source-shopify-native: conditionally discover PII streams

Alex-Bair · Alex-Bair · commit 9beaea70f8f8 · 2025-06-18T15:56:53.000-04:00
Only specific Shopify plans can access PII data that's fetched via the
`customers`, `orders`, and `fulfillment_orders` streams. The connector
now discovers that stream if the Shopify plan is a Plus or higher
account, is a partner development account, or is not one of the known
non-PII plans (Basic &amp; Starter) AND the connector is using an access
token from the user's custom Shopify app (i.e. not our public OAuth app).

Additionally, the name of each resource was moved into the
`ShopifyGraphQLResource` class to reduce a little duplication &amp; make it
easier to filter out specific resources after they're instantiated.
diff --git a/source-shopify-native/source_shopify_native/graphql/abandoned_checkouts.py b/source-shopify-native/source_shopify_native/graphql/abandoned_checkouts.py
@@ -8,6 +8,7 @@
 
 
 class AbandonedCheckouts(ShopifyGraphQLResource):
+    NAME = "abandoned_checkouts"
     QUERY = """
     id
     abandonedCheckoutUrl
diff --git a/source-shopify-native/source_shopify_native/graphql/collections/collections.py b/source-shopify-native/source_shopify_native/graphql/collections/collections.py
@@ -128,6 +128,7 @@ async def process_result(
 
 
 class CustomCollections(_Collections):
+    NAME = "custom_collections"
     COLLECTION_TYPE = "custom"
 
     @staticmethod
@@ -142,6 +143,7 @@ def process_result(
 
 
 class SmartCollections(_Collections):
+    NAME = "smart_collections"
     COLLECTION_TYPE = "smart"
 
     @staticmethod
diff --git a/source-shopify-native/source_shopify_native/graphql/collections/metafields.py b/source-shopify-native/source_shopify_native/graphql/collections/metafields.py
@@ -3,6 +3,7 @@
 
 
 class CustomCollectionMetafields(MetafieldsResource):
+    NAME = "custom_collection_metafields"
     PARENT_ID_KEY = "gid://shopify/Collection/"
 
     @staticmethod
@@ -18,6 +19,7 @@ def build_query(start: datetime, end: datetime) -> str:
 
 
 class SmartCollectionMetafields(MetafieldsResource):
+    NAME = "smart_collection_metafields"
     PARENT_ID_KEY = "gid://shopify/Collection/"
 
     @staticmethod
diff --git a/source-shopify-native/source_shopify_native/graphql/customers/customers.py b/source-shopify-native/source_shopify_native/graphql/customers/customers.py
@@ -6,6 +6,7 @@
 
 
 class Customers(ShopifyGraphQLResource):
+    NAME = "customers"
     QUERY = """
     displayName
     email
diff --git a/source-shopify-native/source_shopify_native/graphql/customers/metafields.py b/source-shopify-native/source_shopify_native/graphql/customers/metafields.py
@@ -3,6 +3,7 @@
 
 
 class CustomerMetafields(MetafieldsResource):
+    NAME = "customer_metafields"
     PARENT_ID_KEY = "gid://shopify/Customer/"
 
     @staticmethod
diff --git a/source-shopify-native/source_shopify_native/graphql/inventory/inventory_items.py b/source-shopify-native/source_shopify_native/graphql/inventory/inventory_items.py
@@ -6,6 +6,7 @@
 
 
 class InventoryItems(ShopifyGraphQLResource):
+    NAME = "inventory_items"
     QUERY = """
     id
     legacyResourceId
diff --git a/source-shopify-native/source_shopify_native/graphql/inventory/inventory_levels.py b/source-shopify-native/source_shopify_native/graphql/inventory/inventory_levels.py
@@ -7,6 +7,7 @@
 
 
 class InventoryLevels(ShopifyGraphQLResource):
+    NAME = "inventory_levels"
     QUERY = """
     inventoryLevels {
         edges {
diff --git a/source-shopify-native/source_shopify_native/graphql/locations/locations.py b/source-shopify-native/source_shopify_native/graphql/locations/locations.py
@@ -6,6 +6,7 @@
 
 
 class Locations(ShopifyGraphQLResource):
+    NAME = "locations"
     QUERY = """
     id
     name
diff --git a/source-shopify-native/source_shopify_native/graphql/locations/metafields.py b/source-shopify-native/source_shopify_native/graphql/locations/metafields.py
@@ -3,6 +3,7 @@
 
 
 class LocationMetafields(MetafieldsResource):
+    NAME = "location_metafields"
     PARENT_ID_KEY = "gid://shopify/Location/"
 
