crawlee 1.0.5b13__py3-none-any.whl → 1.0.5b15__py3-none-any.whl

crawlee/configuration.py

@@ -28,7 +28,9 @@ class Configuration(BaseSettings):
     Settings can also be configured via environment variables, prefixed with `CRAWLEE_`.
     """
 
-    model_config = SettingsConfigDict(validate_by_name=True, validate_by_alias=True)
+    # TODO: https://github.com/pydantic/pydantic-settings/issues/706
+    # Use `SettingsConfigDict(validate_by_name=True, validate_by_alias=True)` when issue is resolved.
+    model_config = SettingsConfigDict(populate_by_name=True)
 
     internal_timeout: Annotated[timedelta | None, Field(alias='crawlee_internal_timeout')] = None
     """Timeout for the internal asynchronous operations."""

crawlee/crawlers/_adaptive_playwright/_adaptive_playwright_crawler.py

@@ -71,7 +71,6 @@ class _NonPersistentStatistics(Statistics):
     async def __aenter__(self) -> Self:
         self._active = True
         await self._state.initialize()
-        self._after_initialize()
         return self
 
     async def __aexit__(

crawlee/statistics/_statistics.py

@@ -1,6 +1,7 @@
 # Inspiration: https://github.com/apify/crawlee/blob/v3.9.2/packages/core/src/crawlers/statistics.ts
 from __future__ import annotations
 
+import asyncio
 import math
 import time
 from datetime import datetime, timedelta, timezone
@@ -84,8 +85,6 @@ class Statistics(Generic[TStatisticsState]):
         self._id = Statistics.__next_id
         Statistics.__next_id += 1
 
-        self._instance_start: datetime | None = None
-
         self.error_tracker = ErrorTracker(
             save_error_snapshots=save_error_snapshots,
             snapshot_kvs_name=persist_state_kvs_name,
@@ -111,6 +110,9 @@ class Statistics(Generic[TStatisticsState]):
         # Flag to indicate the context state.
         self._active = False
 
+        # Pre-existing runtime offset, that can be non-zero when restoring serialized state from KVS.
+        self._runtime_offset = timedelta(seconds=0)
+
     def replace_state_model(self, state_model: type[TNewStatisticsState]) -> Statistics[TNewStatisticsState]:
         """Create near copy of the `Statistics` with replaced `state_model`."""
         new_statistics: Statistics[TNewStatisticsState] = Statistics(
@@ -165,14 +167,17 @@ class Statistics(Generic[TStatisticsState]):
         if self._active:
             raise RuntimeError(f'The {self.__class__.__name__} is already active.')
 
-        self._active = True
-        self._instance_start = datetime.now(timezone.utc)
-
         await self._state.initialize()
-        self._after_initialize()
 
+        self._runtime_offset = self.state.crawler_runtime
+
+        # Start periodic logging and let it print initial state before activation.
         self._periodic_logger.start()
+        await asyncio.sleep(0.01)
+        self._active = True
 
+        self.state.crawler_last_started_at = datetime.now(timezone.utc)
+        self.state.crawler_started_at = self.state.crawler_started_at or self.state.crawler_last_started_at
         return self
 
     async def __aexit__(
@@ -191,14 +196,16 @@ class Statistics(Generic[TStatisticsState]):
 
         if not self.state.crawler_last_started_at:
             raise RuntimeError('Statistics.state.crawler_last_started_at not set.')
-        self.state.crawler_finished_at = datetime.now(timezone.utc)
-        self.state.crawler_runtime += self.state.crawler_finished_at - self.state.crawler_last_started_at
-
-        await self._state.teardown()
 
+        # Stop logging and deactivate the statistics to prevent further changes to crawler_runtime
         await self._periodic_logger.stop()
+        self.state.crawler_finished_at = datetime.now(timezone.utc)
+        self.state.crawler_runtime = (
+            self._runtime_offset + self.state.crawler_finished_at - self.state.crawler_last_started_at
+        )
 
         self._active = False
+        await self._state.teardown()
 
     @property
     def state(self) -> TStatisticsState:
@@ -255,10 +262,19 @@ class Statistics(Generic[TStatisticsState]):
 
         del self._requests_in_progress[request_id_or_key]
 
