Issue parsing files with space in filename

[dperecab]

Whenever there is a space in the route of the file, I keep getting the following error:

File "/usr/local/lib/python3.9/site-packages/unidiff/patch.py", line 460, in __init__ self._parse(data, encoding=encoding, metadata_only=metadata_only) File "/usr/local/lib/python3.9/site-packages/unidiff/patch.py", line 530, in _parse raise UnidiffParseError('Target without source: %s' % line) unidiff.errors.UnidiffParseError: Target without source: +++ b/Audit Repositories/Interfaces/IMaintTransactionRepository.cs

This is an example:
diff --git a/Audit Repositories/Interfaces/IMaintTransactionRepository.cs b/Audit Repositories/Interfaces/IMaintTransactionRepository.cs new file mode 100644 index 0000000..f95add3

or another one:
diff --git a/repairer_allocation_etl/src/test/resources/NCR API GW.postman_collection.json b/repairer_allocation_etl/src/test/resources/NCR API GW.postman_collection.json new file mode 100644 index 0000000..fdd8387

While debugging, I've seen that current_file.target_file is set to "Repositories/Interfaces/IMaintTransactionRepository.cs", while target_file is being correctly set to "b/Audit Repositories/Interfaces/IMaintTransactionRepository.cs"

[krobelus]

On Mon, Feb 21, 2022 at 01:20:35AM -0800, dperecab wrote: Whenever there is a space in the route of the file, I keep getting the following error:

Probably we need to strip a trailing tab character, see commit 1a9eb3b9d5 (git-diff/git-apply: make diff output a bit friendlier to GNU patch (part 2), 2006-09-22) in git/git

[matiasb]

To confirm, was this related to a trailing tab character as mentioned? That would probably break the parsing regex, but I guess there shouldn't be a tab there? If you are still having issues, let me know, and if possible share some specific example and/or how you generate the diff. Thanks!

[dperecab]

It is the space character in the path of the file what is causing the issue.
There is an if in row 529 of patch.py with this condition: current_file.target_file == target_file
When there is a space character in the path of the file, this validation does not match, causing the error

[DevArcana]

Changing the regex to the following seems to fix this type of error:

RE_DIFF_GIT_HEADER = re.compile(
    r'^diff --git (?P<source>a/.+) (?P<target>b/.+)')

Can anyone shed some light on whether that is going to break anything?

[asottile-sentry]

here's a minimal patch which triggers this issue:

diff --git a/has spaces/t.sql b/has spaces/t.sql
new file mode 100644
index 0000000..8a9b485
--- /dev/null
+++ b/has spaces/t.sql
@@ -0,0 +1 @@
+select * FROM t;

0.7.2 seems to be the first version with this issue