crawlee 1.0.2b3__py3-none-any.whl → 1.1.2b7__py3-none-any.whl

crawlee/request_loaders/_sitemap_request_loader.py

@@ -9,7 +9,7 @@ from typing import TYPE_CHECKING, Annotated, Any
 from pydantic import BaseModel, ConfigDict, Field
 from typing_extensions import override
 
-from crawlee import Request
+from crawlee import Request, RequestOptions
 from crawlee._utils.docs import docs_group
 from crawlee._utils.globs import Glob
 from crawlee._utils.recoverable_state import RecoverableState
@@ -18,9 +18,10 @@ from crawlee.request_loaders._request_loader import RequestLoader
 
 if TYPE_CHECKING:
     import re
-    from collections.abc import Sequence
+    from collections.abc import Callable, Sequence
     from types import TracebackType
 
+    from crawlee import RequestTransformAction
     from crawlee.http_clients import HttpClient
     from crawlee.proxy_configuration import ProxyInfo
     from crawlee.storage_clients.models import ProcessedRequest
@@ -90,6 +91,11 @@ class SitemapRequestLoaderState(BaseModel):
 class SitemapRequestLoader(RequestLoader):
     """A request loader that reads URLs from sitemap(s).
 
+    The loader is designed to handle sitemaps that follow the format described in the Sitemaps protocol
+    (https://www.sitemaps.org/protocol.html). It supports both XML and plain text sitemap formats.
+    Note that HTML pages containing links are not supported - those should be handled by regular crawlers
+    and the `enqueue_links` functionality.
+
     The loader fetches and parses sitemaps in the background, allowing crawling to start
     before all URLs are loaded. It supports filtering URLs using glob and regex patterns.
 
@@ -107,6 +113,7 @@ class SitemapRequestLoader(RequestLoader):
         exclude: list[re.Pattern[Any] | Glob] | None = None,
         max_buffer_size: int = 200,
         persist_state_key: str | None = None,
+        transform_request_function: Callable[[RequestOptions], RequestOptions | RequestTransformAction] | None = None,
     ) -> None:
         """Initialize the sitemap request loader.
 
@@ -120,6 +127,9 @@ class SitemapRequestLoader(RequestLoader):
             persist_state_key: A key for persisting the loader's state in the KeyValueStore.
                 When provided, allows resuming from where it left off after interruption.
                 If None, no state persistence occurs.
+            transform_request_function: An optional function to transform requests
+                generated by the loader. It receives `RequestOptions` with `url` and should return either
+                modified `RequestOptions` or a `RequestTransformAction`.
         """
         self._http_client = http_client
         self._sitemap_urls = sitemap_urls
@@ -127,6 +137,7 @@ class SitemapRequestLoader(RequestLoader):
         self._exclude = exclude
         self._proxy_info = proxy_info
         self._max_buffer_size = max_buffer_size
+        self._transform_request_function = transform_request_function
 
         # Synchronization for queue operations
         self._queue_has_capacity = asyncio.Event()
@@ -308,8 +319,15 @@ class SitemapRequestLoader(RequestLoader):
 
             async with self._queue_lock:
                 url = state.url_queue.popleft()
-
-                request = Request.from_url(url)
+                request_option = RequestOptions(url=url)
+                if self._transform_request_function:
+                    transform_request_option = self._transform_request_function(request_option)
+                    if transform_request_option == 'skip':
+                        state.total_count -= 1
+                        continue
+                    if transform_request_option != 'unchanged':
+                        request_option = transform_request_option
+                request = Request.from_url(**request_option)
                 state.in_progress.add(request.url)
                 if len(state.url_queue) < self._max_buffer_size:
                     self._queue_has_capacity.set()

crawlee/sessions/_session_pool.py

@@ -163,7 +163,7 @@ class SessionPool:
     def add_session(self, session: Session) -> None:
         """Add an externally created session to the pool.
 
-        This is intened only for the cases when you want to add a session that was created outside of the pool.
+        This is intended only for the cases when you want to add a session that was created outside of the pool.
         Otherwise, the pool will create new sessions automatically.
 
