Fix source_file path lookup for explain and affected

graphify/affected.py

@@ -54,6 +54,41 @@ def _normalize_label(label: str) -> str:
     return unicodedata.normalize("NFC", label).casefold()
 
 
+def _prefer_file_node(
+    graph: nx.Graph,
+    node_ids: list[str],
+    query: str,
+) -> str | None:
+    """Return the file-level node when a source_file query matches many nodes."""
+    query_basename = _normalize_label(Path(query).name)
+    exact_file_nodes = [
+        node_id
+        for node_id in node_ids
+        if str(graph.nodes[node_id].get("source_location", "")) == "L1"
+        and _normalize_label(str(graph.nodes[node_id].get("label", ""))) == query_basename
+    ]
+    if len(exact_file_nodes) == 1:
+        return exact_file_nodes[0]
+
+    l1_nodes = [
+        node_id
+        for node_id in node_ids
+        if str(graph.nodes[node_id].get("source_location", "")) == "L1"
+    ]
+    if len(l1_nodes) == 1:
+        return l1_nodes[0]
+
+    basename_nodes = [
+        node_id
+        for node_id in node_ids
+        if _normalize_label(str(graph.nodes[node_id].get("label", ""))) == query_basename
+    ]
+    if len(basename_nodes) == 1:
+        return basename_nodes[0]
+
+    return None
+
+
 def resolve_seed(graph: nx.Graph, query: str) -> str | None:
     if query in graph:
         return query
@@ -83,6 +118,10 @@ def resolve_seed(graph: nx.Graph, query: str) -> str | None:
     ]
     if len(exact_source_matches) == 1:
         return exact_source_matches[0]
+    if exact_source_matches:
+        preferred_file_node = _prefer_file_node(graph, exact_source_matches, query)
+        if preferred_file_node is not None:
+            return preferred_file_node
     contains_matches = [
         str(node_id)
         for node_id, data in graph.nodes(data=True)

graphify/serve.py

@@ -152,6 +152,8 @@ def _node_search_text(data: dict, nid: str) -> str:
       `term in label_tokens` branch, where a multi-word `term` can span a token
       boundary that punctuation hides in `norm_label` (e.g. query "foo bar" matches
       label "foo.bar" only via its tokenized form).
+    - `source_tokens` feeds _find_node's exact source-file path lookup, where a
+      query like "app/api/example/route.ts" tokenizes to "app api example route ts".
     - `nid` feeds the whole-query `joined == nid_lower` tier.
 
     NUL separators stop a trigram from spanning two fields (a query never contains
@@ -160,7 +162,8 @@ def _node_search_text(data: dict, nid: str) -> str:
     norm_label = data.get("norm_label") or _strip_diacritics(data.get("label") or "").lower()
     label_tokens = " ".join(_search_tokens(data.get("label") or ""))
     source = (data.get("source_file") or "").lower()
-    return "\x00".join((norm_label, label_tokens, str(nid).lower(), source))
+    source_tokens = " ".join(_search_tokens(data.get("source_file") or ""))
+    return "\x00".join((norm_label, label_tokens, str(nid).lower(), source, source_tokens))
 
 
 def _get_trigram_index(G: nx.Graph) -> dict:
@@ -572,6 +575,7 @@ def _find_node(G: nx.Graph, label: str) -> list[str]:
     term = " ".join(_search_tokens(label))
     if not term:
         return []
+    source_exact: list[str] = []
     exact: list[str] = []
     prefix: list[str] = []
     substring: list[str] = []
@@ -586,8 +590,11 @@ def _find_node(G: nx.Graph, label: str) -> list[str]:
         norm_label = d.get("norm_label") or _strip_diacritics(d.get("label") or "").lower()
         bare_label = norm_label.rstrip("()")
         label_tokens = " ".join(_search_tokens(d.get("label") or ""))
+        source_tokens = " ".join(_search_tokens(d.get("source_file") or ""))
         nid_lower = nid.lower()
-        if term == norm_label or term == bare_label or term == label_tokens or term == nid_lower:
+        if term == source_tokens:
+            source_exact.append(nid)
+        elif term == norm_label or term == bare_label or term == label_tokens or term == nid_lower:
             exact.append(nid)
         elif (
             norm_label.startswith(term)
@@ -598,7 +605,20 @@ def _find_node(G: nx.Graph, label: str) -> list[str]:
             prefix.append(nid)
         elif term in norm_label or term in label_tokens:
             substring.append(nid)
-    return exact + prefix + substring
+
+    if source_exact:
+        query_basename = _strip_diacritics(Path(label).name).lower()
+        preferred = [
+            nid
+            for nid in source_exact
+            if str(G.nodes[nid].get("source_location", "")) == "L1"
+            and _strip_diacritics(str(G.nodes[nid].get("label") or "")).lower()
+            == query_basename
+        ]
+        if len(preferred) == 1:
+            source_exact = preferred + [nid for nid in source_exact if nid != preferred[0]]
+
+    return source_exact + exact + prefix + substring
 
 
 def _filter_blank_stdin() -> None:

tests/test_affected_cli.py

@@ -174,3 +174,71 @@ def test_resolve_seed_bare_name_tie_still_returns_none():
     graph.add_node("b", label="dup()", source_file="pkg/two.py")
 
