Merge pull request #7094 from liuyuzhou666/fix/responses-chat-developer-role-mapping

fix(apicompat): 修复 Responses 降级 Chat Completions 时的 system 消息顺序问题

backend/internal/pkg/apicompat/chatcompletions_responses_bridge.go

@@ -339,7 +339,63 @@ func responsesInputToChatMessagesWithOptions(instructions string, inputRaw json.
 	if err != nil {
 		return nil, err
 	}
-	return normalizeChatMessagesWithToolOutputMedia(built, mediaByCallID), nil
+	normalized := normalizeChatMessagesWithToolOutputMedia(built, mediaByCallID)
+	return normalizeResponsesDerivedChatMessageRoles(normalized), nil
+}
+
+// normalizeResponsesDerivedChatMessageRoles rewrites the Chat Completions
+// message list produced by the Responses bridge so that strict upstreams which
+// only accept system content at the very start of the conversation accept it.
+//
+// The bridge turns the Responses `instructions` field and every role:"developer"
+// item into a system message. Codex always sends both instructions and a leading
+// developer item, and it also injects developer notices into the middle of the
+// message history (for example when the user switches models). Forwarded as-is
+// that produces two leading system messages and mid-conversation system
+// messages, which Qwen-family upstreams reject with
+// "System message must be at the beginning." (HTTP 400).
+//
+// Leading system/developer messages are therefore merged into a single leading
+// system message, while later ones keep their position but are downgraded to
+// user messages so the same text still reaches the model.
+func normalizeResponsesDerivedChatMessageRoles(messages []ChatMessage) []ChatMessage {
+	isInstructionRole := func(role string) bool {
+		return role == "system" || role == "developer"
+	}
+
+	leading := 0
+	for leading < len(messages) && isInstructionRole(messages[leading].Role) {
+		leading++
+	}
+
+	out := make([]ChatMessage, 0, len(messages))
+	switch leading {
+	case 0:
+		// No leading instructions, nothing to merge.
+	case 1:
+		// A single leading prompt is already valid; keep its content byte for
+		// byte instead of round-tripping it through the text merge below.
+		out = append(out, messages[0])
+	default:
+		merged := make([]string, 0, leading)
+		for _, m := range messages[:leading] {
+			if text := strings.TrimSpace(chatMessageContentText(m.Content)); text != "" {
+				merged = append(merged, text)
+			}
+		}
+		if len(merged) > 0 {
+			content, _ := json.Marshal(strings.Join(merged, "\n\n"))
+			out = append(out, ChatMessage{Role: "system", Content: content})
+		}
+	}
+
+	for _, m := range messages[leading:] {
+		if isInstructionRole(m.Role) {
+			m.Role = "user"
+		}
+		out = append(out, m)
+	}
+	return out
 }
 
 // buildChatMessagesFromItems walks the Responses input items and appends the

backend/internal/pkg/apicompat/chatcompletions_responses_bridge_test.go

@@ -96,7 +96,7 @@ func TestResponsesInputToChatMessages_EmptyRoleFallsBackToUser(t *testing.T) {
 	assert.Equal(t, "user", messages[0].Role)
 }
 
-func TestResponsesInputToChatMessages_DeveloperRoleTrimAndCaseInsensitive(t *testing.T) {
+func TestResponsesInputToChatMessages_LeadingDeveloperRolesMergeIntoOneSystem(t *testing.T) {
 	input := json.RawMessage(`[
 		{"role":" Developer ","content":"one"},
 		{"role":"\tDEVELOPER\n","content":"two"}
@@ -104,9 +104,10 @@ func TestResponsesInputToChatMessages_DeveloperRoleTrimAndCaseInsensitive(t *tes
 
 	messages, err := responsesInputToChatMessages("", input)
 	require.NoError(t, err)
-	require.Len(t, messages, 2)
+	require.Len(t, messages, 1)
 
-	assert.Equal(t, []string{"system", "system"}, chatMessageRoles(messages))
+	assert.Equal(t, []string{"system"}, chatMessageRoles(messages))
+	assert.JSONEq(t, `"one\n\ntwo"`, string(messages[0].Content))
 }
 
 func TestResponsesToChatCompletionsRequest_InstructionsAndInputDeveloperRole(t *testing.T) {
@@ -121,12 +122,29 @@ func TestResponsesToChatCompletionsRequest_InstructionsAndInputDeveloperRole(t *
 
 	out, err := ResponsesToChatCompletionsRequest(req)
 	require.NoError(t, err)
-	require.Len(t, out.Messages, 3)
+	require.Len(t, out.Messages, 2)
 
-	assert.Equal(t, []string{"system", "system", "user"}, chatMessageRoles(out.Messages))
-	assert.JSONEq(t, `"Use concise answers."`, string(out.Messages[0].Content))
-	assert.JSONEq(t, `"Prefer JSON."`, string(out.Messages[1].Content))
-	assert.JSONEq(t, `"Hello"`, string(out.Messages[2].Content))
+	assert.Equal(t, []string{"system", "user"}, chatMessageRoles(out.Messages))
+	assert.JSONEq(t, `"Use concise answers.\n\nPrefer JSON."`, string(out.Messages[0].Content))
+	assert.JSONEq(t, `"Hello"`, string(out.Messages[1].Content))
+}
+
+func TestResponsesInputToChatMessages_MidConversationDeveloperBecomesUser(t *testing.T) {
+	input := json.RawMessage(`[
+		{"type":"message","role":"user","content":[{"type":"input_text","text":"hello"}]},
+		{"type":"message","role":"assistant","content":[{"type":"output_text","text":"hi"}]},
+		{"type":"message","role":"developer","content":[{"type":"input_text","text":"<model_switch> switched model"}]},
+		{"type":"message","role":"system","content":[{"type":"input_text","text":"be terse"}]},
+		{"type":"message","role":"user","content":[{"type":"input_text","text":"continue"}]}
+	]`)
+
+	messages, err := responsesInputToChatMessages("", input)
+	require.NoError(t, err)
+	require.Len(t, messages, 5)
+
+	assert.Equal(t, []string{"user", "assistant", "user", "user", "user"}, chatMessageRoles(messages))
+	assert.JSONEq(t, `"<model_switch> switched model"`, string(messages[2].Content))
+	assert.JSONEq(t, `"be terse"`, string(messages[3].Content))
 }
 
 func TestResponsesToChatCompletionsRequest_TextFormatJsonObject(t *testing.T) {

backend/internal/pkg/apicompat/chatcompletions_responses_tool_output_media_test.go

@@ -154,7 +154,7 @@ func TestResponsesToolOutputMedia_InterleavedMessagesFollowMediaBatch(t *testing
 		{"type":"function_call_output","call_id":"call_A","output":[{"type":"input_image","image_url":"data:image/png;base64,AQID"}]}
 	]`)
 
-	require.Equal(t, []string{"assistant", "tool", "user", "system", "user"}, chatMessageRoles(messages))
+	require.Equal(t, []string{"assistant", "tool", "user", "user", "user"}, chatMessageRoles(messages))
 	require.Equal(t, "[Tool output media for call call_A]", chatContentParts(t, messages[2])[0].Text)
 	require.JSONEq(t, `"approval saved"`, string(messages[3].Content))
 	require.JSONEq(t, `"continue"`, string(messages[4].Content))