Skip to content

fix: Add Chat Completions choices data in traces when using streaming #725

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

Draft
wants to merge 1 commit into
base: main
Choose a base branch
from
Draft

fix: Add Chat Completions choices data in traces when using streaming #725

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
58 changes: 57 additions & 1 deletion packages/agents-openai/src/openaiChatCompletionsModel.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -190,7 +190,14 @@
async *getStreamedResponse(
request: ModelRequest,
): AsyncIterable<ResponseStreamEvent> {
const span = request.tracing ? createGenerationSpan() : undefined;
const span = request.tracing
? createGenerationSpan({
data: {
model: this.#model,
model_config: request.modelSettings,
},
})
: undefined;
try {
if (span) {
span.start();
Expand Down Expand Up @@ -218,6 +225,55 @@
event.type === 'response_done' &&
response.usage?.total_tokens === 0
) {
response.choices = event.response.output
.flatMap((o): OpenAI.Chat.Completions.ChatCompletion.Choice[] => {

Check failure on line 229 in packages/agents-openai/src/openaiChatCompletionsModel.ts

View workflow job for this annotation

GitHub Actions / test (24.3.x)

test/openaiChatCompletionsModel.test.ts > OpenAIChatCompletionsModel > populates usage from response_done event when initial usage is zero 

TypeError: Cannot read properties of undefined (reading 'flatMap')
 ❯ OpenAIChatCompletionsModel.getStreamedResponse src/openaiChatCompletionsModel.ts:229:14
 ❯ test/openaiChatCompletionsModel.test.ts:584:24
 ❯ ../agents-core/src/tracing/context.ts:61:20
 ❯ test/openaiChatCompletionsModel.test.ts:583:5

Check failure on line 229 in packages/agents-openai/src/openaiChatCompletionsModel.ts

View workflow job for this annotation

GitHub Actions / test (20)

test/openaiChatCompletionsModel.test.ts > OpenAIChatCompletionsModel > populates usage from response_done event when initial usage is zero 

TypeError: Cannot read properties of undefined (reading 'flatMap')
 ❯ OpenAIChatCompletionsModel.getStreamedResponse src/openaiChatCompletionsModel.ts:229:14
 ❯ test/openaiChatCompletionsModel.test.ts:584:24
 ❯ ../agents-core/src/tracing/context.ts:61:20
 ❯ test/openaiChatCompletionsModel.test.ts:583:5
if (o.type === 'function_call') {
Copy link
Member

@seratch seratch Dec 4, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The logic here requires updates when the platform adds new types. We'd like to explore more robust way to handle the conversion here.

Copy link
Author

@jonjomckay jonjomckay Dec 4, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yeah, I was worried about that too. Do you have any suggestions? This is the first time I've actually looked into this API!

return [
{
finish_reason: 'tool_calls',
index: 0,
logprobs: null,
message: {
content: null,
refusal: null,
role: 'assistant',
tool_calls: [
{
type: 'function',
function: {
arguments: o.arguments,
name: o.name,
},
id: o.callId,
},
],
},
},
];
} else if (o.type === 'message') {
return o.content
.filter((c) => c.type === 'output_text')
.map((c) => {
return {
finish_reason: 'stop',
index: 0,
logprobs: null,
message: {
content: c.text,
refusal: null,
role: 'assistant',
},
};
});
} else {
return [];
}
})
.map((choice, index) => ({
...choice,
index,
}));

response.usage = {
prompt_tokens: event.response.usage.inputTokens,
completion_tokens: event.response.usage.outputTokens,
Expand Down
Loading