:hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP.
Python ab-testingai-analyticsanalyticscdpdata-warehouseexperimentsfeature-flagsjavascriptproduct-analyticspythonreactsession-replaysurveystypescriptweb-analytics
"""
Contract types for data_quality.
Frozen, framework-free values other products need. No Django imports.
"""
from dataclasses import dataclass, field
from typing import Any, Literal, TypeGuard
from uuid import UUID
from posthog.dataclasses import frozen
from .enums import SuiteRunTrigger
CHECK_SUITE_WORKFLOW_NAME = "data-quality-run-suite"
MATERIALIZATION_GATE_ACTIVITY_NAME = "data-quality-materialization-gate"
QualityAuditMode = Literal["skip", "warn", "gate"]
QUALITY_AUDIT_SKIP: QualityAuditMode = "skip"
QUALITY_AUDIT_WARN: QualityAuditMode = "warn"
QUALITY_AUDIT_GATE: QualityAuditMode = "gate"
def is_quality_audit_mode(value: str) -> TypeGuard[QualityAuditMode]:
"""A recorded Temporal history can carry a mode another deploy defined and this one does not."""
return value in (QUALITY_AUDIT_SKIP, QUALITY_AUDIT_WARN, QUALITY_AUDIT_GATE)
@dataclass(frozen=True)
class CheckTypeInfo:
"""One entry of the check-type catalog, so a caller can author config without guessing.
Carries what the registry knows about a type, flattened to plain values -- the spec object
itself stays internal, since it is a compiler, not data.
"""
check_type: str
description: str
requires_column: bool
config_schema: dict[str, Any] = field(default_factory=dict)
@frozen
class RunCheckSuiteInputs:
"""What to run. Exactly one selector is expected; a suite with no matching checks is not an error.
``node_ids`` is the materialization path: the DAG workflow knows nodes, not saved queries, and
the mapping is resolved inside the prepare activity so the workflow stays deterministic.
``table_ids`` is the source-sync path, keyed on the table so it works with or without a DAG.
"""
team_id: int
trigger: SuiteRunTrigger
saved_query_ids: list[str] = field(default_factory=list)
table_ids: list[str] = field(default_factory=list)
metric_ids: list[str] = field(default_factory=list)
posthog_table_ids: list[str] = field(default_factory=list)
check_ids: list[str] = field(default_factory=list)
node_ids: list[str] = field(default_factory=list)
suite_run_id: str | None = None
schedule_id: str | None = None
data_modeling_job_id: str | None = None
created_by_id: int | None = None
# Audits the staged folder rather than the published table. Needs exactly one saved query.
staged_queryable_folder: str | None = None
@dataclass(frozen=True)
class MetricSubject:
id: UUID
name: str
display_name: str
@dataclass(frozen=True)
class SelectableSubject:
"""One thing a caller may author a check on, with enough to fill a picker.
``columns`` is empty for a metric, whose output columns come from running its query.
"""
subject_type: str
id: str
name: str
display_name: str = ""
time_column: str = ""
columns: dict[str, str] = field(default_factory=dict)
editable: bool = True
@dataclass(frozen=True)
class OutputColumn:
name: str
type: str | None