Skip to content

fix(embeddings): check for numpy once per response #3754

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
lazerg wants to merge 2 commits into
base: main
Choose a base branch
from
+36 −1
Open

fix(embeddings): check for numpy once per response #3754

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
b175a3c
fix(embeddings): check for numpy once per response
lazerg Aug 28, 2026
c519bd1
fix(embeddings): only check for numpy when a vector needs decoding
lazerg Aug 28, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 5 additions & 1 deletion src/openai/lib/_parsing/_embeddings.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,11 +20,15 @@ def parse_embedding_response(
if not obj.data:
raise ValueError("No embedding data received")

use_numpy: bool | None = None

for embedding in obj.data:
data = cast(object, embedding.embedding)
if not isinstance(data, str):
continue
if not has_numpy():
if use_numpy is None:
use_numpy = has_numpy()
if not use_numpy:
# use array for base64 optimisation
embedding.embedding = array.array("f", base64.b64decode(data)).tolist()
else:
Expand Down
31 changes: 31 additions & 0 deletions tests/lib/test_embeddings.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -63,6 +63,37 @@ def test_decode_preserves_response_and_non_string_vectors(encoding_format: Omit
assert parsed.model == "text-embedding-3-small"


@pytest.mark.parametrize("encoding_format", [omit, not_given], ids=["omit", "not-given"])
def test_decoder_is_inspected_once(encoding_format: Omit | NotGiven, monkeypatch: pytest.MonkeyPatch) -> None:
calls = 0

def counting_decoder() -> bool:
nonlocal calls
calls += 1
return False

monkeypatch.setattr(embeddings_parser, "has_numpy", counting_decoder)
response = make_response(ENCODED, ENCODED, ENCODED)

parsed = embeddings_parser.parse_embedding_response(response, encoding_format=encoding_format)

assert calls == 1
assert [cast(object, item.embedding) for item in parsed.data] == [VALUES, VALUES, VALUES]


@pytest.mark.parametrize("encoding_format", [omit, not_given], ids=["omit", "not-given"])
def test_float_vectors_skip_the_decoder(encoding_format: Omit | NotGiven, monkeypatch: pytest.MonkeyPatch) -> None:
def unexpected_decoder() -> bool:
raise AssertionError("a response without string vectors must not inspect the decoder")

monkeypatch.setattr(embeddings_parser, "has_numpy", unexpected_decoder)
response = make_response(VALUES, [4.0, 5.0])

parsed = embeddings_parser.parse_embedding_response(response, encoding_format=encoding_format)

assert [cast(object, item.embedding) for item in parsed.data] == [VALUES, [4.0, 5.0]]


@pytest.mark.parametrize("encoding_format", ["float", "base64", None])
@pytest.mark.parametrize("vectors", [(ENCODED,), ("abc",), ()], ids=["encoded", "invalid", "empty"])
def test_explicit_format_is_untouched(
Expand Down