mirror of
https://github.com/bytedance/deer-flow.git
synced 2026-06-09 17:12:01 +00:00
3b105d1e5f
Reasoning models such as MiniMax-M3 inline their chain-of-thought into the
message content as <think>...</think> (reasoning_split defaults to false)
instead of a separate reasoning_content field. The follow-up-suggestions
endpoint extracted the JSON array via find('[') / rfind(']'), which silently
broke whenever the reasoning text contained '[' or ']' — or when long thinking
hit max_tokens and truncated before the array was emitted — returning empty
suggestions.
- Add _strip_think_blocks() and apply it before JSON extraction; it removes
complete <think>...</think> blocks (case-insensitive) and drops an unclosed
<think> left by max_tokens truncation.
- Document the MiniMax thinking toggle in config.example.yaml
(when_thinking_enabled: adaptive / when_thinking_disabled: disabled) so
thinking_enabled=False actually disables reasoning on M3; note that M2.x
models always think and rely on the defensive strip above.
- Tests cover complete/unclosed think blocks, brackets-inside-think, think +
code-fence, and an end-to-end suggestions case reproducing the empty-result
bug.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
172 lines
7.3 KiB
Python
172 lines
7.3 KiB
Python
import asyncio
|
|
from types import SimpleNamespace
|
|
from unittest.mock import AsyncMock, MagicMock
|
|
|
|
from app.gateway.routers import suggestions
|
|
|
|
|
|
def test_strip_markdown_code_fence_removes_wrapping():
|
|
text = '```json\n["a"]\n```'
|
|
assert suggestions._strip_markdown_code_fence(text) == '["a"]'
|
|
|
|
|
|
def test_strip_markdown_code_fence_no_fence_keeps_content():
|
|
text = ' ["a"] '
|
|
assert suggestions._strip_markdown_code_fence(text) == '["a"]'
|
|
|
|
|
|
def test_parse_json_string_list_filters_invalid_items():
|
|
text = '```json\n["a", " ", 1, "b"]\n```'
|
|
assert suggestions._parse_json_string_list(text) == ["a", "b"]
|
|
|
|
|
|
def test_parse_json_string_list_rejects_non_list():
|
|
text = '{"a": 1}'
|
|
assert suggestions._parse_json_string_list(text) is None
|
|
|
|
|
|
def test_strip_think_blocks_removes_complete_block():
|
|
text = "<think>\nreasoning here\n</think>\nanswer"
|
|
assert suggestions._strip_think_blocks(text) == "answer"
|
|
|
|
|
|
def test_strip_think_blocks_is_case_insensitive():
|
|
text = "<Think>reasoning</THINK>\nanswer"
|
|
assert suggestions._strip_think_blocks(text) == "answer"
|
|
|
|
|
|
def test_strip_think_blocks_drops_unclosed_block():
|
|
# Reasoning models truncated at max_tokens emit an unclosed <think>.
|
|
text = "<think>\nreasoning that never finished because tokens ran out"
|
|
assert suggestions._strip_think_blocks(text) == ""
|
|
|
|
|
|
def test_strip_think_blocks_keeps_text_without_think():
|
|
text = '["a", "b"]'
|
|
assert suggestions._strip_think_blocks(text) == '["a", "b"]'
|
|
|
|
|
|
def test_parse_json_string_list_ignores_brackets_inside_think_block():
|
|
# MiniMax-M3 inlines its chain-of-thought as <think>...</think> in content
|
|
# (reasoning_split=false). When that reasoning contains '[' / ']', the old
|
|
# find('[')/rfind(']') logic grabbed the wrong span and parsing failed.
|
|
text = '<think>\nMaybe a list like ["x", "y"] could work. Let me craft 3.\n</think>\n["Q1", "Q2", "Q3"]'
|
|
assert suggestions._parse_json_string_list(text) == ["Q1", "Q2", "Q3"]
|
|
|
|
|
|
def test_parse_json_string_list_strips_think_then_code_fence():
|
|
text = '<think>reasoning</think>\n```json\n["Q1", "Q2"]\n```'
|
|
assert suggestions._parse_json_string_list(text) == ["Q1", "Q2"]
|
|
|
|
|
|
def test_generate_suggestions_strips_inline_think_block(monkeypatch):