+    def _update_crawler_runtime(self) -> None:
+        current_run_duration = (
+            (datetime.now(timezone.utc) - self.state.crawler_last_started_at)
+            if self.state.crawler_last_started_at
+            else timedelta()
+        )
+        self.state.crawler_runtime = current_run_duration + self._runtime_offset
+
     def calculate(self) -> FinalStatistics:
         """Calculate the current statistics."""
-        if self._instance_start is None:
-            raise RuntimeError('The Statistics object is not initialized')
+        if self._active:
+            # Only update state when active. If not, just report the last known runtime.
+            self._update_crawler_runtime()
 
         total_minutes = self.state.crawler_runtime.total_seconds() / 60
         state = self._state.current_value
@@ -291,21 +307,6 @@ class Statistics(Generic[TStatisticsState]):
         else:
             self._periodic_message_logger.info(self._log_message, extra=stats.to_dict())
 
-    def _after_initialize(self) -> None:
-        state = self._state.current_value
-
-        if state.crawler_started_at is None:
-            state.crawler_started_at = datetime.now(timezone.utc)
-
-        if state.stats_persisted_at is not None and state.crawler_last_started_at:
-            self._instance_start = datetime.now(timezone.utc) - (
-                state.stats_persisted_at - state.crawler_last_started_at
-            )
-        elif state.crawler_last_started_at:
-            self._instance_start = state.crawler_last_started_at
-
-        state.crawler_last_started_at = self._instance_start
-
     def _save_retry_count_for_request(self, record: RequestProcessingRecord) -> None:
         retry_count = record.retry_count
         state = self._state.current_value

{crawlee-1.0.5b13.dist-info → crawlee-1.0.5b15.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: crawlee
-Version: 1.0.5b13
+Version: 1.0.5b15
 Summary: Crawlee for Python
 Project-URL: Apify Homepage, https://apify.com
 Project-URL: Changelog, https://crawlee.dev/python/docs/changelog
@@ -231,7 +231,7 @@ Requires-Dist: impit>=0.6.1
 Requires-Dist: more-itertools>=10.2.0
 Requires-Dist: protego>=0.5.0
 Requires-Dist: psutil>=6.0.0
-Requires-Dist: pydantic-settings!=2.7.0,!=2.7.1,!=2.8.0,>=2.2.0
+Requires-Dist: pydantic-settings>=2.12.0
 Requires-Dist: pydantic>=2.11.0
 Requires-Dist: pyee>=9.0.0
 Requires-Dist: tldextract>=5.1.0

