from dataclasses import dataclass, field
from pathlib import Path


@dataclass
class Config:
    # Crawl settings
    default_url: str = "https://example.com"
    default_depth: int = 2
    default_max_pages: int = 50
    max_depth: int = 5
    max_pages_limit: int = 500

    # Chunking settings
    chunk_max_chars: int = 800
    chunk_overlap: int = 100

    # Model settings
    embedding_model: str = "all-MiniLM-L6-v2"
    embedding_model_fallback: str = "BAAI/bge-small-en-v1.5"
    reranker_model: str = "cross-encoder/ms-marco-MiniLM-L-6-v2"
    embedding_dim: int = 384

    # Local LLM settings
    llm_model: str = "Qwen/Qwen2.5-0.5B-Instruct"
    llm_max_tokens: int = 1024
    llm_temperature: float = 0.1

    # Retrieval settings
    top_k_dense: int = 20
    top_k_bm25: int = 20
    top_k_rerank: int = 10
    rrf_k: int = 60

    # Storage settings
    data_dir: Path = field(default_factory=lambda: Path("data"))
    dataset_repo_id: str = "chatbot-data"

    # Paths
    pages_path: str = "pages.parquet"
    chunks_path: str = "chunks.parquet"
    facts_path: str = "facts.parquet"
    faiss_index_path: str = "faiss.index"
    bm25_index_path: str = "bm25_index.pkl"
    metadata_path: str = "metadata.json"

    @property
    def pages_file(self) -> Path:
        return self.data_dir / self.pages_path

    @property
    def chunks_file(self) -> Path:
        return self.data_dir / self.chunks_path

    @property
    def facts_file(self) -> Path:
        return self.data_dir / self.facts_path

    @property
    def faiss_file(self) -> Path:
        return self.data_dir / self.faiss_index_path

    @property
    def bm25_file(self) -> Path:
        return self.data_dir / self.bm25_index_path

    @property
    def metadata_file(self) -> Path:
        return self.data_dir / self.metadata_path