From d530d257939073e234e7472c8c5fc237d7f8559d Mon Sep 17 00:00:00 2001
From: "CJACK." <tetr20071102@gmail.com>
Date: Sun, 22 Mar 2026 00:57:13 +0800
Subject: [PATCH] Expand history-sanitize boundary coverage for stream chunks

---
 .../openai/tool_history_sanitize_test.go      | 61 +++++++++++++++++--
 tests/node/stream-tool-sieve.test.js          | 16 +++++
 2 files changed, 73 insertions(+), 4 deletions(-)

diff --git a/internal/adapter/openai/tool_history_sanitize_test.go b/internal/adapter/openai/tool_history_sanitize_test.go
index 0236b02..02128c9 100644
--- a/internal/adapter/openai/tool_history_sanitize_test.go
+++ b/internal/adapter/openai/tool_history_sanitize_test.go
@@ -11,10 +11,35 @@ func TestSanitizeLeakedToolHistoryRemovesMarkerBlocks(t *testing.T) {
 }
 
 func TestSanitizeLeakedToolHistoryPreservesChunkWhitespace(t *testing.T) {
-	raw := "Hello "
-	got := sanitizeLeakedToolHistory(raw)
-	if got != "Hello " {
-		t.Fatalf("expected trailing whitespace to be preserved, got %q", got)
+	cases := []struct {
+		name string
+		raw  string
+		want string
+	}{
+		{
+			name: "trailing space kept",
+			raw:  "Hello ",
+			want: "Hello ",
+		},
+		{
+			name: "leading newline kept",
+			raw:  "\nworld",
+			want: "\nworld",
+		},
+		{
+			name: "surrounding whitespace around marker is preserved",
+			raw:  "A \n[TOOL_RESULT_HISTORY]\nfunction.name: exec\nfunction.arguments: {}\n[/TOOL_RESULT_HISTORY]\n B",
+			want: "A \n\n B",
+		},
+	}
+
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			got := sanitizeLeakedToolHistory(tc.raw)
+			if got != tc.want {
+				t.Fatalf("unexpected sanitize result, want %q got %q", tc.want, got)
+			}
+		})
 	}
 }
 
@@ -43,3 +68,31 @@ func TestFlushToolSieveDropsToolResultHistoryLeak(t *testing.T) {
 		t.Fatalf("expected result history block to be swallowed, got %+v", flushed)
 	}
 }
+
+func TestProcessToolSieveChunkSplitsResultHistoryBoundary(t *testing.T) {
+	var state toolStreamSieveState
+	parts := []string{
+		"Hello ",
+		"[TOOL_RESULT_HISTORY]\nstatus: already_called\n",
+		"function.name: exec\nfunction.arguments: {}\n[/TOOL_RESULT_HISTORY]",
+		"world",
+	}
+	var events []toolStreamEvent
+	for _, p := range parts {
+		events = append(events, processToolSieveChunk(&state, p, []string{"exec"})...)
+	}
+	events = append(events, flushToolSieve(&state, []string{"exec"})...)
+
+	var text string
+	for _, evt := range events {
+		if evt.Content != "" {
+			text += evt.Content
+		}
+		if len(evt.ToolCalls) > 0 {
+			t.Fatalf("did not expect parsed tool calls from history leak: %+v", evt.ToolCalls)
+		}
+	}
+	if text != "Hello world" {
+		t.Fatalf("expected clean text output preserving boundary spaces, got %q", text)
+	}
+}
diff --git a/tests/node/stream-tool-sieve.test.js b/tests/node/stream-tool-sieve.test.js
index a4c793c..b24e138 100644
--- a/tests/node/stream-tool-sieve.test.js
+++ b/tests/node/stream-tool-sieve.test.js
@@ -260,6 +260,22 @@ test('sieve swallows leaked TOOL_RESULT_HISTORY marker blocks', () => {
   assert.equal(leakedText.includes('[TOOL_RESULT_HISTORY]'), false);
 });
 
+test('sieve preserves text spacing when TOOL_RESULT_HISTORY spans chunks', () => {
+  const events = runSieve(
+    [
+      'Hello ',
+      '[TOOL_RESULT_HISTORY]\nstatus: already_called\n',
+      'function.name: exec\nfunction.arguments: {}\n[/TOOL_RESULT_HISTORY]',
+      'world',
+    ],
+    ['exec'],
+  );
+  const leakedText = collectText(events);
+  const hasToolCall = events.some((evt) => evt.type === 'tool_calls' && evt.calls?.length > 0);
+  assert.equal(hasToolCall, false);
+  assert.equal(leakedText, 'Hello world');
+});
+
 test('sieve intercepts rejected unknown tool payload (no args) without raw leak', () => {
   const events = runSieve(
     ['{"tool_calls":[{"name":"not_in_schema"}]}', '后置正文G。'],