     assert resolve_seed(graph, "dup") is None
+
+
+def test_resolve_seed_source_file_path_prefers_file_level_node():
+    from graphify.affected import resolve_seed
+
+    graph = nx.DiGraph()
+    source_file = "app/api/example/route.ts"
+    graph.add_node(
+        "example_route_get",
+        label="GET()",
+        source_file=source_file,
+        source_location="L42",
+    )
+    graph.add_node(
+        "example_route",
+        label="route.ts",
+        source_file=source_file,
+        source_location="L1",
+    )
+
+    assert resolve_seed(graph, source_file) == "example_route"
+
+
+def test_affected_cli_source_file_path_uses_file_level_node(monkeypatch, tmp_path, capsys):
+    graph = nx.DiGraph()
+    source_file = "app/api/example/route.ts"
+    graph.add_node(
+        "example_route_get",
+        label="GET()",
+        source_file=source_file,
+        source_location="L42",
+    )
+    graph.add_node(
+        "example_route",
+        label="route.ts",
+        source_file=source_file,
+        source_location="L1",
+    )
+    graph.add_node(
+        "consumer",
+        label="consumer.ts",
+        source_file="app/consumer.ts",
+        source_location="L1",
+    )
+    graph.add_edge(
+        "consumer",
+        "example_route",
+        relation="imports_from",
+        context="import",
+        confidence="EXTRACTED",
+    )
+    graph_path = tmp_path / "graph.json"
+    graph_path.write_text(json.dumps(json_graph.node_link_data(graph, edges="links")), encoding="utf-8")
+
+    monkeypatch.setattr(mainmod, "_check_skill_version", lambda _: None)
+    monkeypatch.setattr(
+        mainmod.sys,
+        "argv",
+        ["graphify", "affected", source_file, "--graph", str(graph_path)],
+    )
+
+    mainmod.main()
+
+    out = capsys.readouterr().out
+    assert "Affected nodes for route.ts" in out
+    assert "consumer.ts" in out
+    assert "imports_from" in out
+    assert "No unique node matched" not in out

tests/test_explain_cli.py

@@ -54,3 +54,29 @@ def test_caller_shows_callee_as_outbound(monkeypatch, tmp_path, capsys):
     out = _run(monkeypatch, p, "createPatchHandler", capsys)
     assert "--> validateSanitySession() [calls]" in out
     assert "<-- " not in out
+
+
+def test_explain_source_file_path_prefers_file_level_node(monkeypatch, tmp_path, capsys):
+    source_file = "app/api/example/route.ts"
+    graph_data = {
+        "directed": False, "multigraph": False, "graph": {},
+        "nodes": [
+            {"id": "example_route_get", "label": "GET()",
+             "source_file": source_file, "source_location": "L42", "community": 0},
+            {"id": "example_route", "label": "route.ts",
+             "source_file": source_file, "source_location": "L1", "community": 0},
+        ],
+        "links": [
+            {"source": "example_route", "target": "example_route_get",
+             "relation": "contains", "confidence": "EXTRACTED"},
+        ],
+    }
+    p = tmp_path / "graph.json"
+    p.write_text(json.dumps(graph_data))
+
+    out = _run(monkeypatch, p, source_file, capsys)
+
+    assert "Node: route.ts" in out
+    assert "ID:        example_route" in out
+    assert f"Source:    {source_file} L1" in out
+    assert "Node: GET()" not in out

tests/test_serve.py

@@ -165,13 +165,14 @@ def test_trigrams_basic():
 def test_node_search_text_includes_all_matched_fields():
     G = _make_big_graph()
     text = _node_search_text(G.nodes["punct"], "punct")
-    # norm_label, the tokenized label (label_tokens), nid, and source are all present,
-    # NUL-separated so trigrams can't span fields.
+    # norm_label, tokenized label, nid, raw source, and tokenized source are all
+    # present, NUL-separated so trigrams can't span fields.
     parts = text.split("\x00")
     assert parts[0] == "foo.bar:baz"          # norm_label (punctuation kept)
     assert parts[1] == "foo bar baz"          # label_tokens (tokenized)
     assert parts[2] == "punct"                # nid
     assert parts[3] == "pkg/foobar.py"        # source_file
+    assert parts[4] == "pkg foobar py"        # source_file tokens
 
 
 def test_trigram_candidates_fast_path_fires_for_rare_term():
@@ -227,6 +228,30 @@ def test_find_node_label_tokens_branch_covered_by_index():
     assert _find_node(G, "Foo Bar Baz") == ["punct"]
 
 
+def test_find_node_source_file_path_prefers_file_level_node():
+    G = _make_big_graph()
+    source_file = "app/api/example/route.ts"
+    # Insert the function node first to prove source-file lookup reorders the
+    # file-level node ahead of other nodes from the same file.
+    G.add_node(
+        "example_route_get",
+        label="GET()",
+        source_file=source_file,
+        source_location="L42",
+    )
+    G.add_node(
+        "example_route",
+        label="route.ts",
+        source_file=source_file,
+        source_location="L1",
+    )
+
+    matches = _find_node(G, source_file)
+
+    assert matches[0] == "example_route"
+    assert "example_route_get" in matches
+
+
 def test_trigram_index_cached_and_rebuilt_per_graph():
     G = _make_big_graph()
     idx1 = _get_trigram_index(G)