         Args:

crawlee/statistics/_error_snapshotter.py

@@ -32,7 +32,7 @@ class ErrorSnapshotter:
         """Capture error snapshot and save it to key value store.
 
         It saves the error snapshot directly to a key value store. It can't use `context.get_key_value_store` because
-        it returns `KeyValueStoreChangeRecords` which is commited to the key value store only if the `RequestHandler`
+        it returns `KeyValueStoreChangeRecords` which is committed to the key value store only if the `RequestHandler`
         returned without an exception. ErrorSnapshotter is on the contrary active only when `RequestHandler` fails with
         an exception.
 

crawlee/statistics/_models.py

@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import json
+import warnings
 from dataclasses import asdict, dataclass
 from datetime import datetime, timedelta, timezone
 from typing import Annotated, Any
@@ -76,7 +77,6 @@ class StatisticsState(BaseModel):
     crawler_started_at: Annotated[datetime | None, Field(alias='crawlerStartedAt')] = None
     crawler_last_started_at: Annotated[datetime | None, Field(alias='crawlerLastStartTimestamp')] = None
     crawler_finished_at: Annotated[datetime | None, Field(alias='crawlerFinishedAt')] = None
-    crawler_runtime: Annotated[timedelta_ms, Field(alias='crawlerRuntimeMillis')] = timedelta()
     errors: dict[str, Any] = Field(default_factory=dict)
     retry_errors: dict[str, Any] = Field(alias='retryErrors', default_factory=dict)
     requests_with_status_code: dict[str, int] = Field(alias='requestsWithStatusCode', default_factory=dict)
@@ -93,6 +93,37 @@ class StatisticsState(BaseModel):
         ),
     ] = {}
 
+    # Used to track the crawler runtime, that had already been persisted. This is the runtime from previous runs.
+    _runtime_offset: Annotated[timedelta, Field(exclude=True)] = timedelta()
+
+    def model_post_init(self, /, __context: Any) -> None:
+        self._runtime_offset = self.crawler_runtime or self._runtime_offset
+
+    @property
+    def crawler_runtime(self) -> timedelta:
+        if self.crawler_last_started_at:
+            finished_at = self.crawler_finished_at or datetime.now(timezone.utc)
+            return self._runtime_offset + finished_at - self.crawler_last_started_at
+        return self._runtime_offset
+
+    @crawler_runtime.setter
+    def crawler_runtime(self, value: timedelta) -> None:
+        # Setter for backwards compatibility only, the crawler_runtime is now computed_field, and cant be set manually.
+        # To be removed in v2 release https://github.com/apify/crawlee-python/issues/1567
+        warnings.warn(
+            f"Setting 'crawler_runtime' is deprecated and will be removed in a future version."
+            f' Value {value} will not be used.',
+            DeprecationWarning,
+            stacklevel=2,
+        )
+
+    @computed_field(alias='crawlerRuntimeMillis')
+    def crawler_runtime_for_serialization(self) -> timedelta:
+        if self.crawler_last_started_at:
+            finished_at = self.crawler_finished_at or datetime.now(timezone.utc)
+            return self._runtime_offset + finished_at - self.crawler_last_started_at
+        return self._runtime_offset
+
     @computed_field(alias='requestTotalDurationMillis', return_type=timedelta_ms)  # type: ignore[prop-decorator]
     @property
     def request_total_duration(self) -> timedelta:

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
@@ -17,8 +18,11 @@ from crawlee.statistics import FinalStatistics, StatisticsState
 from crawlee.statistics._error_tracker import ErrorTracker
 
 if TYPE_CHECKING:
+    from collections.abc import Callable, Coroutine
     from types import TracebackType
 
