fix: invalid new tool call creation logic during response handling in OAI-Compat model (#17781)

api/core/model_runtime/model_providers/__base/large_language_model.py

@@ -1,5 +1,6 @@
 import logging
 import time
+import uuid
 from collections.abc import Generator, Sequence
 from typing import Optional, Union
 
@@ -24,6 +25,58 @@ from core.plugin.manager.model import PluginModelManager
 logger = logging.getLogger(__name__)
 
 
+def _gen_tool_call_id() -> str:
+    return f"chatcmpl-tool-{str(uuid.uuid4().hex)}"
+
+
+def _increase_tool_call(
+    new_tool_calls: list[AssistantPromptMessage.ToolCall], existing_tools_calls: list[AssistantPromptMessage.ToolCall]
+):
+    """
+    Merge incremental tool call updates into existing tool calls.
+
+    :param new_tool_calls: List of new tool call deltas to be merged.
+    :param existing_tools_calls: List of existing tool calls to be modified IN-PLACE.
+    """
+
+    def get_tool_call(tool_call_id: str):
+        """
+        Get or create a tool call by ID
+
+        :param tool_call_id: tool call ID
+        :return: existing or new tool call
+        """
+        if not tool_call_id:
+            return existing_tools_calls[-1]
+
+        _tool_call = next((_tool_call for _tool_call in existing_tools_calls if _tool_call.id == tool_call_id), None)
+        if _tool_call is None:
+            _tool_call = AssistantPromptMessage.ToolCall(
+                id=tool_call_id,
+                type="function",
+                function=AssistantPromptMessage.ToolCall.ToolCallFunction(name="", arguments=""),
+            )
+            existing_tools_calls.append(_tool_call)
+
+        return _tool_call
+
+    for new_tool_call in new_tool_calls:
+        # generate ID for tool calls with function name but no ID to track them
+        if new_tool_call.function.name and not new_tool_call.id:
+            new_tool_call.id = _gen_tool_call_id()
+        # get tool call
+        tool_call = get_tool_call(new_tool_call.id)
+        # update tool call
+        if new_tool_call.id:
+            tool_call.id = new_tool_call.id
+        if new_tool_call.type:
+            tool_call.type = new_tool_call.type
+        if new_tool_call.function.name:
+            tool_call.function.name = new_tool_call.function.name
+        if new_tool_call.function.arguments:
+            tool_call.function.arguments += new_tool_call.function.arguments
+
+
 class LargeLanguageModel(AIModel):
     """
     Model class for large language model.
@@ -109,44 +162,13 @@ class LargeLanguageModel(AIModel):
                 system_fingerprint = None
                 tools_calls: list[AssistantPromptMessage.ToolCall] = []
 
-                def increase_tool_call(new_tool_calls: list[AssistantPromptMessage.ToolCall]):
-                    def get_tool_call(tool_name: str):
-                        if not tool_name:
-                            return tools_calls[-1]
-
-                        tool_call = next(
-                            (tool_call for tool_call in tools_calls if tool_call.function.name == tool_name), None
-                        )
-                        if tool_call is None:
-                            tool_call = AssistantPromptMessage.ToolCall(
-                                id="",
-                                type="",
-                                function=AssistantPromptMessage.ToolCall.ToolCallFunction(name=tool_name, arguments=""),
-                            )
-                            tools_calls.append(tool_call)
-
-                        return tool_call
-
-                    for new_tool_call in new_tool_calls:
-                        # get tool call
-                        tool_call = get_tool_call(new_tool_call.function.name)
-                        # update tool call
-                        if new_tool_call.id:
-                            tool_call.id = new_tool_call.id
-                        if new_tool_call.type:
-                            tool_call.type = new_tool_call.type
-                        if new_tool_call.function.name:
-                            tool_call.function.name = new_tool_call.function.name
-                        if new_tool_call.function.arguments:
-                            tool_call.function.arguments += new_tool_call.function.arguments
-
                 for chunk in result:
                     if isinstance(chunk.delta.message.content, str):
                         content += chunk.delta.message.content
                     elif isinstance(chunk.delta.message.content, list):
                         content_list.extend(chunk.delta.message.content)
                     if chunk.delta.message.tool_calls:
-                        increase_tool_call(chunk.delta.message.tool_calls)
+                        _increase_tool_call(chunk.delta.message.tool_calls, tools_calls)
 
                     usage = chunk.delta.usage or LLMUsage.empty_usage()
                     system_fingerprint = chunk.system_fingerprint

