Extract event name and category more accurately (#1695)

* Extract event name and category more accurately, and show it in UI * Simplify method Co-authored-by: Jan-Erik Rediger <jrediger@mozilla.com> * Change the order of name and category to match the new return value --------- Co-authored-by: Jan-Erik Rediger <jrediger@mozilla.com>
mozilla · May 5, 2023 · b1270fd · b1270fd
1 parent 7e7a814
commit b1270fd
Show file tree

Hide file tree

Showing 4 changed files with 42 additions and 6 deletions.
diff --git a/etl/glean_etl.py b/etl/glean_etl.py
@@ -10,7 +10,7 @@
 from .glean import GleanApp
 from .looker import get_looker_explore_metadata_for_metric, get_looker_explore_metadata_for_ping
 from .search import create_metrics_search_js
-from .utils import dump_json
+from .utils import dump_json, get_event_name_and_category
 
 # Various additional sources of metadata
 ANNOTATIONS_URL = os.getenv(
@@ -374,6 +374,12 @@ def write_glean_metadata(output_dir, functions_dir, app_names=None):
                         app_tags_for_objects,
                     )
 
+                    if metric.definition["type"] == "event":
+                        app_metrics[metric.identifier]["event_info"] = {
+                            "name": get_event_name_and_category(metric.identifier)[1],
+                            "category": get_event_name_and_category(metric.identifier)[0],
+                        }
+
                     # sort "send in pings" alphanumerically, except that `metrics`
                     # should always be first if present and `deletion-request`
                     # should be last

diff --git a/etl/looker.py b/etl/looker.py
@@ -3,6 +3,8 @@
 import stringcase
 from furl import furl
 
+from etl.utils import get_event_name_and_category
+
 from .bigquery import get_bigquery_column_name, get_bigquery_ping_table_name
 from .glean import GLEAN_DISTRIBUTION_TYPES
 
@@ -115,7 +117,7 @@ def get_looker_explore_metadata_for_metric(
     if not app.app.get("deprecated") and base_looker_explore:
         looker_metric_link = None
         if metric_type == "event":
-            (metric_category, metric_name) = metric.identifier.split(".", 1)
+            (metric_category, metric_name) = get_event_name_and_category(metric.identifier)
             if base_looker_explore["name"] == "glean_event_counts":
                 looker_metric_link = furl(base_looker_explore["url"]).add(
                     {

diff --git a/etl/utils.py b/etl/utils.py
@@ -41,3 +41,19 @@ def dump_json(data):
     2. We dump the data without spaces (since we want things as small as possible)
     """
     return json.dumps(data, separators=(",", ":"), default=_serialize_sets)
+
+
+def get_event_name_and_category(event_identifier: str):
+    """Get the event name and category from an event identifier.
+
+    Event names cannot contain dots, only the category, so
+    splitting at the right most '.' will extract the category
+    and name, i.e. the name is the last part of the event
+    identifier, and the event category is everything before the
+    event name.
+
+    e.g. "newtab.search.ad.click_metric" ->
+    category = "newtab.search.ad"
+    name = "click_metric"
+    """
+    return event_identifier.rsplit(".", maxsplit=1)
diff --git a/src/pages/MetricDetail.svelte b/src/pages/MetricDetail.svelte
@@ -47,10 +47,9 @@
   const STMO_GLEAN_EVENT_QUERY_URL =
     "https://sql.telemetry.mozilla.org/queries/91780/source";
 
-  const getSTMOQueryURL = (type, columnName, eventName, table) => {
+  const getSTMOQueryURL = (type, columnName, eventInfo, table) => {
     if (type === "event") {
-      const [category, name] = eventName ? eventName.split(".") : [];
-      return `${STMO_GLEAN_EVENT_QUERY_URL}?p_event_category=${category}&p_event_name=${name}&p_table=${table}`;
+      return `${STMO_GLEAN_EVENT_QUERY_URL}?p_event_category=${eventInfo.category}&p_event_name=${eventInfo.name}&p_table=${table}`;
     }
     return `${STMO_GLEAN_QUERY_URL}?p_metric%20location=${columnName}&p_table=${table}`;
   };
@@ -154,6 +153,13 @@
     ping{metric.send_in_pings.length > 1 ? "s" : ""}.
   </p>
 
+  {#if metric.type === "event"}
+    <p>
+      The event category is <code>{metric.event_info.category}</code> and event
+      name is <code>{metric.event_info.name}.</code>
+    </p>
+  {/if}
+
   <MetadataTable
     appName={params.app}
     item={metric}
@@ -324,6 +330,12 @@
               <CopyButton
                 textToCopy={selectedAppVariant.etl.bigquery_column_name}
               />
+            {:else}
+              (event.category=<code>{metric.event_info.category}</code>
+              <CopyButton textToCopy={metric.event_info.category} /> and event.name=<code
+                >{metric.event_info.name}</code
+              >
+              <CopyButton textToCopy={metric.event_info.name} />)
             {/if}
           </div>
         </td>
@@ -343,7 +355,7 @@
               href={getSTMOQueryURL(
                 metric.type,
                 selectedAppVariant.etl.bigquery_column_name,
-                pingData.looker.metric.name,
+                metric.event_info,
                 pingData.bigquery_table
               )}>STMO</a
             >