From 2d2729949f08ad7de074f39f163d001a0b4ac16d Mon Sep 17 00:00:00 2001 From: openhands Date: Sun, 24 Nov 2024 23:02:05 +0000 Subject: [PATCH 01/11] Fix issue #5015: [Bug]: Headless mode awaits for requested user feedback without showing any text for what that feedback should be --- openhands/controller/agent_controller.py | 8 +++++++- tests/unit/test_agent_controller.py | 19 +++++++++++++++++++ 2 files changed, 26 insertions(+), 1 deletion(-) diff --git a/openhands/controller/agent_controller.py b/openhands/controller/agent_controller.py index e0fa0dab0384..e068c9d2e4bd 100644 --- a/openhands/controller/agent_controller.py +++ b/openhands/controller/agent_controller.py @@ -307,7 +307,13 @@ async def _handle_message_action(self, action: MessageAction): if self.get_agent_state() != AgentState.RUNNING: await self.set_agent_state_to(AgentState.RUNNING) elif action.source == EventSource.AGENT and action.wait_for_response: - await self.set_agent_state_to(AgentState.AWAITING_USER_INPUT) + if self.headless_mode: + e = RuntimeError( + f'Agent requested user input in headless mode. Message: {action.content}' + ) + await self._react_to_exception(e) + else: + await self.set_agent_state_to(AgentState.AWAITING_USER_INPUT) def reset_task(self): """Resets the agent's task.""" diff --git a/tests/unit/test_agent_controller.py b/tests/unit/test_agent_controller.py index 9c07969bd090..aca847c0ca80 100644 --- a/tests/unit/test_agent_controller.py +++ b/tests/unit/test_agent_controller.py @@ -355,3 +355,22 @@ async def test_step_max_budget_headless(mock_agent, mock_event_stream): # In headless mode, throttling results in an error assert controller.state.agent_state == AgentState.ERROR await controller.close() + + +@pytest.mark.asyncio +async def test_message_action_user_input_headless(mock_agent, mock_event_stream): + controller = AgentController( + agent=mock_agent, + event_stream=mock_event_stream, + max_iterations=10, + sid='test', + confirmation_mode=False, + headless_mode=True, + ) + controller.state.agent_state = AgentState.RUNNING + message_action = MessageAction(content='Test message', wait_for_response=True) + message_action._source = EventSource.AGENT + await controller.on_event(message_action) + # In headless mode, requesting user input results in an error + assert controller.state.agent_state == AgentState.ERROR + await controller.close() From f81c5802b7a9c8e13b00745b8e72e588ce9c426e Mon Sep 17 00:00:00 2001 From: openhands Date: Mon, 25 Nov 2024 00:29:05 +0000 Subject: [PATCH 02/11] =?UTF-8?q?Fix=20pr=20#5246:=20Fix=20issue=20#5015:?= =?UTF-8?q?=20[Bug]:=20Headless=20mode=20awaits=20for=20requested=20user?= =?UTF-8?q?=20feedb=E2=80=A6?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- openhands/controller/agent_controller.py | 3 +++ 1 file changed, 3 insertions(+) diff --git a/openhands/controller/agent_controller.py b/openhands/controller/agent_controller.py index e068c9d2e4bd..e78580c91ece 100644 --- a/openhands/controller/agent_controller.py +++ b/openhands/controller/agent_controller.py @@ -313,6 +313,9 @@ async def _handle_message_action(self, action: MessageAction): ) await self._react_to_exception(e) else: + # Display the message content to help user understand what input is expected + print(f"\nAgent is requesting input: {action.content}") + print("Request user input >> ", end="", flush=True) await self.set_agent_state_to(AgentState.AWAITING_USER_INPUT) def reset_task(self): From 41c0c0203f708dcc19e32418227ed8c6fd3f51c5 Mon Sep 17 00:00:00 2001 From: openhands Date: Mon, 25 Nov 2024 03:40:33 +0000 Subject: [PATCH 03/11] =?UTF-8?q?Fix=20pr=20#5246:=20Fix=20issue=20#5015:?= =?UTF-8?q?=20[Bug]:=20Headless=20mode=20awaits=20for=20requested=20user?= =?UTF-8?q?=20feedb=E2=80=A6?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- openhands/controller/agent_controller.py | 36 +++++++++++++++++------- 1 file changed, 26 insertions(+), 10 deletions(-) diff --git a/openhands/controller/agent_controller.py b/openhands/controller/agent_controller.py index e78580c91ece..38cb67026057 100644 --- a/openhands/controller/agent_controller.py +++ b/openhands/controller/agent_controller.py @@ -89,6 +89,7 @@ def __init__( is_delegate: bool = False, headless_mode: bool = True, status_callback: Callable | None = None, + fake_user_response_fn: Callable[[str], str] | None = None, ): """Initializes a new instance of the AgentController class. @@ -105,11 +106,14 @@ def __init__( initial_state: The initial state of the controller. is_delegate: Whether this controller is a delegate. headless_mode: Whether the agent is run in headless mode. + status_callback: Callback function for status updates. + fake_user_response_fn: Function to generate fake user responses in headless mode. """ self._step_lock = asyncio.Lock() self.id = sid self.agent = agent self.headless_mode = headless_mode + self.fake_user_response_fn = fake_user_response_fn # subscribe to the event stream self.event_stream = event_stream @@ -308,10 +312,19 @@ async def _handle_message_action(self, action: MessageAction): await self.set_agent_state_to(AgentState.RUNNING) elif action.source == EventSource.AGENT and action.wait_for_response: if self.headless_mode: - e = RuntimeError( - f'Agent requested user input in headless mode. Message: {action.content}' - ) - await self._react_to_exception(e) + # In headless mode, we should use a fake user response if provided + if hasattr(self, 'fake_user_response_fn'): + response = self.fake_user_response_fn(action.content) + self.event_stream.add_event( + MessageAction(content=response), + EventSource.USER, + ) + else: + # If no fake response function is provided, we continue with an empty response + self.event_stream.add_event( + MessageAction(content=''), + EventSource.USER, + ) else: # Display the message content to help user understand what input is expected print(f"\nAgent is requesting input: {action.content}") @@ -622,18 +635,21 @@ async def _handle_traffic_control( else: self.state.traffic_control_state = TrafficControlState.THROTTLING if self.headless_mode: - e = RuntimeError( + # In headless mode, we should stop the agent with an error + self.log( + 'warning', f'Agent reached maximum {limit_type} in headless mode. ' - f'Current {limit_type}: {current_value:.2f}, max {limit_type}: {max_value:.2f}' + f'Current {limit_type}: {current_value:.2f}, max {limit_type}: {max_value:.2f}', ) - await self._react_to_exception(e) + await self.set_agent_state_to(AgentState.ERROR) else: - e = RuntimeError( + self.log( + 'warning', f'Agent reached maximum {limit_type}. ' f'Current {limit_type}: {current_value:.2f}, max {limit_type}: {max_value:.2f}. ' + f'{TRAFFIC_CONTROL_REMINDER}', ) - # FIXME: this isn't really an exception--we should have a different path - await self._react_to_exception(e) + await self.set_agent_state_to(AgentState.PAUSED) stop_step = True return stop_step From 7af3f5fb3dd6bf856be8c911e6cb8ec20f162b8d Mon Sep 17 00:00:00 2001 From: openhands Date: Mon, 25 Nov 2024 04:31:27 +0000 Subject: [PATCH 04/11] =?UTF-8?q?Fix=20pr=20#5246:=20Fix=20issue=20#5015:?= =?UTF-8?q?=20[Bug]:=20Headless=20mode=20awaits=20for=20requested=20user?= =?UTF-8?q?=20feedb=E2=80=A6?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- openhands/controller/agent_controller.py | 10 ++++- tests/unit/test_agent_controller.py | 52 +++++++++++++++++++++++- 2 files changed, 58 insertions(+), 4 deletions(-) diff --git a/openhands/controller/agent_controller.py b/openhands/controller/agent_controller.py index 38cb67026057..f0be5bd45e84 100644 --- a/openhands/controller/agent_controller.py +++ b/openhands/controller/agent_controller.py @@ -108,12 +108,18 @@ def __init__( headless_mode: Whether the agent is run in headless mode. status_callback: Callback function for status updates. fake_user_response_fn: Function to generate fake user responses in headless mode. + If not provided and headless_mode is True, a default function will be used. """ self._step_lock = asyncio.Lock() self.id = sid self.agent = agent self.headless_mode = headless_mode - self.fake_user_response_fn = fake_user_response_fn + + # Set up default fake user response function for headless mode + if headless_mode and fake_user_response_fn is None: + self.fake_user_response_fn = lambda _: "continue" + else: + self.fake_user_response_fn = fake_user_response_fn # subscribe to the event stream self.event_stream = event_stream @@ -313,7 +319,7 @@ async def _handle_message_action(self, action: MessageAction): elif action.source == EventSource.AGENT and action.wait_for_response: if self.headless_mode: # In headless mode, we should use a fake user response if provided - if hasattr(self, 'fake_user_response_fn'): + if self.fake_user_response_fn: response = self.fake_user_response_fn(action.content) self.event_stream.add_event( MessageAction(content=response), diff --git a/tests/unit/test_agent_controller.py b/tests/unit/test_agent_controller.py index aca847c0ca80..cece3e02e715 100644 --- a/tests/unit/test_agent_controller.py +++ b/tests/unit/test_agent_controller.py @@ -359,6 +359,7 @@ async def test_step_max_budget_headless(mock_agent, mock_event_stream): @pytest.mark.asyncio async def test_message_action_user_input_headless(mock_agent, mock_event_stream): + # Test with default fake response controller = AgentController( agent=mock_agent, event_stream=mock_event_stream, @@ -371,6 +372,53 @@ async def test_message_action_user_input_headless(mock_agent, mock_event_stream) message_action = MessageAction(content='Test message', wait_for_response=True) message_action._source = EventSource.AGENT await controller.on_event(message_action) - # In headless mode, requesting user input results in an error - assert controller.state.agent_state == AgentState.ERROR + # In headless mode with default fake response, should