chore: Update to apify-shared v2.0

pyproject.toml

@@ -24,7 +24,7 @@ classifiers = [
 ]
 keywords = ["apify", "api", "client", "automation", "crawling", "scraping"]
 dependencies = [
-    "apify-shared<2.0.0",
+    "apify-shared>=2.0.0,<3.0.0",
     "colorama>=0.4.0",
     "impit>=0.5.3",
     "more_itertools>=10.0.0",

src/apify_client/_http_client.py

@@ -11,7 +11,6 @@
 from urllib.parse import urlencode
 
 import impit
-from apify_shared.utils import ignore_docs
 
 from apify_client._logging import log_context, logger_name
 from apify_client._statistics import Statistics
@@ -21,7 +20,7 @@
 if TYPE_CHECKING:
     from collections.abc import Callable
 
-    from apify_shared.types import JSONSerializable
+    from apify_client._types import JSONSerializable
 
 DEFAULT_BACKOFF_EXPONENTIAL_FACTOR = 2
 DEFAULT_BACKOFF_RANDOM_FACTOR = 1
@@ -30,7 +29,6 @@
 
 
 class _BaseHTTPClient:
-    @ignore_docs
     def __init__(
         self,
         *,

src/apify_client/_types.py

@@ -0,0 +1,42 @@
+from __future__ import annotations
+
+from typing import Any, Generic, TypeVar
+
+JSONSerializable = str | int | float | bool | None | dict[str, Any] | list[Any]
+"""Type for representing json-serializable values. It's close enough to the real thing supported
+by json.parse, and the best we can do until mypy supports recursive types. It was suggested in
+a discussion with (and approved by) Guido van Rossum, so I'd consider it correct enough.
+"""
+
+T = TypeVar('T')
+
+
+class ListPage(Generic[T]):
+    """A single page of items returned from a list() method."""
+
+    items: list[T]
+    """List of returned objects on this page."""
+
+    count: int
+    """Count of the returned objects on this page."""
+
+    offset: int
+    """The limit on the number of returned objects offset specified in the API call."""
+
+    limit: int
+    """The offset of the first object specified in the API call"""
+
+    total: int
+    """Total number of objects matching the API call criteria."""
+
+    desc: bool
+    """Whether the listing is descending or not."""
+
+    def __init__(self: ListPage, data: dict) -> None:
+        """Initialize a ListPage instance from the API response data."""
+        self.items = data.get('items', [])
+        self.offset = data.get('offset', 0)
+        self.limit = data.get('limit', 0)
+        self.count = data['count'] if 'count' in data else len(self.items)
+        self.total = data.get('total', self.offset + self.count)
+        self.desc = data.get('desc', False)

src/apify_client/_utils.py

@@ -2,21 +2,20 @@
 
 import asyncio
 import base64
+import contextlib
+import io
+import json
 import json as jsonlib
 import random
+import re
 import time
 from collections.abc import Callable
+from datetime import datetime, timezone
+from enum import Enum
 from http import HTTPStatus
 from typing import TYPE_CHECKING, Any, TypeVar, cast
 
 import impit
-from apify_shared.utils import (
-    is_content_type_json,
-    is_content_type_text,
-    is_content_type_xml,
-    is_file_or_bytes,
-    maybe_extract_enum_member_value,
-)
 
 from apify_client.errors import InvalidResponseBodyError
 
@@ -29,11 +28,102 @@
 
 PARSE_DATE_FIELDS_MAX_DEPTH = 3
 PARSE_DATE_FIELDS_KEY_SUFFIX = 'At'
-
 RECORD_NOT_FOUND_EXCEPTION_TYPES = ['record-not-found', 'record-or-token-not-found']
 
