Skip to content
This repository was archived by the owner on Jul 17, 2026. It is now read-only.

Commit 580905c

Browse files
sjrldavidsbatista
sjrl
and
davidsbatista
authored
test: Strengthen some of the doc store to mixin tests (deepset-ai#12047)
Co-authored-by: David S. Batista <dsbatista@gmail.com>
1 parent 76d5686 commit 580905c

2 files changed

Lines changed: 52 additions & 40 deletions

File tree

‎haystack/testing/document_store.py‎

Lines changed: 26 additions & 20 deletions
Original file line numberDiff line numberDiff line change
@@ -1028,16 +1028,16 @@ def test_count_unique_metadata_by_filter_all_documents(document_store: DocumentS
10281028
counts = document_store.count_unique_metadata_by_filter( # type:ignore[attr-defined]
10291029
filters={}, metadata_fields=["category", "status", "priority"]
10301030
)
1031-
assert counts["category"] == 3
1032-
assert counts["status"] == 2
1033-
assert counts["priority"] == 3
1031+
assert counts == {"category": 3, "status": 2, "priority": 3}
10341032

10351033
@staticmethod
10361034
def test_count_unique_metadata_by_filter_with_filter(document_store: DocumentStore):
10371035
"""Test count_unique_metadata_by_filter() with a filter."""
1036+
# The filtered-out document carries extra unique values, so the counts only come out right
1037+
# when the store actually applies the filter.
10381038
docs = [
10391039
Document(content="Doc 1", meta={"category": "A", "status": "active", "priority": 1}),
1040-
Document(content="Doc 2", meta={"category": "B", "status": "active", "priority": 2}),
1040+
Document(content="Doc 2", meta={"category": "B", "status": "archived", "priority": 2}),
10411041
Document(content="Doc 3", meta={"category": "A", "status": "inactive", "priority": 1}),
10421042
Document(content="Doc 4", meta={"category": "A", "status": "active", "priority": 3}),
10431043
]
@@ -1047,8 +1047,7 @@ def test_count_unique_metadata_by_filter_with_filter(document_store: DocumentSto
10471047
counts = document_store.count_unique_metadata_by_filter( # type:ignore[attr-defined]
10481048
filters={"field": "meta.category", "operator": "==", "value": "A"}, metadata_fields=["status", "priority"]
10491049
)
1050-
assert counts["status"] == 2
1051-
assert counts["priority"] == 2
1050+
assert counts == {"status": 2, "priority": 2}
10521051

10531052
@staticmethod
10541053
def test_count_unique_metadata_by_filter_with_multiple_filters(document_store: DocumentStore):
@@ -1121,11 +1120,14 @@ class GetMetadataFieldMinMaxTest:
11211120
@staticmethod
11221121
def test_get_metadata_field_min_max_numeric(document_store: DocumentStore):
11231122
"""Test get_metadata_field_min_max() with integer field."""
1123+
# The min and max are deliberately not the first or last written values, so an implementation
1124+
# returning values by insertion order fails. Keeping both 10 and 5 catches implementations that
1125+
# compare numbers as strings ("10" < "5" lexicographically).
11241126
docs = [
1125-
Document(content="Doc 1", meta={"priority": 1}),
1126-
Document(content="Doc 2", meta={"priority": 5}),
1127-
Document(content="Doc 3", meta={"priority": 3}),
1128-
Document(content="Doc 4", meta={"priority": 10}),
1127+
Document(content="Doc 1", meta={"priority": 5}),
1128+
Document(content="Doc 2", meta={"priority": 1}),
1129+
Document(content="Doc 3", meta={"priority": 10}),
1130+
Document(content="Doc 4", meta={"priority": 3}),
11291131
]
11301132
document_store.write_documents(docs)
11311133
assert document_store.count_documents() == 4
@@ -1137,13 +1139,15 @@ def test_get_metadata_field_min_max_numeric(document_store: DocumentStore):
11371139
@staticmethod
11381140
def test_get_metadata_field_min_max_float(document_store: DocumentStore):
11391141
"""Test get_metadata_field_min_max() with float field."""
1142+
# The min and max are deliberately not the first or last written values.
11401143
docs = [
1141-
Document(content="Doc 1", meta={"rating": 0.6}),
1142-
Document(content="Doc 2", meta={"rating": 0.95}),
1143-
Document(content="Doc 3", meta={"rating": 0.8}),
1144+
Document(content="Doc 1", meta={"rating": 0.8}),
1145+
Document(content="Doc 2", meta={"rating": 0.6}),
1146+
Document(content="Doc 3", meta={"rating": 0.95}),
1147+
Document(content="Doc 4", meta={"rating": 0.7}),
11441148
]
11451149
document_store.write_documents(docs)
1146-
assert document_store.count_documents() == 3
1150+
assert document_store.count_documents() == 4
11471151