|
|
# End-to-end: model returns thinking inline followed by the JSON array.
|
|
req = suggestions.SuggestionsRequest(
|
|
messages=[
|
|
suggestions.SuggestionMessage(role="user", content="介绍深度学习"),
|
|
suggestions.SuggestionMessage(role="assistant", content="深度学习是机器学习的分支。"),
|
|
],
|
|
n=3,
|
|
model_name=None,
|
|
)
|
|
content = '<think>\nThe user asked about deep learning. Options: maybe [1] frameworks, [2] math basics.\n</think>\n["深度学习和机器学习的区别？", "常用框架有哪些？", "需要什么数学基础？"]'
|
|
fake_model = MagicMock()
|
|
fake_model.ainvoke = AsyncMock(return_value=MagicMock(content=content))
|
|
monkeypatch.setattr(suggestions, "create_chat_model", lambda **kwargs: fake_model)
|
|
|
|
result = asyncio.run(suggestions.generate_suggestions.__wrapped__("t1", req, request=None, config=SimpleNamespace()))
|
|
|
|
assert result.suggestions == ["深度学习和机器学习的区别？", "常用框架有哪些？", "需要什么数学基础？"]
|
|
|
|
|
|
def test_format_conversation_formats_roles():
|
|
messages = [
|
|
suggestions.SuggestionMessage(role="User", content="Hi"),
|
|
suggestions.SuggestionMessage(role="assistant", content="Hello"),
|
|
suggestions.SuggestionMessage(role="system", content="note"),
|
|
]
|
|
assert suggestions._format_conversation(messages) == "User: Hi\nAssistant: Hello\nsystem: note"
|
|
|
|
|
|
def test_generate_suggestions_parses_and_limits(monkeypatch):
|
|
req = suggestions.SuggestionsRequest(
|
|
messages=[
|
|
suggestions.SuggestionMessage(role="user", content="Hi"),
|
|
suggestions.SuggestionMessage(role="assistant", content="Hello"),
|
|
],
|
|
n=3,
|
|
model_name=None,
|
|
)
|
|
fake_model = MagicMock()
|
|
fake_model.ainvoke = AsyncMock(return_value=MagicMock(content='```json\n["Q1", "Q2", "Q3", "Q4"]\n```'))
|
|
monkeypatch.setattr(suggestions, "create_chat_model", lambda **kwargs: fake_model)
|
|
|
|
# Bypass the require_permission decorator (which needs request +
|
|
# thread_store) — these tests cover the parsing logic.
|
|
result = asyncio.run(suggestions.generate_suggestions.__wrapped__("t1", req, request=None, config=SimpleNamespace()))
|
|
|
|
assert result.suggestions == ["Q1", "Q2", "Q3"]
|
|
fake_model.ainvoke.assert_awaited_once()
|
|
assert fake_model.ainvoke.await_args.kwargs["config"] == {"run_name": "suggest_agent"}
|
|
|
|
|
|
def test_generate_suggestions_parses_list_block_content(monkeypatch):
|
|
req = suggestions.SuggestionsRequest(
|
|
messages=[
|
|
suggestions.SuggestionMessage(role="user", content="Hi"),
|
|
suggestions.SuggestionMessage(role="assistant", content="Hello"),
|
|
],
|
|
n=2,
|
|
model_name=None,
|
|
)
|
|
fake_model = MagicMock()
|
|
fake_model.ainvoke = AsyncMock(return_value=MagicMock(content=[{"type": "text", "text": '```json\n["Q1", "Q2"]\n```'}]))
|
|
monkeypatch.setattr(suggestions, "create_chat_model", lambda **kwargs: fake_model)
|
|
|
|
# Bypass the require_permission decorator (which needs request +
|
|
# thread_store) — these tests cover the parsing logic.
|
|
result = asyncio.run(suggestions.generate_suggestions.__wrapped__("t1", req, request=None, config=SimpleNamespace()))
|
|
|
|
assert result.suggestions == ["Q1", "Q2"]
|
|
fake_model.ainvoke.assert_awaited_once()
|
|
assert fake_model.ainvoke.await_args.kwargs["config"] == {"run_name": "suggest_agent"}
|
|
|
|
|
|
def test_generate_suggestions_parses_output_text_block_content(monkeypatch):
|
|
req = suggestions.SuggestionsRequest(
|
|
messages=[
|
|
suggestions.SuggestionMessage(role="user", content="Hi"),
|
|
suggestions.SuggestionMessage(role="assistant", content="Hello"),
|
|
],
|
|
n=2,
|
|
model_name=None,
|
|
)
|
|
fake_model = MagicMock()
|
|
fake_model.ainvoke = AsyncMock(return_value=MagicMock(content=[{"type": "output_text", "text": '```json\n["Q1", "Q2"]\n```'}]))
|
|
monkeypatch.setattr(suggestions, "create_chat_model", lambda **kwargs: fake_model)
|
|
|
|
# Bypass the require_permission decorator (which needs request +
|
|
# thread_store) — these tests cover the parsing logic.
|
|
result = asyncio.run(suggestions.generate_suggestions.__wrapped__("t1", req, request=None, config=SimpleNamespace()))
|
|
|
|
assert result.suggestions == ["Q1", "Q2"]
|
|
fake_model.ainvoke.assert_awaited_once()
|
|
assert fake_model.ainvoke.await_args.kwargs["config"] == {"run_name": "suggest_agent"}
|
|
|
|
|
|
def test_generate_suggestions_returns_empty_on_model_error(monkeypatch):
|
|
req = suggestions.SuggestionsRequest(
|
|
messages=[suggestions.SuggestionMessage(role="user", content="Hi")],
|
|
n=2,
|
|
model_name=None,
|
|
)
|
|
fake_model = MagicMock()
|
|
fake_model.ainvoke = AsyncMock(side_effect=RuntimeError("boom"))
|
|
monkeypatch.setattr(suggestions, "create_chat_model", lambda **kwargs: fake_model)
|
|
|
|
# Bypass the require_permission decorator (which needs request +
|
|
# thread_store) — these tests cover the parsing logic.
|
|
result = asyncio.run(suggestions.generate_suggestions.__wrapped__("t1", req, request=None, config=SimpleNamespace()))
|
|
|
|
assert result.suggestions == []
|