Sanitize leaked tool-call wire format in assistant text

Author: CJackHwang

internal/adapter/openai/tool_history_sanitize.go

@@ -6,12 +6,18 @@ import (
 
 var leakedToolHistoryPattern = regexp.MustCompile(`(?is)\[TOOL_CALL_HISTORY\][\s\S]*?\[/TOOL_CALL_HISTORY\]|\[TOOL_RESULT_HISTORY\][\s\S]*?\[/TOOL_RESULT_HISTORY\]`)
 var emptyJSONFencePattern = regexp.MustCompile("(?is)```json\\s*```")
+var leakedToolCallArrayPattern = regexp.MustCompile(`(?is)\[\{\s*"function"\s*:\s*\{[\s\S]*?\}\s*,\s*"id"\s*:\s*"call[^"]*"\s*,\s*"type"\s*:\s*"function"\s*}\]`)
+var leakedToolResultBlobPattern = regexp.MustCompile(`(?is)<\s*\|\s*tool\s*\|\s*>\s*\{[\s\S]*?"tool_call_id"\s*:\s*"call[^"]*"\s*}`)
+var leakedMetaMarkerPattern = regexp.MustCompile(`(?is)<\s*\|\s*(?:assistant|tool|end_of_sentence|end_of_thinking)\s*\|\s*>`)
 
 func sanitizeLeakedToolHistory(text string) string {
 	if text == "" {
 		return text
 	}
 	out := leakedToolHistoryPattern.ReplaceAllString(text, "")
 	out = emptyJSONFencePattern.ReplaceAllString(out, "")
+	out = leakedToolCallArrayPattern.ReplaceAllString(out, "")
+	out = leakedToolResultBlobPattern.ReplaceAllString(out, "")
+	out = leakedMetaMarkerPattern.ReplaceAllString(out, "")
 	return out
 }

internal/adapter/openai/tool_history_sanitize_test.go

@@ -77,6 +77,22 @@ func TestFlushToolSieveDropsToolResultHistoryLeak(t *testing.T) {
 	}
 }
 
+func TestSanitizeLeakedToolHistoryRemovesLeakedWireToolCallAndResult(t *testing.T) {
+	raw := "开始\n[{\"function\":{\"arguments\":\"{\\\"command\\\":\\\"java -version\\\"}\",\"name\":\"exec\"},\"id\":\"callb9a321\",\"type\":\"function\"}]< | Tool | >{\"content\":\"openjdk version 21\",\"tool_call_id\":\"callb9a321\"}\n结束"
+	got := sanitizeLeakedToolHistory(raw)
+	if got != "开始\n\n结束" {
+		t.Fatalf("unexpected sanitize result for leaked wire format: %q", got)
+	}
+}
+
+func TestSanitizeLeakedToolHistoryRemovesStandaloneMetaMarkers(t *testing.T) {
+	raw := "A<| end_of_sentence |><| Assistant |>B<| end_of_thinking |>C"
+	got := sanitizeLeakedToolHistory(raw)
+	if got != "ABC" {
+		t.Fatalf("unexpected sanitize result for meta markers: %q", got)
+	}
+}
+
 func TestProcessToolSieveChunkSplitsResultHistoryBoundary(t *testing.T) {
 	var state toolStreamSieveState
 	parts := []string{