src/embeddings.py

"""
embeddings.py — single source of the embedding function used by BOTH ingest.py
and search.py (they must match, or query vectors won't align with the index).

Uses Chroma's built-in ONNX all-MiniLM-L6-v2 (onnxruntime). This deliberately
avoids torch / sentence-transformers, whose unsigned DLLs get blocked by Windows
App Control on managed devices. onnxruntime's DLLs are signed and pass.
"""

from chromadb.utils import embedding_functions

_ef = None


def get_embedding_function():
    """Return a shared ONNX MiniLM embedding function (cached)."""
    global _ef
    if _ef is None:
        # DefaultEmbeddingFunction == ONNXMiniLM_L6_V2 in current chromadb.
        _ef = embedding_functions.DefaultEmbeddingFunction()
    return _ef