The file-size rule fires at >1500 lines. In the slopbucket corpus the bloated/ directory is 8 files that are each exactly 356 lines / ~29KB of repetitive assignments — they sit just under the line threshold and pass clean.
Evidence (pinned to e6793a3e26c394cb116711196e9335078086fe1b):
Line count is easy to game. Suggested additions: a bytes-per-file threshold, and/or a repetition ratio (fraction of lines matching a repeated structural template) so high-volume boilerplate is caught regardless of line count.
The
file-sizerule fires at >1500 lines. In theslopbucketcorpus thebloated/directory is 8 files that are each exactly 356 lines / ~29KB of repetitive assignments — they sit just under the line threshold and pass clean.Evidence (pinned to
e6793a3e26c394cb116711196e9335078086fe1b):bloated/bloated_transform_0.py— 356 lines, 350 of themresult["<rand>_N"] = "<word salad>"bloated_transform_1.py…_7.pyLine count is easy to game. Suggested additions: a bytes-per-file threshold, and/or a repetition ratio (fraction of lines matching a repeated structural template) so high-volume boilerplate is caught regardless of line count.