fix node label lookup normalization (#1353)

Fixes #1338 (Unicode NFD/NFC): serve._find_node now matches tokenized labels; affected.resolve_seed NFC-normalizes + casefolds. Reviewed: full suite 2087 passed, CLI smoke clean, no regressions. Thanks @balloon72.

graphify/affected.py

@@ -4,6 +4,7 @@ from collections import deque
 from dataclasses import dataclass
 from pathlib import Path
 from typing import Iterable
+import unicodedata
 
 import networkx as nx
 
@@ -45,18 +46,22 @@ def _format_location(data: dict) -> str:
 
 def _bare_name(label: str) -> str:
     """Lowercased label with the callable decoration (trailing "()") removed."""
-    label = label.lower()
+    label = _normalize_label(label)
     return label[:-2] if label.endswith("()") else label
 
 
+def _normalize_label(label: str) -> str:
+    return unicodedata.normalize("NFC", label).casefold()
+
+
 def resolve_seed(graph: nx.Graph, query: str) -> str | None:
     if query in graph:
         return query
-    query_lower = query.lower()
+    query_lower = _normalize_label(query)
     exact_label_matches = [
         str(node_id)
         for node_id, data in graph.nodes(data=True)
-        if str(data.get("label", "")).lower() == query_lower
+        if _normalize_label(str(data.get("label", ""))) == query_lower
     ]
     if len(exact_label_matches) == 1:
         return exact_label_matches[0]
@@ -74,14 +79,14 @@ def resolve_seed(graph: nx.Graph, query: str) -> str | None:
     exact_source_matches = [
         str(node_id)
         for node_id, data in graph.nodes(data=True)
-        if str(data.get("source_file", "")).lower() == query_lower
+        if _normalize_label(str(data.get("source_file", ""))) == query_lower
     ]
     if len(exact_source_matches) == 1:
         return exact_source_matches[0]
     contains_matches = [
         str(node_id)
         for node_id, data in graph.nodes(data=True)
-        if query_lower in str(data.get("label", "")).lower()
+        if query_lower in _normalize_label(str(data.get("label", "")))
     ]
     if len(contains_matches) == 1:
         return contains_matches[0]

graphify/serve.py

@@ -464,12 +464,18 @@ def _find_node(G: nx.Graph, label: str) -> list[str]:
     for nid, d in G.nodes(data=True):
         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 ""))
         nid_lower = nid.lower()
-        if term == norm_label or term == bare_label or term == nid_lower:
+        if term == norm_label or term == bare_label or term == label_tokens or term == nid_lower:
             exact.append(nid)
-        elif norm_label.startswith(term) or bare_label.startswith(term) or nid_lower.startswith(term):
+        elif (
+            norm_label.startswith(term)
+            or bare_label.startswith(term)
+            or label_tokens.startswith(term)
+            or nid_lower.startswith(term)
+        ):
             prefix.append(nid)
-        elif term in norm_label:
+        elif term in norm_label or term in label_tokens:
             substring.append(nid)
     return exact + prefix + substring
 

tests/test_affected_cli.py

@@ -145,6 +145,27 @@ def test_resolve_seed_decorated_query_matches_bare_label():
     assert resolve_seed(graph, "Foo()") == "a"
 
 
+def test_resolve_seed_matches_unicode_normalized_label():
+    import unicodedata
+
+    from graphify.affected import resolve_seed
+
+    graph = nx.DiGraph()
+    graph.add_node("a", label="Auditoría", source_file="pkg/auditoria.py")
+
+    assert resolve_seed(graph, unicodedata.normalize("NFD", "Auditoría")) == "a"
+
+
+def test_resolve_seed_preserves_distinct_accents():
+    from graphify.affected import resolve_seed
+
+    graph = nx.DiGraph()
+    graph.add_node("a", label="resume", source_file="pkg/resume.py")
+    graph.add_node("b", label="résumé", source_file="pkg/resume_accented.py")
+
+    assert resolve_seed(graph, "resume") == "a"
+
+
 def test_resolve_seed_bare_name_tie_still_returns_none():
     from graphify.affected import resolve_seed
 

tests/test_serve.py

@@ -123,6 +123,13 @@ def test_find_node_ignores_trailing_punctuation():
     assert _find_node(G, "extract?") == ["n1"]
 
 
+def test_find_node_matches_full_punctuated_unicode_label():
+    G = nx.Graph()
+    G.add_node("n1", label="Skill /auditar — Auditoría inquisitiva de enlaces")
+
+    assert _find_node(G, "Skill /auditar — Auditoría inquisitiva de enlaces") == ["n1"]
+
+
 def test_query_terms_strips_search_punctuation():
     assert _query_terms("what calls extract?") == ["what", "calls", "extract"]