From 7904412ff2825343520e7e2e9541e89933cfac6e Mon Sep 17 00:00:00 2001
From: Sylvester Kaczmarek
 <16242628+sylvesterkaczmarek@users.noreply.github.com>
Date: Sat, 15 Aug 2026 19:35:00 +0100
Subject: [PATCH] [#17740][fix] Preserve streamed content after wrapped Qwen
 tool calls

Signed-off-by: Sylvester Kaczmarek <16242628+sylvesterkaczmarek@users.noreply.github.com>
---
 .../serve/tool_parser/base_tool_parser.py     |  4 +-
 ...test_qwen3_tool_parser_trailing_content.py | 42 +++++++++++++++++++
 2 files changed, 45 insertions(+), 1 deletion(-)
 create mode 100644 tests/unittest/llmapi/apps/test_qwen3_tool_parser_trailing_content.py

diff --git a/tensorrt_llm/serve/tool_parser/base_tool_parser.py b/tensorrt_llm/serve/tool_parser/base_tool_parser.py
index ece736749d2f..3a08179f982f 100644
--- a/tensorrt_llm/serve/tool_parser/base_tool_parser.py
+++ b/tensorrt_llm/serve/tool_parser/base_tool_parser.py
@@ -135,7 +135,9 @@ def parse_streaming_increment(self, new_text: str,
         # or it is the start of a new tool call after a tool call separator, when there is a previous tool call
         if not (self.has_tool_call(current_text) or
                 (self.current_tool_id > 0
-                 and current_text.startswith(self.tool_call_separator))):
+                 and current_text.startswith(self.tool_call_separator)
+                 and not (self.eot_token
+                          and current_text.startswith(self.eot_token)))):
             # Only clear buffer if we're sure no tool call is starting
             if not self._ends_with_partial_token(self._buffer, self.bot_token):
                 normal_text = self._buffer
diff --git a/tests/unittest/llmapi/apps/test_qwen3_tool_parser_trailing_content.py b/tests/unittest/llmapi/apps/test_qwen3_tool_parser_trailing_content.py
new file mode 100644
index 000000000000..4d1e4a2752ae
--- /dev/null
+++ b/tests/unittest/llmapi/apps/test_qwen3_tool_parser_trailing_content.py
@@ -0,0 +1,42 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+
+import pytest
+
+from tensorrt_llm.serve.openai_protocol import (
+    ChatCompletionToolsParam,
+    FunctionDefinition,
+)
+from tensorrt_llm.serve.tool_parser.qwen3_tool_parser import Qwen3ToolParser
+
+pytestmark = pytest.mark.cpu_only
+
+
+def test_streaming_wrapped_form_preserves_text_after_tool_call():
+    tools = [
+        ChatCompletionToolsParam(
+            type="function",
+            function=FunctionDefinition(
+                name="get_weather",
+                description="Get the current weather",
+                parameters={
+                    "type": "object",
+                    "properties": {"location": {"type": "string"}},
+                    "required": ["location"],
+                },
+            ),
+        )
+    ]
+    parser = Qwen3ToolParser()
+
+    parser.parse_streaming_increment("<tool_call>\n", tools)
+    parser.parse_streaming_increment(
+        '{"name":"get_weather","arguments":{"location":"Paris"}}', tools
+    )
+    parser.parse_streaming_increment("\n</tool_call>", tools)
+
+    result = parser.parse_streaming_increment(" It is sunny.", tools)
+
+    assert result.normal_text == " It is sunny."
+    assert result.calls == []
+    assert parser._buffer == ""