make browser download timeout configurable for blocks and tasks (#3619)

skyvern/forge/agent.py

@@ -18,6 +18,7 @@ from playwright.async_api import Page
 from skyvern import analytics
 from skyvern.config import settings
 from skyvern.constants import (
+    BROWSER_DOWNLOAD_TIMEOUT,
     BROWSER_DOWNLOADING_SUFFIX,
     DEFAULT_MAX_SCREENSHOT_SCROLLS,
     GET_DOWNLOADED_FILES_TIMEOUT,
@@ -199,6 +200,7 @@ class ForgeAgent:
             extra_http_headers=workflow_run.extra_http_headers,
             browser_address=workflow_run.browser_address,
             browser_session_id=workflow_run.browser_session_id,
+            download_timeout=task_block.download_timeout,
         )
         LOG.info(
             "Created a new task for workflow run",
@@ -217,6 +219,7 @@ class ForgeAgent:
             organization_id=task.organization_id,
             status=TaskStatus.running,
         )
+
         step = await app.DATABASE.create_step(
             task.task_id,
             order=0,
@@ -500,7 +503,10 @@ class ForgeAgent:
                         step_id=step.step_id,
                     )
                     try:
-                        await wait_for_download_finished(downloading_files=downloading_files)
+                        await wait_for_download_finished(
+                            downloading_files=downloading_files,
+                            timeout=task_block.download_timeout or BROWSER_DOWNLOAD_TIMEOUT,
+                        )
                     except DownloadFileMaxWaitingTime as e:
                         LOG.warning(
                             "There're several long-time downloading files, these files might be broken",

skyvern/forge/sdk/db/client.py

@@ -172,6 +172,7 @@ class AgentDB:
         extra_http_headers: dict[str, str] | None = None,
         browser_session_id: str | None = None,
         browser_address: str | None = None,
+        download_timeout: float | None = None,
     ) -> Task:
         try:
             async with self.Session() as session:
@@ -203,6 +204,7 @@ class AgentDB:
                     extra_http_headers=extra_http_headers,
                     browser_session_id=browser_session_id,
                     browser_address=browser_address,
+                    download_timeout=download_timeout,
                 )
                 session.add(new_task)
                 await session.commit()

skyvern/forge/sdk/db/models.py

@@ -109,6 +109,7 @@ class TaskModel(Base):
     )
     model = Column(JSON, nullable=True)
     browser_address = Column(String, nullable=True)
+    download_timeout = Column(Numeric, nullable=True)
 
 
 class StepModel(Base):

skyvern/forge/sdk/db/utils.py

@@ -158,6 +158,7 @@ def convert_to_task(task_obj: TaskModel, debug_enabled: bool = False, workflow_p
         max_screenshot_scrolls=task_obj.max_screenshot_scrolling_times,
         browser_session_id=task_obj.browser_session_id,
         browser_address=task_obj.browser_address,
+        download_timeout=task_obj.download_timeout,
     )
     return task
 

skyvern/forge/sdk/schemas/tasks.py

@@ -113,6 +113,11 @@ class TaskBase(BaseModel):
         description="The CDP address for the task.",
         examples=["http://127.0.0.1:9222", "ws://127.0.0.1:9222/devtools/browser/1234567890"],
     )
+    download_timeout: float | None = Field(
+        default=None,
+        description="The maximum time to wait for downloads to complete, in minutes. If not set, defaults to BROWSER_DOWNLOAD_TIMEOUT minutes.",
+        examples=[15.0],
+    )
 
 
 class TaskRequest(TaskBase):

skyvern/forge/sdk/workflow/models/block.py

@@ -422,6 +422,7 @@ class BaseTaskBlock(Block):
     cache_actions: bool = False
     complete_verification: bool = True
     include_action_history_in_verification: bool = False
+    download_timeout: float | None = None  # minutes
 
     def get_all_parameters(
         self,
@@ -631,6 +632,7 @@ class BaseTaskBlock(Block):
                         failure_reason=str(e),
                     )
                     raise e
+
                 try:
                     # add screenshot artifact for the first task
                     screenshot = await browser_state.take_fullpage_screenshot(

skyvern/forge/sdk/workflow/service.py

@@ -2490,6 +2490,7 @@ class WorkflowService:
                 cache_actions=block_yaml.cache_actions,
                 complete_on_download=True,
                 complete_verification=True,
+                download_timeout=block_yaml.download_timeout,
             )
         elif block_yaml.block_type == BlockType.TaskV2:
             return TaskV2Block(