chore(saas/hubspot): create contacts with default domain

api/integrations/lead_tracking/hubspot/client.py

@@ -3,6 +3,7 @@
 import hubspot
 from django.conf import settings
 from hubspot.crm.companies import (
+    PublicObjectSearchRequest,
     SimplePublicObjectInput,
     SimplePublicObjectInputForCreate,
 )
@@ -14,9 +15,9 @@
 
 
 class HubspotClient:
-    def __init__(self) -> None:
+    def __init__(self, client: hubspot.Client = None) -> None:
         access_token = settings.HUBSPOT_ACCESS_TOKEN
-        self.client = hubspot.Client.create(access_token=access_token)
+        self.client = client or hubspot.Client.create(access_token=access_token)
 
     def get_contact(self, user: FFAdminUser) -> None | dict:
         public_object_id = BatchReadInputSimplePublicObjectId(
@@ -67,20 +68,55 @@ def create_contact(self, user: FFAdminUser, hubspot_company_id: str) -> dict:
         )
         return response.to_dict()
 
+    def get_company_by_domain(self, domain: str) -> dict | None:
+        """
+        Domain should be unique in Hubspot by design, so we should only ever have
+        0 or 1 results.
+        """
+        public_object_search_request = PublicObjectSearchRequest(
+            filter_groups=[
+                {
+                    "filters": [
+                        {"value": domain, "propertyName": "domain", "operator": "EQ"}
+                    ]
+                }
+            ]
+        )
+
+        response = self.client.crm.companies.search_api.do_search(
+            public_object_search_request=public_object_search_request,
+        )
+
+        results = response.to_dict()["results"]
+        if not results:
+            return None
+
+        if len(results) > 1:
+            logger.error("Multiple companies exist in Hubspot for domain %s.", domain)
+
+        return results[0]
+
     def create_company(
         self,
         name: str,
-        active_subscription: str,
-        organisation_id: int,
-        domain: str | None,
+        active_subscription: str = None,
+        organisation_id: int = None,
+        domain: str | None = None,
     ) -> dict:
         properties = {
             "name": name,
             "active_subscription": active_subscription,
             "orgid": str(organisation_id),
         }
+
         if domain:
             properties["domain"] = domain
+        if active_subscription:
+            properties["active_subscription"] = active_subscription
+
+        # hubspot doesn't allow null values for numeric fields, so we
+        # set this to -1 for auto generated organisations.
+        properties["orgid"] = organisation_id or -1
 
         simple_public_object_input_for_create = SimplePublicObjectInputForCreate(
             properties=properties,

api/integrations/lead_tracking/hubspot/lead_tracker.py

@@ -50,46 +50,50 @@ def should_track(user: FFAdminUser) -> bool:
 
         return True
 
-    def create_lead(self, user: FFAdminUser, organisation: Organisation) -> None:
+    def create_lead(self, user: FFAdminUser, organisation: Organisation = None) -> None:
         contact_data = self.client.get_contact(user)
 
         if contact_data:
             # The user is already present in the system as a lead
             # for an existing organisation, so return early.
             return
 
-        hubspot_id = self.get_or_create_organisation_hubspot_id(organisation, user)
+        hubspot_id = self.get_or_create_organisation_hubspot_id(user, organisation)
 
         response = self.client.create_contact(user, hubspot_id)
 
         HubspotLead.objects.create(user=user, hubspot_id=response["id"])
 
     def get_or_create_organisation_hubspot_id(
-        self, organisation: Organisation, user: FFAdminUser
+        self, user: FFAdminUser, organisation: Organisation = None
     ) -> str:
         """
         Return the Hubspot API's id for an organisation.
         """
-        if getattr(organisation, "hubspot_organisation", None):
+        if organisation and getattr(organisation, "hubspot_organisation", None):
             return organisation.hubspot_organisation.hubspot_id
 
         if user.email_domain in settings.HUBSPOT_IGNORE_ORGANISATION_DOMAINS:
             domain = None
         else:
             domain = user.email_domain
-        response = self.client.create_company(
-            name=organisation.name,
-            active_subscription=organisation.subscription.plan,
-            organisation_id=organisation.id,
-            domain=domain,
-        )
 
-        # Store the organisation data in the database since we are
-        # unable to look them up via a unique identifier.
-        HubspotOrganisation.objects.create(
-            organisation=organisation,
-            hubspot_id=response["id"],
-        )
+        if organisation:
+            response = self.client.create_company(
+                name=organisation.name,
+                active_subscription=organisation.subscription.plan,
+                organisation_id=organisation.id,
+                domain=domain,
+            )
+
+            # Store the organisation data in the database since we are
+            # unable to look them up via a unique identifier.
+            HubspotOrganisation.objects.create(
+                organisation=organisation,
+                hubspot_id=response["id"],
+            )
+        else:
+            response = self._get_or_create_company_by_domain(domain)
 
         return response["id"]
 
@@ -112,5 +116,15 @@ def update_company_active_subscription(
 
         return response
 
+    def _get_or_create_company_by_domain(self, domain: str) -> dict:
+        # try to find the organisation via it's domain
+        company = self.client.get_company_by_domain(domain)
+        if not company:
+            # otherwise create a new organisation with the domain
+            # as the name.
+            company = self.client.create_company(name=domain)
+
+        return company
+
     def _get_client(self) -> HubspotClient:
         return HubspotClient()

api/integrations/lead_tracking/hubspot/tasks.py

@@ -4,7 +4,7 @@
 
 
 @register_task_handler()
-def track_hubspot_lead(user_id: int, organisation_id: int) -> None:
+def track_hubspot_lead(user_id: int, organisation_id: int = None) -> None:
     assert settings.ENABLE_HUBSPOT_LEAD_TRACKING
 