11481152
result = document_store.get_metadata_field_min_max("rating") # type:ignore[attr-defined]
11491153

@@ -1173,15 +1177,16 @@ def test_get_metadata_field_min_max_empty_collection(document_store: DocumentSto
11731177
@staticmethod
11741178
def test_get_metadata_field_min_max_meta_prefix(document_store: DocumentStore):
11751179
"""Test get_metadata_field_min_max() with field names that include 'meta.' prefix."""
1180+
# The min and max of each field are deliberately not the first or last written values.
11761181
docs = [
1177-
Document(content="Doc 1", meta={"priority": 1, "age": 10}),
1178-
Document(content="Doc 2", meta={"priority": 5, "age": 20}),
1179-
Document(content="Doc 3", meta={"priority": 3, "age": 15}),
1180-
Document(content="Doc 4", meta={"priority": 10, "age": 5}),
1182+
Document(content="Doc 1", meta={"priority": 5, "age": 10}),
1183+
Document(content="Doc 2", meta={"priority": 1, "age": 20}),
1184+
Document(content="Doc 3", meta={"priority": 10, "age": 15}),
1185+
Document(content="Doc 4", meta={"priority": 3, "age": 5}),
11811186
Document(content="Doc 6", meta={"rating": 10.5}),
11821187
Document(content="Doc 7", meta={"rating": 20.3}),
1183-
Document(content="Doc 8", meta={"rating": 15.7}),
1184-
Document(content="Doc 9", meta={"rating": 5.2}),
1188+
Document(content="Doc 8", meta={"rating": 5.2}),
1189+
Document(content="Doc 9", meta={"rating": 15.7}),
11851190
]
11861191
document_store.write_documents(docs)
11871192

@@ -1233,6 +1238,7 @@ def test_get_metadata_field_unique_values_basic(document_store: DocumentStore):
12331238

12341239
values = result[0] if isinstance(result, tuple) else result
12351240
assert isinstance(values, list)
1241+
assert len(values) == 3 # the returned values must not contain duplicates
12361242
assert set(values) == {"A", "B", "C"}
12371243
if isinstance(result, tuple) and len(result) >= 2 and isinstance(result[1], int):
12381244
assert result[1] == 3

‎haystack/testing/document_store_async.py‎

Lines changed: 26 additions & 20 deletions
Original file line numberDiff line numberDiff line change
@@ -277,17 +277,17 @@ async def test_count_unique_metadata_by_filter_async_all_documents(document_stor
277277
counts = await document_store.count_unique_metadata_by_filter_async( # type:ignore[attr-defined]
278278
filters={}, metadata_fields=["category", "status", "priority"]
279279
)
280-
assert counts["category"] == 3
281-
assert counts["status"] == 2
282-
assert counts["priority"] == 3
280+
assert counts == {"category": 3, "status": 2, "priority": 3}
283281

284282
@staticmethod
285283
@pytest.mark.asyncio
286284
async def test_count_unique_metadata_by_filter_async_with_filter(document_store: AsyncDocumentStore):
287285
"""Test count_unique_metadata_by_filter_async() with a filter."""
286+
# The filtered-out document carries extra unique values, so the counts only come out right
287+
# when the store actually applies the filter.
288288
docs = [
289289
Document(content="Doc 1", meta={"category": "A", "status": "active", "priority": 1}),
290-
Document(content="Doc 2", meta={"category": "B", "status": "active", "priority": 2}),
290+
Document(content="Doc 2", meta={"category": "B", "status": "archived", "priority": 2}),
291291
Document(content="Doc 3", meta={"category": "A", "status": "inactive", "priority": 1}),
292292
Document(content="Doc 4", meta={"category": "A", "status": "active", "priority": 3}),
293293
]
@@ -297,8 +297,7 @@ async def test_count_unique_metadata_by_filter_async_with_filter(document_store:
297297
counts = await document_store.count_unique_metadata_by_filter_async( # type:ignore[attr-defined]
298298
filters={"field": "meta.category", "operator": "==", "value": "A"}, metadata_fields=["status", "priority"]
299299
)
300-
assert counts["status"] == 2
301-
assert counts["priority"] == 2
300+
assert counts == {"status": 2, "priority": 2}
302301

