airbyte-cdk 6.31.2.dev0__py3-none-any.whl → 6.32.0__py3-none-any.whl

airbyte_cdk/sources/streams/concurrent/clamping.py

@@ -0,0 +1,99 @@
+from abc import ABC
+from datetime import datetime, timedelta
+from enum import Enum
+from typing import Callable
+
+from airbyte_cdk.sources.streams.concurrent.cursor_types import CursorValueType
+
+
+class ClampingStrategy(ABC):
+    def clamp(self, value: CursorValueType) -> CursorValueType:
+        raise NotImplementedError()
+
+
+class NoClamping(ClampingStrategy):
+    def clamp(self, value: CursorValueType) -> CursorValueType:
+        return value
+
+
+class ClampingEndProvider:
+    def __init__(
+        self,
+        clamping_strategy: ClampingStrategy,
+        end_provider: Callable[[], CursorValueType],
+        granularity: timedelta,
+    ) -> None:
+        self._clamping_strategy = clamping_strategy
+        self._end_provider = end_provider
+        self._granularity = granularity
+
+    def __call__(self) -> CursorValueType:
+        return self._clamping_strategy.clamp(self._end_provider()) - self._granularity
+
+
+class DayClampingStrategy(ClampingStrategy):
+    def __init__(self, is_ceiling: bool = True) -> None:
+        self._is_ceiling = is_ceiling
+
+    def clamp(self, value: datetime) -> datetime:  # type: ignore  # datetime implements method from CursorValueType
+        return_value = value.replace(hour=0, minute=0, second=0, microsecond=0)
+        if self._is_ceiling:
+            return return_value + timedelta(days=1)
+        return return_value
+
+
+class MonthClampingStrategy(ClampingStrategy):
+    def __init__(self, is_ceiling: bool = True) -> None:
+        self._is_ceiling = is_ceiling
+
+    def clamp(self, value: datetime) -> datetime:  # type: ignore  # datetime implements method from CursorValueType
+        return_value = value.replace(hour=0, minute=0, second=0, microsecond=0)
+        needs_to_round = value.day != 1
+        if not needs_to_round:
+            return return_value
+
+        return self._ceil(return_value) if self._is_ceiling else return_value.replace(day=1)
+
+    def _ceil(self, value: datetime) -> datetime:
+        return value.replace(
+            year=value.year + 1 if value.month == 12 else value.year,
+            month=(value.month % 12) + 1,
+            day=1,
+            hour=0,
+            minute=0,
+            second=0,
+            microsecond=0,
+        )
+
+
+class Weekday(Enum):
+    """
+    These integer values map to the same ones used by the Datetime.date.weekday() implementation
+    """
+
+    MONDAY = 0
+    TUESDAY = 1
+    WEDNESDAY = 2
+    THURSDAY = 3
+    FRIDAY = 4
+    SATURDAY = 5
+    SUNDAY = 6
+
+
+class WeekClampingStrategy(ClampingStrategy):
+    def __init__(self, day_of_week: Weekday, is_ceiling: bool = True) -> None:
+        self._day_of_week = day_of_week.value
+        self._is_ceiling = is_ceiling
+
+    def clamp(self, value: datetime) -> datetime:  # type: ignore  # datetime implements method from CursorValueType
+        days_diff_to_ceiling = (
+            7 - (value.weekday() - self._day_of_week)
+            if value.weekday() > self._day_of_week
+            else abs(value.weekday() - self._day_of_week)
+        )
+        delta = (
+            timedelta(days_diff_to_ceiling)
+            if self._is_ceiling
+            else timedelta(days_diff_to_ceiling - 7)
+        )
+        return value.replace(hour=0, minute=0, second=0, microsecond=0) + delta

airbyte_cdk/sources/streams/concurrent/cursor.py

@@ -13,7 +13,6 @@ from typing import (
     Mapping,
     MutableMapping,
     Optional,
-    Protocol,
     Tuple,
     Union,
 )
@@ -21,6 +20,8 @@ from typing import (
 from airbyte_cdk.sources.connector_state_manager import ConnectorStateManager
 from airbyte_cdk.sources.message import MessageRepository
 from airbyte_cdk.sources.streams import NO_CURSOR_STATE_KEY
+from airbyte_cdk.sources.streams.concurrent.clamping import ClampingStrategy, NoClamping
+from airbyte_cdk.sources.streams.concurrent.cursor_types import CursorValueType, GapType
 from airbyte_cdk.sources.streams.concurrent.partitions.partition import Partition
 from airbyte_cdk.sources.streams.concurrent.partitions.stream_slicer import StreamSlicer
 from airbyte_cdk.sources.streams.concurrent.state_converters.abstract_stream_state_converter import (
@@ -35,36 +36,6 @@ def _extract_value(mapping: Mapping[str, Any], path: List[str]) -> Any:
     return functools.reduce(lambda a, b: a[b], path, mapping)
 
