector_index)
if new_query_engine:
query_engine = new_query_engine
current_model = model_name
return status_message
def main():
global query_engine, chunks_df, reranker, vector_index, current_model
GOOGLE_API_KEY = os.getenv("GOOGLE_API_KEY", "")
if GOOGLE_API_KEY:
log_message("Использование Google API для модели генерации текста")
else:
log_message("Google API ключ не найден, использование локальной модели")
log_message("Запуск AIEXP - AI Expert для нормативной документации")
query_engine, chunks_df, reranker, vector_index, chunk_info = initialize_system(
repo_id=HF_REPO_ID,
hf_token=HF_TOKEN,
download_dir=DOWNLOAD_DIR,
json_files_dir=JSON_FILES_DIR,
table_data_dir=TABLE_DATA_DIR,
image_data_dir=IMAGE_DATA_DIR,
use_json_instead_csv=True,
)
if query_engine:
log_message("Запуск веб-интерфейса")
demo = create_demo_interface(
answer_question_func=main_answer_question,
switch_model_func=main_switch_model,
current_model=current_model,
chunk_info=chunk_info
)
demo.api = "retrieve_chunks"
demo.queue()
demo.launch(
server_name="0.0.0.0",
server_port=7860,
share=True,
debug=False
)
else:
log_message("Невозможно запустить приложение из-за ошибки инициализации")
sys.exit(1)
if __name__ == "__main__":
main() |