solve

Author: Mantisus

CHANGELOG.md

@@ -11,11 +11,13 @@ All notable changes to this project will be documented in this file.
 - Add `block_requests` helper for `PlaywrightCrawler` ([#919](https://github.com/apify/crawlee-python/pull/919)) ([1030459](https://github.com/apify/crawlee-python/commit/103045994908f80cffee5ccfff91a040e0042f48)) by [@Mantisus](https://github.com/Mantisus), closes [#848](https://github.com/apify/crawlee-python/issues/848)
 - Return request handlers from decorator methods to allow further decoration ([#934](https://github.com/apify/crawlee-python/pull/934)) ([9ec0aae](https://github.com/apify/crawlee-python/commit/9ec0aae54e2a340d29c893567ae80bf8bd4510a9)) by [@mylank](https://github.com/mylank)
 - Add `transform_request_function` for `enqueue_links` ([#923](https://github.com/apify/crawlee-python/pull/923)) ([6b15957](https://github.com/apify/crawlee-python/commit/6b159578f612251e6d2253a72b6521430f4f9b09)) by [@Mantisus](https://github.com/Mantisus), closes [#894](https://github.com/apify/crawlee-python/issues/894)
+- Add `time_remaining_secs` property to `MIGRATING` event data ([#940](https://github.com/apify/crawlee-python/pull/940)) ([b44501b](https://github.com/apify/crawlee-python/commit/b44501bcadbd12673a8f47aa92f12da8e404f60b)) by [@fnesveda](https://github.com/fnesveda)
 
 ### 🐛 Bug Fixes
 
 - Fix crawler not retrying user handler if there was timeout in the handler ([#909](https://github.com/apify/crawlee-python/pull/909)) ([f4090ef](https://github.com/apify/crawlee-python/commit/f4090ef0ea0281d53dab16a77ceea2ef6ac43d76)) by [@Pijukatel](https://github.com/Pijukatel), closes [#907](https://github.com/apify/crawlee-python/issues/907)
 - Optimize memory consumption for `HttpxHttpClient`, fix proxy handling ([#905](https://github.com/apify/crawlee-python/pull/905)) ([d7ad480](https://github.com/apify/crawlee-python/commit/d7ad480834263ae0480049cb0a8db4dfc3946d8d)) by [@Mantisus](https://github.com/Mantisus), closes [#895](https://github.com/apify/crawlee-python/issues/895)
+- Fix `BrowserPool` and `PlaywrightBrowserPlugin` closure ([#932](https://github.com/apify/crawlee-python/pull/932)) ([997543d](https://github.com/apify/crawlee-python/commit/997543d2fa5afba49929f4407ee95d7a4933a50d)) by [@Mantisus](https://github.com/Mantisus)
 
 
 <!-- git-cliff-unreleased-end -->

src/crawlee/_utils/models.py

@@ -17,6 +17,14 @@ def _timedelta_to_ms(td: timedelta | None) -> float | None:
     return int(round(td.total_seconds() * 1000))
 
 
+def _timedelta_to_secs(td: timedelta | None) -> float | None:
+    if td == timedelta.max:
+        return float('inf')
+    if td is None:
+        return td
+    return td.total_seconds()
+
+
 _number_parser = TypeAdapter(float)
 
 
@@ -35,4 +43,23 @@ def _timedelta_from_ms(value: float | timedelta | Any | None, handler: Callable[
     return timedelta(milliseconds=value)
 
 
+def _timedelta_from_secs(
+    value: float | timedelta | Any | None,
+    handler: Callable[[Any], timedelta],
+) -> timedelta | None:
+    if value == float('inf'):
+        return timedelta.max
+
+    # If the value is a string-encoded number, decode it
+    if isinstance(value, str):
+        with suppress(ValidationError):
+            value = _number_parser.validate_python(value)
+
+    if not isinstance(value, (int, float)):
+        return handler(value)
+
+    return timedelta(seconds=value)
+
+
 timedelta_ms = Annotated[timedelta, PlainSerializer(_timedelta_to_ms), WrapValidator(_timedelta_from_ms)]
+timedelta_secs = Annotated[timedelta, PlainSerializer(_timedelta_to_secs), WrapValidator(_timedelta_from_secs)]

src/crawlee/browsers/_browser_pool.py

@@ -204,6 +204,8 @@ async def __aexit__(
 
         for browser in self._active_browsers + self._inactive_browsers:
             await browser.close(force=True)
+        self._active_browsers.clear()
+        self._inactive_browsers.clear()
 
         for plugin in self._plugins:
             await plugin.__aexit__(exc_type, exc_value, exc_traceback)

src/crawlee/browsers/_playwright_browser_plugin.py

@@ -135,6 +135,7 @@ async def __aexit__(
             raise RuntimeError(f'The {self.__class__.__name__} is not active.')
 
         await self._playwright_context_manager.__aexit__(exc_type, exc_value, exc_traceback)
+        self._playwright_context_manager = async_playwright()
         self._active = False
 
     @override

src/crawlee/events/_types.py

@@ -7,6 +7,7 @@
 from pydantic import BaseModel, ConfigDict, Field
 
 from crawlee._utils.docs import docs_group
+from crawlee._utils.models import timedelta_secs
 from crawlee._utils.system import CpuInfo, MemoryUsageInfo
 
 
@@ -59,6 +60,10 @@ class EventMigratingData(BaseModel):
 
     model_config = ConfigDict(populate_by_name=True)
 
+    # The remaining time in seconds before the migration is forced and the process is killed
+    # Optional because it's not present when the event handler is called manually
+    time_remaining: Annotated[timedelta_secs | None, Field(alias='timeRemainingSecs')] = None
+
 
 @docs_group('Event payloads')
 class EventAbortingData(BaseModel):