 
-class GapType(Protocol):
-    """
-    This is the representation of gaps between two cursor values. Examples:
-    * if cursor values are datetimes, GapType is timedelta
-    * if cursor values are integer, GapType will also be integer
-    """
-
-    pass
-
-
-class CursorValueType(Protocol):
-    """Protocol for annotating comparable types."""
-
-    @abstractmethod
-    def __lt__(self: "CursorValueType", other: "CursorValueType") -> bool:
-        pass
-
-    @abstractmethod
-    def __ge__(self: "CursorValueType", other: "CursorValueType") -> bool:
-        pass
-
-    @abstractmethod
-    def __add__(self: "CursorValueType", other: GapType) -> "CursorValueType":
-        pass
-
-    @abstractmethod
-    def __sub__(self: "CursorValueType", other: GapType) -> "CursorValueType":
-        pass
-
-
 class CursorField:
     def __init__(self, cursor_field_key: str) -> None:
         self.cursor_field_key = cursor_field_key
@@ -172,6 +143,7 @@ class ConcurrentCursor(Cursor):
         lookback_window: Optional[GapType] = None,
         slice_range: Optional[GapType] = None,
         cursor_granularity: Optional[GapType] = None,
+        clamping_strategy: ClampingStrategy = NoClamping(),
     ) -> None:
         self._stream_name = stream_name
         self._stream_namespace = stream_namespace
@@ -193,10 +165,13 @@ class ConcurrentCursor(Cursor):
         self._cursor_granularity = cursor_granularity
         # Flag to track if the logger has been triggered (per stream)
         self._should_be_synced_logger_triggered = False
+        self._clamping_strategy = clamping_strategy
 
     @property
     def state(self) -> MutableMapping[str, Any]:
-        return self._concurrent_state
+        return self._connector_state_converter.convert_to_state_message(
+            self.cursor_field, self._concurrent_state
+        )
 
     @property
     def cursor_field(self) -> CursorField:
@@ -241,10 +216,10 @@ class ConcurrentCursor(Cursor):
         return self._connector_state_converter.parse_value(self._cursor_field.extract_value(record))
 
     def close_partition(self, partition: Partition) -> None:
-        slice_count_before = len(self.state.get("slices", []))
+        slice_count_before = len(self._concurrent_state.get("slices", []))
         self._add_slice_to_state(partition)
         if slice_count_before < len(
-            self.state["slices"]
+            self._concurrent_state["slices"]
         ):  # only emit if at least one slice has been processed
             self._merge_partitions()
             self._emit_state_message()
@@ -256,11 +231,11 @@ class ConcurrentCursor(Cursor):
         )
 
         if self._slice_boundary_fields:
-            if "slices" not in self.state:
+            if "slices" not in self._concurrent_state:
                 raise RuntimeError(
                     f"The state for stream {self._stream_name} should have at least one slice to delineate the sync start time, but no slices are present. This is unexpected. Please contact Support."
                 )
-            self.state["slices"].append(
+            self._concurrent_state["slices"].append(
                 {
                     self._connector_state_converter.START_KEY: self._extract_from_slice(
                         partition, self._slice_boundary_fields[self._START_BOUNDARY]
@@ -288,7 +263,7 @@ class ConcurrentCursor(Cursor):
                     "expected. Please contact the Airbyte team."
                 )
 
