Function build_merge

graphify/build.py:735–909

Load existing graph.json, merge new chunks into it, and save back. Re-extracted files REPLACE their prior contribution: any source_file present in new_chunks is dropped from the loaded graph before merging, so a changed file's stale nodes/edges don't accumulate. Files absent from new_ch

(
    new_chunks: list[dict],
    graph_path: str | Path | None = None,
    prune_sources: list[str] | None = None,
    *,
    directed: bool = False,
    dedup: bool = True,
    dedup_llm_backend: str | None = None,
    root: str | Path | None = None,
)

def build_merge(
    new_chunks: list[dict],
    graph_path: str | Path | None = None,
    prune_sources: list[str] | None = None,
    *,
    directed: bool = False,
    dedup: bool = True,
    dedup_llm_backend: str | None = None,
    root: str | Path | None = None,
) -> nx.Graph:
    """Load existing graph.json, merge new chunks into it, and save back.

    Re-extracted files REPLACE their prior contribution: any source_file present
    in new_chunks is dropped from the loaded graph before merging, so a changed
    file's stale nodes/edges don't accumulate. Files absent from new_chunks are
    preserved unchanged; deleted files are removed via prune_sources.
    Safe to call repeatedly.
    root: if given, absolute source_file paths in new_chunks are made relative (#932).
    """
    graph_path = Path(graph_path if graph_path is not None else _default_graph_json())
    if graph_path.exists():
        # Read JSON directly instead of going through node_link_graph().
        # The latter rebuilds an undirected nx.Graph and then enumerating
        # edges() yields endpoints based on node insertion order, which
        # silently flips directional edges (e.g. `calls`) when the callee
        # was inserted before the caller. The _src/_tgt direction-preserving
        # attrs are popped before saving in export.py, so going through the
        # NetworkX round-trip loses direction permanently (#760).
        from graphify.security import check_graph_file_size_cap
        check_graph_file_size_cap(graph_path)
        try:
            data = json.loads(graph_path.read_text(encoding="utf-8"))
        except (json.JSONDecodeError, OSError) as exc:
            raise RuntimeError(
                f"Cannot read {graph_path} for incremental merge: {exc}. "
                "Delete the file and run a full rebuild."
            ) from exc
        links_key = "links" if "links" in data else "edges"
        existing_nodes = list(data.get("nodes", []))
        existing_edges = list(data.get(links_key, []))
        existing_hyperedges = list(data.get("hyperedges", []))
        had_graph = True
    else:
        existing_nodes = []
        existing_edges = []
        existing_hyperedges = []
        had_graph = False

    # Effective root for relativizing absolute source_file / prune paths back to the
    # stored relative source_file keys. When the caller passes root we use it;
    # otherwise fall back to the graph's recorded scan root, so absolute
    # prune_sources and new-chunk paths still match even when a caller omits root
    # (#1571 — the skill's --update runbook calls build_merge without root, so
    # absolute deleted-file paths never matched the relative node keys and their
    # nodes survived as ghosts).
    _eff_root = (
        str(Path(root).resolve()) if root is not None
        else _infer_merge_root(graph_path)