Merge pull request router-for-me#3657 from catoncat/fix/responses-input-id-dedupe-orphaned-output

fix(openai): keep referenced tool call when deduping websocket input IDs

Author: luispater

sdk/api/handlers/openai/openai_responses_websocket.go

@@ -728,14 +728,63 @@ func dedupeInputItemsByID(rawArray string) (string, error) {
 		return "", errUnmarshal
 	}
 
-	lastIndexByID := make(map[string]int, len(items))
+	// Parse each item's type, id and call_id once; gjson is a scan-based
+	// parser, so reusing this metadata avoids rescanning every item in each of
+	// the loops below as the conversation history grows.
+	type itemMetadata struct {
+		itemType string
+		id       string
+		callID   string
+	}
+	meta := make([]itemMetadata, len(items))
 	for i, item := range items {
 		if len(item) == 0 {
 			continue
 		}
-		itemID := strings.TrimSpace(gjson.GetBytes(item, "id").String())
-		if itemID != "" {
-			lastIndexByID[itemID] = i
+		res := gjson.GetManyBytes(item, "type", "id", "call_id")
+		meta[i] = itemMetadata{
+			itemType: strings.TrimSpace(res[0].String()),
+			id:       strings.TrimSpace(res[1].String()),
+			callID:   strings.TrimSpace(res[2].String()),
+		}
+	}
+
+	// Collect the call_ids that are still referenced by tool-call output
+	// items. When several input items share the same id, the one we keep must
+	// preserve any call_id that has a matching output; otherwise the upstream
+	// rejects the request with "No tool call found for function call output".
+	referencedCallIDs := make(map[string]struct{}, len(items))
+	for i := range items {
+		switch meta[i].itemType {
+		case "function_call_output", "custom_tool_call_output":
+			if meta[i].callID != "" {
+				referencedCallIDs[meta[i].callID] = struct{}{}
+			}
+		}
+	}
+
+	// For each id, choose the index to keep. The default is the last
+	// occurrence (matching the original dedupe behavior), but we never replace
+	// an item whose call_id still has a matching output with one that does not.
+	// This keeps a single item per id while ensuring retained tool calls stay
+	// paired with their outputs.
+	keepIndexByID := make(map[string]int, len(items))
+	keepReferencedByID := make(map[string]bool, len(items))
+	for i := range items {
+		itemID := meta[i].id
+		if itemID == "" {
+			continue
+		}
+		_, referenced := referencedCallIDs[meta[i].callID]
+		referenced = referenced && meta[i].callID != ""
+		if _, seen := keepIndexByID[itemID]; !seen {
+			keepIndexByID[itemID] = i
+			keepReferencedByID[itemID] = referenced
+			continue
+		}
+		if referenced || !keepReferencedByID[itemID] {
+			keepIndexByID[itemID] = i
+			keepReferencedByID[itemID] = referenced
 		}
 	}
 
@@ -744,9 +793,9 @@ func dedupeInputItemsByID(rawArray string) (string, error) {
 		if len(item) == 0 {
 			continue
 		}
-		itemID := strings.TrimSpace(gjson.GetBytes(item, "id").String())
+		itemID := meta[i].id
 		if itemID != "" {
-			if lastIndexByID[itemID] != i {
+			if keepIndexByID[itemID] != i {
 				continue
 			}
 		}

sdk/api/handlers/openai/openai_responses_websocket_test.go

@@ -1845,6 +1845,29 @@ func TestDedupeResponsesWebsocketInputItemsByIDAfterRepair(t *testing.T) {
 	}
 }
 
+func TestDedupeResponsesWebsocketInputItemsByIDKeepsReferencedToolCall(t *testing.T) {
+	// Two function_call items share the same id but carry different call_ids
+	// (e.g. the upstream reused the item id across a re-sent/repaired call).
+	// Only the first call_id has a matching function_call_output. Deduping by
+	// id must keep the referenced call so the output is not orphaned, which
+	// previously triggered an upstream 400 "No tool call found for function
+	// call output with call_id ...".
+	payload := []byte(`{"input":[{"type":"function_call","id":"fc-1","call_id":"call-1","name":"exec_command"},{"type":"function_call","id":"fc-1","call_id":"call-2","name":"exec_command"},{"type":"function_call_output","id":"fco-1","call_id":"call-1"}]}`)
+
+	deduped := dedupeResponsesWebsocketInputItemsByID(payload)
+
+	items := gjson.GetBytes(deduped, "input").Array()
+	if len(items) != 2 {
+		t.Fatalf("deduped input len = %d, want 2: %s", len(items), deduped)
+	}
+	if items[0].Get("id").String() != "fc-1" ||
+		items[0].Get("call_id").String() != "call-1" ||
+		items[1].Get("id").String() != "fco-1" ||
+		items[1].Get("call_id").String() != "call-1" {
+		t.Fatalf("unexpected deduped input: %s", deduped)
+	}
+}
+
 func TestResponsesWebsocketCompactionResetsTurnStateOnCustomToolTranscriptReplacement(t *testing.T) {
 	gin.SetMode(gin.TestMode)