 T = TypeVar('T')
 StopRetryingType = Callable[[], None]
+ListOrDict = TypeVar('ListOrDict', list, dict)
+
+
+def filter_out_none_values_recursively(dictionary: dict) -> dict:
+    """Return copy of the dictionary, recursively omitting all keys for which values are None."""
+    return cast('dict', filter_out_none_values_recursively_internal(dictionary))
+
+
+def filter_out_none_values_recursively_internal(
+    dictionary: dict,
+    *,
+    remove_empty_dicts: bool | None = None,
+) -> dict | None:
+    """Recursively filters out None values from a dictionary.
+
+    Unfortunately, it's necessary to have an internal function for the correct result typing,
+    without having to create complicated overloads
+    """
+    result = {}
+    for k, v in dictionary.items():
+        if isinstance(v, dict):
+            v = filter_out_none_values_recursively_internal(  # noqa: PLW2901
+                v, remove_empty_dicts=remove_empty_dicts is True or remove_empty_dicts is None
+            )
+        if v is not None:
+            result[k] = v
+    if not result and remove_empty_dicts:
+        return None
+    return result
+
+
+def parse_date_fields(data: ListOrDict, max_depth: int = PARSE_DATE_FIELDS_MAX_DEPTH) -> ListOrDict:
+    """Recursively parse date fields in a list or dictionary up to the specified depth."""
+    if max_depth < 0:
+        return data
+
+    if isinstance(data, list):
+        return [parse_date_fields(item, max_depth - 1) for item in data]
+
+    if isinstance(data, dict):
+
+        def parse(key: str, value: object) -> object:
+            parsed_value = value
+            if key.endswith(PARSE_DATE_FIELDS_KEY_SUFFIX) and isinstance(value, str):
+                with contextlib.suppress(ValueError):
+                    parsed_value = datetime.strptime(value, '%Y-%m-%dT%H:%M:%S.%fZ').replace(tzinfo=timezone.utc)
+            elif isinstance(value, dict):
+                parsed_value = parse_date_fields(value, max_depth - 1)
+            elif isinstance(value, list):
+                parsed_value = parse_date_fields(value, max_depth)
+            return parsed_value
+
+        return {key: parse(key, value) for (key, value) in data.items()}
+
+    return data
+
+
+def is_content_type_json(content_type: str) -> bool:
+    """Check if the given content type is JSON."""
+    return bool(re.search(r'^application/json', content_type, flags=re.IGNORECASE))
+
+
+def is_content_type_xml(content_type: str) -> bool:
+    """Check if the given content type is XML."""
+    return bool(re.search(r'^application/.*xml$', content_type, flags=re.IGNORECASE))
+
+
+def is_content_type_text(content_type: str) -> bool:
+    """Check if the given content type is text."""
+    return bool(re.search(r'^text/', content_type, flags=re.IGNORECASE))
+
+
+def is_file_or_bytes(value: Any) -> bool:
+    """Check if the input value is a file-like object or bytes.
+
+    The check for IOBase is not ideal, it would be better to use duck typing,
+    but then the check would be super complex, judging from how the 'requests' library does it.
+    This way should be good enough for the vast majority of use cases, if it causes issues, we can improve it later.
+    """
+    return isinstance(value, (bytes, bytearray, io.IOBase))
+
+
+def json_dumps(obj: Any) -> str:
+    """Dump JSON to a string with the correct settings and serializer."""
+    return json.dumps(obj, ensure_ascii=False, indent=2, default=str)
+
+
+def maybe_extract_enum_member_value(maybe_enum_member: Any) -> Any:
+    """Extract the value of an enumeration member if it is an Enum, otherwise return the original value."""
+    if isinstance(maybe_enum_member, Enum):
+        return maybe_enum_member.value
+    return maybe_enum_member
 
 
 def to_safe_id(id: str) -> str:

src/apify_client/client.py

@@ -1,7 +1,5 @@
 from __future__ import annotations
 
-from apify_shared.utils import ignore_docs
-
 from apify_client._http_client import HTTPClient, HTTPClientAsync
 from apify_client._statistics import Statistics
 from apify_client.clients import (
@@ -61,7 +59,6 @@
 class _BaseApifyClient:
     http_client: HTTPClient | HTTPClientAsync
 
-    @ignore_docs
     def __init__(
         self,
         token: str | None = None,

src/apify_client/clients/base/actor_job_base_client.py

@@ -7,9 +7,8 @@
 from datetime import datetime, timezone
 
 from apify_shared.consts import ActorJobStatus
-from apify_shared.utils import ignore_docs, parse_date_fields
 
-from apify_client._utils import catch_not_found_or_throw, pluck_data
+from apify_client._utils import catch_not_found_or_throw, parse_date_fields, pluck_data
 from apify_client.clients.base.resource_client import ResourceClient, ResourceClientAsync
 from apify_client.errors import ApifyApiError
 
@@ -19,7 +18,6 @@
 DEFAULT_WAIT_WHEN_JOB_NOT_EXIST_SEC = 3
 
 
-@ignore_docs
 class ActorJobBaseClient(ResourceClient):
     """Base sub-client class for Actor runs and Actor builds."""
 
