Skip to content

fix: handle missing optional fields in non-streaming citation conversion #2098

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

Merged
mkmeral merged 1 commit into from
Apr 9, 2026
Merged

fix: handle missing optional fields in non-streaming citation conversion #2098

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
16 changes: 9 additions & 7 deletions src/strands/models/bedrock.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -966,13 +966,15 @@ def _convert_non_streaming_to_streaming(self, response: dict[str, Any]) -> Itera
}

for citation in content["citationsContent"]["citations"]:
# Then emit citation metadata (for structure)

citation_metadata: CitationsDelta = {
"title": citation["title"],
"location": citation["location"],
"sourceContent": citation["sourceContent"],
}
# Emit citation metadata, only including fields that are present
# Nova grounding may omit title/sourceContent
citation_metadata: CitationsDelta = {}
if "title" in citation:
citation_metadata["title"] = citation["title"]
if "location" in citation:
citation_metadata["location"] = citation["location"]
if "sourceContent" in citation:
citation_metadata["sourceContent"] = citation["sourceContent"]
yield {"contentBlockDelta": {"delta": {"citation": citation_metadata}}}

# Yield contentBlockStop event
Expand Down
141 changes: 141 additions & 0 deletions tests/strands/models/test_bedrock.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2823,3 +2823,144 @@ def test_guardrail_latest_message_disabled_does_not_wrap(model):

assert "text" in formatted
assert "guardContent" not in formatted


@pytest.mark.asyncio
async def test_non_streaming_citations_with_missing_optional_fields(bedrock_client, model, alist):
"""Test that _convert_non_streaming_to_streaming handles citations missing optional fields.

Nova grounding returns citations with only url/domain but no title field. The conversion
should not crash with KeyError when optional fields like title, location, or sourceContent
are missing from the citation response.
"""
# Simulate a non-streaming response with citations missing the 'title' field
# This is what Nova grounding returns: url+domain in location, no title
non_streaming_response = {
"output": {
"message": {
"role": "assistant",
"content": [
{
"citationsContent": {
"content": [{"text": "Top shoe brands include Nike and Adidas."}],
"citations": [
{
"location": {
"web": {
"url": "https://example.com/shoes",
"domain": "example.com",
}
},
},
],
}
}
],
}
},
"stopReason": "end_turn",
"usage": {"inputTokens": 10, "outputTokens": 20},
}

events = list(model._convert_non_streaming_to_streaming(non_streaming_response))

# Should have: messageStart, contentBlockDelta (text + citation), contentBlockStop, messageStop, metadata
citation_deltas = [
e for e in events if "contentBlockDelta" in e and "citation" in e.get("contentBlockDelta", {}).get("delta", {})
]
assert len(citation_deltas) == 1

citation = citation_deltas[0]["contentBlockDelta"]["delta"]["citation"]
# title should NOT be present since the source didn't have it
assert "title" not in citation
# location should be present
assert "location" in citation
# sourceContent should NOT be present since the source didn't have it
assert "sourceContent" not in citation


@pytest.mark.asyncio
async def test_non_streaming_citations_with_all_fields_present(bedrock_client, model, alist):
"""Test that _convert_non_streaming_to_streaming correctly includes all fields when present."""
non_streaming_response = {
"output": {
"message": {
"role": "assistant",
"content": [
{
"citationsContent": {
"content": [{"text": "Nike is a top shoe brand."}],
"citations": [
{
"title": "Top Shoe Brands",
"location": {
"web": {
"url": "https://example.com/shoes",
"domain": "example.com",
}
},
"sourceContent": [{"text": "Nike is a leading brand"}],
},
],
}
}
],
}
},
"stopReason": "end_turn",
"usage": {"inputTokens": 10, "outputTokens": 20},
}

events = list(model._convert_non_streaming_to_streaming(non_streaming_response))

citation_deltas = [
e for e in events if "contentBlockDelta" in e and "citation" in e.get("contentBlockDelta", {}).get("delta", {})
]
assert len(citation_deltas) == 1

citation = citation_deltas[0]["contentBlockDelta"]["delta"]["citation"]
assert citation["title"] == "Top Shoe Brands"
assert citation["location"] == {"web": {"url": "https://example.com/shoes", "domain": "example.com"}}
assert citation["sourceContent"] == [{"text": "Nike is a leading brand"}]


@pytest.mark.asyncio
async def test_non_streaming_citations_with_only_location(bedrock_client, model, alist):
"""Test citations with only location field (no title, no sourceContent)."""
non_streaming_response = {
"output": {
"message": {
"role": "assistant",
"content": [
{
"citationsContent": {
"citations": [
{
"location": {
"web": {
"url": "https://example.com",
"domain": "example.com",
}
},
},
],
}
}
],
}
},
"stopReason": "end_turn",
"usage": {"inputTokens": 5, "outputTokens": 10},
}

events = list(model._convert_non_streaming_to_streaming(non_streaming_response))

citation_deltas = [
e for e in events if "contentBlockDelta" in e and "citation" in e.get("contentBlockDelta", {}).get("delta", {})
]
assert len(citation_deltas) == 1

citation = citation_deltas[0]["contentBlockDelta"]["delta"]["citation"]
assert citation["location"] == {"web": {"url": "https://example.com", "domain": "example.com"}}
assert "title" not in citation
assert "sourceContent" not in citation
Loading