From f82d00af74e8874861397ac6a05f35988cc45a52 Mon Sep 17 00:00:00 2001
From: Bruce MacDonald <brucewmacdonald@gmail.com>
Date: Thu, 18 Dec 2025 14:07:31 -0800
Subject: [PATCH] openai: do not separate images into their own messages

- Changed the case []any: branch to accumulate text and images into slices first, then create a single api.Message at the end
- Multiple text parts are joined with \n\n
- All images are collected into a single Images slice
---
 openai/openai.go      | 29 ++++++++++++++++-------------
 openai/openai_test.go | 36 ++++++++++++++++++++++++++----------
 2 files changed, 42 insertions(+), 23 deletions(-)

diff --git a/openai/openai.go b/openai/openai.go
index 9dcba3000..5b4662a7e 100644
--- a/openai/openai.go
+++ b/openai/openai.go
@@ -463,6 +463,8 @@ func FromChatRequest(r ChatCompletionRequest) (*api.ChatRequest, error) {
 			}
 			messages = append(messages, api.Message{Role: msg.Role, Content: content, Thinking: msg.Reasoning, ToolCalls: toolCalls, ToolName: toolName, ToolCallID: msg.ToolCallID})
 		case []any:
+			var texts []string
+			var images []api.ImageData
 			for _, c := range content {
 				data, ok := c.(map[string]any)
 				if !ok {
@@ -474,7 +476,7 @@ func FromChatRequest(r ChatCompletionRequest) (*api.ChatRequest, error) {
 					if !ok {
 						return nil, errors.New("invalid message format")
 					}
-					messages = append(messages, api.Message{Role: msg.Role, Content: text})
+					texts = append(texts, text)
 				case "image_url":
 					var url string
 					if urlMap, ok := data["image_url"].(map[string]any); ok {
@@ -492,23 +494,24 @@ func FromChatRequest(r ChatCompletionRequest) (*api.ChatRequest, error) {
 						return nil, err
 					}
 
-					messages = append(messages, api.Message{Role: msg.Role, Images: []api.ImageData{img}})
+					images = append(images, img)
 				default:
 					return nil, errors.New("invalid message format")
 				}
 			}
-			// since we might have added multiple messages above, if we have tools
-			// calls we'll add them to the last message
-			if len(messages) > 0 && len(msg.ToolCalls) > 0 {
-				toolCalls, err := FromCompletionToolCall(msg.ToolCalls)
-				if err != nil {
-					return nil, err
-				}
-				messages[len(messages)-1].ToolCalls = toolCalls
-				messages[len(messages)-1].ToolName = toolName
-				messages[len(messages)-1].ToolCallID = msg.ToolCallID
-				messages[len(messages)-1].Thinking = msg.Reasoning
+			toolCalls, err := FromCompletionToolCall(msg.ToolCalls)
+			if err != nil {
+				return nil, err
 			}
+			messages = append(messages, api.Message{
+				Role:       msg.Role,
+				Content:    strings.Join(texts, "\n\n"),
+				Images:     images,
+				Thinking:   msg.Reasoning,
+				ToolCalls:  toolCalls,
+				ToolName:   toolName,
+				ToolCallID: msg.ToolCallID,
+			})
 		default:
 			// content is only optional if tool calls are present
 			if msg.ToolCalls == nil {
diff --git a/openai/openai_test.go b/openai/openai_test.go
index 51e243dec..683c2e447 100644
--- a/openai/openai_test.go
+++ b/openai/openai_test.go
@@ -41,7 +41,7 @@ func TestFromChatRequest_Basic(t *testing.T) {
 	}
 }
 
-func TestFromChatRequest_WithImage(t *testing.T) {
+func TestFromChatRequest_MultiPartContent(t *testing.T) {
 	imgData, _ := base64.StdEncoding.DecodeString(image)
 
 	req := ChatCompletionRequest{
@@ -50,7 +50,12 @@ func TestFromChatRequest_WithImage(t *testing.T) {
 			{
 				Role: "user",
 				Content: []any{
-					map[string]any{"type": "text", "text": "Hello"},
+					map[string]any{"type": "text", "text": "First part."},
+					map[string]any{"type": "text", "text": "Second part."},
+					map[string]any{
+						"type":      "image_url",
+						"image_url": map[string]any{"url": prefix + image},
+					},
 					map[string]any{
 						"type":      "image_url",
 						"image_url": map[string]any{"url": prefix + image},
@@ -65,20 +70,31 @@ func TestFromChatRequest_WithImage(t *testing.T) {
 		t.Fatalf("unexpected error: %v", err)
 	}
 
-	if len(result.Messages) != 2 {
-		t.Fatalf("expected 2 messages, got %d", len(result.Messages))
+	// Multi-part content array should produce a single message per OpenAI spec
+	if len(result.Messages) != 1 {
+		t.Fatalf("expected 1 message, got %d", len(result.Messages))
 	}
 
-	if result.Messages[0].Content != "Hello" {
-		t.Errorf("expected first message content 'Hello', got %q", result.Messages[0].Content)
+	msg := result.Messages[0]
+	if msg.Role != "user" {
+		t.Errorf("expected role 'user', got %q", msg.Role)
 	}
 
-	if len(result.Messages[1].Images) != 1 {
-		t.Fatalf("expected 1 image, got %d", len(result.Messages[1].Images))
+	// Multiple text parts should be joined
+	expectedContent := "First part.\n\nSecond part."
+	if msg.Content != expectedContent {
+		t.Errorf("expected content %q, got %q", expectedContent, msg.Content)
 	}
 
-	if string(result.Messages[1].Images[0]) != string(imgData) {
-		t.Error("image data mismatch")
+	// Multiple images should be in the same message
+	if len(msg.Images) != 2 {
+		t.Fatalf("expected 2 images, got %d", len(msg.Images))
+	}
+
+	for i, img := range msg.Images {
+		if string(img) != string(imgData) {
+			t.Errorf("image %d data mismatch", i)
+		}
 	}
 }