From 7e3cdc83710b7494852e95aaa5b1a5db1a8fb02a Mon Sep 17 00:00:00 2001
From: nguyenha935 <nguyenthanhha935@gmail.com>
Date: Thu, 2 Jul 2026 16:24:25 +0700
Subject: [PATCH] fix(providers): preserve streamed content whitespace (#1324)

Co-authored-by: GoClaw Operator <operator@goclaw>
---
 internal/providers/adapter_openai.go          |  6 +--
 internal/providers/adapter_openai_test.go     | 14 ++++++
 .../providers/control_output_normalizer.go    |  1 -
 .../control_output_normalizer_test.go         | 46 +++++++++++++++++++
 4 files changed, 62 insertions(+), 5 deletions(-)

diff --git a/internal/providers/adapter_openai.go b/internal/providers/adapter_openai.go
index d8334dc5..6e744441 100644
--- a/internal/providers/adapter_openai.go
+++ b/internal/providers/adapter_openai.go
@@ -112,10 +112,8 @@ func (a *OpenAIAdapter) FromStreamChunk(data []byte) (*StreamChunk, error) {
 
 	// Text content
 	if delta.Content != "" {
-		normalized := normalizeControlOutput(delta.Content, "", nil)
-		sc.Content = normalized.Content
-		sc.Thinking = strings.TrimSpace(strings.Join([]string{sc.Thinking, normalized.Thinking}, "\n"))
-		hasContent = sc.Content != "" || sc.Thinking != ""
+		sc.Content = delta.Content
+		hasContent = true
 	}
 
 	if !hasContent {
diff --git a/internal/providers/adapter_openai_test.go b/internal/providers/adapter_openai_test.go
index c7cfcb4f..b237ed4a 100644
--- a/internal/providers/adapter_openai_test.go
+++ b/internal/providers/adapter_openai_test.go
@@ -185,6 +185,20 @@ func TestOpenAIAdapter_FromStreamChunk(t *testing.T) {
 		}
 	})
 
+	t.Run("content delta preserves leading whitespace", func(t *testing.T) {
+		raw := []byte(`{"choices":[{"delta":{"content":" xin lỗi"}}]}`)
+		sc, err := a.FromStreamChunk(raw)
+		if err != nil {
+			t.Fatalf("err: %v", err)
+		}
+		if sc == nil {
+			t.Fatal("want stream chunk, got nil")
+		}
+		if sc.Content != " xin lỗi" {
+			t.Errorf("want leading whitespace preserved, got %q", sc.Content)
+		}
+	})
+
 	t.Run("reasoning_content delta", func(t *testing.T) {
 		raw := []byte(`{"choices":[{"delta":{"reasoning_content":"thinking..."}}]}`)
 		sc, err := a.FromStreamChunk(raw)
diff --git a/internal/providers/control_output_normalizer.go b/internal/providers/control_output_normalizer.go
index 095f8b33..c4c0a206 100644
--- a/internal/providers/control_output_normalizer.go
+++ b/internal/providers/control_output_normalizer.go
@@ -70,7 +70,6 @@ func normalizeControlOutput(content, thinking string, tools []ToolDefinition) co
 	out.Content, extracted = extractThinkingTags(out.Content)
 	out.Thinking = joinThinking(out.Thinking, extracted)
 
-	out.Content = strings.TrimSpace(out.Content)
 	return out
 }
 
diff --git a/internal/providers/control_output_normalizer_test.go b/internal/providers/control_output_normalizer_test.go
index b830ecaf..c671e089 100644
--- a/internal/providers/control_output_normalizer_test.go
+++ b/internal/providers/control_output_normalizer_test.go
@@ -286,6 +286,52 @@ func TestOpenAIProviderChatStream_NormalizesSplitKimiTextToolCallBeforeEmittingC
 	}
 }
 
+func TestOpenAIProviderChatStream_PreservesWhitespaceAcrossContentDeltas(t *testing.T) {
+	events := []string{
+		`data: {"choices":[{"delta":{"content":"Dạ"}}]}` + "\n\n",
+		`data: {"choices":[{"delta":{"content":" em"}}]}` + "\n\n",
+		`data: {"choices":[{"delta":{"content":" xin"}}]}` + "\n\n",
+		`data: {"choices":[{"delta":{"content":" lỗi"}}]}` + "\n\n",
+		`data: {"choices":[{"delta":{"content":" anh"}}]}` + "\n\n",
+		`data: {"choices":[{"finish_reason":"stop","delta":{}}]}` + "\n\n",
+	}
+	srv := newOpenAISSEServer(t, events)
+	defer srv.Close()
+
+	p := NewOpenAIProvider("openai-compatible", "sk", srv.URL, "model")
+	var streamed strings.Builder
+	resp, err := p.ChatStream(context.Background(), ChatRequest{
+		Messages: []Message{{Role: "user", Content: "say it"}},
+	}, func(chunk StreamChunk) {
+		streamed.WriteString(chunk.Content)
+	})
+	if err != nil {
+		t.Fatalf("ChatStream: %v", err)
+	}
+
+	const want = "Dạ em xin lỗi anh"
+	if resp.Content != want {
+		t.Fatalf("Content = %q, want %q", resp.Content, want)
+	}
+	if got := streamed.String(); got != want {
+		t.Fatalf("streamed content = %q, want %q", got, want)
+	}
+}
+
+func TestControlOutputNormalizer_PreservesWhitespaceAroundNormalText(t *testing.T) {
+	n := newControlOutputNormalizer(nil)
+	var got strings.Builder
+	got.WriteString(n.Append("Dạ").Content)
+	got.WriteString(n.Append(" em").Content)
+	got.WriteString(n.Append(" xin").Content)
+	got.WriteString(n.Append(" lỗi").Content)
+	got.WriteString(n.Finish().Content)
+
+	if got.String() != "Dạ em xin lỗi" {
+		t.Fatalf("normalized stream content = %q, want %q", got.String(), "Dạ em xin lỗi")
+	}
+}
+
 func newOpenAIJSONServer(t *testing.T, body string) *httptest.Server {
 	t.Helper()
 	return httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {