Resolving 'bad object' errors due to rewritten history change in submodules due to Git LFS migration about git-lfs HOT 2 CLOSED

Hey,

This isn't surprising to me based on how submodules work. A submodule in a Git repository consists of an entry in a tree object pointing to an external commit (of mode 160000), plus an entry in .gitmodules that says where to access the repository. Because the tree entry refers to a specific object ID, if you remove that object in the submodule, Git will complain about being unable to check it out.

We don't have automatic functionality for rewriting superprojects, but you can use the --object-map option to make a map of the old and new object IDs and then rewrite the superproject using that map with something like git filter-repo or git filter-branch. However, there's no way to avoid rewriting the superproject history here; if you don't, the submodules can't be accessed in the history.

from git-lfs.

Thank you for the idea @bk2204. I have created a python script exactly does that and it looks like it is working as intended. Here it is the script and what i did in case someone may need this in the future.

I got the map exported with git lfs migrate import --everything --object-map=map and then run the script below in the super project as follows: py fixSuperProject.py "./mySubmodule/map" "mySubmodule". Lastly, I pushed my changes with the --force option, that's another important part otherwise you get an error about merging which is not what you want.

import sys
import subprocess

if len(sys.argv) < 2:
    print("Usage: py fixSuperProject.py <object_map_file> <sbmodule_dir_with_respect_to_git_dir>")
    sys.exit(1)

object_map_file = sys.argv[1]
sbmodule_dir_with_respect_to_git_dir = sys.argv[2] # Submodule folder with respect to super project .git folder

id_map = {}

try:
    with open(object_map_file, "r") as f:
        for line in f:
            parts = line.strip().split(",")
            old_id, new_id = parts
            id_map[old_id] = new_id
except FileNotFoundError:
    print(f"File not found: {object_map_file}")

# Write the command to a file, as it can get very long
with open("callback_script.py", "w") as f:
    f.write(f"""
blob_map = {id_map}
for change in commit.file_changes:
    if change.mode is not None:
        if change.mode.decode("utf-8") == "160000" and change.filename.decode("utf-8") == "{sbmodule_dir_with_respect_to_git_dir}":
            if change.blob_id is not None:
                new_blob_id = blob_map.get(change.blob_id.decode("utf-8"))
                if new_blob_id is not None:
                    change.blob_id = new_blob_id.encode("utf-8")
""")

callback_script_path = 'callback_script.py'
command = ['git', 'filter-repo', '--commit-callback', f'{callback_script_path}', '--force']
subprocess.run(command)

from git-lfs.