Function embed_nodes_for_user

skills/wiki-store/scripts/wiki_db.py:352–403

Embed wiki nodes with the configured embedding provider.

(
    user_id: str,
    node_ids: Optional[List[str]] = None,
    *,
    force: bool = False,
    limit: int = 500,
)

def embed_nodes_for_user(
    user_id: str,
    node_ids: Optional[List[str]] = None,
    *,
    force: bool = False,
    limit: int = 500,
) -> Dict[str, Any]:
    """Embed wiki nodes with the configured embedding provider."""
    init_wiki_schema()
    from paperflow.providers import build_embedding_provider

    params: List[Any] = [user_id]
    where = "WHERE user_id = ?"
    if node_ids:
        placeholders = ",".join("?" for _ in node_ids)
        where += f" AND node_id IN ({placeholders})"
        params.extend(node_ids)
    if not force:
        where += " AND embedding IS NULL"
    params.append(max(1, int(limit)))

    conn = db_ops.get_connection()
    rows = conn.execute(
        f"""
        SELECT * FROM wiki_nodes
        {where}
        ORDER BY updated_at DESC, id DESC
        LIMIT ?
        """,
        params,
    ).fetchall()
    if not rows:
        conn.close()
        return {"embedded": 0, "model": None}

    nodes = _rows_to_nodes(rows)
    provider = build_embedding_provider()
    texts = [_node_embedding_text(node) for node in nodes]
    vectors = provider.embed_batch(texts)
    model_name = f"{provider.name}:{provider.model}"
    for node, vector in zip(nodes, vectors):
        conn.execute(
            """
            UPDATE wiki_nodes
            SET embedding = ?, embedding_model = ?, updated_at = ?
            WHERE user_id = ? AND node_id = ?
            """,
            (_vector_to_blob(vector), model_name, _now(), user_id, node["node_id"]),
        )
    conn.commit()
    conn.close()
    return {"embedded": len(nodes), "model": model_name}


def upsert_edge(