+    from crawlee.storages import KeyValueStore
+
 TStatisticsState = TypeVar('TStatisticsState', bound=StatisticsState, default=StatisticsState)
 TNewStatisticsState = TypeVar('TNewStatisticsState', bound=StatisticsState, default=StatisticsState)
 logger = getLogger(__name__)
@@ -70,6 +74,7 @@ class Statistics(Generic[TStatisticsState]):
         persistence_enabled: bool | Literal['explicit_only'] = False,
         persist_state_kvs_name: str | None = None,
         persist_state_key: str | None = None,
+        persist_state_kvs_factory: Callable[[], Coroutine[None, None, KeyValueStore]] | None = None,
         log_message: str = 'Statistics',
         periodic_message_logger: Logger | None = None,
         log_interval: timedelta = timedelta(minutes=1),
@@ -80,8 +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,
@@ -92,9 +95,10 @@ class Statistics(Generic[TStatisticsState]):
 
         self._state = RecoverableState(
             default_state=state_model(stats_id=self._id),
-            persist_state_key=persist_state_key or f'SDK_CRAWLER_STATISTICS_{self._id}',
+            persist_state_key=persist_state_key or f'__CRAWLER_STATISTICS_{self._id}',
             persistence_enabled=persistence_enabled,
             persist_state_kvs_name=persist_state_kvs_name,
+            persist_state_kvs_factory=persist_state_kvs_factory,
             logger=logger,
         )
 
@@ -110,8 +114,8 @@ class Statistics(Generic[TStatisticsState]):
         """Create near copy of the `Statistics` with replaced `state_model`."""
         new_statistics: Statistics[TNewStatisticsState] = Statistics(
             persistence_enabled=self._state._persistence_enabled,  # noqa: SLF001
-            persist_state_kvs_name=self._state._persist_state_kvs_name,  # noqa: SLF001
             persist_state_key=self._state._persist_state_key,  # noqa: SLF001
+            persist_state_kvs_factory=self._state._persist_state_kvs_factory,  # noqa: SLF001
             log_message=self._log_message,
             periodic_message_logger=self._periodic_message_logger,
             state_model=state_model,
@@ -125,6 +129,7 @@ class Statistics(Generic[TStatisticsState]):
         persistence_enabled: bool = False,
         persist_state_kvs_name: str | None = None,
         persist_state_key: str | None = None,
+        persist_state_kvs_factory: Callable[[], Coroutine[None, None, KeyValueStore]] | None = None,
         log_message: str = 'Statistics',
         periodic_message_logger: Logger | None = None,
         log_interval: timedelta = timedelta(minutes=1),
@@ -136,6 +141,7 @@ class Statistics(Generic[TStatisticsState]):
             persistence_enabled=persistence_enabled,
             persist_state_kvs_name=persist_state_kvs_name,
             persist_state_key=persist_state_key,
+            persist_state_kvs_factory=persist_state_kvs_factory,
             log_message=log_message,
             periodic_message_logger=periodic_message_logger,
             log_interval=log_interval,
@@ -158,14 +164,17 @@ 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()
+        # Reset `crawler_finished_at` to indicate a new run in progress.
+        self.state.crawler_finished_at = None
 
+        # 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__(
@@ -182,13 +191,14 @@ class Statistics(Generic[TStatisticsState]):
         if not self._active:
             raise RuntimeError(f'The {self.__class__.__name__} is not active.')
 
-        self._state.current_value.crawler_finished_at = datetime.now(timezone.utc)
-
-        await self._state.teardown()
+        if not self.state.crawler_last_started_at:
+            raise RuntimeError('Statistics.state.crawler_last_started_at not set.')
 
+        # 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._active = False
+        await self._state.teardown()
 
     @property
     def state(self) -> TStatisticsState:
@@ -247,11 +257,7 @@ class Statistics(Generic[TStatisticsState]):
 
     def calculate(self) -> FinalStatistics:
         """Calculate the current statistics."""
-        if self._instance_start is None:
-            raise RuntimeError('The Statistics object is not initialized')
-
-        crawler_runtime = datetime.now(timezone.utc) - self._instance_start
-        total_minutes = crawler_runtime.total_seconds() / 60
+        total_minutes = self.state.crawler_runtime.total_seconds() / 60
         state = self._state.current_value
         serialized_state = state.model_dump(by_alias=False)
 
@@ -262,7 +268,7 @@ class Statistics(Generic[TStatisticsState]):
             requests_failed_per_minute=math.floor(state.requests_failed / total_minutes) if total_minutes else 0,
             request_total_duration=state.request_total_finished_duration + state.request_total_failed_duration,
             requests_total=state.requests_failed + state.requests_finished,
-            crawler_runtime=crawler_runtime,
+            crawler_runtime=state.crawler_runtime,
             requests_finished=state.requests_finished,
             requests_failed=state.requests_failed,
             retry_histogram=serialized_state['request_retry_histogram'],
@@ -282,21 +288,6 @@ 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/storage_clients/__init__.py