continue running + assert controller.state.agent_state == AgentState.RUNNING + mock_event_stream.add_event.assert_called_once() + args = mock_event_stream.add_event.call_args[0] + assert isinstance(args[0], MessageAction) + assert args[0].content == "continue" + await controller.close() + + # Test with custom fake response + mock_event_stream.reset_mock() + custom_response = "custom response" + controller = AgentController( + agent=mock_agent, + event_stream=mock_event_stream, + max_iterations=10, + sid='test', + confirmation_mode=False, + headless_mode=True, + fake_user_response_fn=lambda _: custom_response, + ) + controller.state.agent_state = AgentState.RUNNING + message_action = MessageAction(content='Test message', wait_for_response=True) + message_action._source = EventSource.AGENT + await controller.on_event(message_action) + # In headless mode with custom fake response, should continue running + assert controller.state.agent_state == AgentState.RUNNING + mock_event_stream.add_event.assert_called_once() + args = mock_event_stream.add_event.call_args[0] + assert isinstance(args[0], MessageAction) + assert args[0].content == custom_response + await controller.close() + +@pytest.mark.asyncio +async def test_message_action_user_input_non_headless(mock_agent, mock_event_stream): + controller = AgentController( + agent=mock_agent, + event_stream=mock_event_stream, + max_iterations=10, + sid='test', + confirmation_mode=False, + headless_mode=False, + ) + controller.state.agent_state = AgentState.RUNNING + message_action = MessageAction(content='Test message', wait_for_response=True) + message_action._source = EventSource.AGENT + await controller.on_event(message_action) + # In non-headless mode, should wait for user input + assert controller.state.agent_state == AgentState.AWAITING_USER_INPUT + mock_event_stream.add_event.assert_not_called() await controller.close() From d893bea83fc06cf2ee043dd4a4b63e64aee27ad7 Mon Sep 17 00:00:00 2001 From: openhands Date: Mon, 25 Nov 2024 04:41:47 +0000 Subject: [PATCH 05/11] =?UTF-8?q?Fix=20pr=20#5246:=20Fix=20issue=20#5015:?= =?UTF-8?q?=20[Bug]:=20Headless=20mode=20awaits=20for=20requested=20user?= =?UTF-8?q?=20feedb=E2=80=A6?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- frontend/public/config.json | 2 +- .../agenthub/codeact_agent/codeact_agent.py | 4 +- openhands/controller/agent_controller.py | 11 +- openhands/events/action/message.py | 1 + openhands/events/serialization/action.py | 2 +- openhands/resolver/patching/__init__.py | 4 +- openhands/resolver/patching/apply.py | 34 +-- openhands/resolver/patching/exceptions.py | 2 +- openhands/resolver/patching/patch.py | 160 ++++++------ openhands/resolver/patching/snippets.py | 2 +- .../all-hands-ai___openhands-resolver.txt | 2 +- .../repo_instructions/rbren___rss-parser.txt | 2 +- .../prompts/resolve/basic-with-tests.jinja | 2 +- .../resolver/prompts/resolve/basic.jinja | 2 +- openhands/runtime/utils/command.py | 2 +- tests/unit/resolver/test_guess_success.py | 30 +-- .../test_issue_handler_error_handling.py | 65 ++--- tests/unit/resolver/test_issue_references.py | 22 +- .../resolver/test_pr_handler_guess_success.py | 238 +++++++++--------- tests/unit/resolver/test_pr_title_escaping.py | 109 ++++---- tests/unit/test_agent_controller.py | 5 +- 21 files changed, 359 insertions(+), 342 deletions(-) diff --git a/frontend/public/config.json b/frontend/public/config.json index 94900dcbaf31..7dbb7e1d966c 100644 --- a/frontend/public/config.json +++ b/frontend/public/config.json @@ -2,4 +2,4 @@ "APP_MODE": "oss", "GITHUB_CLIENT_ID": "", "POSTHOG_CLIENT_KEY": "phc_3ESMmY9SgqEAGBB6sMGK5ayYHkeUuknH2vP6FmWH9RA" -} \ No newline at end of file +} diff --git a/openhands/agenthub/codeact_agent/codeact_agent.py b/openhands/agenthub/codeact_agent/codeact_agent.py index 39b9e69247be..6743de87ade6 100644 --- a/openhands/agenthub/codeact_agent/codeact_agent.py +++ b/openhands/agenthub/codeact_agent/codeact_agent.py @@ -187,7 +187,9 @@ def get_action_message( ) ] elif isinstance(action, CmdRunAction) and action.source == 'user': - content = [TextContent(text=f'User executed the command:\n{action.command}')] + content = [ + TextContent(text=f'User executed the command:\n{action.command}') + ] return [ Message( role='user', diff --git a/openhands/controller/agent_controller.py b/openhands/controller/agent_controller.py index f0be5bd45e84..3b3daed8793d 100644 --- a/openhands/controller/agent_controller.py +++ b/openhands/controller/agent_controller.py @@ -68,6 +68,7 @@ class AgentController: delegate: 'AgentController | None' = None _pending_action: Action | None = None _closed: bool = False + fake_user_response_fn: Callable[[str], str] | None = None filter_out: ClassVar[tuple[type[Event], ...]] = ( NullAction, NullObservation, @@ -114,10 +115,10 @@ def __init__( self.id = sid self.agent = agent self.headless_mode = headless_mode - + # Set up default fake user response function for headless mode if headless_mode and fake_user_response_fn is None: - self.fake_user_response_fn = lambda _: "continue" + self.fake_user_response_fn = lambda _: 'continue' else: self.fake_user_response_fn = fake_user_response_fn @@ -319,7 +320,7 @@ async def _handle_message_action(self, action: MessageAction): elif action.source == EventSource.AGENT and action.wait_for_response: if self.headless_mode: # In headless mode, we should use a fake user response if provided - if self.fake_user_response_fn: + if self.fake_user_response_fn is not None: response = self.fake_user_response_fn(action.content) self.event_stream.add_event( MessageAction(content=response), @@ -333,8 +334,8 @@ async def _handle_message_action(self, action: MessageAction): ) else: # Display the message content to help user understand what input is expected - print(f"\nAgent is requesting input: {action.content}") - print("Request user input >> ", end="", flush=True) + print(f'\nAgent is requesting input: {action.content}') + print('Request user input >> ', end='', flush=True) await self.set_agent_state_to(AgentState.AWAITING_USER_INPUT) def reset_task(self): diff --git a/openhands/events/action/message.py b/openhands/events/action/message.py index 86d7c439e936..d86526419664 100644 --- a/openhands/events/action/message.py +++ b/openhands/events/action/message.py @@ -24,6 +24,7 @@ def images_urls(self): @images_urls.setter def images_urls(self, value): self.image_urls = value + def __str__(self) -> str: ret = f'**MessageAction** (source={self.source})\n' ret += f'CONTENT: {self.content}' diff --git a/openhands/events/serialization/action.py b/openhands/events/serialization/action.py index defac3b5dda6..f34b4b0ec0cf 100644 --- a/openhands/events/serialization/action.py +++ b/openhands/events/serialization/action.py @@ -69,7 +69,7 @@ def action_from_dict(action: dict) -> Action: # images_urls has been renamed to image_urls if 'images_urls' in args: args['image_urls'] = args.pop('images_urls') - + try: decoded_action = action_class(**args) if 'timeout' in action: diff --git a/openhands/resolver/patching/__init__.py b/openhands/resolver/patching/__init__.py index 5c31f160a0a0..165a623af537 100644 --- a/openhands/resolver/patching/__init__.py +++ b/openhands/resolver/patching/__init__.py @@ -1,6 +1,6 @@ # -*- coding: utf-8 -*- -from .patch import parse_patch from .apply import apply_diff +from .patch import parse_patch -__all__ = ["parse_patch", "apply_diff"] +__all__ = ['parse_patch', 'apply_diff'] diff --git a/openhands/resolver/patching/apply.py b/openhands/resolver/patching/apply.py index f13e814292cb..24f2266f56cf 100644 --- a/openhands/resolver/patching/apply.py +++ b/openhands/resolver/patching/apply.py @@ -10,33 +10,33 @@ def _apply_diff_with_subprocess(diff, lines, reverse=False): # call out to patch program - patchexec = which("patch") + patchexec = which('patch') if not patchexec: - raise SubprocessException("cannot find patch program", code=-1) + raise SubprocessException('cannot find patch program', code=-1) tempdir = tempfile.gettempdir() - filepath = os.path.join(tempdir, "wtp-" + str(hash(diff.header))) - oldfilepath = filepath + ".old" - newfilepath = filepath + ".new" - rejfilepath = filepath + ".rej" - patchfilepath = filepath + ".patch" - with open(oldfilepath, "w") as f: - f.write("\n".join(lines) + "\n") + filepath = os.path.join(tempdir, 'wtp-' + str(hash(diff.header))) + oldfilepath = filepath + '.old' + newfilepath = filepath + '.new' + rejfilepath = filepath + '.rej' + patchfilepath = filepath + '.patch' + with open(oldfilepath, 'w') as f: + f.write('\n'.join(lines) + '\n') - with open(patchfilepath, "w") as f: + with open(patchfilepath, 'w') as f: f.write(diff.text) args = [ patchexec, - "--reverse" if reverse else "--forward", - "--quiet", - "--no-backup-if-mismatch", - "-o", + '--reverse' if reverse else '--forward', + '--quiet', + '--no-backup-if-mismatch', + '-o', newfilepath, - "-i", + '-i', patchfilepath, - "-r", + '-r', rejfilepath, oldfilepath, ] @@ -58,7 +58,7 @@ def _apply_diff_with_subprocess(diff, lines, reverse=False): # do this last to ensure files get cleaned up if ret != 0: - raise SubprocessException("patch program failed", code=ret) + raise SubprocessException('patch program failed', code=ret) return lines, rejlines diff --git a/openhands/resolver/patching/exceptions.py b/openhands/resolver/patching/exceptions.py index 594b079e8365..30653c56da18 100644 --- a/openhands/resolver/patching/exceptions.py +++ b/openhands/resolver/patching/exceptions.py @@ -7,7 +7,7 @@ def __init__(self, msg, hunk=None): self.hunk = hunk if hunk is not None: super(HunkException, self).__init__( - "{msg}, in hunk #{n}".format(msg=msg, n=hunk) + '{msg}, in hunk #{n}'.format(msg=msg, n=hunk) ) else: super(HunkException, self).__init__(msg) diff --git a/openhands/resolver/patching/patch.py b/openhands/resolver/patching/patch.py index c0304e06543b..7e3b98ed0883 100644 --- a/openhands/resolver/patching/patch.py +++ b/openhands/resolver/patching/patch.py @@ -8,67 +8,67 @@ from .snippets import findall_regex, split_by_regex header = namedtuple( - "header", - "index_path old_path old_version new_path new_version", + 'header', + 'index_path old_path old_version new_path new_version', ) -diffobj = namedtuple("diffobj", "header changes text") -Change = namedtuple("Change", "old new line hunk") +diffobj = namedtuple('diffobj', 'header changes text') +Change = namedtuple('Change', 'old new line hunk') -file_timestamp_str = "(.+?)(?:\t|:| +)(.*)" +file_timestamp_str = '(.+?)(?:\t|:| +)(.*)' # .+? was previously [^:\t\n\r\f\v]+ # general diff regex -diffcmd_header = re.compile("^diff.* (.+) (.+)$") -unified_header_index = re.compile("^Index: (.+)$") -unified_header_old_line = re.compile(r"^--- " + file_timestamp_str + "$") -unified_header_new_line = re.compile(r"^\+\+\+ " + file_timestamp_str + "$") -unified_hunk_start = re.compile(r"^@@ -(\d+),?(\d*) \+(\d+),?(\d*) @@(.*)$") -unified_change = re.compile("^([-+ ])(.*)$") - -context_header_old_line = re.compile(r"^\*\*\* " + file_timestamp_str + "$") -context_header_new_line = re.compile("^--- " + file_timestamp_str + "$") -context_hunk_start = re.compile(r"^\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*$") -context_hunk_old = re.compile(r"^\*\*\* (\d+),?(\d*) \*\*\*\*$") -context_hunk_new = re.compile(r"^--- (\d+),?(\d*) ----$") -context_change = re.compile("^([-+ !]) (.*)$") - -ed_hunk_start = re.compile(r"^(\d+),?(\d*)([acd])$") -ed_hunk_end = re.compile("^.$") +diffcmd_header = re.compile('^diff.* (.+) (.+)$') +unified_header_index = re.compile('^Index: (.+)$') +unified_header_old_line = re.compile(r'^--- ' + file_timestamp_str + '$') +unified_header_new_line = re.compile(r'^\+\+\+ ' + file_timestamp_str + '$') +unified_hunk_start = re.compile(r'^@@ -(\d+),?(\d*) \+(\d+),?(\d*) @@(.*)$') +unified_change = re.compile('^([-+ ])(.*)$') + +context_header_old_line = re.compile(r'^\*\*\* ' + file_timestamp_str + '$') +context_header_new_line = re.compile('^--- ' + file_timestamp_str + '$') +context_hunk_start = re.compile(r'^\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*$') +context_hunk_old = re.compile(r'^\*\*\* (\d+),?(\d*) \*\*\*\*$') +context_hunk_new = re.compile(r'^--- (\d+),?(\d*) ----$') +context_change = re.compile('^([-+ !]) (.*)$') + +ed_hunk_start = re.compile(r'^(\d+),?(\d*)([acd])$') +ed_hunk_end = re.compile('^.$') # much like forward ed, but no 'c' type -rcs_ed_hunk_start = re.compile(r"^([ad])(\d+) ?(\d*)$") +rcs_ed_hunk_start = re.compile(r'^([ad])(\d+) ?(\d*)$') -default_hunk_start = re.compile(r"^(\d+),?(\d*)([acd])(\d+),?(\d*)$") -default_hunk_mid = re.compile("^---$") -default_change = re.compile("^([><]) (.*)$") +default_hunk_start = re.compile(r'^(\d+),?(\d*)([acd])(\d+),?(\d*)$') +default_hunk_mid = re.compile('^---$') +default_change = re.compile('^([><]) (.*)$') # Headers # git has a special index header and no end part -git_diffcmd_header = re.compile("^diff --git a/(.+) b/(.+)$") -git_header_index = re.compile(r"^index ([a-f0-9]+)..([a-f0-9]+) ?(\d*)$") -git_header_old_line = re.compile("^--- (.+)$") -git_header_new_line = re.compile(r"^\+\+\+ (.+)$") -git_header_file_mode = re.compile(r"^(new|deleted) file mode \d{6}$") -git_header_binary_file = re.compile("^Binary files (.+) and (.+) differ") -git_binary_patch_start = re.compile(r"^GIT binary patch$") -git_binary_literal_start = re.compile(r"^literal (\d+)$") -git_binary_delta_start = re.compile(r"^delta (\d+)$") -base85string = re.compile(r"^[0-9A-Za-z!#$%&()*+;<=>?@^_`{|}~-]+$") - -bzr_header_index = re.compile("=== (.+)") +git_diffcmd_header = re.compile('^diff --git a/(.+) b/(.+)$') +git_header_index = re.compile(r'^index ([a-f0-9]+)..([a-f0-9]+) ?(\d*)$') +git_header_old_line = re.compile('^--- (.+)$') +git_header_new_line = re.compile(r'^\+\+\+ (.+)$') +git_header_file_mode = re.compile(r'^(new|deleted) file mode \d{6}$') +git_header_binary_file = re.compile('^Binary files (.+) and (.+) differ') +git_binary_patch_start = re.compile(r'^GIT binary patch$') +git_binary_literal_start = re.compile(r'^literal (\d+)$') +git_binary_delta_start = re.compile(r'^delta (\d+)$') +base85string = re.compile(r'^[0-9A-Za-z!#$%&()*+;<=>?@^_`{|}~-]+$') + +bzr_header_index = re.compile('=== (.+)') bzr_header_old_line = unified_header_old_line bzr_header_new_line = unified_header_new_line svn_header_index = unified_header_index -svn_header_timestamp_version = re.compile(r"\((?:working copy|revision (\d+))\)") -svn_header_timestamp = re.compile(r".*(\(.*\))$") +svn_header_timestamp_version = re.compile(r'\((?:working copy|revision (\d+))\)') +svn_header_timestamp = re.compile(r'.*(\(.*\))$') cvs_header_index = unified_header_index -cvs_header_rcs = re.compile(r"^RCS file: (.+)(?:,\w{1}$|$)") -cvs_header_timestamp = re.compile(r"(.+)\t([\d.]+)") -cvs_header_timestamp_colon = re.compile(r":([\d.]+)\t(.+)") -old_cvs_diffcmd_header = re.compile("^diff.* (.+):(.*) (.+):(.*)$") +cvs_header_rcs = re.compile(r'^RCS file: (.+)(?:,\w{1}$|$)') +cvs_header_timestamp = re.compile(r'(.+)\t([\d.]+)') +cvs_header_timestamp_colon = re.compile(r':([\d.]+)\t(.+)') +old_cvs_diffcmd_header = re.compile('^diff.* (.+):(.*) (.+):(.*)$') def parse_patch(text): @@ -97,7 +97,7 @@ def parse_patch(text): break for diff in diffs: - difftext = "\n".join(diff) + "\n" + difftext = '\n'.join(diff) + '\n' h = parse_header(diff) d = parse_diff(diff) if h or d: @@ -133,10 +133,10 @@ def parse_scm_header(text): if res: old_path = res.old_path new_path = res.new_path - if old_path.startswith("a/"): + if old_path.startswith('a/'): old_path = old_path[2:] - if new_path.startswith("b/"): + if new_path.startswith('b/'): new_path = new_path[2:] return header( @@ -240,10 +240,10 @@ def parse_git_header(text): new_path = binary.group(2) if old_path and new_path: - if old_path.startswith("a/"): + if old_path.startswith('a/'): old_path = old_path[2:] - if new_path.startswith("b/"): + if new_path.startswith('b/'): new_path = new_path[2:] return header( index_path=None, @@ -256,19 +256,19 @@ def parse_git_header(text): # if we go through all of the text without finding our normal info, # use the cmd if available if cmd_old_path and cmd_new_path and old_version and new_version: - if cmd_old_path.startswith("a/"): + if cmd_old_path.startswith('a/'): cmd_old_path = cmd_old_path[2:] - if cmd_new_path.startswith("b/"): + if cmd_new_path.startswith('b/'): cmd_new_path = cmd_new_path[2:] return header( index_path=None, # wow, I kind of hate this: # assume /dev/null if the versions are zeroed out - old_path="/dev/null" if old_version == "0000000" else cmd_old_path, + old_path='/dev/null' if old_version == '0000000' else cmd_old_path, old_version=old_version, - new_path="/dev/null" if new_version == "0000000" else cmd_new_path, + new_path='/dev/null' if new_version == '0000000' else cmd_new_path, new_version=new_version, ) @@ -569,10 +569,10 @@ def parse_default_diff(text): kind = c.group(1) line = c.group(2) - if kind == "<" and (r != old_len or r == 0): + if kind == '<' and (r != old_len or r == 0): changes.append(Change(old + r, None, line, hunk_n)) r += 1 - elif kind == ">" and (i != new_len or i == 0): + elif kind == '>' and (i != new_len or i == 0): changes.append(Change(None, new + i, line, hunk_n)) i += 1 @@ -627,13 +627,13 @@ def parse_unified_diff(text): kind = c.group(1) line = c.group(2) - if kind == "-" and (r != old_len or r == 0): + if kind == '-' and (r != old_len or r == 0): changes.append(Change(old + r, None, line, hunk_n)) r += 1 - elif kind == "+" and (i != new_len or i == 0): + elif kind == '+' and (i != new_len or i == 0): changes.append(Change(None, new + i, line, hunk_n)) i += 1 - elif kind == " ": + elif kind == ' ': if r != old_len and i != new_len: changes.append(Change(old + r, new + i, line, hunk_n)) r += 1 @@ -667,7 +667,7 @@ def parse_context_diff(text): k = 0 parts = split_by_regex(hunk, context_hunk_new) if len(parts) != 2: - raise exceptions.ParseException("Context diff invalid", hunk_n) + raise exceptions.ParseException('Context diff invalid', hunk_n) old_hunk = parts[0] new_hunk = parts[1] @@ -695,7 +695,7 @@ def parse_context_diff(text): # now have old and new set, can start processing? if len(old_hunk) > 0 and len(new_hunk) == 0: - msg = "Got unexpected change in removal hunk: " + msg = 'Got unexpected change in removal hunk: ' # only removes left? while len(old_hunk) > 0: c = context_change.match(old_hunk[0]) @@ -707,22 +707,22 @@ def parse_context_diff(text): kind = c.group(1) line = c.group(2) - if kind == "-" and (j != old_len or j == 0): + if kind == '-' and (j != old_len or j == 0): changes.append(Change(old + j, None, line, hunk_n)) j += 1 - elif kind == " " and ( + elif kind == ' ' and ( (j != old_len and k != new_len) or (j == 0 or k == 0) ): changes.append(Change(old + j, new + k, line, hunk_n)) j += 1 k += 1 - elif kind == "+" or kind == "!": + elif kind == '+' or kind == '!': raise exceptions.ParseException(msg + kind, hunk_n) continue if len(old_hunk) == 0 and len(new_hunk) > 0: - msg = "Got unexpected change in removal hunk: " + msg = 'Got unexpected change in removal hunk: ' # only insertions left? while len(new_hunk) > 0: c = context_change.match(new_hunk[0]) @@ -734,16 +734,16 @@ def parse_context_diff(text): kind = c.group(1) line = c.group(2) - if kind == "+" and (k != new_len or k == 0): + if kind == '+' and (k != new_len or k == 0): changes.append(Change(None, new + k, line, hunk_n)) k += 1 - elif kind == " " and ( + elif kind == ' ' and ( (j != old_len and k != new_len) or (j == 0 or k == 0) ): changes.append(Change(old + j, new + k, line, hunk_n)) j += 1 k += 1 - elif kind == "-" or kind == "!": + elif kind == '-' or kind == '!': raise exceptions.ParseException(msg + kind, hunk_n) continue @@ -765,17 +765,17 @@ def parse_context_diff(text): if not (oc or nc): del old_hunk[0] del new_hunk[0] - elif okind == " " and nkind == " " and oline == nline: + elif okind == ' ' and nkind == ' ' and oline == nline: changes.append(Change(old + j, new + k, oline, hunk_n)) j += 1 k += 1 del old_hunk[0] del new_hunk[0] - elif okind == "-" or okind == "!" and (j != old_len or j == 0): + elif okind == '-' or okind == '!' and (j != old_len or j == 0): changes.append(Change(old + j, None, oline, hunk_n)) j += 1 del old_hunk[0] - elif nkind == "+" or nkind == "!" and (k != new_len or k == 0): + elif nkind == '+' or nkind == '!' and (k != new_len or k == 0): changes.append(Change(None, new + k, nline, hunk_n)) k += 1 del new_hunk[0] @@ -821,7 +821,7 @@ def parse_ed_diff(text): old_end = int(o.group(2)) if len(o.group(2)) else old hunk_kind = o.group(3) - if hunk_kind == "d": + if hunk_kind == 'd': k = 0 while old_end >= old: changes.append(Change(old + k, None, None, hunk_n)) @@ -832,7 +832,7 @@ def parse_ed_diff(text): while len(hunk) > 0: e = ed_hunk_end.match(hunk[0]) - if not e and hunk_kind == "c": + if not e and hunk_kind == 'c': k = 0 while old_end >= old: changes.append(Change(old + k, None, None, hunk_n)) @@ -852,7 +852,7 @@ def parse_ed_diff(text): ) i += 1 j += 1 - if not e and hunk_kind == "a": + if not e and hunk_kind == 'a': changes.append( Change( None, @@ -900,7 +900,7 @@ def parse_rcs_ed_diff(text): old = int(o.group(2)) size = int(o.group(3)) - if hunk_kind == "a": + if hunk_kind == 'a': old += total_change_size + 1 total_change_size += size while size > 0 and len(hunk) > 0: @@ -910,7 +910,7 @@ def parse_rcs_ed_diff(text): del hunk[0] - elif hunk_kind == "d": + elif hunk_kind == 'd': total_change_size -= size while size > 0: changes.append(Change(old + j, None, None, hunk_n)) @@ -938,8 +938,8 @@ def parse_git_binary_diff(text): # the sizes are used as latch-up new_size = 0 old_size = 0 - old_encoded = "" - new_encoded = "" + old_encoded = '' + new_encoded = '' for line in lines: if cmd_old_path is None and cmd_new_path is None: hm = git_diffcmd_header.match(line) @@ -978,11 +978,11 @@ def parse_git_binary_diff(text): change = Change(None, 0, added_data, None) changes.append(change) new_size = 0 - new_encoded = "" + new_encoded = '' else: # Invalid line format new_size = 0 - new_encoded = "" + new_encoded = '' # the second is removed file if old_size == 0: @@ -1006,10 +1006,10 @@ def parse_git_binary_diff(text): change = Change(0, None, None, removed_data) changes.append(change) old_size = 0 - old_encoded = "" + old_encoded = '' else: # Invalid line format old_size = 0 - old_encoded = "" + old_encoded = '' return changes diff --git a/openhands/resolver/patching/snippets.py b/openhands/resolver/patching/snippets.py index 710b1191b560..f9d9e620d0f7 100644 --- a/openhands/resolver/patching/snippets.py +++ b/openhands/resolver/patching/snippets.py @@ -54,7 +54,7 @@ def is_exe(fpath): if is_exe(program): return program else: - for path in os.environ["PATH"].split(os.pathsep): + for path in os.environ['PATH'].split(os.pathsep): path = path.strip('"') exe_file = os.path.join(path, program) if is_exe(exe_file): diff --git a/openhands/resolver/prompts/repo_instructions/all-hands-ai___openhands-resolver.txt b/openhands/resolver/prompts/repo_instructions/all-hands-ai___openhands-resolver.txt index ca040d591683..e4bc3b165d28 100644 --- a/openhands/resolver/prompts/repo_instructions/all-hands-ai___openhands-resolver.txt +++ b/openhands/resolver/prompts/repo_instructions/all-hands-ai___openhands-resolver.txt @@ -1,4 +1,4 @@ This is a Python repo for openhands-resolver, a library that attempts to resolve github issues with the AI agent OpenHands. - Setup: `poetry install --with test --with dev` -- Testing: `poetry run pytest tests/test_*.py` \ No newline at end of file +- Testing: `poetry run pytest tests/test_*.py` diff --git a/openhands/resolver/prompts/repo_instructions/rbren___rss-parser.txt b/openhands/resolver/prompts/repo_instructions/rbren___rss-parser.txt index b6e8fba1a200..5ef52d64c35d 100644 --- a/openhands/resolver/prompts/repo_instructions/rbren___rss-parser.txt +++ b/openhands/resolver/prompts/repo_instructions/rbren___rss-parser.txt @@ -1,4 +1,4 @@ This is a node repo for an RSS parser. - Setup: `yes | npm install` - Testing: `SKIP_BROWSER_TESTS=1 npm test` -- Writing Tests: Add to the `test` directory. \ No newline at end of file +- Writing Tests: Add to the `test` directory. diff --git a/openhands/resolver/prompts/resolve/basic-with-tests.jinja b/openhands/resolver/prompts/resolve/basic-with-tests.jinja index 54c35910ec62..595489c42841 100644 --- a/openhands/resolver/prompts/resolve/basic-with-tests.jinja +++ b/openhands/resolver/prompts/resolve/basic-with-tests.jinja @@ -14,4 +14,4 @@ For all changes to actual application code (e.g. in Python or Javascript), add a Run the tests, and if they pass you are done! You do NOT need to write new tests if there are only changes to documentation or configuration files. -When you think you have fixed the issue through code changes, please call the finish action to end the interaction. \ No newline at end of file +When you think you have fixed the issue through code changes, please call the finish action to end the interaction. diff --git a/openhands/resolver/prompts/resolve/basic.jinja b/openhands/resolver/prompts/resolve/basic.jinja index b3bec7ef7f53..a5bb806cc4df 100644 --- a/openhands/resolver/prompts/resolve/basic.jinja +++ b/openhands/resolver/prompts/resolve/basic.jinja @@ -10,4 +10,4 @@ You SHOULD INCLUDE PROPER INDENTATION in your edit commands.{% if repo_instructi Some basic information about this repository: {{ repo_instruction }}{% endif %} -When you think you have fixed the issue through code changes, please finish the interaction. \ No newline at end of file +When you think you have fixed the issue through code changes, please finish the interaction. diff --git a/openhands/runtime/utils/command.py b/openhands/runtime/utils/command.py index 35a1252336c0..3a32d45fb7e1 100644 --- a/openhands/runtime/utils/command.py +++ b/openhands/runtime/utils/command.py @@ -38,7 +38,7 @@ def get_remote_startup_command( '-20', # Highest priority 'sh', '-c', - f'echo -1000 > /proc/self/oom_score_adj && exec {cmd_str}' + f'echo -1000 > /proc/self/oom_score_adj && exec {cmd_str}', ] else: # If not root, run with normal priority diff --git a/tests/unit/resolver/test_guess_success.py b/tests/unit/resolver/test_guess_success.py index 9bf3da2b3d02..d6b0e946adda 100644 --- a/tests/unit/resolver/test_guess_success.py +++ b/tests/unit/resolver/test_guess_success.py @@ -1,22 +1,22 @@ -from openhands.resolver.issue_definitions import IssueHandler -from openhands.resolver.github_issue import GithubIssue -from openhands.events.action.message import MessageAction from openhands.core.config import LLMConfig +from openhands.events.action.message import MessageAction +from openhands.resolver.github_issue import GithubIssue +from openhands.resolver.issue_definitions import IssueHandler def test_guess_success_multiline_explanation(): # Mock data issue = GithubIssue( - owner="test", - repo="test", + owner='test', + repo='test', number=1, - title="Test Issue", - body="Test body", + title='Test Issue', + body='Test body', thread_comments=None, review_comments=None, ) - history = [MessageAction(content="Test message")] - llm_config = LLMConfig(model="test", api_key="test") + history = [MessageAction(content='Test message')] + llm_config = LLMConfig(model='test', api_key='test') # Create a mock response with multi-line explanation mock_response = """--- success @@ -31,7 +31,7 @@ def test_guess_success_multiline_explanation(): Automatic fix generated by OpenHands 🙌""" # Create a handler instance - handler = IssueHandler("test", "test", "test") + handler = IssueHandler('test', 'test', 'test') # Mock the litellm.completion call def mock_completion(*args, **kwargs): @@ -61,11 +61,11 @@ def __init__(self, content): # Verify the results assert success is True - assert "The PR successfully addressed the issue by:" in explanation - assert "Fixed bug A" in explanation - assert "Added test B" in explanation - assert "Updated documentation C" in explanation - assert "Automatic fix generated by OpenHands" in explanation + assert 'The PR successfully addressed the issue by:' in explanation + assert 'Fixed bug A' in explanation + assert 'Added test B' in explanation + assert 'Updated documentation C' in explanation + assert 'Automatic fix generated by OpenHands' in explanation finally: # Restore the original function litellm.completion = original_completion diff --git a/tests/unit/resolver/test_issue_handler_error_handling.py b/tests/unit/resolver/test_issue_handler_error_handling.py index 54adff3466fd..2b04e3b13111 100644 --- a/tests/unit/resolver/test_issue_handler_error_handling.py +++ b/tests/unit/resolver/test_issue_handler_error_handling.py @@ -1,94 +1,97 @@ -import pytest +from unittest.mock import MagicMock, patch + import requests -from unittest.mock import patch, MagicMock from openhands.resolver.issue_definitions import PRHandler -from openhands.resolver.github_issue import ReviewThread def test_handle_nonexistent_issue_reference(): - handler = PRHandler("test-owner", "test-repo", "test-token") - + handler = PRHandler('test-owner', 'test-repo', 'test-token') + # Mock the requests.get to simulate a 404 error mock_response = MagicMock() - mock_response.raise_for_status.side_effect = requests.exceptions.HTTPError("404 Client Error: Not Found") - + mock_response.raise_for_status.side_effect = requests.exceptions.HTTPError( + '404 Client Error: Not Found' + ) + with patch('requests.get', return_value=mock_response): # Call the method with a non-existent issue reference result = handler._PRHandler__get_context_from_external_issues_references( closing_issues=[], closing_issue_numbers=[], - issue_body="This references #999999", # Non-existent issue + issue_body='This references #999999', # Non-existent issue review_comments=[], review_threads=[], - thread_comments=None + thread_comments=None, ) - + # The method should return an empty list since the referenced issue couldn't be fetched assert result == [] def test_handle_rate_limit_error(): - handler = PRHandler("test-owner", "test-repo", "test-token") - + handler = PRHandler('test-owner', 'test-repo', 'test-token') + # Mock the requests.get to simulate a rate limit error mock_response = MagicMock() mock_response.raise_for_status.side_effect = requests.exceptions.HTTPError( - "403 Client Error: Rate Limit Exceeded" + '403 Client Error: Rate Limit Exceeded' ) - + with patch('requests.get', return_value=mock_response): # Call the method with an issue reference result = handler._PRHandler__get_context_from_external_issues_references( closing_issues=[], closing_issue_numbers=[], - issue_body="This references #123", + issue_body='This references #123', review_comments=[], review_threads=[], - thread_comments=None + thread_comments=None, ) - + # The method should return an empty list since the request was rate limited assert result == [] def test_handle_network_error(): - handler = PRHandler("test-owner", "test-repo", "test-token") - + handler = PRHandler('test-owner', 'test-repo', 'test-token') + # Mock the requests.get to simulate a network error - with patch('requests.get', side_effect=requests.exceptions.ConnectionError("Network Error")): + with patch( + 'requests.get', side_effect=requests.exceptions.ConnectionError('Network Error') + ): # Call the method with an issue reference result = handler._PRHandler__get_context_from_external_issues_references( closing_issues=[], closing_issue_numbers=[], - issue_body="This references #123", + issue_body='This references #123', review_comments=[], review_threads=[], - thread_comments=None + thread_comments=None, ) - + # The method should return an empty list since the network request failed assert result == [] def test_successful_issue_reference(): - handler = PRHandler("test-owner", "test-repo", "test-token") - + handler = PRHandler('test-owner', 'test-repo', 'test-token') + # Mock a successful response mock_response = MagicMock() mock_response.raise_for_status.return_value = None - mock_response.json.return_value = {"body": "This is the referenced issue body"} - + mock_response.json.return_value = {'body': 'This is the referenced issue body'} + with patch('requests.get', return_value=mock_response): # Call the method with an issue reference result = handler._PRHandler__get_context_from_external_issues_references( closing_issues=[], closing_issue_numbers=[], - issue_body="This references #123", + issue_body='This references #123', review_comments=[], review_threads=[], - thread_comments=None + thread_comments=None, ) - + # The method should return a list with the referenced issue body - assert result == ["This is the referenced issue body"] \ No newline at end of file + assert result == ['This is the referenced issue body'] diff --git a/tests/unit/resolver/test_issue_references.py b/tests/unit/resolver/test_issue_references.py index e4da644983db..1252f8555540 100644 --- a/tests/unit/resolver/test_issue_references.py +++ b/tests/unit/resolver/test_issue_references.py @@ -2,13 +2,13 @@ def test_extract_issue_references(): - handler = IssueHandler("test-owner", "test-repo", "test-token") + handler = IssueHandler('test-owner', 'test-repo', 'test-token') # Test basic issue reference - assert handler._extract_issue_references("Fixes #123") == [123] + assert handler._extract_issue_references('Fixes #123') == [123] # Test multiple issue references - assert handler._extract_issue_references("Fixes #123, #456") == [123, 456] + assert handler._extract_issue_references('Fixes #123, #456') == [123, 456] # Test issue references in code blocks should be ignored assert handler._extract_issue_references(""" @@ -22,13 +22,21 @@ def func(): """) == [789] # Test issue references in inline code should be ignored - assert handler._extract_issue_references("This `#123` should be ignored but #456 should be extracted") == [456] + assert handler._extract_issue_references( + 'This `#123` should be ignored but #456 should be extracted' + ) == [456] # Test issue references in URLs should be ignored - assert handler._extract_issue_references("Check http://example.com/#123 but #456 should be extracted") == [456] + assert handler._extract_issue_references( + 'Check http://example.com/#123 but #456 should be extracted' + ) == [456] # Test issue references in markdown links should be extracted - assert handler._extract_issue_references("[Link to #123](http://example.com) and #456") == [123, 456] + assert handler._extract_issue_references( + '[Link to #123](http://example.com) and #456' + ) == [123, 456] # Test issue references with text around them - assert handler._extract_issue_references("Issue #123 is fixed and #456 is pending") == [123, 456] + assert handler._extract_issue_references( + 'Issue #123 is fixed and #456 is pending' + ) == [123, 456] diff --git a/tests/unit/resolver/test_pr_handler_guess_success.py b/tests/unit/resolver/test_pr_handler_guess_success.py index bc29fbe2632e..e7e7705e8747 100644 --- a/tests/unit/resolver/test_pr_handler_guess_success.py +++ b/tests/unit/resolver/test_pr_handler_guess_success.py @@ -1,39 +1,39 @@ import json -from unittest.mock import patch, MagicMock +from unittest.mock import MagicMock, patch -from openhands.resolver.issue_definitions import PRHandler -from openhands.resolver.github_issue import GithubIssue, ReviewThread -from openhands.events.action.message import MessageAction from openhands.core.config import LLMConfig +from openhands.events.action.message import MessageAction +from openhands.resolver.github_issue import GithubIssue, ReviewThread +from openhands.resolver.issue_definitions import PRHandler def test_guess_success_review_threads_litellm_call(): """Test that the litellm.completion() call for review threads contains the expected content.""" # Create a PR handler instance - handler = PRHandler("test-owner", "test-repo", "test-token") + handler = PRHandler('test-owner', 'test-repo', 'test-token') # Create a mock issue with review threads issue = GithubIssue( - owner="test-owner", - repo="test-repo", + owner='test-owner', + repo='test-repo', number=1, - title="Test PR", - body="Test Body", + title='Test PR', + body='Test Body', thread_comments=None, - closing_issues=["Issue 1 description", "Issue 2 description"], + closing_issues=['Issue 1 description', 'Issue 2 description'], review_comments=None, review_threads=[ ReviewThread( - comment="Please fix the formatting\n---\nlatest feedback:\nAdd docstrings", - files=["/src/file1.py", "/src/file2.py"], + comment='Please fix the formatting\n---\nlatest feedback:\nAdd docstrings', + files=['/src/file1.py', '/src/file2.py'], ), ReviewThread( - comment="Add more tests\n---\nlatest feedback:\nAdd test cases", - files=["/tests/test_file.py"], + comment='Add more tests\n---\nlatest feedback:\nAdd test cases', + files=['/tests/test_file.py'], ), ], - thread_ids=["1", "2"], - head_branch="test-branch", + thread_ids=['1', '2'], + head_branch='test-branch', ) # Create mock history with a detailed response @@ -47,7 +47,7 @@ def test_guess_success_review_threads_litellm_call(): ] # Create mock LLM config - llm_config = LLMConfig(model="test-model", api_key="test-key") + llm_config = LLMConfig(model='test-model', api_key='test-key') # Mock the LLM response mock_response = MagicMock() @@ -64,7 +64,7 @@ def test_guess_success_review_threads_litellm_call(): ] # Test the guess_success method - with patch("litellm.completion") as mock_completion: + with patch('litellm.completion') as mock_completion: mock_completion.return_value = mock_response success, success_list, explanation = handler.guess_success( issue, history, llm_config @@ -75,63 +75,63 @@ def test_guess_success_review_threads_litellm_call(): # Check first call first_call = mock_completion.call_args_list[0] - first_prompt = first_call[1]["messages"][0]["content"] + first_prompt = first_call[1]['messages'][0]['content'] assert ( - "Issue descriptions:\n" - + json.dumps(["Issue 1 description", "Issue 2 description"], indent=4) + 'Issue descriptions:\n' + + json.dumps(['Issue 1 description', 'Issue 2 description'], indent=4) in first_prompt ) assert ( - "Feedback:\nPlease fix the formatting\n---\nlatest feedback:\nAdd docstrings" + 'Feedback:\nPlease fix the formatting\n---\nlatest feedback:\nAdd docstrings' in first_prompt ) assert ( - "Files locations:\n" - + json.dumps(["/src/file1.py", "/src/file2.py"], indent=4) + 'Files locations:\n' + + json.dumps(['/src/file1.py', '/src/file2.py'], indent=4) in first_prompt ) - assert "Last message from AI agent:\n" + history[0].content in first_prompt + assert 'Last message from AI agent:\n' + history[0].content in first_prompt # Check second call second_call = mock_completion.call_args_list[1] - second_prompt = second_call[1]["messages"][0]["content"] + second_prompt = second_call[1]['messages'][0]['content'] assert ( - "Issue descriptions:\n" - + json.dumps(["Issue 1 description", "Issue 2 description"], indent=4) + 'Issue descriptions:\n' + + json.dumps(['Issue 1 description', 'Issue 2 description'], indent=4) in second_prompt ) assert ( - "Feedback:\nAdd more tests\n---\nlatest feedback:\nAdd test cases" + 'Feedback:\nAdd more tests\n---\nlatest feedback:\nAdd test cases' in second_prompt ) assert ( - "Files locations:\n" + json.dumps(["/tests/test_file.py"], indent=4) + 'Files locations:\n' + json.dumps(['/tests/test_file.py'], indent=4) in second_prompt ) - assert "Last message from AI agent:\n" + history[0].content in second_prompt + assert 'Last message from AI agent:\n' + history[0].content in second_prompt def test_guess_success_thread_comments_litellm_call(): """Test that the litellm.completion() call for thread comments contains the expected content.""" # Create a PR handler instance - handler = PRHandler("test-owner", "test-repo", "test-token") + handler = PRHandler('test-owner', 'test-repo', 'test-token') # Create a mock issue with thread comments issue = GithubIssue( - owner="test-owner", - repo="test-repo", + owner='test-owner', + repo='test-repo', number=1, - title="Test PR", - body="Test Body", + title='Test PR', + body='Test Body', thread_comments=[ - "Please improve error handling", - "Add input validation", - "latest feedback:\nHandle edge cases", + 'Please improve error handling', + 'Add input validation', + 'latest feedback:\nHandle edge cases', ], - closing_issues=["Issue 1 description", "Issue 2 description"], + closing_issues=['Issue 1 description', 'Issue 2 description'], review_comments=None, thread_ids=None, - head_branch="test-branch", + head_branch='test-branch', ) # Create mock history with a detailed response @@ -145,7 +145,7 @@ def test_guess_success_thread_comments_litellm_call(): ] # Create mock LLM config - llm_config = LLMConfig(model="test-model", api_key="test-key") + llm_config = LLMConfig(model='test-model', api_key='test-key') # Mock the LLM response mock_response = MagicMock() @@ -162,7 +162,7 @@ def test_guess_success_thread_comments_litellm_call(): ] # Test the guess_success method - with patch("litellm.completion") as mock_completion: + with patch('litellm.completion') as mock_completion: mock_completion.return_value = mock_response success, success_list, explanation = handler.guess_success( issue, history, llm_config @@ -171,77 +171,77 @@ def test_guess_success_thread_comments_litellm_call(): # Verify the litellm.completion() call mock_completion.assert_called_once() call_args = mock_completion.call_args - prompt = call_args[1]["messages"][0]["content"] + prompt = call_args[1]['messages'][0]['content'] # Check prompt content assert ( - "Issue descriptions:\n" - + json.dumps(["Issue 1 description", "Issue 2 description"], indent=4) + 'Issue descriptions:\n' + + json.dumps(['Issue 1 description', 'Issue 2 description'], indent=4) in prompt ) - assert "PR Thread Comments:\n" + "\n---\n".join(issue.thread_comments) in prompt - assert "Last message from AI agent:\n" + history[0].content in prompt + assert 'PR Thread Comments:\n' + '\n---\n'.join(issue.thread_comments) in prompt + assert 'Last message from AI agent:\n' + history[0].content in prompt def test_check_feedback_with_llm(): """Test the _check_feedback_with_llm helper function.""" # Create a PR handler instance - handler = PRHandler("test-owner", "test-repo", "test-token") + handler = PRHandler('test-owner', 'test-repo', 'test-token') # Create mock LLM config - llm_config = LLMConfig(model="test-model", api_key="test-key") + llm_config = LLMConfig(model='test-model', api_key='test-key') # Test cases for different LLM responses test_cases = [ { - "response": "--- success\ntrue\n--- explanation\nChanges look good", - "expected": (True, "Changes look good"), + 'response': '--- success\ntrue\n--- explanation\nChanges look good', + 'expected': (True, 'Changes look good'), }, { - "response": "--- success\nfalse\n--- explanation\nNot all issues fixed", - "expected": (False, "Not all issues fixed"), + 'response': '--- success\nfalse\n--- explanation\nNot all issues fixed', + 'expected': (False, 'Not all issues fixed'), }, { - "response": "Invalid response format", - "expected": ( + 'response': 'Invalid response format', + 'expected': ( False, - "Failed to decode answer from LLM response: Invalid response format", + 'Failed to decode answer from LLM response: Invalid response format', ), }, { - "response": "--- success\ntrue\n--- explanation\nMultiline\nexplanation\nhere", - "expected": (True, "Multiline\nexplanation\nhere"), + 'response': '--- success\ntrue\n--- explanation\nMultiline\nexplanation\nhere', + 'expected': (True, 'Multiline\nexplanation\nhere'), }, ] for case in test_cases: # Mock the LLM response mock_response = MagicMock() - mock_response.choices = [MagicMock(message=MagicMock(content=case["response"]))] + mock_response.choices = [MagicMock(message=MagicMock(content=case['response']))] # Test the function - with patch("litellm.completion", return_value=mock_response): + with patch('litellm.completion', return_value=mock_response): success, explanation = handler._check_feedback_with_llm( - "test prompt", llm_config + 'test prompt', llm_config ) - assert (success, explanation) == case["expected"] + assert (success, explanation) == case['expected'] def test_check_review_thread(): """Test the _check_review_thread helper function.""" # Create a PR handler instance - handler = PRHandler("test-owner", "test-repo", "test-token") + handler = PRHandler('test-owner', 'test-repo', 'test-token') # Create test data review_thread = ReviewThread( - comment="Please fix the formatting\n---\nlatest feedback:\nAdd docstrings", - files=["/src/file1.py", "/src/file2.py"], + comment='Please fix the formatting\n---\nlatest feedback:\nAdd docstrings', + files=['/src/file1.py', '/src/file2.py'], ) issues_context = json.dumps( - ["Issue 1 description", "Issue 2 description"], indent=4 + ['Issue 1 description', 'Issue 2 description'], indent=4 ) - last_message = "I have fixed the formatting and added docstrings" - llm_config = LLMConfig(model="test-model", api_key="test-key") + last_message = 'I have fixed the formatting and added docstrings' + llm_config = LLMConfig(model='test-model', api_key='test-key') # Mock the LLM response mock_response = MagicMock() @@ -258,7 +258,7 @@ def test_check_review_thread(): ] # Test the function - with patch("litellm.completion") as mock_completion: + with patch('litellm.completion') as mock_completion: mock_completion.return_value = mock_response success, explanation = handler._check_review_thread( review_thread, issues_context, last_message, llm_config @@ -267,37 +267,37 @@ def test_check_review_thread(): # Verify the litellm.completion() call mock_completion.assert_called_once() call_args = mock_completion.call_args - prompt = call_args[1]["messages"][0]["content"] + prompt = call_args[1]['messages'][0]['content'] # Check prompt content - assert "Issue descriptions:\n" + issues_context in prompt - assert "Feedback:\n" + review_thread.comment in prompt + assert 'Issue descriptions:\n' + issues_context in prompt + assert 'Feedback:\n' + review_thread.comment in prompt assert ( - "Files locations:\n" + json.dumps(review_thread.files, indent=4) in prompt + 'Files locations:\n' + json.dumps(review_thread.files, indent=4) in prompt ) - assert "Last message from AI agent:\n" + last_message in prompt + assert 'Last message from AI agent:\n' + last_message in prompt # Check result assert success is True - assert explanation == "Changes look good" + assert explanation == 'Changes look good' def test_check_thread_comments(): """Test the _check_thread_comments helper function.""" # Create a PR handler instance - handler = PRHandler("test-owner", "test-repo", "test-token") + handler = PRHandler('test-owner', 'test-repo', 'test-token') # Create test data thread_comments = [ - "Please improve error handling", - "Add input validation", - "latest feedback:\nHandle edge cases", + 'Please improve error handling', + 'Add input validation', + 'latest feedback:\nHandle edge cases', ] issues_context = json.dumps( - ["Issue 1 description", "Issue 2 description"], indent=4 + ['Issue 1 description', 'Issue 2 description'], indent=4 ) - last_message = "I have added error handling and input validation" - llm_config = LLMConfig(model="test-model", api_key="test-key") + last_message = 'I have added error handling and input validation' + llm_config = LLMConfig(model='test-model', api_key='test-key') # Mock the LLM response mock_response = MagicMock() @@ -314,7 +314,7 @@ def test_check_thread_comments(): ] # Test the function - with patch("litellm.completion") as mock_completion: + with patch('litellm.completion') as mock_completion: mock_completion.return_value = mock_response success, explanation = handler._check_thread_comments( thread_comments, issues_context, last_message, llm_config @@ -323,34 +323,34 @@ def test_check_thread_comments(): # Verify the litellm.completion() call mock_completion.assert_called_once() call_args = mock_completion.call_args - prompt = call_args[1]["messages"][0]["content"] + prompt = call_args[1]['messages'][0]['content'] # Check prompt content - assert "Issue descriptions:\n" + issues_context in prompt - assert "PR Thread Comments:\n" + "\n---\n".join(thread_comments) in prompt - assert "Last message from AI agent:\n" + last_message in prompt + assert 'Issue descriptions:\n' + issues_context in prompt + assert 'PR Thread Comments:\n' + '\n---\n'.join(thread_comments) in prompt + assert 'Last message from AI agent:\n' + last_message in prompt # Check result assert success is True - assert explanation == "Changes look good" + assert explanation == 'Changes look good' def test_check_review_comments(): """Test the _check_review_comments helper function.""" # Create a PR handler instance - handler = PRHandler("test-owner", "test-repo", "test-token") + handler = PRHandler('test-owner', 'test-repo', 'test-token') # Create test data review_comments = [ - "Please improve code readability", - "Add comments to complex functions", - "Follow PEP 8 style guide", + 'Please improve code readability', + 'Add comments to complex functions', + 'Follow PEP 8 style guide', ] issues_context = json.dumps( - ["Issue 1 description", "Issue 2 description"], indent=4 + ['Issue 1 description', 'Issue 2 description'], indent=4 ) - last_message = "I have improved code readability and added comments" - llm_config = LLMConfig(model="test-model", api_key="test-key") + last_message = 'I have improved code readability and added comments' + llm_config = LLMConfig(model='test-model', api_key='test-key') # Mock the LLM response mock_response = MagicMock() @@ -367,7 +367,7 @@ def test_check_review_comments(): ] # Test the function - with patch("litellm.completion") as mock_completion: + with patch('litellm.completion') as mock_completion: mock_completion.return_value = mock_response success, explanation = handler._check_review_comments( review_comments, issues_context, last_message, llm_config @@ -376,39 +376,39 @@ def test_check_review_comments(): # Verify the litellm.completion() call mock_completion.assert_called_once() call_args = mock_completion.call_args - prompt = call_args[1]["messages"][0]["content"] + prompt = call_args[1]['messages'][0]['content'] # Check prompt content - assert "Issue descriptions:\n" + issues_context in prompt - assert "PR Review Comments:\n" + "\n---\n".join(review_comments) in prompt - assert "Last message from AI agent:\n" + last_message in prompt + assert 'Issue descriptions:\n' + issues_context in prompt + assert 'PR Review Comments:\n' + '\n---\n'.join(review_comments) in prompt + assert 'Last message from AI agent:\n' + last_message in prompt # Check result assert success is True - assert explanation == "Changes look good" + assert explanation == 'Changes look good' def test_guess_success_review_comments_litellm_call(): """Test that the litellm.completion() call for review comments contains the expected content.""" # Create a PR handler instance - handler = PRHandler("test-owner", "test-repo", "test-token") + handler = PRHandler('test-owner', 'test-repo', 'test-token') # Create a mock issue with review comments issue = GithubIssue( - owner="test-owner", - repo="test-repo", + owner='test-owner', + repo='test-repo', number=1, - title="Test PR", - body="Test Body", + title='Test PR', + body='Test Body', thread_comments=None, - closing_issues=["Issue 1 description", "Issue 2 description"], + closing_issues=['Issue 1 description', 'Issue 2 description'], review_comments=[ - "Please improve code readability", - "Add comments to complex functions", - "Follow PEP 8 style guide", + 'Please improve code readability', + 'Add comments to complex functions', + 'Follow PEP 8 style guide', ], thread_ids=None, - head_branch="test-branch", + head_branch='test-branch', ) # Create mock history with a detailed response @@ -422,7 +422,7 @@ def test_guess_success_review_comments_litellm_call(): ] # Create mock LLM config - llm_config = LLMConfig(model="test-model", api_key="test-key") + llm_config = LLMConfig(model='test-model', api_key='test-key') # Mock the LLM response mock_response = MagicMock() @@ -439,7 +439,7 @@ def test_guess_success_review_comments_litellm_call(): ] # Test the guess_success method - with patch("litellm.completion") as mock_completion: + with patch('litellm.completion') as mock_completion: mock_completion.return_value = mock_response success, success_list, explanation = handler.guess_success( issue, history, llm_config @@ -448,13 +448,13 @@ def test_guess_success_review_comments_litellm_call(): # Verify the litellm.completion() call mock_completion.assert_called_once() call_args = mock_completion.call_args - prompt = call_args[1]["messages"][0]["content"] + prompt = call_args[1]['messages'][0]['content'] # Check prompt content assert ( - "Issue descriptions:\n" - + json.dumps(["Issue 1 description", "Issue 2 description"], indent=4) + 'Issue descriptions:\n' + + json.dumps(['Issue 1 description', 'Issue 2 description'], indent=4) in prompt ) - assert "PR Review Comments:\n" + "\n---\n".join(issue.review_comments) in prompt - assert "Last message from AI agent:\n" + history[0].content in prompt + assert 'PR Review Comments:\n' + '\n---\n'.join(issue.review_comments) in prompt + assert 'Last message from AI agent:\n' + history[0].content in prompt diff --git a/tests/unit/resolver/test_pr_title_escaping.py b/tests/unit/resolver/test_pr_title_escaping.py index 03f2b7104807..45dd523b036a 100644 --- a/tests/unit/resolver/test_pr_title_escaping.py +++ b/tests/unit/resolver/test_pr_title_escaping.py @@ -1,45 +1,46 @@ -from openhands.resolver.github_issue import GithubIssue -from openhands.resolver.send_pull_request import make_commit import os -import tempfile import subprocess +import tempfile + +from openhands.resolver.github_issue import GithubIssue +from openhands.resolver.send_pull_request import make_commit def test_commit_message_with_quotes(): # Create a temporary directory and initialize git repo with tempfile.TemporaryDirectory() as temp_dir: - subprocess.run(["git", "init", temp_dir], check=True) + subprocess.run(['git', 'init', temp_dir], check=True) # Create a test file and add it to git - test_file = os.path.join(temp_dir, "test.txt") - with open(test_file, "w") as f: - f.write("test content") + test_file = os.path.join(temp_dir, 'test.txt') + with open(test_file, 'w') as f: + f.write('test content') - subprocess.run(["git", "-C", temp_dir, "add", "test.txt"], check=True) + subprocess.run(['git', '-C', temp_dir, 'add', 'test.txt'], check=True) # Create a test issue with problematic title issue = GithubIssue( - owner="test-owner", - repo="test-repo", + owner='test-owner', + repo='test-repo', number=123, title="Issue with 'quotes' and \"double quotes\" and ", - body="Test body", + body='Test body', labels=[], assignees=[], - state="open", - created_at="2024-01-01T00:00:00Z", - updated_at="2024-01-01T00:00:00Z", + state='open', + created_at='2024-01-01T00:00:00Z', + updated_at='2024-01-01T00:00:00Z', closed_at=None, head_branch=None, thread_ids=None, ) # Make the commit - make_commit(temp_dir, issue, "issue") + make_commit(temp_dir, issue, 'issue') # Get the commit message result = subprocess.run( - ["git", "-C", temp_dir, "log", "-1", "--pretty=%B"], + ['git', '-C', temp_dir, 'log', '-1', '--pretty=%B'], capture_output=True, text=True, check=True, @@ -48,7 +49,7 @@ def test_commit_message_with_quotes(): # The commit message should contain the quotes without excessive escaping expected = "Fix issue #123: Issue with 'quotes' and \"double quotes\" and " - assert commit_msg == expected, f"Expected: {expected}\nGot: {commit_msg}" + assert commit_msg == expected, f'Expected: {expected}\nGot: {commit_msg}' def test_pr_title_with_quotes(monkeypatch): @@ -56,39 +57,39 @@ def test_pr_title_with_quotes(monkeypatch): class MockResponse: def __init__(self, status_code=201): self.status_code = status_code - self.text = "" + self.text = '' def json(self): - return {"html_url": "https://github.com/test/test/pull/1"} + return {'html_url': 'https://github.com/test/test/pull/1'} def raise_for_status(self): pass def mock_post(*args, **kwargs): # Verify that the PR title is not over-escaped - data = kwargs.get("json", {}) - title = data.get("title", "") + data = kwargs.get('json', {}) + title = data.get('title', '') expected = "Fix issue #123: Issue with 'quotes' and \"double quotes\" and " assert ( title == expected - ), f"PR title was incorrectly escaped.\nExpected: {expected}\nGot: {title}" + ), f'PR title was incorrectly escaped.\nExpected: {expected}\nGot: {title}' return MockResponse() class MockGetResponse: def __init__(self, status_code=200): self.status_code = status_code - self.text = "" + self.text = '' def json(self): - return {"default_branch": "main"} + return {'default_branch': 'main'} def raise_for_status(self): pass - monkeypatch.setattr("requests.post", mock_post) - monkeypatch.setattr("requests.get", lambda *args, **kwargs: MockGetResponse()) + monkeypatch.setattr('requests.post', mock_post) + monkeypatch.setattr('requests.get', lambda *args, **kwargs: MockGetResponse()) monkeypatch.setattr( - "openhands.resolver.send_pull_request.branch_exists", + 'openhands.resolver.send_pull_request.branch_exists', lambda *args, **kwargs: False, ) @@ -97,69 +98,69 @@ def raise_for_status(self): def mock_run(*args, **kwargs): print(f"Running command: {args[0] if args else kwargs.get('args', [])}") - if isinstance(args[0], list) and args[0][0] == "git": - if "push" in args[0]: + if isinstance(args[0], list) and args[0][0] == 'git': + if 'push' in args[0]: return subprocess.CompletedProcess( - args[0], returncode=0, stdout="", stderr="" + args[0], returncode=0, stdout='', stderr='' ) return original_run(*args, **kwargs) return original_run(*args, **kwargs) - monkeypatch.setattr("subprocess.run", mock_run) + monkeypatch.setattr('subprocess.run', mock_run) # Create a temporary directory and initialize git repo with tempfile.TemporaryDirectory() as temp_dir: - print("Initializing git repo...") - subprocess.run(["git", "init", temp_dir], check=True) + print('Initializing git repo...') + subprocess.run(['git', 'init', temp_dir], check=True) # Add these lines to configure git subprocess.run( - ["git", "-C", temp_dir, "config", "user.name", "Test User"], check=True + ['git', '-C', temp_dir, 'config', 'user.name', 'Test User'], check=True ) subprocess.run( - ["git", "-C", temp_dir, "config", "user.email", "test@example.com"], + ['git', '-C', temp_dir, 'config', 'user.email', 'test@example.com'], check=True, ) # Create a test file and add it to git - test_file = os.path.join(temp_dir, "test.txt") - with open(test_file, "w") as f: - f.write("test content") + test_file = os.path.join(temp_dir, 'test.txt') + with open(test_file, 'w') as f: + f.write('test content') - print("Adding and committing test file...") - subprocess.run(["git", "-C", temp_dir, "add", "test.txt"], check=True) + print('Adding and committing test file...') + subprocess.run(['git', '-C', temp_dir, 'add', 'test.txt'], check=True) subprocess.run( - ["git", "-C", temp_dir, "commit", "-m", "Initial commit"], check=True + ['git', '-C', temp_dir, 'commit', '-m', 'Initial commit'], check=True ) # Create a test issue with problematic title - print("Creating test issue...") + print('Creating test issue...') issue = GithubIssue( - owner="test-owner", - repo="test-repo", + owner='test-owner', + repo='test-repo', number=123, title="Issue with 'quotes' and \"double quotes\" and ", - body="Test body", + body='Test body', labels=[], assignees=[], - state="open", - created_at="2024-01-01T00:00:00Z", - updated_at="2024-01-01T00:00:00Z", + state='open', + created_at='2024-01-01T00:00:00Z', + updated_at='2024-01-01T00:00:00Z', closed_at=None, head_branch=None, thread_ids=None, ) # Try to send a PR - this will fail if the title is incorrectly escaped - print("Sending PR...") - from openhands.resolver.send_pull_request import send_pull_request + print('Sending PR...') from openhands.core.config import LLMConfig + from openhands.resolver.send_pull_request import send_pull_request send_pull_request( github_issue=issue, - github_token="dummy-token", - github_username="test-user", + github_token='dummy-token', + github_username='test-user', patch_dir=temp_dir, - llm_config=LLMConfig(model="test-model", api_key="test-key"), - pr_type="ready", + llm_config=LLMConfig(model='test-model', api_key='test-key'), + pr_type='ready', ) diff --git a/tests/unit/test_agent_controller.py b/tests/unit/test_agent_controller.py index cece3e02e715..99e3820da683 100644 --- a/tests/unit/test_agent_controller.py +++ b/tests/unit/test_agent_controller.py @@ -377,12 +377,12 @@ async def test_message_action_user_input_headless(mock_agent, mock_event_stream) mock_event_stream.add_event.assert_called_once() args = mock_event_stream.add_event.call_args[0] assert isinstance(args[0], MessageAction) - assert args[0].content == "continue" + assert args[0].content == 'continue' await controller.close() # Test with custom fake response mock_event_stream.reset_mock() - custom_response = "custom response" + custom_response = 'custom response' controller = AgentController( agent=mock_agent, event_stream=mock_event_stream, @@ -404,6 +404,7 @@ async def test_message_action_user_input_headless(mock_agent, mock_event_stream) assert args[0].content == custom_response await controller.close() + @pytest.mark.asyncio async def test_message_action_user_input_non_headless(mock_agent, mock_event_stream): controller = AgentController( From dfa625ea36ab5e38461b6fb7c4c28a9174e13370 Mon Sep 17 00:00:00 2001 From: openhands Date: Tue, 26 Nov 2024 04:38:42 +0000 Subject: [PATCH 06/11] =?UTF-8?q?Fix=20pr=20#5246:=20Fix=20issue=20#5015:?= =?UTF-8?q?=20[Bug]:=20Headless=20mode=20awaits=20for=20requested=20user?= =?UTF-8?q?=20feedb=E2=80=A6?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- openhands/controller/agent_controller.py | 15 ++++++--------- 1 file changed, 6 insertions(+), 9 deletions(-) diff --git a/openhands/controller/agent_controller.py b/openhands/controller/agent_controller.py index 83081e7342c6..f08c182122e6 100644 --- a/openhands/controller/agent_controller.py +++ b/openhands/controller/agent_controller.py @@ -646,21 +646,18 @@ async def _handle_traffic_control( else: self.state.traffic_control_state = TrafficControlState.THROTTLING if self.headless_mode: - # In headless mode, we should stop the agent with an error - self.log( - 'warning', + e = RuntimeError( f'Agent reached maximum {limit_type} in headless mode. ' - f'Current {limit_type}: {current_value:.2f}, max {limit_type}: {max_value:.2f}', + f'Current {limit_type}: {current_value:.2f}, max {limit_type}: {max_value:.2f}' ) - await self.set_agent_state_to(AgentState.ERROR) + await self._react_to_exception(e) else: - self.log( - 'warning', + e = RuntimeError( f'Agent reached maximum {limit_type}. ' f'Current {limit_type}: {current_value:.2f}, max {limit_type}: {max_value:.2f}. ' - f'{TRAFFIC_CONTROL_REMINDER}', ) - await self.set_agent_state_to(AgentState.PAUSED) + # FIXME: this isn't really an exception--we should have a different path + await self._react_to_exception(e) stop_step = True return stop_step From 05dad65bdad18811d399c64816b575ab0e11817c Mon Sep 17 00:00:00 2001 From: Engel Nyst Date: Tue, 26 Nov 2024 05:45:29 +0100 Subject: [PATCH 07/11] Update openhands/controller/agent_controller.py --- openhands/controller/agent_controller.py | 1 - 1 file changed, 1 deletion(-) diff --git a/openhands/controller/agent_controller.py b/openhands/controller/agent_controller.py index f08c182122e6..dda5a7b041c3 100644 --- a/openhands/controller/agent_controller.py +++ b/openhands/controller/agent_controller.py @@ -338,7 +338,6 @@ async def _handle_message_action(self, action: MessageAction) -> None: EventSource.USER, ) else: - # Display the message content to help user understand what input is expected print(f'\nAgent is requesting input: {action.content}') print('Request user input >> ', end='', flush=True) await self.set_agent_state_to(AgentState.AWAITING_USER_INPUT) From da818e24f621f853ed1c2a5a14567047435e18aa Mon Sep 17 00:00:00 2001 From: Engel Nyst Date: Tue, 26 Nov 2024 05:45:49 +0100 Subject: [PATCH 08/11] Update openhands/controller/agent_controller.py --- openhands/controller/agent_controller.py | 1 - 1 file changed, 1 deletion(-) diff --git a/openhands/controller/agent_controller.py b/openhands/controller/agent_controller.py index dda5a7b041c3..2bdf9f0c44ad 100644 --- a/openhands/controller/agent_controller.py +++ b/openhands/controller/agent_controller.py @@ -338,7 +338,6 @@ async def _handle_message_action(self, action: MessageAction) -> None: EventSource.USER, ) else: - print(f'\nAgent is requesting input: {action.content}') print('Request user input >> ', end='', flush=True) await self.set_agent_state_to(AgentState.AWAITING_USER_INPUT) From 200d70d5d5ae60ad025f1c7190e3d1ed9512b0d5 Mon Sep 17 00:00:00 2001 From: Engel Nyst Date: Tue, 26 Nov 2024 05:46:11 +0100 Subject: [PATCH 09/11] Update openhands/controller/agent_controller.py --- openhands/controller/agent_controller.py | 1 - 1 file changed, 1 deletion(-) diff --git a/openhands/controller/agent_controller.py b/openhands/controller/agent_controller.py index 2bdf9f0c44ad..5897ef1ef5a9 100644 --- a/openhands/controller/agent_controller.py +++ b/openhands/controller/agent_controller.py @@ -338,7 +338,6 @@ async def _handle_message_action(self, action: MessageAction) -> None: EventSource.USER, ) else: - print('Request user input >> ', end='', flush=True) await self.set_agent_state_to(AgentState.AWAITING_USER_INPUT) def reset_task(self) -> None: From 127636b5ec9cf101898eeb3496d25826b93c5e04 Mon Sep 17 00:00:00 2001 From: openhands Date: Fri, 29 Nov 2024 22:59:58 +0000 Subject: [PATCH 10/11] =?UTF-8?q?Fix=20pr=20#5246:=20Fix=20issue=20#5015:?= =?UTF-8?q?=20[Bug]:=20Headless=20mode=20awaits=20for=20requested=20user?= =?UTF-8?q?=20feedb=E2=80=A6?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- tests/unit/test_agent_controller.py | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/tests/unit/test_agent_controller.py b/tests/unit/test_agent_controller.py index 99e3820da683..65252c67da4e 100644 --- a/tests/unit/test_agent_controller.py +++ b/tests/unit/test_agent_controller.py @@ -421,5 +421,9 @@ async def test_message_action_user_input_non_headless(mock_agent, mock_event_str await controller.on_event(message_action) # In non-headless mode, should wait for user input assert controller.state.agent_state == AgentState.AWAITING_USER_INPUT - mock_event_stream.add_event.assert_not_called() + # Verify that only the state change event is added, but no message action + mock_event_stream.add_event.assert_called_once_with( + mock_event_stream.add_event.call_args[0][0], + EventSource.ENVIRONMENT + ) await controller.close() From 38deae59b6a34b606b141ee71d358cf705bcc3fd Mon Sep 17 00:00:00 2001 From: openhands Date: Fri, 29 Nov 2024 23:15:47 +0000 Subject: [PATCH 11/11] =?UTF-8?q?Fix=20pr=20#5246:=20Fix=20issue=20#5015:?= =?UTF-8?q?=20[Bug]:=20Headless=20mode=20awaits=20for=20requested=20user?= =?UTF-8?q?=20feedb=E2=80=A6?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- tests/unit/test_agent_controller.py | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-) diff --git a/tests/unit/test_agent_controller.py b/tests/unit/test_agent_controller.py index 65252c67da4e..d332fe63e51b 100644 --- a/tests/unit/test_agent_controller.py +++ b/tests/unit/test_agent_controller.py @@ -13,6 +13,7 @@ from openhands.events import Event, EventSource, EventStream, EventStreamSubscriber from openhands.events.action import ChangeAgentStateAction, CmdRunAction, MessageAction from openhands.events.observation import ( + AgentStateChangedObservation, ErrorObservation, ) from openhands.events.serialization import event_to_dict @@ -421,9 +422,11 @@ async def test_message_action_user_input_non_headless(mock_agent, mock_event_str await controller.on_event(message_action) # In non-headless mode, should wait for user input assert controller.state.agent_state == AgentState.AWAITING_USER_INPUT - # Verify that only the state change event is added, but no message action - mock_event_stream.add_event.assert_called_once_with( - mock_event_stream.add_event.call_args[0][0], - EventSource.ENVIRONMENT - ) + # Verify that an AgentStateChangedObservation is added with the correct state + mock_event_stream.add_event.assert_called_once() + args = mock_event_stream.add_event.call_args[0] + assert len(args) == 2 + assert isinstance(args[0], AgentStateChangedObservation) + assert args[0].agent_state == AgentState.AWAITING_USER_INPUT + assert args[1] == EventSource.ENVIRONMENT await controller.close()