47 lines
1.5 KiB
Python
47 lines
1.5 KiB
Python
"""Загрузка документов из директории в векторное хранилище."""
|
|
from __future__ import annotations
|
|
|
|
import argparse
|
|
from pathlib import Path
|
|
|
|
from knowledge_base import KnowledgeBase
|
|
|
|
SUPPORTED_SUFFIXES = {".txt", ".md", ".markdown"}
|
|
|
|
|
|
def load_directory(knowledge_dir: Path, kb: KnowledgeBase | None = None) -> int:
|
|
kb = kb or KnowledgeBase()
|
|
if not knowledge_dir.exists():
|
|
knowledge_dir.mkdir(parents=True, exist_ok=True)
|
|
print(f"Создана пустая директория: {knowledge_dir}")
|
|
return 0
|
|
|
|
total_chunks = 0
|
|
files = [p for p in knowledge_dir.rglob("*") if p.suffix.lower() in SUPPORTED_SUFFIXES]
|
|
if not files:
|
|
print(f"Нет файлов {SUPPORTED_SUFFIXES} в {knowledge_dir}")
|
|
return 0
|
|
|
|
for file_path in sorted(files):
|
|
chunks = kb.add_file(file_path)
|
|
total_chunks += chunks
|
|
print(f" + {file_path.name}: {chunks} чанк(ов)")
|
|
|
|
print(f"Готово: {len(files)} файл(ов), {total_chunks} чанк(ов).")
|
|
return total_chunks
|
|
|
|
|
|
def main() -> None:
|
|
parser = argparse.ArgumentParser(description="Инициализация базы знаний из папки")
|
|
parser.add_argument(
|
|
"--dir",
|
|
default="./knowledge",
|
|
help="Путь к директории с документами (.txt, .md)",
|
|
)
|
|
args = parser.parse_args()
|
|
load_directory(Path(args.dir))
|
|
|
|
|
|
if __name__ == "__main__":
|
|
main()
|