[CI] Collect labels and time merged for LLVM pull requests (llvm#740)

Begin collecting data for when pull requests are merged into main, as
well as what labels those pull requests are tagged with.

For the time being, all pull requests we're concerned with should have
their `mergedAt` field set. However, it is likely that we will also be
interested in pull requests that haven't been merged sometime in the
near future.

Author: jriv01

premerge/bigquery_schema/llvm_pull_requests_table_schema.json

@@ -11,6 +11,12 @@
     "mode": "NULLABLE",
     "description": "Time the pull request was created at, as a Unix timestamp."
   },
+  {
+    "name": "merged_at_timestamp_seconds",
+    "type": "INTEGER",
+    "mode": "NULLABLE",
+    "description": "Time the pull request was merged at, as a Unix timestamp."
+  },
   {
     "name": "pull_request_number",
     "type": "INTEGER",
@@ -22,5 +28,11 @@
     "type": "STRING",
     "mode": "NULLABLE",
     "description": "SHA of the commit associated with the pull request."
+  },
+  {
+    "name": "labels",
+    "type": "STRING",
+    "mode": "REPEATED",
+    "description": "Labels associated with the pull request."
   }
 ]

premerge/ops-container/process_llvm_commits.py

@@ -45,6 +45,12 @@
           author {{ login }}
           number
           createdAt
+          mergedAt
+          labels(first: 25) {{
+            nodes {{
+              name
+            }}
+          }}
           reviews(last: 100) {{
             nodes {{
               state
@@ -66,7 +72,9 @@ class LLVMCommitData:
   commit_sha: str
   commit_timestamp_seconds: int
   diff: list[dict[str, int | str]]
-  commit_author: str | None = None  # Username of author is unknown until API call
+  commit_author: str | None = (
+      None  # Username of author is unknown until API call
+  )
   associated_pull_request: int | None = None
   is_revert: bool = False
   pull_request_reverted: int | None = None
@@ -78,7 +86,9 @@ class LLVMPullRequestData:
   pull_request_number: int
   pull_request_author: str
   pull_request_timestamp_seconds: int
+  merged_at_timestamp_seconds: int
   associated_commit: str
+  labels: list[str]
 
 
 @dataclasses.dataclass
@@ -255,16 +265,26 @@ def extract_pull_request_data(
       )
 
     # Convert ISO timestamp to Unix timestamp, in seconds
-    unix_timestamp = int(
+    create_unix_timestamp = int(
         datetime.datetime.fromisoformat(pull_request["createdAt"]).timestamp()
     )
+    if pull_request["mergedAt"] is not None:
+      merge_unix_timestamp = int(
+          datetime.datetime.fromisoformat(pull_request["mergedAt"]).timestamp()
+      )
+    else:
+      merge_unix_timestamp = None
 
     pull_request_data.append(
         LLVMPullRequestData(
             pull_request_number=pull_request["number"],
             pull_request_author=author_login,
-            pull_request_timestamp_seconds=unix_timestamp,
+            pull_request_timestamp_seconds=create_unix_timestamp,
+            merged_at_timestamp_seconds=merge_unix_timestamp,
             associated_commit=commit_sha.removeprefix("commit_"),
+            labels=[
+                label["name"] for label in pull_request["labels"]["nodes"]
+            ],
         )
     )
 
@@ -354,7 +374,6 @@ def query_github_graphql_api(
       },
       json={"query": query},
   )
-
   # Exit if API call fails
   # A failed API call means a large batch of data is missing and will not be
   # reflected in the dashboard. The dashboard will silently misrepresent

premerge/ops-container/process_llvm_commits_test.py

@@ -73,8 +73,10 @@ def _create_pull_request_api_data(
       self,
       pull_request_number: int,
       created_at: str | None = None,
+      merged_at: str | None = None,
       pull_request_author: str | None = None,
       reviews: list[dict[str, Any]] | None = None,
+      labels: list[str] | None = None,
   ) -> dict[str, Any]:
     """Create a GitHub API response for a pull request."""
     return {
@@ -83,7 +85,9 @@ def _create_pull_request_api_data(
             {'login': pull_request_author} if pull_request_author else None
         ),
         'createdAt': created_at,
+        'mergedAt': merged_at,
         'reviews': {'nodes': reviews or []},
+        'labels': {'nodes': [{'name': label} for label in labels or []]},
     }
 
   def _create_review_api_data(
@@ -321,10 +325,14 @@ def test_extract_pull_request_data(self):
     """Test extracting pull request data from GitHub API data."""
     created_at = datetime.datetime(2020, 1, 1, tzinfo=datetime.timezone.utc)
     created_at_iso = created_at.isoformat()
+    merged_at = datetime.datetime(2020, 1, 2, tzinfo=datetime.timezone.utc)
+    merged_at_iso = merged_at.isoformat()
     pull_request_api_data = self._create_pull_request_api_data(
         pull_request_number=12345,
         created_at=created_at_iso,
+        merged_at=merged_at_iso,
         pull_request_author='pull_request_author',
+        labels=['llvm:test-label'],
     )
     commit_api_data = {
         'commit_abcdef': self._create_commit_api_data(
@@ -345,7 +353,12 @@ def test_extract_pull_request_data(self):
         pull_request_data[0].pull_request_timestamp_seconds,
         created_at.timestamp(),
     )
+    self.assertEqual(
+        pull_request_data[0].merged_at_timestamp_seconds,
+        merged_at.timestamp(),
+    )
     self.assertEqual(pull_request_data[0].associated_commit, 'abcdef')
+    self.assertIn('llvm:test-label', pull_request_data[0].labels)
 
   def test_extract_pull_request_data_with_missing_author(self):
     """Test extracting pull request data from GitHub API data."""