chore(data-warehouse): Disabled the geographic_stats table by default for google ads

frontend/src/scenes/data-warehouse/new/sourceWizardLogic.tsx

@@ -960,7 +960,7 @@ export const sourceWizardLogic = kea<sourceWizardLogicType>([
 
                     if (schema.sync_type === null) {
                         showToast = true
-                        schema.should_sync = true
+                        schema.should_sync = schema.should_sync_default ?? true
 
                         // Use incremental if available
                         if (schema.incremental_available || schema.append_available) {

frontend/src/scenes/data-warehouse/settings/source/Schemas.tsx

@@ -657,6 +657,8 @@ const SyncMethodModal = ({ schema }: { schema: ExternalDataSourceSchema }): JSX.
                     schema={{
                         table: currentSyncMethodModalSchema.name,
                         should_sync: currentSyncMethodModalSchema.should_sync,
+                        description: currentSyncMethodModalSchema.description,
+                        should_sync_default: currentSyncMethodModalSchema.should_sync_default ?? true,
                         sync_type: currentSyncMethodModalSchema.sync_type,
                         sync_time_of_day: currentSyncMethodModalSchema.sync_time_of_day ?? '00:00:00',
                         incremental_field: currentSyncMethodModalSchema.incremental_field ?? null,

frontend/src/types.ts

@@ -5528,6 +5528,7 @@ export interface ExternalDataSourceSyncSchema {
     append_available: boolean
     supports_webhooks: boolean
     description?: string | null
+    should_sync_default: boolean
 }
 
 export interface ExternalDataSourceSchema extends SimpleExternalDataSourceSchema {
@@ -5541,6 +5542,7 @@ export interface ExternalDataSourceSchema extends SimpleExternalDataSourceSchema
     incremental_field_type: string | null
     sync_frequency: DataWarehouseSyncInterval
     description?: string | null
+    should_sync_default?: boolean
 }
 
 export enum ExternalDataSchemaStatus {

posthog/temporal/data_imports/sources/common/schema.py

@@ -14,3 +14,4 @@ class SourceSchema:
     columns: list[tuple[str, str, bool]] = field(default_factory=list)
     foreign_keys: list[tuple[str, str, str]] = field(default_factory=list)
     description: str | None = None
+    should_sync_default: bool = True

posthog/temporal/data_imports/sources/google_ads/google_ads.py

@@ -259,12 +259,16 @@ def __init__(
         *args,
         requires_filter: bool,
         primary_key: list[str],
+        should_sync_default: bool,
+        description: str | None,
         partition_keys: list[str] | None = None,
         extra_where: str | None = None,
         **kwargs,
     ):
         self.requires_filter = requires_filter
         self.primary_key = [pkey.replace(".", "_") for pkey in primary_key]
+        self.should_sync_default = should_sync_default
+        self.description = description
         self.partition_keys = [pkey.replace(".", "_") for pkey in partition_keys] if partition_keys else None
         self.extra_where = extra_where
         super().__init__(*args, **kwargs)
@@ -303,6 +307,9 @@ def get_schemas(config: GoogleAdsSourceConfigUnion, team_id: int) -> TableSchema
         extra_where = typing.cast(str | None, resource_contents.get("extra_where", None))
         partition_keys = typing.cast(list[str] | None, resource_contents.get("partition_keys", None))
 
+        should_sync_default = resource_contents.get("should_sync_default", True)
+        description = resource_contents.get("description", None)
+
         columns = []
 
         for field_name in field_names:
@@ -335,6 +342,8 @@ def get_schemas(config: GoogleAdsSourceConfigUnion, team_id: int) -> TableSchema
             partition_keys=partition_keys,
             columns=columns,
             parents=None,
+            should_sync_default=should_sync_default,
+            description=description,
         )
         table_schemas[table_alias] = table
 

posthog/temporal/data_imports/sources/google_ads/schemas.py

@@ -636,7 +636,8 @@
             "customer.id",
             "segments.date",
         ],
-        "filter_field_names": [("segments.date", IncrementalFieldType.Date)],
+        "should_sync_default": False,
+        "description": "This table can load hundreds of millions of rows for active campaigns. We don't recommend enabling this table unless you know you need it.",
         "extra_where": "metrics.impressions > 0",
         "partition_keys": ["segments.date", "campaign.id", "geographic_view.country_criterion_id"],
         "field_names": [

posthog/temporal/data_imports/sources/google_ads/source.py

@@ -78,8 +78,10 @@ def get_schemas(
                     {"label": column_name, "type": column_type, "field": column_name, "field_type": column_type}
                     for column_name, column_type in ads_incremental_fields.get(endpoint, [])
                 ],
+                description=endpoint_config.description,
+                should_sync_default=endpoint_config.should_sync_default,
             )
-            for endpoint in google_ads_schemas.keys()
+            for endpoint, endpoint_config in google_ads_schemas.items()
         ]
 
         if names is not None:

products/data_warehouse/backend/api/external_data_source.py

@@ -846,6 +846,7 @@ def database_schema(self, request: Request, *arg: Any, **kwargs: Any):
                 "rows": schema.row_count,
                 "supports_webhooks": schema.supports_webhooks,
                 "description": schema.description,
+                "should_sync_default": schema.should_sync_default,
             }
             for schema in schemas
         ]

products/data_warehouse/backend/api/test/test_external_data_source.py

@@ -1505,6 +1505,7 @@ def test_internal_postgres(
                 {
                     "table": "table_1",
                     "should_sync": False,
+                    "should_sync_default": True,
                     "description": None,
                     "rows": 42,
                     "incremental_fields": [
@@ -1556,6 +1557,7 @@ def test_internal_postgres(
                 {
                     "table": "table_1",
                     "should_sync": False,
+                    "should_sync_default": True,
                     "description": None,
                     "rows": 42,
                     "incremental_fields": [