fix: save context state in result for AdaptivePlaywrightCrawler after isolated processing in SubCrawler (#1488)

### Description

- save context state in result for `AdaptivePlaywrightCrawler` after
isolated processing in `SubCrawler`

### Issues

- Closes: #1483

Author: Mantisus

src/crawlee/crawlers/_adaptive_playwright/_adaptive_playwright_crawler.py

@@ -315,7 +315,7 @@ async def get_input_state(
                 ),
                 logger=self._logger,
             )
-            return SubCrawlerRun(result=result)
+            return SubCrawlerRun(result=result, run_context=context_linked_to_result)
         except Exception as e:
             return SubCrawlerRun(exception=e)
 
@@ -371,7 +371,8 @@ async def _run_request_handler(self, context: BasicCrawlingContext) -> None:
                 self.track_http_only_request_handler_runs()
 
                 static_run = await self._crawl_one(rendering_type='static', context=context)
-                if static_run.result and self.result_checker(static_run.result):
+                if static_run.result and static_run.run_context and self.result_checker(static_run.result):
+                    self._update_context_from_copy(context, static_run.run_context)
                     self._context_result_map[context] = static_run.result
                     return
                 if static_run.exception:
@@ -402,13 +403,10 @@ async def _run_request_handler(self, context: BasicCrawlingContext) -> None:
         if pw_run.exception is not None:
             raise pw_run.exception
 
-        if pw_run.result:
-            self._context_result_map[context] = pw_run.result
-
+        if pw_run.result and pw_run.run_context:
             if should_detect_rendering_type:
                 detection_result: RenderingType
                 static_run = await self._crawl_one('static', context=context, state=old_state_copy)
-
                 if static_run.result and self.result_comparator(static_run.result, pw_run.result):
                     detection_result = 'static'
                 else:
@@ -417,6 +415,9 @@ async def _run_request_handler(self, context: BasicCrawlingContext) -> None:
                 context.log.debug(f'Detected rendering type {detection_result} for {context.request.url}')
                 self.rendering_type_predictor.store_result(context.request, detection_result)
 
+            self._update_context_from_copy(context, pw_run.run_context)
+            self._context_result_map[context] = pw_run.result
+
     def pre_navigation_hook(
         self,
         hook: Callable[[AdaptivePlaywrightPreNavCrawlingContext], Awaitable[None]] | None = None,
@@ -451,8 +452,32 @@ def track_browser_request_handler_runs(self) -> None:
     def track_rendering_type_mispredictions(self) -> None:
         self.statistics.state.rendering_type_mispredictions += 1
 
+    def _update_context_from_copy(self, context: BasicCrawlingContext, context_copy: BasicCrawlingContext) -> None:
+        """Update mutable fields of `context` from `context_copy`.
+
+        Uses object.__setattr__ to bypass frozen dataclass restrictions,
+        allowing state synchronization after isolated crawler execution.
+        """
+        updating_attributes = {
+            'request': ('headers', 'user_data'),
+            'session': ('_user_data', '_usage_count', '_error_score', '_cookies'),
+        }
+
+        for attr, sub_attrs in updating_attributes.items():
+            original_sub_obj = getattr(context, attr)
+            copy_sub_obj = getattr(context_copy, attr)
+
+            # Check that both sub objects are not None
+            if original_sub_obj is None or copy_sub_obj is None:
+                continue
+
+            for sub_attr in sub_attrs:
+                new_value = getattr(copy_sub_obj, sub_attr)
+                object.__setattr__(original_sub_obj, sub_attr, new_value)
+
 
 @dataclass(frozen=True)
 class SubCrawlerRun:
     result: RequestHandlerRunResult | None = None
     exception: Exception | None = None
+    run_context: BasicCrawlingContext | None = None

tests/unit/crawlers/_adaptive_playwright/test_adaptive_playwright_crawler.py

@@ -29,9 +29,10 @@
 from crawlee.crawlers._adaptive_playwright._adaptive_playwright_crawling_context import (
     AdaptiveContextError,
 )
+from crawlee.sessions import SessionPool
 from crawlee.statistics import Statistics
 from crawlee.storage_clients import SqlStorageClient
-from crawlee.storages import KeyValueStore
+from crawlee.storages import KeyValueStore, RequestQueue
 
 if TYPE_CHECKING:
     from collections.abc import AsyncGenerator, Iterator
@@ -730,6 +731,84 @@ async def request_handler(context: AdaptivePlaywrightCrawlingContext) -> None:
     mocked_h3_handler.assert_called_once_with(None)
 
 
+@pytest.mark.parametrize(
+    'test_input',
+    [
+        pytest.param(
+            TestInput(
+                expected_pw_count=0,
+                expected_static_count=2,
+                rendering_types=cycle(['static']),
+                detection_probability_recommendation=cycle([0]),
+            ),
+            id='Static only',
+        ),
+        pytest.param(
+            TestInput(
+                expected_pw_count=2,
+                expected_static_count=0,
+                rendering_types=cycle(['client only']),
+                detection_probability_recommendation=cycle([0]),
+            ),
+            id='Client only',
+        ),
+        pytest.param(
+            TestInput(
+                expected_pw_count=2,
+                expected_static_count=2,
+                rendering_types=cycle(['static', 'client only']),
+                detection_probability_recommendation=cycle([1]),
+            ),
+            id='Enforced rendering type detection',
+        ),
+    ],
+)
+async def test_change_context_state_after_handling(test_input: TestInput, server_url: URL) -> None:
+    """Test that context state is saved after handling the request."""
+    predictor = _SimpleRenderingTypePredictor(
+        rendering_types=test_input.rendering_types,
+        detection_probability_recommendation=test_input.detection_probability_recommendation,
+    )
+
+    request_queue = await RequestQueue.open(name='state-test')
+    used_session_id = None
+
+    async with SessionPool() as session_pool:
+        crawler = AdaptivePlaywrightCrawler.with_beautifulsoup_static_parser(
+            rendering_type_predictor=predictor,
+            session_pool=session_pool,
+            request_manager=request_queue,
+        )
+
+        @crawler.router.default_handler
+        async def request_handler(context: AdaptivePlaywrightCrawlingContext) -> None:
+            nonlocal used_session_id
+
+            if context.session is not None:
+                used_session_id = context.session.id
+                context.session.user_data['session_state'] = True
+
+            if isinstance(context.request.user_data['request_state'], list):
+                context.request.user_data['request_state'].append('handler')
+
+        request = Request.from_url(str(server_url), user_data={'request_state': ['initial']})
+
+        await crawler.run([request])
+
+        assert used_session_id is not None
+
+        session = await session_pool.get_session_by_id(used_session_id)
+        check_request = await request_queue.get_request(request.unique_key)
+
+        assert session is not None
+        assert check_request is not None
+        assert session.user_data.get('session_state') is True
+        # Check that request user data was updated in the handler and only onse.
+        assert check_request.user_data.get('request_state') == ['initial', 'handler']
+
+        await request_queue.drop()
+
+
 async def test_adaptive_playwright_crawler_with_sql_storage(test_urls: list[str], tmp_path: Path) -> None:
     """Tests that AdaptivePlaywrightCrawler can be initialized with SqlStorageClient."""
     storage_dir = tmp_path / 'test_table.db'