@@ -13,9 +13,13 @@ _install_import_hook(__name__)
 with _try_import(__name__, 'SqlStorageClient'):
     from ._sql import SqlStorageClient
 
+with _try_import(__name__, 'RedisStorageClient'):
+    from ._redis import RedisStorageClient
+
 __all__ = [
     'FileSystemStorageClient',
     'MemoryStorageClient',
+    'RedisStorageClient',
     'SqlStorageClient',
     'StorageClient',
 ]

crawlee/storage_clients/_file_system/_dataset_client.py

@@ -134,7 +134,7 @@ class FileSystemDatasetClient(DatasetClient):
                     continue
 
                 try:
-                    file = await asyncio.to_thread(path_to_metadata.open)
+                    file = await asyncio.to_thread(path_to_metadata.open, 'r', encoding='utf-8')
                     try:
                         file_content = json.load(file)
                         metadata = DatasetMetadata(**file_content)
@@ -163,7 +163,7 @@ class FileSystemDatasetClient(DatasetClient):
 
             # If the dataset directory exists, reconstruct the client from the metadata file.
             if path_to_dataset.exists() and path_to_metadata.exists():
-                file = await asyncio.to_thread(open, path_to_metadata)
+                file = await asyncio.to_thread(open, path_to_metadata, 'r', encoding='utf-8')
                 try:
                     file_content = json.load(file)
                 finally:

crawlee/storage_clients/_file_system/_key_value_store_client.py

@@ -133,7 +133,7 @@ class FileSystemKeyValueStoreClient(KeyValueStoreClient):
                     continue
 
                 try:
-                    file = await asyncio.to_thread(path_to_metadata.open)
+                    file = await asyncio.to_thread(path_to_metadata.open, 'r', encoding='utf-8')
                     try:
                         file_content = json.load(file)
                         metadata = KeyValueStoreMetadata(**file_content)
@@ -162,7 +162,7 @@ class FileSystemKeyValueStoreClient(KeyValueStoreClient):
 
             # If the key-value store directory exists, reconstruct the client from the metadata file.
             if path_to_kvs.exists() and path_to_metadata.exists():
-                file = await asyncio.to_thread(open, path_to_metadata)
+                file = await asyncio.to_thread(open, path_to_metadata, 'r', encoding='utf-8')
                 try:
                     file_content = json.load(file)
                 finally:
@@ -239,7 +239,7 @@ class FileSystemKeyValueStoreClient(KeyValueStoreClient):
         # Read the metadata file
         async with self._lock:
             try:
-                file = await asyncio.to_thread(open, record_metadata_filepath)
+                file = await asyncio.to_thread(open, record_metadata_filepath, 'r', encoding='utf-8')
             except FileNotFoundError:
                 logger.warning(f'Metadata file disappeared for key "{key}", aborting get_value')
                 return None

crawlee/storage_clients/_file_system/_request_queue_client.py

@@ -31,6 +31,7 @@ if TYPE_CHECKING:
     from collections.abc import Sequence
 
     from crawlee.configuration import Configuration
+    from crawlee.storages import KeyValueStore
 
 logger = getLogger(__name__)
 
@@ -92,6 +93,7 @@ class FileSystemRequestQueueClient(RequestQueueClient):
         metadata: RequestQueueMetadata,
         path_to_rq: Path,
         lock: asyncio.Lock,
+        recoverable_state: RecoverableState[RequestQueueState],
     ) -> None:
         """Initialize a new instance.
 
@@ -114,12 +116,7 @@ class FileSystemRequestQueueClient(RequestQueueClient):
         self._is_empty_cache: bool | None = None
         """Cache for is_empty result: None means unknown, True/False is cached state."""
 