303302
@staticmethod
304303
@pytest.mark.asyncio
@@ -622,11 +621,14 @@ class GetMetadataFieldMinMaxAsyncTest:
622621
@pytest.mark.asyncio
623622
async def test_get_metadata_field_min_max_numeric_async(document_store: AsyncDocumentStore):
624623
"""Test get_metadata_field_min_max_async() with integer field."""
624+
# The min and max are deliberately not the first or last written values, so an implementation
625+
# returning values by insertion order fails. Keeping both 10 and 5 catches implementations that
626+
# compare numbers as strings ("10" < "5" lexicographically).
625627
docs = [
626-
Document(content="Doc 1", meta={"priority": 1}),
627-
Document(content="Doc 2", meta={"priority": 5}),
628-
Document(content="Doc 3", meta={"priority": 3}),
629-
Document(content="Doc 4", meta={"priority": 10}),
628+
Document(content="Doc 1", meta={"priority": 5}),
629+
Document(content="Doc 2", meta={"priority": 1}),
630+
Document(content="Doc 3", meta={"priority": 10}),
631+
Document(content="Doc 4", meta={"priority": 3}),
630632
]
631633
await document_store.write_documents_async(docs)
632634
assert await document_store.count_documents_async() == 4
@@ -639,13 +641,15 @@ async def test_get_metadata_field_min_max_numeric_async(document_store: AsyncDoc
639641
@pytest.mark.asyncio
640642
async def test_get_metadata_field_min_max_float_async(document_store: AsyncDocumentStore):
641643
"""Test get_metadata_field_min_max_async() with float field."""
644+
# The min and max are deliberately not the first or last written values.
642645
docs = [
643-
Document(content="Doc 1", meta={"rating": 0.6}),
644-
Document(content="Doc 2", meta={"rating": 0.95}),
645-
Document(content="Doc 3", meta={"rating": 0.8}),
646+
Document(content="Doc 1", meta={"rating": 0.8}),
647+
Document(content="Doc 2", meta={"rating": 0.6}),
648+
Document(content="Doc 3", meta={"rating": 0.95}),
649+
Document(content="Doc 4", meta={"rating": 0.7}),
646650
]
647651
await document_store.write_documents_async(docs)
648-
assert await document_store.count_documents_async() == 3
652+
assert await document_store.count_documents_async() == 4
649653

650654
result = await document_store.get_metadata_field_min_max_async("rating") # type:ignore[attr-defined]
651655

@@ -678,15 +682,16 @@ async def test_get_metadata_field_min_max_empty_collection_async(document_store:
678682
@pytest.mark.asyncio
679683
async def test_get_metadata_field_min_max_meta_prefix_async(document_store: AsyncDocumentStore):
680684
"""Test get_metadata_field_min_max_async() with field names that include 'meta.' prefix."""
685+
# The min and max of each field are deliberately not the first or last written values.
681686
docs = [
682-
Document(content="Doc 1", meta={"priority": 1, "age": 10}),
683-
Document(content="Doc 2", meta={"priority": 5, "age": 20}),
684-
Document(content="Doc 3", meta={"priority": 3, "age": 15}),
685-
Document(content="Doc 4", meta={"priority": 10, "age": 5}),
687+
Document(content="Doc 1", meta={"priority": 5, "age": 10}),
688+
Document(content="Doc 2", meta={"priority": 1, "age": 20}),
689+
Document(content="Doc 3", meta={"priority": 10, "age": 15}),
690+
Document(content="Doc 4", meta={"priority": 3, "age": 5}),
686691
Document(content="Doc 6", meta={"rating": 10.5}),
687692
Document(content="Doc 7", meta={"rating": 20.3}),
688-
Document(content="Doc 8", meta={"rating": 15.7}),
689-
Document(content="Doc 9", meta={"rating": 5.2}),
693+
Document(content="Doc 8", meta={"rating": 5.2}),
694+
Document(content="Doc 9", meta={"rating": 15.7}),
690695
]
691696
await document_store.write_documents_async(docs)
692697

@@ -739,6 +744,7 @@ async def test_get_metadata_field_unique_values_basic_async(document_store: Asyn
739744

740745
values = result[0] if isinstance(result, tuple) else result
741746
assert isinstance(values, list)
747+
assert len(values) == 3 # the returned values must not contain duplicates
742748
assert set(values) == {"A", "B", "C"}
743749
if isinstance(result, tuple) and len(result) >= 2 and isinstance(result[1], int):
744750
assert result[1] == 3

0 commit comments

Comments
 (0)