DENG-3540 Add `event_id` as a primary key for event view (#960)

* Add `event_id` as a synthetic pk for event view

* ci fix

* Generate an event_id dimension for the final view. Update test.

* comments

* review comments

generator/views/events_view.py

@@ -76,11 +76,9 @@ class EventsView(View):
 
         # set document_id as primary key if it exists in the underlying table
         # this will allow one_to_many joins
-        document_id_field = self.get_document_id(dimensions, "events")
-        if document_id_field is not None:
-            view_defn["dimensions"] = [
-                {"name": document_id_field, "primary_key": "yes"}
-            ]
+        event_id_dimension = self.generate_event_id_dimension(dimensions)
+        if event_id_dimension is not None:
+            view_defn["dimensions"] = [event_id_dimension]
 
         return {
             "includes": [f"{self.tables[0]['events_table_view']}.view.lkml"],
@@ -104,3 +102,17 @@ class EventsView(View):
             )
 
         return measures
+
+    def generate_event_id_dimension(
+        self, dimensions: list[dict]
+    ) -> Optional[Dict[str, str]]:
+        """Generate the event_id dimension to be used as a primary key for a one to many join."""
+        event_id = self.select_dimension("event_id", dimensions, "events")
+        if event_id:
+            event_id_field = event_id["name"]
+            return {
+                "name": "event_id",
+                "primary_key": "yes",
+                "sql": f"${{{event_id_field}}}",
+            }
+        return None

generator/views/view.py

@@ -104,26 +104,39 @@ class View(object):
 
     def get_client_id(self, dimensions: List[dict], table: str) -> Optional[str]:
         """Return the first field that looks like a client identifier."""
-        client_id_fields = [
-            d["name"]
-            for d in dimensions
-            if d["name"] in {"client_id", "client_info__client_id", "context_id"}
-        ]
-        if not client_id_fields:
-            # Some pings purposely disinclude client_ids, e.g. firefox installer
-            return None
-        if len(client_id_fields) > 1:
-            raise ClickException(f"Duplicate client_id dimension in {table!r}")
-        return client_id_fields[0]
+        client_id_fields = self.select_dimension(
+            {"client_id", "client_info__client_id", "context_id"},
+            dimensions,
+            table,
+        )
+        # Some pings purposely disinclude client_ids, e.g. firefox installer
+        return client_id_fields["name"] if client_id_fields else None
 
     def get_document_id(self, dimensions: List[dict], table: str) -> Optional[str]:
         """Return the first field that looks like a document_id."""
-        document_id_fields = [
-            d["name"] for d in dimensions if d["name"] in {"document_id"}
-        ]
-        if not document_id_fields:
-            # Some pings purposely disinclude client_ids, e.g. firefox installer
-            return None
-        if len(document_id_fields) > 1:
-            raise ClickException(f"Duplicate document_id dimension in {table!r}")
-        return document_id_fields[0]
+        document_id = self.select_dimension("document_id", dimensions, table)
+        return document_id["name"] if document_id else None
+
+    def select_dimension(
+        self,
+        dimension_names: str | set[str],
+        dimensions: List[dict],
+        table: str,
+    ) -> Optional[dict[str, str]]:
+        """
+        Return the first field that matches dimension name.
+
+        Throws if the query set is greater than one and more than one item is selected.
+        """
+        if isinstance(dimension_names, str):
+            dimension_names = {dimension_names}
+        selected = [d for d in dimensions if d["name"] in dimension_names]
+        if selected:
+            # there should only be one dimension selected from the set
+            # if there are multiple options in the dimention_names set.
+            if len(dimension_names) > 1 and len(selected) > 1:
+                raise ClickException(
+                    f"Duplicate {'/'.join(dimension_names)} dimension in {table!r}"
+                )
+            return selected[0]
+        return None

tests/test_events.py

@@ -160,6 +160,13 @@ def test_view_lookml(events_view):
                         "sql": "${client_info__client_id}",
                     },
                 ],
+                "dimensions": [
+                    {
+                        "name": "event_id",
+                        "primary_key": "yes",
+                        "sql": "${event_id}",
+                    },
+                ],
             },
         ],
     }
@@ -168,7 +175,8 @@ def test_view_lookml(events_view):
         [
             SchemaField(
                 "client_info", "RECORD", fields=[SchemaField("client_id", "STRING")]
-            )
+            ),
+            SchemaField("event_id", "STRING"),
         ]
     )
     actual = events_view.to_lookml(mock_bq_client, None)