feat: Assistants streaming

[coolbaluk]

Supersedes #731

Based on the original fork of @tanzyy96 just brought up to date.

We've been running this for a couple of weeks and has served us well.

Example usage:

              stream, err = client.CreateThreadAndStream(ctx, openai.CreateThreadAndRunRequest{
			RunRequest: openai.RunRequest{
				AssistantID: AssistantID,
			},
			Thread: openai.ThreadRequest{
				Messages: Messages,
			},
		})
		
		defer stream.Close()
		
		for {
		        resp, err = stream.Recv()
		        if errors.Is(err, io.EOF) {
			        break
		        }
                 }
		

[codecov[bot]]

Codecov Report

Attention: Patch coverage is 80.00000% with 12 lines in your changes missing coverage. Please review.

Project coverage is 98.24%. Comparing base (774fc9d) to head (c649497). Report is 47 commits behind head on master.

Files with missing lines	Patch %	Lines
run.go	80.00%	6 Missing and 6 partials :warning:

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #737      +/-   ##
==========================================
- Coverage   98.46%   98.24%   -0.22%     
==========================================
  Files          24       26       +2     
  Lines        1364     1478     +114     
==========================================
+ Hits         1343     1452     +109     
+ Misses         15       14       -1     
- Partials        6       12       +6     

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[opvexe]

may i ask question?
how to talk in next context  use last thread_id?

	cc := openai.NewClientWithConfig(config)

	stream, _ := cc.CreateRunStreaming(context.Background(), "thread_VEPTeIWj1umjdFyUd0Aj4lb2", openai.RunRequest{
		AssistantID: "asst_IuAlZbLkuIgcky26QPB2TQy0",
	})

	//stream, _ := cc.CreateThreadAndStream(context.Background(), openai.CreateThreadAndRunRequest{
	//	RunRequest: openai.RunRequest{
	//		AssistantID: "asst_IuAlZbLkuIgcky26QPB2TQy0",
	//	},
	//	Thread: openai.ThreadRequest{
	//		Messages: []openai.ThreadMessage{
	//			{
	//				Role:    openai.ThreadMessageRoleUser,
	//				Content: "我刚问了什么？",
	//			},
	//		},
	//	},
	//})

	defer stream.Close()

	for {
		resp, err := stream.Recv()
		if errors.Is(err, io.EOF) {
			break
		}

		t.Log("thread_id", resp.ID)

		for _, content := range resp.Delta.Content {
			t.Log(content.Text.Value)
		}
	}

[tanzyy96]

may i ask question?
how to talk in next context  use last thread_id?

	cc := openai.NewClientWithConfig(config)

	stream, _ := cc.CreateRunStreaming(context.Background(), "thread_VEPTeIWj1umjdFyUd0Aj4lb2", openai.RunRequest{
		AssistantID: "asst_IuAlZbLkuIgcky26QPB2TQy0",
	})

	//stream, _ := cc.CreateThreadAndStream(context.Background(), openai.CreateThreadAndRunRequest{
	//	RunRequest: openai.RunRequest{
	//		AssistantID: "asst_IuAlZbLkuIgcky26QPB2TQy0",
	//	},
	//	Thread: openai.ThreadRequest{
	//		Messages: []openai.ThreadMessage{
	//			{
	//				Role:    openai.ThreadMessageRoleUser,
	//				Content: "我刚问了什么？",
	//			},
	//		},
	//	},
	//})

	defer stream.Close()

	for {
		resp, err := stream.Recv()
		if errors.Is(err, io.EOF) {
			break
		}

		t.Log("thread_id", resp.ID)

		for _, content := range resp.Delta.Content {
			t.Log(content.Text.Value)
		}
	}

There's an API for inserting message in a thread. Example below:

_, err = a.client.CreateMessage(ctx, threadId, openai.MessageRequest{
    Role:    openai.ChatMessageRoleUser,
    Content: messageText,
})
if err != nil {
    logger.Error("failed to create message", zap.Error(err))
    return err
}

outStream, err = a.client.CreateRunStreaming(ctx, threadId, openai.RunRequest{
    AssistantID: assistantId,
})

[opvexe]

may i ask question?
how to talk in next context  use last thread_id?

	cc := openai.NewClientWithConfig(config)

	stream, _ := cc.CreateRunStreaming(context.Background(), "thread_VEPTeIWj1umjdFyUd0Aj4lb2", openai.RunRequest{
		AssistantID: "asst_IuAlZbLkuIgcky26QPB2TQy0",
	})

	//stream, _ := cc.CreateThreadAndStream(context.Background(), openai.CreateThreadAndRunRequest{
	//	RunRequest: openai.RunRequest{
	//		AssistantID: "asst_IuAlZbLkuIgcky26QPB2TQy0",
	//	},
	//	Thread: openai.ThreadRequest{
	//		Messages: []openai.ThreadMessage{
	//			{
	//				Role:    openai.ThreadMessageRoleUser,
	//				Content: "我刚问了什么？",
	//			},
	//		},
	//	},
	//})

	defer stream.Close()

	for {
		resp, err := stream.Recv()
		if errors.Is(err, io.EOF) {
			break
		}

		t.Log("thread_id", resp.ID)

		for _, content := range resp.Delta.Content {
			t.Log(content.Text.Value)
		}
	}

There's an API for inserting message in a thread. Example below:

_, err = a.client.CreateMessage(ctx, threadId, openai.MessageRequest{
    Role:    openai.ChatMessageRoleUser,
    Content: messageText,
})
if err != nil {
    logger.Error("failed to create message", zap.Error(err))
    return err
}

outStream, err = a.client.CreateRunStreaming(ctx, threadId, openai.RunRequest{
    AssistantID: assistantId,
})

thanks very much

[liushaobo-maker]

合并进去了吗？现在很需要这个

[coolbaluk]

@sashabaranov

Hi Sasha,

What is needed to get this one through review ?

I believe it's the most straightforward of the streaming implementations and will give a good base for further development.

Can confirm that has been extensively ran with production workloads for a while by us.

[opvexe]

i have use office openai sdk https://github.com/openai/openai-go/blob/main/api.md