     @staticmethod
diff --git a/source-shopify-native/source_shopify_native/graphql/orders/agreements.py b/source-shopify-native/source_shopify_native/graphql/orders/agreements.py
@@ -8,6 +8,7 @@
 
 
 class OrderAgreements(ShopifyGraphQLResource):
+    NAME = "order_agreements"
     QUERY = """
     agreements {
         edges {
diff --git a/source-shopify-native/source_shopify_native/graphql/orders/fulfillment_orders.py b/source-shopify-native/source_shopify_native/graphql/orders/fulfillment_orders.py
@@ -7,6 +7,7 @@
 
 
 class FulfillmentOrders(ShopifyGraphQLResource):
+    NAME = "fulfillment_orders"
     QUERY = """
     id
     fulfillmentOrders {
diff --git a/source-shopify-native/source_shopify_native/graphql/orders/fulfillments.py b/source-shopify-native/source_shopify_native/graphql/orders/fulfillments.py
@@ -6,6 +6,7 @@
 
 
 class Fulfillments(ShopifyGraphQLResource):
+    NAME = "fulfillments"
     QUERY = """
     fulfillments {
         id
diff --git a/source-shopify-native/source_shopify_native/graphql/orders/metafields.py b/source-shopify-native/source_shopify_native/graphql/orders/metafields.py
@@ -3,6 +3,7 @@
 
 
 class OrderMetafields(MetafieldsResource):
+    NAME = "order_metafields"
     PARENT_ID_KEY = "gid://shopify/Order/"
 
     @staticmethod
diff --git a/source-shopify-native/source_shopify_native/graphql/orders/orders.py b/source-shopify-native/source_shopify_native/graphql/orders/orders.py
@@ -8,6 +8,7 @@
 
 
 class Orders(ShopifyGraphQLResource):
+    NAME = "orders"
     QUERY = """
     app {
         id
diff --git a/source-shopify-native/source_shopify_native/graphql/orders/refunds.py b/source-shopify-native/source_shopify_native/graphql/orders/refunds.py
@@ -7,6 +7,7 @@
 
 
 class OrderRefunds(ShopifyGraphQLResource):
+    NAME = "order_refunds"
     QUERY = """
     displayFinancialStatus
     displayFulfillmentStatus
diff --git a/source-shopify-native/source_shopify_native/graphql/orders/risks.py b/source-shopify-native/source_shopify_native/graphql/orders/risks.py
@@ -6,6 +6,7 @@
 
 
 class OrderRisks(ShopifyGraphQLResource):
+    NAME = "order_risks"
     QUERY = """
     risk {
         recommendation
diff --git a/source-shopify-native/source_shopify_native/graphql/orders/transactions.py b/source-shopify-native/source_shopify_native/graphql/orders/transactions.py
@@ -7,6 +7,7 @@
 
 
 class OrderTransactions(ShopifyGraphQLResource):
+    NAME = "order_transactions"
     QUERY = """
     transactions {
         id
diff --git a/source-shopify-native/source_shopify_native/graphql/products/media.py b/source-shopify-native/source_shopify_native/graphql/products/media.py
@@ -7,6 +7,7 @@
 
 
 class ProductMedia(ShopifyGraphQLResource):
+    NAME = "product_media"
     QUERY = """
     media(query:"media_type:IMAGE") {
         edges {
diff --git a/source-shopify-native/source_shopify_native/graphql/products/metafields.py b/source-shopify-native/source_shopify_native/graphql/products/metafields.py
@@ -3,6 +3,7 @@
 
 
 class ProductMetafields(MetafieldsResource):
+    NAME = "product_metafields"
     PARENT_ID_KEY = "gid://shopify/Product/"
 
