sorter: cross-system name adoption needs token containment; sys-retry keeps the higher score

tests/test_sorter_units.py

@@ -84,19 +84,43 @@ def test_prov_dir_rehome():
 def test_prov_weak_tier():
     """Below-threshold dir hit is accepted as low confidence."""
     root = Path(tempfile.mkdtemp(prefix="sorter_test_src_"))
-    pack = root / "Multi Pack"
-    # 'guardian squad extras' scores below 1.0 but high vs 'Guardian Squad'
-    write(pack / "guardian squad extras" / "g.stl")
+    # pack name carries a keyword so the probe resolves system+faction -
+    # only then does the same-faction weak tier apply
+    pack = root / "Aeldari Pack"
+    write(pack / "guardian weapon team" / "g.stl")  # 0.59 vs Guardian Squad
     library = make_lib()
-    items = prov_items(root, library)
+    (root / "extract_manifest.jsonl").write_text(json.dumps({
+        "archive": str(root / "Aeldari Pack.zip"),
+        "dest": "Aeldari Pack", "unit": "Aeldari Pack.zip",
+    }), encoding="utf-8")
+    files = sorted((root / "Aeldari Pack").rglob("*"))
+    items = sort_plan(files, library, get_catalog(), provenance={
+        "Aeldari Pack": "Aeldari Pack.zip"}, root=root)
     it = next(i for i in items if i.name == "g.stl")
-    assert it.match_kind in ("weak", "catalog"), it.match_kind
-    if it.match_kind == "weak":
-        assert it.confidence == "low"
+    assert it.match_kind == "weak", (it.match_kind, it.confidence)
+    assert it.confidence == "low"
     assert "Guardian" in str(it.target)
     shutil.rmtree(root); shutil.rmtree(library)
 
 
+def test_cross_system_needs_token_containment():
+    """'Aeldari Rangers' must not adopt Middle-earth 'Dwarf Rangers'."""
+    root = Path(tempfile.mkdtemp(prefix="sorter_test_src_"))
+    pack = root / "MQVxKzh"  # hash name: system stays unknown
+    write(pack / "Aeldari Rangers" / "Luminary - Arms -  (1).stl")
+    write(pack / "14 Warp spiders" / "sp.stl")
+    library = make_lib()
+    files = sorted((root / "MQVxKzh").rglob("*"))
+    items = sort_plan(files, library, get_catalog(), provenance={
+        "MQVxKzh": "MQVxKzh.zip"}, root=root)
+    by_name = {i.name: i for i in items}
+    lum = by_name["Luminary - Arms -  (1).stl"]
+    assert "Middle-Earth" not in str(lum.target), lum.target
+    spider = by_name["sp.stl"]
+    assert "Warp Spiders" in str(spider.target), spider.target
+    shutil.rmtree(root); shutil.rmtree(library)
+
+
 def test_nonprov_rehome():
     """Loose noise file inside a named dir re-homes to the dir's unit."""
     root = Path(tempfile.mkdtemp(prefix="sorter_test_src_"))
@@ -140,7 +164,7 @@ def load_catalog_path():
 def main():
     catalog = load_catalog_path()
     for fn in (test_prov_dir_rehome, test_prov_weak_tier, test_nonprov_rehome,
-               test_apply_weak_doc):
+               test_cross_system_needs_token_containment, test_apply_weak_doc):
         fn()
         print("OK", fn.__name__)