-            self.state["slices"].append(
+            self._concurrent_state["slices"].append(
                 {
                     self._connector_state_converter.START_KEY: self.start,
                     self._connector_state_converter.END_KEY: most_recent_cursor_value,
@@ -300,9 +275,7 @@ class ConcurrentCursor(Cursor):
         self._connector_state_manager.update_state_for_stream(
             self._stream_name,
             self._stream_namespace,
-            self._connector_state_converter.convert_to_state_message(
-                self._cursor_field, self.state
-            ),
+            self.state,
         )
         state_message = self._connector_state_manager.create_state_message(
             self._stream_name, self._stream_namespace
@@ -310,7 +283,9 @@ class ConcurrentCursor(Cursor):
         self._message_repository.emit_message(state_message)
 
     def _merge_partitions(self) -> None:
-        self.state["slices"] = self._connector_state_converter.merge_intervals(self.state["slices"])
+        self._concurrent_state["slices"] = self._connector_state_converter.merge_intervals(
+            self._concurrent_state["slices"]
+        )
 
     def _extract_from_slice(self, partition: Partition, key: str) -> CursorValueType:
         try:
@@ -347,36 +322,42 @@ class ConcurrentCursor(Cursor):
         if self._start is not None and self._is_start_before_first_slice():
             yield from self._split_per_slice_range(
                 self._start,
-                self.state["slices"][0][self._connector_state_converter.START_KEY],
+                self._concurrent_state["slices"][0][self._connector_state_converter.START_KEY],
                 False,
             )
 
-        if len(self.state["slices"]) == 1:
+        if len(self._concurrent_state["slices"]) == 1:
             yield from self._split_per_slice_range(
                 self._calculate_lower_boundary_of_last_slice(
-                    self.state["slices"][0][self._connector_state_converter.END_KEY]
+                    self._concurrent_state["slices"][0][self._connector_state_converter.END_KEY]
                 ),
                 self._end_provider(),
                 True,
             )
-        elif len(self.state["slices"]) > 1:
-            for i in range(len(self.state["slices"]) - 1):
+        elif len(self._concurrent_state["slices"]) > 1:
+            for i in range(len(self._concurrent_state["slices"]) - 1):
                 if self._cursor_granularity:
                     yield from self._split_per_slice_range(
-                        self.state["slices"][i][self._connector_state_converter.END_KEY]
+                        self._concurrent_state["slices"][i][self._connector_state_converter.END_KEY]
                         + self._cursor_granularity,
-                        self.state["slices"][i + 1][self._connector_state_converter.START_KEY],
+                        self._concurrent_state["slices"][i + 1][
+                            self._connector_state_converter.START_KEY
+                        ],
                         False,
                     )
                 else:
                     yield from self._split_per_slice_range(
-                        self.state["slices"][i][self._connector_state_converter.END_KEY],
-                        self.state["slices"][i + 1][self._connector_state_converter.START_KEY],
+                        self._concurrent_state["slices"][i][
+                            self._connector_state_converter.END_KEY
+                        ],
+                        self._concurrent_state["slices"][i + 1][
+                            self._connector_state_converter.START_KEY
+                        ],
                         False,
                     )
             yield from self._split_per_slice_range(
                 self._calculate_lower_boundary_of_last_slice(
-                    self.state["slices"][-1][self._connector_state_converter.END_KEY]
+                    self._concurrent_state["slices"][-1][self._connector_state_converter.END_KEY]
                 ),
                 self._end_provider(),
                 True,
@@ -387,7 +368,8 @@ class ConcurrentCursor(Cursor):
     def _is_start_before_first_slice(self) -> bool:
         return (
             self._start is not None
-            and self._start < self.state["slices"][0][self._connector_state_converter.START_KEY]
+            and self._start
+            < self._concurrent_state["slices"][0][self._connector_state_converter.START_KEY]
         )
 
     def _calculate_lower_boundary_of_last_slice(
@@ -408,10 +390,12 @@ class ConcurrentCursor(Cursor):
 
         lower = max(lower, self._start) if self._start else lower
         if not self._slice_range or self._evaluate_upper_safely(lower, self._slice_range) >= upper:
+            clamped_lower = self._clamping_strategy.clamp(lower)
+            clamped_upper = self._clamping_strategy.clamp(upper)
             start_value, end_value = (
-                (lower, upper - self._cursor_granularity)
+                (clamped_lower, clamped_upper - self._cursor_granularity)
                 if self._cursor_granularity and not upper_is_end
-                else (lower, upper)
+                else (clamped_lower, clamped_upper)
             )
             yield StreamSlice(
                 partition={},
@@ -433,11 +417,21 @@ class ConcurrentCursor(Cursor):
                 )
                 has_reached_upper_boundary = current_upper_boundary >= upper
 
