Class HashEmbedding

paperflow/providers/embedding.py:127–153

Deterministic offline backend. Used for tests and when no real provider is configured. Embeddings are not semantically meaningful, but they are stable and unit-norm.

class HashEmbedding:
    """Deterministic offline backend.

    Used for tests and when no real provider is configured. Embeddings are
    not semantically meaningful, but they are stable and unit-norm.
    """

    name = "hash"

    def __init__(self, dimensions: int = 768, model: str = "hash") -> None:
        self.model = model
        self.dimensions = dimensions

    def embed(self, text: str) -> List[float]:
        digest = hashlib.sha256((text or "").encode("utf-8")).digest()
        vector: List[float] = []
        for index in range(self.dimensions):
            byte_index = index % len(digest)
            bit = (digest[byte_index] >> (index % 8)) & 1
            vector.append(1.0 if bit else -1.0)
        norm = sum(value * value for value in vector) ** 0.5
        if norm > 0:
            vector = [value / norm for value in vector]
        return vector

    def embed_batch(self, texts: Iterable[str]) -> List[List[float]]:
        return [self.embed(text) for text in texts]


def _is_placeholder(value: Optional[str]) -> bool: