github/OpenHands
1
0
Fork 0
mirror of https://github.com/OpenHands/OpenHands.git synced 2026-03-22 13:47:19 +08:00
Code Issues Packages Projects Releases Wiki Activity

Fix issue #8372: Implement browser screenshot saving functionality (#8383)

Browse Source 
Co-authored-by: openhands <openhands@all-hands.dev>
Co-authored-by: Graham Neubig <neubig@gmail.com>
This commit is contained in:
Xingyao Wang
2025-05-11 15:51:18 +08:00
committed by GitHub
parent a17c57d82e
commit 3d02c0c3a3
6 changed files with 124 additions and 49 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
openhands/events/observation/browse.py
View File

@@ -14,6 +14,7 @@ class BrowserOutputObservation(Observation):
url: str
trigger_by_action: str
screenshot: str = field(repr=False, default='') # don't show in repr
screenshot_path: str | None = field(default=None) # path to saved screenshot file
set_of_marks: str = field(default='', repr=False) # don't show in repr
error: bool = False
observation: str = ObservationType.BROWSE
@@ -49,6 +50,8 @@ class BrowserOutputObservation(Observation):
f'Last browser action error: {self.last_browser_action_error}\n'
f'Focused element bid: {self.focused_element_bid}\n'
)
if self.screenshot_path:
ret += f'Screenshot saved to: {self.screenshot_path}\n'
ret += '--- Agent Observation ---\n'
ret += self.get_agent_obs_text()
return ret
@@ -57,7 +60,14 @@ class BrowserOutputObservation(Observation):
"""Get a concise text that will be shown to the agent."""
if self.trigger_by_action == ActionType.BROWSE_INTERACTIVE:
text = f'[Current URL: {self.url}]\n'
text += f'[Focused element bid: {self.focused_element_bid}]\n\n'
text += f'[Focused element bid: {self.focused_element_bid}]\n'
# Add screenshot path information if available
if self.screenshot_path:
text += f'[Screenshot saved to: {self.screenshot_path}]\n'
text += '\n'
if self.error:
text += (
'================ BEGIN error message ===============\n'
@@ -85,6 +95,7 @@ class BrowserOutputObservation(Observation):
elif self.trigger_by_action == ActionType.BROWSE:
text = f'[Current URL: {self.url}]\n'
if self.error:
text += (
'================ BEGIN error message ===============\n'