Feat: accept cron schedule for analyzers

tests/monitor/manager/test_monitor_setup.py

@@ -273,4 +273,61 @@ def test_set_non_iso_data_readiness_raises(monitor_setup) -> None:
 
     with pytest.raises(ValueError):
         monitor_setup.data_readiness_duration = "Some non-conformant string"
-
+
+
+def test_cron_schedule_for_analyzer(monitor_setup) -> None:
+    monitor_setup.config = FixedThresholdsConfig(
+        metric=DatasetMetric.classification_accuracy,
+        upper=0.75
+    )
+    monitor_setup.schedule = CronSchedule(cron="0 0 * * *")
+    monitor_setup.apply()
+
+    assert monitor_setup.analyzer.schedule == CronSchedule(
+        cron="0 0 * * *"
+    )
+
+    monitor_setup.schedule = CronSchedule(cron="0 0 * * 1-5")
+    monitor_setup.apply()
+
+    assert monitor_setup.analyzer.schedule == CronSchedule(
+        cron="0 0 * * 1-5"
+    )
+
+    monitor_setup.schedule = CronSchedule(cron="0 0 * * 6,0")
+    monitor_setup.apply()
+
+    assert monitor_setup.analyzer.schedule == CronSchedule(
+        cron="0 0 * * 6,0"
+    )
+
+    monitor_setup.schedule = CronSchedule(cron="0 9-17 * * *")
+    monitor_setup.apply()
+
+    assert monitor_setup.analyzer.schedule == CronSchedule(
+        cron="0 9-17 * * *"
+    )
+
+    monitor_setup.schedule = CronSchedule(cron="0 9,10,17 * * *") 
+    monitor_setup.apply()
+
+    assert monitor_setup.analyzer.schedule == CronSchedule(
+        cron="0 9,10,17 * * *"
+    )
+
+    monitor_setup.schedule = CronSchedule(cron="0,1 9,10,17 1,2,3 2,4,5 2,4")
+    monitor_setup.apply()
+
+    assert monitor_setup.analyzer.schedule == CronSchedule(
+        cron="0,1 9,10,17 1,2,3 2,4,5 2,4"
+    )
+
+    # All below Must fail
+
+    monitor_setup.schedule = CronSchedule(cron="* * * * *") # Every minute
+    with pytest.raises(ValueError):
+        monitor_setup.apply()
+
+    monitor_setup.schedule = CronSchedule(cron="0 0 * * * *") # Too many fields
+    with pytest.raises(ValueError):
+        monitor_setup.apply()

whylabs_toolkit/helpers/cron_validators.py

@@ -0,0 +1,53 @@
+from dataclasses import dataclass
+
+
+@dataclass
+class SplitCron:
+    day_of_week: str
+    month: str
+    day_of_month: str
+    hour: str
+    minute: str
+
+
+def split_cron_expression(cron: str) -> SplitCron:
+    """Split the cron expression into its components."""
+    cron_slots = cron.split(" ")
+    if len(cron_slots) != 5:
+        raise ValueError("CronSchedule must have 5 fields.")
+    return SplitCron(
+        day_of_week=cron_slots[0],
+        month=cron_slots[1],
+        day_of_month=cron_slots[2],
+        hour=cron_slots[3],
+        minute=cron_slots[4],
+    )
+
+
+def _is_not_less_granular_than_1_hour(split_cron: SplitCron) -> bool:
+    """Check if the cron expression is less granular than 1 hour."""
+    # Specific days checks
+    if split_cron.minute != "*" and split_cron.minute != "0":
+        return True
+    if split_cron.hour != "*" and split_cron.hour != "0":
+        return True
+    if split_cron.day_of_month != "*" and split_cron.day_of_month != "1":
+        return True
+    if split_cron.month != "*" and split_cron.month != "1":
+        return True
+    if split_cron.day_of_week != "*" and split_cron.day_of_week != "1":
+        return True
+
+    # Check range
+    for field in (split_cron.day_of_week, split_cron.month, split_cron.day_of_month, split_cron.hour):
+        for item in field.split(","):
+            if "-" in item:
+                start, end = map(int, item.split("-"))
+                if end - start > 0:
+                    return False
+    return False
+
+
+def validate_cron_expression(cron: str) -> bool:
+    split_cron = split_cron_expression(cron)
+    return _is_not_less_granular_than_1_hour(split_cron=split_cron)

whylabs_toolkit/monitor/manager/monitor_setup.py

@@ -31,7 +31,7 @@ def __init__(self, monitor_id: str, dataset_id: Optional[str] = None, config: Co
 
         self._monitor_mode: Optional[Union[EveryAnomalyMode, DigestMode]] = None
         self._monitor_actions: Optional[List[Union[GlobalAction, EmailRecipient, SlackWebhook, PagerDuty]]] = None
-        self._analyzer_schedule: Optional[FixedCadenceSchedule] = None
+        self._analyzer_schedule: Optional[Union[FixedCadenceSchedule, CronSchedule]] = None
         self._target_matrix: Optional[Union[ColumnMatrix, DatasetMatrix]] = None
         self._analyzer_config: Optional[
             Union[

whylabs_toolkit/monitor/models/analyzer/analyzer.py

@@ -1,7 +1,7 @@
 """Schema for analyses."""
 from typing import Any, Dict, List, Optional, Union
 
-from pydantic import BaseModel, Field, constr
+from pydantic import BaseModel, Field, constr, validator
 
 from whylabs_toolkit.monitor.models.commons import NoExtrasBaseModel
 
@@ -22,6 +22,7 @@
     DisjunctionConfig,
 )
 from .targets import ColumnMatrix, DatasetMatrix
+from whylabs_toolkit.helpers.cron_validators import validate_cron_expression
 
 
 class Analyzer(NoExtrasBaseModel):
@@ -57,8 +58,8 @@ class Analyzer(NoExtrasBaseModel):
     ] = Field(  # noqa F722
         None, description="A list of tags that are associated with the analyzer."
     )
-    # disabling CronSchedule as it can be tricky on the BE
-    schedule: Optional[FixedCadenceSchedule] = Field(  # Optional[Union[CronSchedule, FixedCadenceSchedule]] = Field(
+
+    schedule: Optional[Union[FixedCadenceSchedule, CronSchedule]] = Field(
         None,
         description="A schedule for running the analyzer. If not set, the analyzer's considered disabled",
     )
@@ -100,6 +101,15 @@ class Analyzer(NoExtrasBaseModel):
         "monthly data.",
     )
 
+    @validator("schedule", pre=True, always=True)
+    def validate_schedule(
+        cls, v: Optional[Union[FixedCadenceSchedule, CronSchedule]]
+    ) -> Optional[Union[FixedCadenceSchedule, CronSchedule]]:
+        """Validate the schedule."""
+        if isinstance(v, CronSchedule) and not validate_cron_expression(v.cron):
+            raise ValueError("CronSchedule must be no less granular than 1 hour and must have 5 fields.")
+        return v
+
     # NOT YET IMPLEMENTED:
     # ExperimentalConfig,
     # ColumnListChangeConfig,

whylabs_toolkit/monitor/models/commons.py

@@ -50,7 +50,6 @@ class CronSchedule(NoExtrasBaseModel):
     exclusionRanges: Optional[List[TimeRange]] = Field(
         title="ExclusionRanges", description="The ranges of dates during which this Analyzer is NOT run."
     )
-    # TODO: support other mode of configuring scheduling
 
 
 class Cadence(str, Enum):