api/tests/unit_tests/core/model_runtime/__base/test_increase_tool_call.py

@@ -0,0 +1,99 @@
+from unittest.mock import MagicMock, patch
+
+from core.model_runtime.entities.message_entities import AssistantPromptMessage
+from core.model_runtime.model_providers.__base.large_language_model import _increase_tool_call
+
+ToolCall = AssistantPromptMessage.ToolCall
+
+# CASE 1: Single tool call
+INPUTS_CASE_1 = [
+    ToolCall(id="1", type="function", function=ToolCall.ToolCallFunction(name="func_foo", arguments="")),
+    ToolCall(id="", type="function", function=ToolCall.ToolCallFunction(name="", arguments='{"arg1": ')),
+    ToolCall(id="", type="function", function=ToolCall.ToolCallFunction(name="", arguments='"value"}')),
+]
+EXPECTED_CASE_1 = [
+    ToolCall(
+        id="1", type="function", function=ToolCall.ToolCallFunction(name="func_foo", arguments='{"arg1": "value"}')
+    ),
+]
+
+# CASE 2: Tool call sequences where IDs are anchored to the first chunk (vLLM/SiliconFlow ...)
+INPUTS_CASE_2 = [
+    ToolCall(id="1", type="function", function=ToolCall.ToolCallFunction(name="func_foo", arguments="")),
+    ToolCall(id="", type="function", function=ToolCall.ToolCallFunction(name="", arguments='{"arg1": ')),
+    ToolCall(id="", type="function", function=ToolCall.ToolCallFunction(name="", arguments='"value"}')),
+    ToolCall(id="2", type="function", function=ToolCall.ToolCallFunction(name="func_bar", arguments="")),
+    ToolCall(id="", type="function", function=ToolCall.ToolCallFunction(name="", arguments='{"arg2": ')),
+    ToolCall(id="", type="function", function=ToolCall.ToolCallFunction(name="", arguments='"value"}')),
+]
+EXPECTED_CASE_2 = [
+    ToolCall(
+        id="1", type="function", function=ToolCall.ToolCallFunction(name="func_foo", arguments='{"arg1": "value"}')
+    ),
+    ToolCall(
+        id="2", type="function", function=ToolCall.ToolCallFunction(name="func_bar", arguments='{"arg2": "value"}')
+    ),
+]
+
+# CASE 3: Tool call sequences where IDs are anchored to every chunk (SGLang ...)
+INPUTS_CASE_3 = [
+    ToolCall(id="1", type="function", function=ToolCall.ToolCallFunction(name="func_foo", arguments="")),
+    ToolCall(id="1", type="function", function=ToolCall.ToolCallFunction(name="", arguments='{"arg1": ')),
+    ToolCall(id="1", type="function", function=ToolCall.ToolCallFunction(name="", arguments='"value"}')),
+    ToolCall(id="2", type="function", function=ToolCall.ToolCallFunction(name="func_bar", arguments="")),
+    ToolCall(id="2", type="function", function=ToolCall.ToolCallFunction(name="", arguments='{"arg2": ')),
+    ToolCall(id="2", type="function", function=ToolCall.ToolCallFunction(name="", arguments='"value"}')),
+]
+EXPECTED_CASE_3 = [
+    ToolCall(
+        id="1", type="function", function=ToolCall.ToolCallFunction(name="func_foo", arguments='{"arg1": "value"}')
+    ),
+    ToolCall(
+        id="2", type="function", function=ToolCall.ToolCallFunction(name="func_bar", arguments='{"arg2": "value"}')
+    ),
+]
+
+# CASE 4: Tool call sequences with no IDs
+INPUTS_CASE_4 = [
+    ToolCall(id="", type="function", function=ToolCall.ToolCallFunction(name="func_foo", arguments="")),
+    ToolCall(id="", type="function", function=ToolCall.ToolCallFunction(name="", arguments='{"arg1": ')),
+    ToolCall(id="", type="function", function=ToolCall.ToolCallFunction(name="", arguments='"value"}')),
+    ToolCall(id="", type="function", function=ToolCall.ToolCallFunction(name="func_bar", arguments="")),
+    ToolCall(id="", type="function", function=ToolCall.ToolCallFunction(name="", arguments='{"arg2": ')),
+    ToolCall(id="", type="function", function=ToolCall.ToolCallFunction(name="", arguments='"value"}')),
+]
+EXPECTED_CASE_4 = [
+    ToolCall(
+        id="RANDOM_ID_1",
+        type="function",
+        function=ToolCall.ToolCallFunction(name="func_foo", arguments='{"arg1": "value"}'),
+    ),
+    ToolCall(
+        id="RANDOM_ID_2",
+        type="function",
+        function=ToolCall.ToolCallFunction(name="func_bar", arguments='{"arg2": "value"}'),
+    ),
+]
+
+
+def _run_case(inputs: list[ToolCall], expected: list[ToolCall]):
+    actual = []
+    _increase_tool_call(inputs, actual)
+    assert actual == expected
+
+
+def test__increase_tool_call():
+    # case 1:
+    _run_case(INPUTS_CASE_1, EXPECTED_CASE_1)
+
+    # case 2:
+    _run_case(INPUTS_CASE_2, EXPECTED_CASE_2)
+
+    # case 3:
+    _run_case(INPUTS_CASE_3, EXPECTED_CASE_3)
+
+    # case 4:
+    mock_id_generator = MagicMock()
+    mock_id_generator.side_effect = [_exp_case.id for _exp_case in EXPECTED_CASE_4]
+    with patch("core.model_runtime.model_providers.__base.large_language_model._gen_tool_call_id", mock_id_generator):
+        _run_case(INPUTS_CASE_4, EXPECTED_CASE_4)