Describe the bug
While looking through sbin/anonymize_autograding_logs.py, I noticed that it only handles autograding log entries whose job path matches term/course/submissions/gradeable/user/version. Valid log entries for generated_output jobs, such as term/course/generated_output/gradeable, do not match that format and are silently skipped during anonymization.
As a result, anonymized autograding logs can lose valid entries.
Expected behavior
The anonymizer should preserve valid generated_output log entries instead of dropping them. Since those entries do not include a student username, they could either be copied through unchanged or handled separately, but they should still appear in the output.
To Reproduce
Steps to reproduce the behavior:
- Create a file containing this valid autograding log line:
2026-03-30 12:00:00 | ABC123 | | zip | f24/sample/generated_output/lab1 | wait: 1.000 sec | ok
- Run:
python3 sbin/anonymize_autograding_logs.py file /tmp/generated_input.txt /tmp/generated_output.txt XX
- Open
/tmp/generated_output.txt
- See that the output file is empty, even though the input line is a valid log entry
Configuration
Screenshots
N/A
Additional context
This script is used to anonymize autograding logs before sharing them for debugging or performance experiments. While reviewing/testing the script, I noticed it only supports submissions/... paths. I then checked the autograder code and found that valid generated_output/... job names are also written to the autograding log, and confirmed with a small repro that those entries are silently dropped by the anonymizer.
Describe the bug
While looking through
sbin/anonymize_autograding_logs.py, I noticed that it only handles autograding log entries whose job path matchesterm/course/submissions/gradeable/user/version. Valid log entries forgenerated_outputjobs, such asterm/course/generated_output/gradeable, do not match that format and are silently skipped during anonymization.As a result, anonymized autograding logs can lose valid entries.
Expected behavior
The anonymizer should preserve valid
generated_outputlog entries instead of dropping them. Since those entries do not include a student username, they could either be copied through unchanged or handled separately, but they should still appear in the output.To Reproduce
Steps to reproduce the behavior:
2026-03-30 12:00:00 | ABC123 | | zip | f24/sample/generated_output/lab1 | wait: 1.000 sec | okpython3 sbin/anonymize_autograding_logs.py file /tmp/generated_input.txt /tmp/generated_output.txt XX/tmp/generated_output.txtConfiguration
Screenshots
N/A
Additional context
This script is used to anonymize autograding logs before sharing them for debugging or performance experiments. While reviewing/testing the script, I noticed it only supports
submissions/...paths. I then checked the autograder code and found that validgenerated_output/...job names are also written to the autograding log, and confirmed with a small repro that those entries are silently dropped by the anonymizer.