+                clamped_upper = (
+                    self._clamping_strategy.clamp(current_upper_boundary)
+                    if current_upper_boundary != upper
+                    else current_upper_boundary
+                )
+                clamped_lower = self._clamping_strategy.clamp(current_lower_boundary)
+                if clamped_lower >= clamped_upper:
+                    # clamping collapsed both values which means that it is time to stop processing
+                    # FIXME should this be replace by proper end_provider
+                    break
                 start_value, end_value = (
-                    (current_lower_boundary, current_upper_boundary - self._cursor_granularity)
+                    (clamped_lower, clamped_upper - self._cursor_granularity)
                     if self._cursor_granularity
                     and (not upper_is_end or not has_reached_upper_boundary)
-                    else (current_lower_boundary, current_upper_boundary)
+                    else (clamped_lower, clamped_upper)
                 )
                 yield StreamSlice(
                     partition={},
@@ -450,7 +444,7 @@ class ConcurrentCursor(Cursor):
                         ]: self._connector_state_converter.output_format(end_value),
                     },
                 )
-                current_lower_boundary = current_upper_boundary
+                current_lower_boundary = clamped_upper
                 if current_upper_boundary >= upper:
                     stop_processing = True
 

airbyte_cdk/sources/streams/concurrent/cursor_types.py

@@ -0,0 +1,32 @@
+from abc import abstractmethod
+from typing import Protocol
+
+
+class GapType(Protocol):
+    """
+    This is the representation of gaps between two cursor values. Examples:
+    * if cursor values are datetimes, GapType is timedelta
+    * if cursor values are integer, GapType will also be integer
+    """
+
+    pass
+
+
+class CursorValueType(Protocol):
+    """Protocol for annotating comparable types."""
+
+    @abstractmethod
+    def __lt__(self: "CursorValueType", other: "CursorValueType") -> bool:
+        pass
+
+    @abstractmethod
+    def __ge__(self: "CursorValueType", other: "CursorValueType") -> bool:
+        pass
+
+    @abstractmethod
+    def __add__(self: "CursorValueType", other: GapType) -> "CursorValueType":
+        pass
+
+    @abstractmethod
+    def __sub__(self: "CursorValueType", other: GapType) -> "CursorValueType":
+        pass

airbyte_cdk/sources/streams/concurrent/state_converters/datetime_stream_state_converter.py

@@ -6,9 +6,6 @@ from abc import abstractmethod
 from datetime import datetime, timedelta, timezone
 from typing import Any, Callable, List, MutableMapping, Optional, Tuple
 
-import pendulum
-from pendulum.datetime import DateTime
-
 # FIXME We would eventually like the Concurrent package do be agnostic of the declarative package. However, this is a breaking change and
 #  the goal in the short term is only to fix the issue we are seeing for source-declarative-manifest.
 from airbyte_cdk.sources.declarative.datetime.datetime_parser import DatetimeParser
@@ -17,6 +14,7 @@ from airbyte_cdk.sources.streams.concurrent.state_converters.abstract_stream_sta
     AbstractStreamStateConverter,
     ConcurrencyCompatibleStateType,
 )
+from airbyte_cdk.utils.datetime_helpers import AirbyteDateTime, ab_datetime_now, ab_datetime_parse
 
 
 class DateTimeStreamStateConverter(AbstractStreamStateConverter):
@@ -36,7 +34,7 @@ class DateTimeStreamStateConverter(AbstractStreamStateConverter):
 
     @classmethod
     def get_end_provider(cls) -> Callable[[], datetime]:
-        return lambda: datetime.now(timezone.utc)
+        return ab_datetime_now
 
     @abstractmethod
     def increment(self, timestamp: datetime) -> datetime: ...
@@ -136,10 +134,10 @@ class EpochValueConcurrentStreamStateConverter(DateTimeStreamStateConverter):
         return int(timestamp.timestamp())
 
     def parse_timestamp(self, timestamp: int) -> datetime:
-        dt_object = pendulum.from_timestamp(timestamp)
-        if not isinstance(dt_object, DateTime):
+        dt_object = AirbyteDateTime.fromtimestamp(timestamp, timezone.utc)
+        if not isinstance(dt_object, AirbyteDateTime):
             raise ValueError(
-                f"DateTime object was expected but got {type(dt_object)} from pendulum.parse({timestamp})"
+                f"AirbyteDateTime object was expected but got {type(dt_object)} from AirbyteDateTime.fromtimestamp({timestamp})"
             )
         return dt_object
 
@@ -169,14 +167,25 @@ class IsoMillisConcurrentStreamStateConverter(DateTimeStreamStateConverter):
     def increment(self, timestamp: datetime) -> datetime:
         return timestamp + self._cursor_granularity
 
-    def output_format(self, timestamp: datetime) -> Any:
-        return timestamp.strftime("%Y-%m-%dT%H:%M:%S.%f")[:-3] + "Z"
+    def output_format(self, timestamp: datetime) -> str:
+        """Format datetime with milliseconds always included.
+
+        Args:
+            timestamp: The datetime to format.
+
+        Returns:
+            str: ISO8601/RFC3339 formatted string with milliseconds.
+        """
+        dt = AirbyteDateTime.from_datetime(timestamp)
+        # Always include milliseconds, even if zero
+        millis = dt.microsecond // 1000 if dt.microsecond else 0
+        return f"{dt.year:04d}-{dt.month:02d}-{dt.day:02d}T{dt.hour:02d}:{dt.minute:02d}:{dt.second:02d}.{millis:03d}Z"
 
     def parse_timestamp(self, timestamp: str) -> datetime:
-        dt_object = pendulum.parse(timestamp)
-        if not isinstance(dt_object, DateTime):
+        dt_object = ab_datetime_parse(timestamp)
+        if not isinstance(dt_object, AirbyteDateTime):
             raise ValueError(
-                f"DateTime object was expected but got {type(dt_object)} from pendulum.parse({timestamp})"
+                f"AirbyteDateTime object was expected but got {type(dt_object)} from parse({timestamp})"
             )
         return dt_object
 
@@ -184,7 +193,7 @@ class IsoMillisConcurrentStreamStateConverter(DateTimeStreamStateConverter):
 class CustomFormatConcurrentStreamStateConverter(IsoMillisConcurrentStreamStateConverter):
     """
     Datetime State converter that emits state according to the supplied datetime format. The converter supports reading
-    incoming state in any valid datetime format via Pendulum.
+    incoming state in any valid datetime format using AirbyteDateTime parsing utilities.
     """
 
     def __init__(

airbyte_cdk/sources/streams/core.py

@@ -223,17 +223,17 @@ class Stream(ABC):
                     record_counter += 1
 
                     checkpoint_interval = self.state_checkpoint_interval
-                    checkpoint = checkpoint_reader.get_checkpoint()
                     if (
                         should_checkpoint
                         and checkpoint_interval
                         and record_counter % checkpoint_interval == 0
-                        and checkpoint is not None
                     ):
-                        airbyte_state_message = self._checkpoint_state(
-                            checkpoint, state_manager=state_manager
-                        )
-                        yield airbyte_state_message
+                        checkpoint = checkpoint_reader.get_checkpoint()
+                        if checkpoint:
+                            airbyte_state_message = self._checkpoint_state(
+                                checkpoint, state_manager=state_manager
+                            )
+                            yield airbyte_state_message
 
                     if internal_config.is_limit_reached(record_counter):
                         break

airbyte_cdk/sources/streams/http/http.py

@@ -423,8 +423,6 @@ class HttpStream(Stream, CheckpointMixin, ABC):
         stream_slice: Optional[Mapping[str, Any]] = None,
         stream_state: Optional[Mapping[str, Any]] = None,
     ) -> Iterable[StreamData]:
-        partition, _, _ = self._extract_slice_fields(stream_slice=stream_slice)
-
         stream_state = stream_state or {}
         pagination_complete = False
         next_page_token = None
@@ -438,6 +436,7 @@ class HttpStream(Stream, CheckpointMixin, ABC):
 
         cursor = self.get_cursor()
         if cursor and isinstance(cursor, SubstreamResumableFullRefreshCursor):
+            partition, _, _ = self._extract_slice_fields(stream_slice=stream_slice)
             # Substreams checkpoint state by marking an entire parent partition as completed so that on the subsequent attempt
             # after a failure, completed parents are skipped and the sync can make progress
             cursor.close_slice(StreamSlice(cursor_slice={}, partition=partition))