From 3655f13666ba9f38966de26c12f559cc4761ab4c Mon Sep 17 00:00:00 2001
From: Gabe Joseph <gjoseph92@gmail.com>
Date: Wed, 7 Sep 2022 03:05:09 -0600
Subject: [PATCH] Fix test report after queuing job added (#7012)

---
 continuous_integration/scripts/test_report.py | 41 ++++++++++++-------
 1 file changed, 27 insertions(+), 14 deletions(-)

diff --git a/continuous_integration/scripts/test_report.py b/continuous_integration/scripts/test_report.py
index f8afeae2530..6f37d8b630d 100644
--- a/continuous_integration/scripts/test_report.py
+++ b/continuous_integration/scripts/test_report.py
@@ -136,23 +136,31 @@ def get_jobs(run, session):
             cache[url] = jobs
 
     df_jobs = pandas.DataFrame.from_records(jobs)
+    # Interpolate the `$TEST_ID` variable from the job name.
+    # Somehow the job ID is not part of the workflow schema and we have no other way to later join
+    # this to the JXML results.
+
     name_components = (
         df_jobs.name.str.extract(r"test \((.+)\)", expand=False)
         .dropna()
         .str.split(", ", expand=True)
-        .set_axis(["OS", "python_version", "queuing", "partition"], axis="columns")
-        .assign(
-            # We later need to join on this. Somehow the job ID is not part of the workflow schema and we have no other way to join
-            suite_name=lambda df: df["OS"]
-            + "-"
-            + df["python_version"]
-            + "-"
-            + df["queuing"]
-            + "-"
-            + df["partition"].str.replace(" ", "")
-        )
     )
-    return pandas.concat([df_jobs, name_components], axis="columns")
+    if len(name_components.columns) == 4:
+        name_components.columns = ["OS", "python_version", "queuing", "partition"]
+    elif len(name_components.columns) == 3:
+        # Migration: handle older jobs without the `queuing` configuration.
+        # This branch can be removed after 2022-12-07.
+        name_components.columns = ["OS", "python_version", "partition"]
+    else:
+        raise ValueError(f"Job names must have 3 or 4 components:\n{name_components!r}")
+
+    # See `Set $TEST_ID` step in `tests.yaml`
+    name_components["partition"] = name_components.partition.str.replace(" ", "")
+
+    df_jobs["suite_name"] = name_components.iloc[:, 0].str.cat(
+        name_components.iloc[:, 1:], sep="-"
+    )
+    return df_jobs
 
 
 def get_workflow_run_listing(
@@ -206,7 +214,12 @@ def suite_from_name(name: str) -> str:
     can have matrix partitions, pytest marks, etc. Basically,
     just lop off the front of the name to get the suite.
     """
-    return "-".join(name.split("-")[:4])
+    parts = name.split("-")
+    if len(parts) == 4:  # [OS, 'latest', py_version, $PARTITION_LABEL]
+        # Migration: handle older jobs without the `queuing` configuration.
+        # This branch can be removed after 2022-12-07.
+        parts.insert(3, "no_queue")
+    return "-".join(parts[:4])
 
 
 def download_and_parse_artifact(
@@ -354,7 +367,7 @@ def download_and_parse_artifacts(
                 html_url = jobs_df[jobs_df["suite_name"] == a["name"]].html_url.unique()
                 assert (
                     len(html_url) == 1
-                ), f"Artifact suit name {a['name']} did not match any jobs dataframe {jobs_df['suite_name'].unique()}"
+                ), f"Artifact suite name {a['name']} did not match any jobs dataframe:\n{jobs_df['suite_name'].unique()}"
                 html_url = html_url[0]
                 assert html_url is not None
                 df2 = df.assign(