mirror of
https://github.com/tiennm99/DocsGPT.git
synced 2026-10-15 04:17:56 +00:00
fix(graphrag): drop entities whose name normalizes to nothing
``canonical_name`` answers "" for a punctuation-only name, which callers are meant to read as "no entity" -- ``_resolve_endpoint`` already does. Entity extraction did not, and nodes merge on that key, so every such entity in a source collapsed onto one shared node that belonged to none of them.
This commit is contained in:
1 parent
2b6d4d509e
commit
ef5b718f37
2 files changed
+29
-2
No files matched your search
@@ -1085,6 +1085,27 @@ class TestSummaryCountFailure:
|
||||
assert store.count_nodes.call_args.kwargs.get("strict") is True
|
||||
|
||||
|
||||
@pytest.mark.unit
|
||||
class TestEntityNormalization:
|
||||
"""An entity whose name normalizes to nothing is not an entity.
|
||||
|
||||
``canonical_name`` answers "" for a punctuation-only name, and nodes are
|
||||
merged on that key, so keeping them collapses every such entity onto one
|
||||
shared node. ``_resolve_endpoint`` already drops them on the relationship
|
||||
side.
|
||||
"""
|
||||
|
||||
@pytest.mark.parametrize("name", ["!!!", "--", "?", " *** "])
|
||||
def test_a_name_that_normalizes_to_nothing_is_dropped(self, name):
|
||||
assert extraction_module._build_entities([{"name": name}]) == []
|
||||
|
||||
def test_real_names_survive(self):
|
||||
built = extraction_module._build_entities(
|
||||
[{"name": "Quill Store"}, {"name": "!!!"}, {"name": "Alder"}]
|
||||
)
|
||||
assert [e["normalized_name"] for e in built] == ["quill store", "alder"]
|
||||
|
||||
|
||||
@pytest.mark.unit
|
||||
class TestParsing:
|
||||
def test_parses_embedded_json(self):
|
||||
|
||||
Reference in new issue
Block a user