     @staticmethod
diff --git a/source-shopify-native/source_shopify_native/graphql/products/products.py b/source-shopify-native/source_shopify_native/graphql/products/products.py
@@ -7,6 +7,7 @@
 
 
 class Products(ShopifyGraphQLResource):
+    NAME = "products"
     QUERY = """
     title
     bodyHtml
diff --git a/source-shopify-native/source_shopify_native/graphql/products/variants.py b/source-shopify-native/source_shopify_native/graphql/products/variants.py
@@ -7,6 +7,7 @@
 
 
 class ProductVariants(ShopifyGraphQLResource):
+    NAME = "product_variants"
     QUERY = """
     variants {
         edges {
diff --git a/source-shopify-native/source_shopify_native/models.py b/source-shopify-native/source_shopify_native/models.py
@@ -196,9 +196,53 @@ class BulkOperationRunQuery(BaseModel, extra="forbid"):
     data: Data
 
 
+# Names of Shopify plan types. Some plan types do not have access
+# to certain resources (ex: BASIC and STARTER plans cannot access PII, like customer data).
+class PlanName(StrEnum):
+    STARTER = "Starter"
+    BASIC = "Basic"
+    SHOPIFY = "Shopify"
+    ADVANCED = "Advanced"
+    PLUS = "Plus"
+    SHOPIFY_PLUS = "Shopify Plus"
+
+
+class ShopDetails(BaseModel, extra="allow"):
+    class Data(BaseModel, extra="forbid"):
+        class Shop(BaseModel, extra="forbid"):
+            class Plan(BaseModel, extra="forbid"):
+                # The displayName field will be deprecated in the future,
+                # but its replacement publicDisplayName is not available
+                # on the current API version 2025-04.
+                displayName: str
+                partnerDevelopment: bool
+                shopifyPlus: bool
+
+            plan: Plan
+
+        shop: Shop
+
+    data: Data
+
+    @staticmethod
+    def query() -> str:
+        return """
+        {
+            shop {
+                plan {
+                    displayName
+                    partnerDevelopment
+                    shopifyPlus
+                }
+            }
+        }
+        """
+
+
 class ShopifyGraphQLResource(BaseDocument, extra="allow"):
     QUERY: ClassVar[str] = ""
     FRAGMENTS: ClassVar[list[str]] = []
+    NAME: ClassVar[str] = ""  # Add NAME class variable for resource name
 
     id: str
 
diff --git a/source-shopify-native/source_shopify_native/resources.py b/source-shopify-native/source_shopify_native/resources.py
@@ -17,42 +17,82 @@
 
 from .models import (
     OAUTH2_SPEC,
+    AccessToken,
     EndpointConfig,
     ShopifyGraphQLResource,
+    ShopDetails,
+    PlanName,
 )
 from .api import (
     fetch_incremental,
 )
 
 AUTHORIZATION_HEADER = "X-Shopify-Access-Token"
 
