Fix issue #19: emtpy new git file #86
Conversation
|
Hi @matiasb, Do you mind taking a look here? It's a kinda annoying issue for us. We use your library in a CI, and it's blind for empty new files currently. |
|
Hi @Felixoid, thanks, happy new year for you too! Thanks for the PR 👍 |
| @@ -473,8 +475,10 @@ def _parse(self, diff, encoding, metadata_only): | |||
| # check for a git file rename | |||
| is_diff_git_header = RE_DIFF_GIT_HEADER.match(line) | |||
| if is_diff_git_header: | |||
| if patch_info is None: | |||
| patch_info = PatchInfo() | |||
| if current_file is not None and not is_file_appended: | |||
There was a problem hiding this comment.
I think we could refactor this to avoid introducing the is_file_appended flag by always setting up a PatchedFile when a git header is found (and some minor changes to accommodate for that), will check.
There was a problem hiding this comment.
AFAI remember I was in a rush of fixing it. Most probably, there is a more elegant way to fix it. I am 100% sure you are more confident with code! =)
|
Merging, thanks 👍 |
This is needed to fix a bug with it not noticing new empty files being added. The fix first appears in v0.7.1[[1]] [[2]] It was reported in the pip project[[3]]. [1]: matiasb/python-unidiff#19 [2]: matiasb/python-unidiff#86 [3]: pypa/pip#11969
Hello. I've faced the issue today in ClickHouse/ClickHouse#32911 that the empty new files aren't fetched. It means, our CI may not work correctly for some other edge cases.
Here I've implemented the parsing for such diffs and tests for them:
diff --git a/boo.bin b/boo.bin new file mode 100644 index 0000000..ae000000The PR fixes #19