GitPython list all files affected by a certain commit

Question

I am using this for loop to loop through all commits:

repo = Repo("C:/Users/shiro/Desktop/lucene-solr/")
for commit in list(repo.iter_commits()):
 print commit.files_list # how to do that ?

How can I get a list with the files affected from this specific commit ?

Answer 1

Try it

for commit in list(repo.iter_commits()):
 commit.stats.files

• nicbou: Keep in mind that this method is very slow. It runs git diff --numstat, which calculates a lot more than just the file list.
• BCS: FWIW: that that does strange things with renames/moves.

Answer 2

from git import Repo
repo = Repo('/home/worldmind/test.git/')
prev = repo.commit('30c55d43d143189698bebb759143ed72e766aaa9')
curr = repo.commit('5f5eb0a3446628ef0872170bd989f4e2fa760277')
diff_index = prev.diff(curr)
for diff in diff_index:
 print(diff.change_type)
 print(f"{diff.a_path} -> {diff.b_path}")

• BCS: Things get more complicated when curr has more than one parent. Dealing with that will likely involve something like for prev in curr.parents: but exactly what to do from there is a function of what you are trying to do.
• BCS: There is also the question of what to do when curr is an initial commit and doesn't have a parent.
• BCS: After some more tinkering, it looks like commit.diff(None) is useful (e.g. with initial commits?) in some cases. But I haven't yet played around with it enough to know the details.

Answer 3

commit.stats.files works, but it's very slow. It will take several seconds to process a large commit.

This is much faster:

repo = Repo("C:/Users/shiro/Desktop/lucene-solr/")
for commit in list(repo.iter_commits()):
 print(self.repo.git.show(commit.hexsha, name_only=True).split('\n'))

• BCS: That works but also generates some "human readable" lines that will need to be filtered.

Answer 4

I solved this problem for SCM Workbench. The important file is:

https://github.com/barry-scott/scm-workbench/blob/master/Source/Git/wb_git_project.py

Look at cmdCommitLogForFile() and its helper __addCommitChangeInformation().

The trick is to diff the tree objects.

• Craig S. Anderson: StackOverflow highly discourages people from putting answers in linked pages. Linked pages often get moved or removed, then the answer is gone.
• muuvmuuv: Can you please update your answer. I'm searching for the same and don't want to search this big file when you already have found the solution there. I want to get all files that are committed currently not all commits since the start of the repo.