Update summarize_fail.py

scripts/tools/summarize_fail.py

@@ -1,19 +1,25 @@
+import logging
 import os
 import pandas as pd
 import subprocess
 
-subprocess.run("gh run list -R project-chip/connectedhomeip -b master -s failure --json databaseId,displayTitle,workflowName > run_list.json", shell=True)
-df = pd.read_json("run_list.json")
-df.columns = ["ID", "Pull Request", "Workflow"]
-print("Recent Failures:")
-print(df.to_string(index=False))
-df.to_csv("recent_fails.csv", index=False)
-print()
-print("Percentage Frequency:")
-frequency = df["Workflow"].value_counts(normalize=True).mul(100).astype(str).reset_index(name="Percentage")
-print(frequency.to_string(index=False))
-frequency.to_csv("recent_fails_frequency.csv")
-print()
-for failId in df["ID"].tolist():
-    os.makedirs(f"fail_logs/{failId}")  # Resolve this to actual run name later
-    subprocess.run(f"gh run view -R project-chip/connectedhomeip {failId} --log-failed > fail_logs/{failId}/raw_output.txt", shell=True)
+def process_fail(id, pr, workflow):
+    os.makedirs(f"logs/{pr}/{workflow}")
+    subprocess.run(f"gh run view -R project-chip/connectedhomeip {id} --log-failed > logs/{pr}/{workflow}/fail_logs.txt", shell=True)
+
+def main():
+    logging.info("Gathering recent run failure information into run_list.json.")
+    subprocess.run("gh run list -R project-chip/connectedhomeip -b master -s failure --json databaseId,displayTitle,workflowName > run_list.json", shell=True)
+    df = pd.read_json("run_list.json")
+    df.columns = ["ID", "Pull Request", "Workflow"]
+    logging.info("Recent Failures:")
+    logging.info(df.to_string(index=False))
+    df.to_csv("recent_fails.csv", index=False)
+    logging.info("Percentage Frequency of Workflow Failures:")
+    frequency = df["Workflow"].value_counts(normalize=True).mul(100).astype(str).reset_index(name="Percentage")
+    logging.info(frequency.to_string(index=False))
+    frequency.to_csv("recent_workflow_fails_frequency.csv")
+    df.apply(lambda row: process_fail(row["ID"], row["Pull Request"], row["Workflow"]), axis=1)
+
+if __name__ == "__main__":
+    main()