@@ -74,7 +72,6 @@ def _abort(self, *, gracefully: bool | None = None) -> dict:
         return parse_date_fields(pluck_data(jsonlib.loads(response.text)))
 
 
-@ignore_docs
 class ActorJobBaseClientAsync(ResourceClientAsync):
     """Base async sub-client class for Actor runs and Actor builds."""
 

src/apify_client/clients/base/base_client.py

@@ -2,8 +2,6 @@
 
 from typing import TYPE_CHECKING, Any
 
-from apify_shared.utils import ignore_docs
-
 from apify_client._logging import WithLogDetailsClient
 from apify_client._utils import to_safe_id
 
@@ -45,14 +43,12 @@ def _sub_resource_init_options(self, **kwargs: Any) -> dict:
         }
 
 
-@ignore_docs
 class BaseClient(_BaseBaseClient):
     """Base class for sub-clients."""
 
     http_client: HTTPClient
     root_client: ApifyClient
 
-    @ignore_docs
     def __init__(
         self,
         *,
@@ -88,14 +84,12 @@ def __init__(
             self.url = f'{self.url}/{self.safe_id}'
 
 
-@ignore_docs
 class BaseClientAsync(_BaseBaseClient):
     """Base class for async sub-clients."""
 
     http_client: HTTPClientAsync
     root_client: ApifyClientAsync
 
-    @ignore_docs
     def __init__(
         self,
         *,

src/apify_client/clients/base/resource_client.py

@@ -2,14 +2,11 @@
 
 import json as jsonlib
 
-from apify_shared.utils import ignore_docs, parse_date_fields
-
-from apify_client._utils import catch_not_found_or_throw, pluck_data
+from apify_client._utils import catch_not_found_or_throw, parse_date_fields, pluck_data
 from apify_client.clients.base.base_client import BaseClient, BaseClientAsync
 from apify_client.errors import ApifyApiError
 
 
-@ignore_docs
 class ResourceClient(BaseClient):
     """Base class for sub-clients manipulating a single resource."""
 
@@ -53,7 +50,6 @@ def _delete(self, timeout_secs: int | None = None) -> None:
             catch_not_found_or_throw(exc)
 
 
-@ignore_docs
 class ResourceClientAsync(BaseClientAsync):
     """Base class for async sub-clients manipulating a single resource."""
 

src/apify_client/clients/base/resource_collection_client.py

@@ -3,9 +3,7 @@
 import json as jsonlib
 from typing import Any, Generic, TypeVar
 
-from apify_shared.utils import ignore_docs, parse_date_fields
-
-from apify_client._utils import pluck_data
+from apify_client._utils import parse_date_fields, pluck_data
 from apify_client.clients.base.base_client import BaseClient, BaseClientAsync
 
 T = TypeVar('T')
@@ -32,7 +30,6 @@ class ListPage(Generic[T]):
     desc: bool
     """Whether the listing is descending or not"""
 
-    @ignore_docs
     def __init__(self, data: dict) -> None:
         """Initialize a ListPage instance from the API response data."""
         self.items = data.get('items', [])
@@ -43,7 +40,6 @@ def __init__(self, data: dict) -> None:
         self.desc = data.get('desc', False)
 
 
-@ignore_docs
 class ResourceCollectionClient(BaseClient):
     """Base class for sub-clients manipulating a resource collection."""
 
@@ -77,7 +73,6 @@ def _get_or_create(self, name: str | None = None, resource: dict | None = None)
         return parse_date_fields(pluck_data(jsonlib.loads(response.text)))
 
 
-@ignore_docs
 class ResourceCollectionClientAsync(BaseClientAsync):
     """Base class for async sub-clients manipulating a resource collection."""
 

src/apify_client/clients/resource_clients/actor.py

@@ -3,14 +3,14 @@
 import json as jsonlib
 from typing import TYPE_CHECKING, Any, Literal
 
-from apify_shared.utils import (
+from apify_client._utils import (
+    encode_key_value_store_record_value,
+    encode_webhook_list_to_base64,
     filter_out_none_values_recursively,
-    ignore_docs,
     maybe_extract_enum_member_value,
     parse_date_fields,
+    pluck_data,
 )
-
-from apify_client._utils import encode_key_value_store_record_value, encode_webhook_list_to_base64, pluck_data
 from apify_client.clients.base import ResourceClient, ResourceClientAsync
 from apify_client.clients.resource_clients.actor_version import ActorVersionClient, ActorVersionClientAsync
 from apify_client.clients.resource_clients.actor_version_collection import (
@@ -98,7 +98,6 @@ def get_actor_representation(
 class ActorClient(ResourceClient):
     """Sub-client for manipulating a single Actor."""
 
-    @ignore_docs
     def __init__(self, *args: Any, **kwargs: Any) -> None:
         resource_path = kwargs.pop('resource_path', 'acts')
         super().__init__(*args, resource_path=resource_path, **kwargs)
@@ -503,7 +502,6 @@ def validate_input(
 class ActorClientAsync(ResourceClientAsync):
     """Async sub-client for manipulating a single Actor."""
 
-    @ignore_docs
     def __init__(self, *args: Any, **kwargs: Any) -> None:
         resource_path = kwargs.pop('resource_path', 'acts')
         super().__init__(*args, resource_path=resource_path, **kwargs)