Function index_stats

tools/kb_semantic.py:673–726

Return a status dict for display in /status or setup output.

()

def index_stats() -> dict:
    """Return a status dict for display in /status or setup output."""
    chunk_dir = KB_ROOT / "embeddings"
    if not chunk_dir.exists():
        return {
            "chunk_files": 0, "total_chunks": 0,
            "has_semantic": False, "backend": "none",
            "kb_root": str(KB_ROOT),
        }

    chunk_files = list(chunk_dir.glob("*.chunks.json"))
    total = 0
    for cf in chunk_files:
        try:
            with open(cf) as f:
                total += len(json.load(f))
        except Exception:
            pass

    has_vec = (chunk_dir / "vectors.npy").exists()
    meta = {}
    mep = chunk_dir / "vectors.meta.json"
    if mep.exists():
        try:
            with open(mep) as f:
                meta = json.load(f)
        except Exception:
            pass

    index_backend = meta.get("backend", "unknown") if has_vec else None
    dim = meta.get("dim", "?") if has_vec else None

    if has_vec and index_backend == "llama-server":
        backend = f"hybrid BM25 + llama-embeddings ({dim}d, RRF)"
    elif has_vec and index_backend == "fastembed":
        backend = f"hybrid BM25 + fastembed ({dim}d, RRF)"
    elif has_vec and index_backend == "sentence-transformers":
        backend = f"hybrid BM25 + sentence-transformers ({dim}d, RRF)"
    elif HAS_FASTEMBED:
        backend = "fastembed available (run build_semantic_index)"
    elif HAS_SENTENCE_TRANSFORMERS:
        backend = "sentence-transformers available (run build_semantic_index)"
    else:
        backend = "BM25 keyword (no vector index — run setup_skills.sh with llama-server active)"

    return {
        "chunk_files":            len(chunk_files),
        "total_chunks":           total,
        "has_semantic":           has_vec,
        "backend":                backend,
        "llama_embed_available":  check_llama_embeddings(),
        "sentence_transformers":  HAS_SENTENCE_TRANSFORMERS,
        "kb_root":                str(KB_ROOT),
    }