Rerank.py failed with latest change on "multi-file edits"

[HejiaZ2023]

Current rerank.py calls normalize_patch which supposes original_file_content to be a str, but it's already a list now after supportation for multi-file edits. An exception is thrown because of this.

python agentless/repair/rerank.py --patch_folder results/swe-bench-lite/repair_sample_1/,results/swe-bench-lite/repair_sample_2/,results/swe-bench-lite/repair_sample_3/,results/swe-bench-lite/repair_sample_4/ \
                                  --num_samples 40 \
                                  --deduplicate \
                                  --regression \
                                  --reproduction

Traceback (most recent call last):
  File "/.../Agentless/agentless/repair/rerank.py", line 337, in <module>
    main()
  File "/.../Agentless/agentless/repair/rerank.py", line 329, in main
    normalize_patches(args)
  File "/.../Agentless/agentless/repair/rerank.py", line 308, in normalize_patches
    normalized_patch = normalize_patch(
                       ^^^^^^^^^^^^^^^^
  File "/.../Agentless/agentless/util/postprocess_data.py", line 425, in normalize_patch
    new_content = fake_git_apply("playground", edited_file, old_content, patch)
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/.../Agentless/agentless/util/postprocess_data.py", line 187, in fake_git_apply
    f.write(old_content)
TypeError: write() argument must be str, not list

Agentless Commit ID I Used: 35ab2a6

BTW, would you mind to share the commit ID used in your submission on SWE-bench/experiments#133 ? With that, I would be able to continue working on it before a fix is available.

Thanks for you time and contribution to the open source community!

[brutalsavage]

Hi, I have pushed the fix that should address this issue.

You will need to remove the _processed.jsonl files and regenerate them again using --post_process flag as well as --select_id to select the patch number. (Note: you do not need to re run any LLM generations)

[HejiaZ2023]

Thank you for the response! I have tested the solution on my end, and it works perfectly.
Thanks for helping in resolving this issue so quickly!