{crawlee-1.0.5b13.dist-info → crawlee-1.0.5b15.dist-info}/RECORD

@@ -6,7 +6,7 @@ crawlee/_log_config.py,sha256=VyxoEfWCq_9fyicmmJbjiZ5KC91onMcAtX2L4oKX4m4,5999
 crawlee/_request.py,sha256=XliqiaL5Gp3fIDqHVVw0GF35VydXOtg6wJIkeaLcAwk,16458
 crawlee/_service_locator.py,sha256=SJ8ABYtclBl7rz8kfZ2jZkIgKq5oNIoGT7WmN8ApTzo,5058
 crawlee/_types.py,sha256=DAmfSv5W1dt3nJhJ8z-02gDaE06fdEizNKUlHpsd2_A,29129
-crawlee/configuration.py,sha256=KG_XDkPe1VaYfaIu41nICvMjfHbDKM0h4-YTi3DkyRY,7917
+crawlee/configuration.py,sha256=DWS2z1FC6Ua93W2tStK3R1ZKZbZjVQYWGiGFbZFaRtA,8064
 crawlee/errors.py,sha256=RhFNA_uT615nVBHf9TylpX5YWwtDuHUUEV8LPT4CYa4,3878
 crawlee/proxy_configuration.py,sha256=rqf67yerXvLvraBaAHW04nvf5ECze3wMQbK7LlqXucM,10386
 crawlee/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -58,7 +58,7 @@ crawlee/crawlers/_abstract_http/_abstract_http_parser.py,sha256=Y5o_hiW_0mQAte5G
 crawlee/crawlers/_abstract_http/_http_crawling_context.py,sha256=Rno_uJ8ivmyRxFQv2MyY_z9B5WPHSEd5MAPz31_1ZIo,2179
 crawlee/crawlers/_abstract_http/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 crawlee/crawlers/_adaptive_playwright/__init__.py,sha256=LREq9WR9BKsE8S8lSsEhlCoNjQaLhlJ9yo8y_6a8o4c,1072
-crawlee/crawlers/_adaptive_playwright/_adaptive_playwright_crawler.py,sha256=ME90JLkScWj_ynUymA59f832vEvvVpkP01cYfEc8m-Y,22895
+crawlee/crawlers/_adaptive_playwright/_adaptive_playwright_crawler.py,sha256=nPFB9Q_3xQDJprb24NIQO53gf56J8wXjbM9C-58iiZ8,22862
 crawlee/crawlers/_adaptive_playwright/_adaptive_playwright_crawler_statistics.py,sha256=_At8T8S3JLGPA-1AeCFGrpE-FuCDW9sazrXt9U0tK6U,1048
 crawlee/crawlers/_adaptive_playwright/_adaptive_playwright_crawling_context.py,sha256=9FlHIUC05IzUhJsVldQvpnDnj1jk8GJpqC98mPLN_fw,10431
 crawlee/crawlers/_adaptive_playwright/_rendering_type_predictor.py,sha256=TM4mkbIN_059jUyCG8Z6XAb_FBLClIKw7z-aDvjon2I,10834
@@ -148,7 +148,7 @@ crawlee/statistics/__init__.py,sha256=lXAsHNkeRZQBffW1B7rERarivXIUJveNlcKTGOXQZY
 crawlee/statistics/_error_snapshotter.py,sha256=g-roZgkJ-glyStZL7gXrOhrpdZvZ686W9lR43uZjPao,3279
 crawlee/statistics/_error_tracker.py,sha256=x9Yw1TuyEptjwgPPJ4gIom-0oVjawcNReQDsHH2nZ3w,8553
 crawlee/statistics/_models.py,sha256=SFWYpT3r1c4XugU8nrm0epTpcM5_0fS1mXi9fnbhGJ8,5237
-crawlee/statistics/_statistics.py,sha256=d6z5XxXm-an4M_8TierOPpSB78vxqxwvUFCewIEmiK4,12786
+crawlee/statistics/_statistics.py,sha256=AnxbVq6w8fuiRumUJMznhTtQmtONyF4pzqrFYgO4yjo,13076
 crawlee/storage_clients/__init__.py,sha256=X3M6Z_WAOJ3M9I8JhGhJDnrtbCOmM27DpGAzgt87R2A,874
 crawlee/storage_clients/models.py,sha256=gfW_kpSCOBuoTBIW0N7tb3FUv7BgD3keZADS7pyT4_I,6586
 crawlee/storage_clients/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -199,8 +199,8 @@ crawlee/storages/_request_queue.py,sha256=bjBOGbpMaGUsqJPVB-JD2VShziPAYMI-GvWKKp
 crawlee/storages/_storage_instance_manager.py,sha256=72n0YlPwNpSQDJSPf4TxnI2GvIK6L-ZiTmHRbFcoVU0,8164
 crawlee/storages/_utils.py,sha256=Yz-5tEBYKYCFJemYT29--uGJqoJLApLDLgPcsnbifRw,439
 crawlee/storages/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-crawlee-1.0.5b13.dist-info/METADATA,sha256=hDy8VXBEMs_osj0eBKSqOntcEHFSN1cHHTeWwcugAgg,29445
-crawlee-1.0.5b13.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-crawlee-1.0.5b13.dist-info/entry_points.txt,sha256=1p65X3dA-cYvzjtlxLL6Kn1wpY-3uEDVqJLp53uNPeo,45
-crawlee-1.0.5b13.dist-info/licenses/LICENSE,sha256=AsFjHssKjj4LGd2ZCqXn6FBzMqcWdjQre1byPPSypVw,11355
-crawlee-1.0.5b13.dist-info/RECORD,,
+crawlee-1.0.5b15.dist-info/METADATA,sha256=DrXwvY89KKTek0NBwVSk_d-lhRsKyjZO1X-CKwe1q7A,29422
+crawlee-1.0.5b15.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+crawlee-1.0.5b15.dist-info/entry_points.txt,sha256=1p65X3dA-cYvzjtlxLL6Kn1wpY-3uEDVqJLp53uNPeo,45
+crawlee-1.0.5b15.dist-info/licenses/LICENSE,sha256=AsFjHssKjj4LGd2ZCqXn6FBzMqcWdjQre1byPPSypVw,11355
+crawlee-1.0.5b15.dist-info/RECORD,,