support-bot-saas/backend/app/services/chunking.py
ed0ss 0063df0a87
Some checks are pending
CI / lint-backend (push) Waiting to run
CI / lint-frontend (push) Waiting to run
CI / build-backend (push) Blocked by required conditions
CI / build-frontend (push) Blocked by required conditions
Реструктуризация проекта в SaaS: FastAPI бэкенд, React SPA фронтенд, Helm-чарт, CI/CD
- Бэкенд: FastAPI + SQLAlchemy async + PostgreSQL, JWT-аутентификация,
  Qdrant multi-tenant векторное хранилище, Celery воркер
- Фронтенд: React + Vite + TypeScript SPA (логин, регистрация,
  дашборд, поиск по коллекциям, настройки)
- Инфраструктура: docker-compose, Helm-чарт, GitHub Actions CI,
  Prometheus/Grafana мониторинг
- Embedding через OpenRouter (nvidia/llama-nemotron-embed-vl-1b-v2:free)
- LLM: google/gemma-4-31b-it:free через OpenRouter
2026-06-20 17:40:27 +03:00

83 lines
2.4 KiB
Python
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

import re
from typing import Optional
NON_SOLUTION = re.compile(
r"(уточнит|приложите скрин|какая ошибка|с какой проблемой"
r"|нет обратной связи|запрос завершу|открыть его снова"
r"|откройте новый|обратиться в службу поддержки"
r"|свяжитесь с технической"
r"|напишите нам|позвоните нам)", re.I
)
def clean_text(text: str) -> str:
text = re.sub(r'\[/?q\].*?\[/q\]', '', text, flags=re.DOTALL | re.IGNORECASE)
text = re.sub(r'https?://\S+', '', text)
text = re.sub(r'\s+', ' ', text).strip()
return text
def has_real_solution(messages: list[dict]) -> Optional[str]:
support_texts = []
for m in messages:
if m["role"] == "support":
cleaned = clean_text(m["text"])
if cleaned:
support_texts.append(cleaned)
if not support_texts:
return None
combined = "\n".join(support_texts)
if len(combined) < 100:
return None
last = support_texts[-1]
if NON_SOLUTION.search(last):
return None
if last.strip().endswith("?") and len(last) < 150:
return None
return combined
def get_client_dialogue(messages: list[dict]) -> str:
texts = []
for m in messages:
if m["role"] == "client":
cleaned = clean_text(m["text"])
if cleaned:
texts.append(cleaned)
return "\n".join(texts)
def build_chunk(ticket: dict) -> Optional[dict]:
msgs = ticket.get("messages") or []
if not msgs:
return None
description = clean_text(ticket.get("description", ""))
solution = has_real_solution(msgs)
if not solution:
return None
client_msgs = get_client_dialogue(msgs)
search_parts = []
if description:
search_parts.append(description)
if client_msgs:
search_parts.append(client_msgs)
search_text = "\n".join(search_parts).strip()
if not search_text:
return None
full_text = (
f"Категория: {ticket['category']}\n"
f"Проблема: {description}\n"
f"Решение: {solution}"
)
return {
"ticket_id": ticket["ticket_id"],
"client": ticket["client"],
"category": ticket["category"],
"search_text": search_text,
"full_text": full_text,
"tenant_id": None,
}