import hashlib
import math
import os

from backend.core.config import settings


class HashEmbedder:
    """Small deterministic fallback embedder for local MVP retrieval."""

    def __init__(self, dimensions: int = 384):
        self.dimensions = dimensions

    def embed(self, text: str) -> list[float]:
        vector = [0.0] * self.dimensions
        for token in text.lower().split():
            digest = hashlib.sha1(token.encode("utf-8")).digest()
            idx = int.from_bytes(digest[:2], "big") % self.dimensions
            sign = 1.0 if digest[2] % 2 else -1.0
            vector[idx] += sign
        norm = math.sqrt(sum(value * value for value in vector)) or 1.0
        return [value / norm for value in vector]


class HuggingFaceEmbedder:
    def __init__(self, model_name: str, dimensions: int = 384):
        self.model_name = model_name
        self.dimensions = dimensions
        if settings.hf_api_key:
            os.environ["HUGGINGFACEHUB_API_TOKEN"] = settings.hf_api_key
        try:
            from sentence_transformers import SentenceTransformer
        except ImportError as exc:
            raise ImportError(
                "sentence-transformers is required for Hugging Face embeddings. Install it or set use_local_embeddings=True."
            ) from exc
        self.model = SentenceTransformer(model_name)

    def embed(self, text: str) -> list[float]:
        embeddings = self.model.encode([text], normalize_embeddings=True)
        return embeddings[0].tolist()


def _build_embedder() -> object:
    if settings.use_local_embeddings:
        return HashEmbedder(dimensions=settings.embedding_dimensions)

    try:
        return HuggingFaceEmbedder(settings.embedding_model, dimensions=settings.embedding_dimensions)
    except Exception:
        return HashEmbedder(dimensions=settings.embedding_dimensions)


embedder = _build_embedder()