[CI] Base data staleness off activity instead of creation (#829)

Instead of strictly using age to determine when to stop updating data
for a pull request, use the time since the most recent pull request
activity.

This will address cases where pull requests can have ongoing reviews and
discussions for months prior to approval and submission.

Author: jriv01

premerge/ops-container/amend_pull_request_data.py

@@ -92,7 +92,7 @@ def mark_stale_pull_request_data_in_bigquery(
     pull_request_state = 'OPEN'
     AND TIMESTAMP_DIFF(
         CURRENT_TIMESTAMP(),
-        TIMESTAMP_SECONDS(pull_request_timestamp_seconds),
+        TIMESTAMP_SECONDS(last_updated_at_timestamp_seconds),
         DAY
     ) > @cutoff_age_days
   """

premerge/ops-container/operational_metrics_lib.py

@@ -22,6 +22,7 @@
 number
 state
 createdAt
+updatedAt
 mergedAt
 label_events: timelineItems(last: 10, itemTypes: [LABELED_EVENT]) {
   nodes {
@@ -76,6 +77,7 @@ class LLVMPullRequestData:
   pull_request_title: str
   pull_request_state: str
   pull_request_timestamp_seconds: int
+  last_updated_at_timestamp_seconds: int
   merged_at_timestamp_seconds: int | None
   associated_commit: str | None
   labels: list[dict[str, str]]
@@ -204,6 +206,9 @@ def parse_pull_request_data(
   create_unix_timestamp = int(
       datetime.datetime.fromisoformat(pull_request["createdAt"]).timestamp()
   )
+  updated_unix_timestamp = int(
+      datetime.datetime.fromisoformat(pull_request["updatedAt"]).timestamp()
+  )
   if pull_request["mergedAt"] is not None:
     merge_unix_timestamp = int(
         datetime.datetime.fromisoformat(pull_request["mergedAt"]).timestamp()
@@ -241,6 +246,7 @@ def parse_pull_request_data(
       pull_request_title=pull_request["title"],
       pull_request_state=pull_request["state"],
       pull_request_timestamp_seconds=create_unix_timestamp,
+      last_updated_at_timestamp_seconds=updated_unix_timestamp,
       merged_at_timestamp_seconds=merge_unix_timestamp,
       associated_commit=associated_commit,
       labels=labels,

premerge/ops-container/process_llvm_commits_test.py

@@ -63,6 +63,7 @@ def _create_pull_request_api_data(
       pull_request_number: int,
       pull_request_title: str,
       created_at: str | None = None,
+      updated_at: str | None = None,
       merged_at: str | None = None,
       pull_request_author: str | None = None,
       reviews: list[dict[str, Any]] | None = None,
@@ -78,6 +79,7 @@ def _create_pull_request_api_data(
         'title': pull_request_title,
         'state': 'MERGED',
         'createdAt': created_at,
+        'updatedAt': updated_at,
         'mergedAt': merged_at,
         'reviews': {'nodes': reviews or []},
         'label_events': {
@@ -286,6 +288,7 @@ def test_extract_pull_request_data(self):
         pull_request_number=12345,
         pull_request_title='[TEST] Title',
         created_at=created_at_iso,
+        updated_at=merged_at_iso,
         merged_at=merged_at_iso,
         pull_request_author='pull_request_author',
         labels=['llvm:test-label'],
@@ -311,6 +314,10 @@ def test_extract_pull_request_data(self):
         pull_request_data[0].pull_request_timestamp_seconds,
         created_at.timestamp(),
     )
+    self.assertEqual(
+        pull_request_data[0].last_updated_at_timestamp_seconds,
+        merged_at.timestamp(),
+    )
     self.assertEqual(
         pull_request_data[0].merged_at_timestamp_seconds,
         merged_at.timestamp(),
@@ -330,6 +337,7 @@ def test_extract_pull_request_data_with_missing_author(self):
         pull_request_number=12345,
         pull_request_title='[TEST] Title',
         created_at=created_at_iso,
+        updated_at=created_at_iso,
     )
     commit_api_data = {
         'commit_abcdef': self._create_commit_api_data(