-        self._state = RecoverableState[RequestQueueState](
-            default_state=RequestQueueState(),
-            persist_state_key=f'__RQ_STATE_{self._metadata.id}',
-            persistence_enabled=True,
-            logger=logger,
-        )
+        self._state = recoverable_state
         """Recoverable state to maintain request ordering, in-progress status, and handled status."""
 
     @override
@@ -136,6 +133,22 @@ class FileSystemRequestQueueClient(RequestQueueClient):
         """The full path to the request queue metadata file."""
         return self.path_to_rq / METADATA_FILENAME
 
+    @classmethod
+    async def _create_recoverable_state(cls, id: str, configuration: Configuration) -> RecoverableState:
+        async def kvs_factory() -> KeyValueStore:
+            from crawlee.storage_clients import FileSystemStorageClient  # noqa: PLC0415 avoid circular import
+            from crawlee.storages import KeyValueStore  # noqa: PLC0415 avoid circular import
+
+            return await KeyValueStore.open(storage_client=FileSystemStorageClient(), configuration=configuration)
+
+        return RecoverableState[RequestQueueState](
+            default_state=RequestQueueState(),
+            persist_state_key=f'__RQ_STATE_{id}',
+            persist_state_kvs_factory=kvs_factory,
+            persistence_enabled=True,
+            logger=logger,
+        )
+
     @classmethod
     async def open(
         cls,
@@ -184,7 +197,7 @@ class FileSystemRequestQueueClient(RequestQueueClient):
                     continue
 
                 try:
-                    file = await asyncio.to_thread(path_to_metadata.open)
+                    file = await asyncio.to_thread(path_to_metadata.open, 'r', encoding='utf-8')
                     try:
                         file_content = json.load(file)
                         metadata = RequestQueueMetadata(**file_content)
@@ -194,6 +207,9 @@ class FileSystemRequestQueueClient(RequestQueueClient):
                                 metadata=metadata,
                                 path_to_rq=rq_base_path / rq_dir,
                                 lock=asyncio.Lock(),
+                                recoverable_state=await cls._create_recoverable_state(
+                                    id=id, configuration=configuration
+                                ),
                             )
                             await client._state.initialize()
                             await client._discover_existing_requests()
@@ -216,7 +232,7 @@ class FileSystemRequestQueueClient(RequestQueueClient):
 
             # If the RQ directory exists, reconstruct the client from the metadata file.
             if path_to_rq.exists() and path_to_metadata.exists():
-                file = await asyncio.to_thread(open, path_to_metadata)
+                file = await asyncio.to_thread(open, path_to_metadata, 'r', encoding='utf-8')
                 try:
                     file_content = json.load(file)
                 finally:
@@ -230,6 +246,7 @@ class FileSystemRequestQueueClient(RequestQueueClient):
                     metadata=metadata,
                     path_to_rq=path_to_rq,
                     lock=asyncio.Lock(),
+                    recoverable_state=await cls._create_recoverable_state(id=metadata.id, configuration=configuration),
                 )
 
                 await client._state.initialize()
@@ -254,6 +271,7 @@ class FileSystemRequestQueueClient(RequestQueueClient):
                     metadata=metadata,
                     path_to_rq=path_to_rq,
                     lock=asyncio.Lock(),
+                    recoverable_state=await cls._create_recoverable_state(id=metadata.id, configuration=configuration),
                 )
                 await client._state.initialize()
                 await client._update_metadata()
@@ -757,7 +775,7 @@ class FileSystemRequestQueueClient(RequestQueueClient):
         """
         # Open the request file.
         try:
-            file = await asyncio.to_thread(open, file_path)
+            file = await asyncio.to_thread(open, file_path, 'r', encoding='utf-8')
         except FileNotFoundError:
             logger.warning(f'Request file "{file_path}" not found.')
             return None

crawlee/storage_clients/_redis/__init__.py

@@ -0,0 +1,6 @@
+from ._dataset_client import RedisDatasetClient
+from ._key_value_store_client import RedisKeyValueStoreClient
+from ._request_queue_client import RedisRequestQueueClient
+from ._storage_client import RedisStorageClient
+
+__all__ = ['RedisDatasetClient', 'RedisKeyValueStoreClient', 'RedisRequestQueueClient', 'RedisStorageClient']