-INCREMENTAL_RESOURCES: list[tuple[str, type[ShopifyGraphQLResource]]] = [
-    ("abandoned_checkouts", gql.AbandonedCheckouts),
-    ("customers", gql.Customers),
-    ("customer_metafields", gql.CustomerMetafields),
-    ("products", gql.Products),
-    ("product_media", gql.ProductMedia),
-    ("product_metafields", gql.ProductMetafields),
-    ("product_variants", gql.ProductVariants),
-    ("fulfillment_orders", gql.FulfillmentOrders),
-    ("fulfillments", gql.Fulfillments),
-    ("orders", gql.Orders),
-    ("order_agreements", gql.OrderAgreements),
-    ("order_metafields", gql.OrderMetafields),
-    ("order_transactions", gql.OrderTransactions),
-    ("order_refunds", gql.OrderRefunds),
-    ("order_risks", gql.OrderRisks),
-    ("inventory_items", gql.InventoryItems),
-    ("inventory_levels", gql.InventoryLevels),
-    ("custom_collections", gql.CustomCollections),
-    ("smart_collections", gql.SmartCollections),
-    ("custom_collection_metafields", gql.CustomCollectionMetafields),
-    ("smart_collection_metafields", gql.SmartCollectionMetafields),
-    ("locations", gql.Locations),
-    ("location_metafields", gql.LocationMetafields),
+INCREMENTAL_RESOURCES: list[type[ShopifyGraphQLResource]] = [
+    gql.AbandonedCheckouts,
+    gql.Customers,
+    gql.CustomerMetafields,
+    gql.Products,
+    gql.ProductMedia,
+    gql.ProductMetafields,
+    gql.ProductVariants,
+    gql.FulfillmentOrders,
+    gql.Fulfillments,
+    gql.Orders,
+    gql.OrderAgreements,
+    gql.OrderMetafields,
+    gql.OrderTransactions,
+    gql.OrderRefunds,
+    gql.OrderRisks,
+    gql.InventoryItems,
+    gql.InventoryLevels,
+    gql.CustomCollections,
+    gql.SmartCollections,
+    gql.CustomCollectionMetafields,
+    gql.SmartCollectionMetafields,
+    gql.Locations,
+    gql.LocationMetafields,
 ]
 
 
+PII_RESOURCES: list[type[ShopifyGraphQLResource]] = [
+    gql.Customers,
+    gql.Orders,
+    gql.FulfillmentOrders,
+]
+
+
+async def _can_access_pii(
+    http: HTTPMixin,
+    url: str,
+    log: Logger,
+) -> bool:
+    response = ShopDetails.model_validate_json(
+        await http.request(
+            log, url, method="POST", json={"query": ShopDetails.query()}
+        )
+    )
+
+    plan = response.data.shop.plan
+
+    if plan.partnerDevelopment or plan.shopifyPlus:
+        return True
+
+    match plan.displayName:
+        case PlanName.BASIC | PlanName.STARTER:
+            return False
+        case _:
+            if plan.displayName in PlanName:
+                return True
+            else:
+                log.warning(
+                    f"Shopify plan '{plan.displayName}' is not recognized. "
+                    f"Assuming access to PII is supported on the {plan.displayName} plan."
+                )
+                return True
+
+
 def _incremental_resources(
     http: HTTPMixin,
     config: EndpointConfig,
@@ -82,17 +122,17 @@ def open(
 
     return [
         Resource(
-            name=name,
+            name=model.NAME,
             key=["/id"],
             model=ShopifyGraphQLResource,
             open=functools.partial(open, model),
             initial_state=ResourceState(
                 inc=ResourceState.Incremental(cursor=config.start_date),
             ),
-            initial_config=ResourceConfig(name=name, interval=timedelta(minutes=5)),
+            initial_config=ResourceConfig(name=model.NAME, interval=timedelta(minutes=5)),
             schema_inference=True,
         )
-        for name, model in INCREMENTAL_RESOURCES
+        for model in INCREMENTAL_RESOURCES
     ]
 
 
@@ -134,6 +174,19 @@ async def all_resources(
     if should_cancel_ongoing_job:
         await bulk_job_manager.cancel_current()
 
-    return [
+
+    resources = [
         *_incremental_resources(http, config, bulk_job_manager),
     ]
+
+    # If the user is authenticating with an access token from their custom Shopify app,
+    # they may not be able to access certain PII data.
+    # https://help.shopify.com/en/manual/apps/app-types/custom-apps
+    # https://community.shopify.com/c/shopify-discussions/no-more-customer-pii-in-custom-app-integrations-for-shopify/td-p/2496209
+    if isinstance(config.credentials, AccessToken) and not await _can_access_pii(http, bulk_job_manager.url, log):
+        for model in PII_RESOURCES:
+            resources = [
+                r for r in resources if r.name != model.NAME
+            ]
+
+    return resources