     # Avoid circular imports.
@@ -18,10 +18,15 @@ def track_hubspot_lead(user_id: int, organisation_id: int) -> None:
     if not HubspotLeadTracker.should_track(user):
         return
 
-    organisation = Organisation.objects.get(id=organisation_id)
-
     hubspot_lead_tracker = HubspotLeadTracker()
-    hubspot_lead_tracker.create_lead(user=user, organisation=organisation)
+
+    create_lead_kwargs = {"user": user}
+    if organisation_id:
+        create_lead_kwargs["organisation"] = Organisation.objects.get(
+            id=organisation_id
+        )
+
+    hubspot_lead_tracker.create_lead(**create_lead_kwargs)
 
 
 @register_task_handler()
@@ -35,3 +40,24 @@ def update_hubspot_active_subscription(subscription_id: int) -> None:
     subscription = Subscription.objects.get(id=subscription_id)
     hubspot_lead_tracker = HubspotLeadTracker()
     hubspot_lead_tracker.update_company_active_subscription(subscription)
+
+
+@register_task_handler()
+def track_hubspot_lead_without_organisation(user_id: int) -> None:
+    """
+    The Hubspot logic relies on users joining or creating an organisation
+    to be tracked. This should cover most use cases, but for users that
+    sign up but don't join or create an organisation we still want to be
+    able to track them.
+    """
+
+    from users.models import FFAdminUser
+
+    user = FFAdminUser.objects.get(id=user_id)
+    if hasattr(user, "hubspot_lead"):
+        # Since this task is designed to be delayed, there's a chance
+        # that the user will have joined an organisation and thus been
+        # tracked in hubspot already. If so, do nothing.
+        return
+
+    track_hubspot_lead(user.id)

api/tests/unit/integrations/lead_tracking/hubspot/_hubspot_responses.py

@@ -0,0 +1,106 @@
+from datetime import datetime
+
+from dateutil.tz import tzlocal
+
+
+class DummyHubspotResponse:
+    """
+    Dummy class to replicate the to_dict() method of the Hubspot response classes.
+    """
+
+    def __init__(self, data: dict) -> None:
+        self.data = data
+
+    def to_dict(self) -> dict:
+        return self.data
+
+
+def generate_get_company_by_domain_response(
+    name: str, domain: str
+) -> DummyHubspotResponse:
+    """
+    Generate a sample response given by the Hubspot API when searching for a company by domain.
+
+    This response was retrieved from the API directly, and then modified to allow us to set the
+    certain properties dynamically.
+    """
+
+    return DummyHubspotResponse(
+        data={
+            "paging": None,
+            "results": [
+                {
+                    "archived": False,
+                    "archived_at": None,
+                    "created_at": datetime(
+                        2024, 1, 25, 21, 48, 28, 655000, tzinfo=tzlocal()
+                    ),
+                    "id": "9765318341",
+                    "properties": {
+                        "createdate": "2024-01-25T21:48:28.655Z",
+                        "domain": domain,
+                        "hs_lastmodifieddate": "2024-04-23T15:54:13.336Z",
+                        "hs_object_id": "9765318341",
+                        "name": name,
+                    },
+                    "properties_with_history": None,
+                    "updated_at": datetime(
+                        2024, 4, 23, 15, 54, 13, 336000, tzinfo=tzlocal()
+                    ),
+                }
+            ],
+            "total": 1,
+        }
+    )
+
+
+def generate_get_company_by_domain_response_no_results() -> DummyHubspotResponse:
+    """
+    Generate a sample response given by the Hubspot API when searching for a company by domain
+    but no results are returned.
+
+    This response was retrieved from the API directly and hard coded here for simplicity.
+    """
+
+    return DummyHubspotResponse(
+        data={
+            "paging": None,
+            "results": [],
+            "total": 0,
+        }
+    )
+
+
+def generate_create_company_response(
+    name: str, domain: str = None, organisation_id: int = None
+) -> DummyHubspotResponse:
+    """
+    Generate a sample response given by the Hubspot API when creating a company.
+
+    This response was retrieved from the API directly, and then modified to allow us to set the
+    properties dynamically.
+    """
+
+    return DummyHubspotResponse(
+        data={
+            "archived": False,
+            "archived_at": None,
+            "created_at": datetime(2024, 4, 23, 17, 36, 50, 158000, tzinfo=tzlocal()),
+            "id": "11349198823",
+            "properties": {
+                "active_subscription": None,
+                "createdate": "2024-04-23T17:36:50.158Z",
+                "domain": domain,
+                "hs_lastmodifieddate": "2024-04-23T17:36:50.158Z",
+                "hs_object_id": "11349198823",
+                "hs_object_source": "INTEGRATION",
+                "hs_object_source_id": "2902325",
+                "hs_object_source_label": "INTEGRATION",
+                "name": name,
+                "orgid": organisation_id,
+                "website": domain,
+            },
+            "properties_with_history": None,
+            "updated_at": datetime(2024, 4, 23, 17, 36, 50, 158000, tzinfo=tzlocal()),
+        }
+    )