Exclude folder using exclamation mark ('!') doesn't work

[wangxuepeng]

I'm using pattern
"""
!test1/
*.txt
"""

to scan a folder, from gitingore description
(http://git-scm.com/book/en/v2/Git-Basics-Recording-Changes-to-the-Repository)

ignore all files in the build/ directory

build/

this should exclude everything under test1 folder, but the run result shows the line dose not work, the result still show all contents under test 1 folder.

[An Example Folder Structure]
d:\dev\eclipse_workspace\test_scan\src\test1\a.txt
d:\dev\eclipse_workspace\test_scan\src\test1\b.txt
d:\dev\eclipse_workspace\test_scan\src\test1\c\c1.txt
d:\dev\eclipse_workspace\test_scan\src\test2\a.txt
d:\dev\eclipse_workspace\test_scan\src\test2\b.txt
d:\dev\eclipse_workspace\test_scan\src\test2\c\c1.txt

[Code]
SCAN_PATTERN = """
!test1/
*.txt
"""

spec = pathspec.PathSpec.from_lines(pathspec.GitIgnorePattern, SCAN_PATTERN.splitlines())
spec.match_tree('d:\dev\eclipse_workspace\test_scan')

[Run Result]
d:\dev\eclipse_workspace\test_scan\src\test1\a.txt
d:\dev\eclipse_workspace\test_scan\src\test1\b.txt
d:\dev\eclipse_workspace\test_scan\src\test1\c\c1.txt
d:\dev\eclipse_workspace\test_scan\src\test2\a.txt
d:\dev\eclipse_workspace\test_scan\src\test2\b.txt
d:\dev\eclipse_workspace\test_scan\src\test2\c\c1.txt

Can you please have a look? Thanks.

[cpburnz]

@029xue Try switching the order of your patterns to:

*.txt
!test1/

If you have !test1/ first, the files under that directory will be ignored first, but with *.txt following it will then re-include the .txt files within test1/.

A second possibility is that the Windows paths are not matching properly. Can you try just using:

!test1/

And let me know if that excludes the test1/ directory.

[wangxuepeng]

Hi,
Thanks for the quick response.

I tested
*.txt
!test1/

and it shows no difference with

!test1/
*.txt

and, if just leave
!test1/

The result is empty, no any files is selected.

[cpburnz]

The underlying issue is that Windows file paths are not being matched properly because Windows tends to use \ instead of / to separate directories. I will work on fix for this.

[wangxuepeng]

Thanks very much, waiting your updates :)

[cpburnz]

@029xue I made some changes, and with my testing it should work now. Please note that the order of your patterns should be:

*.txt
!test1/

To first include all *.txt files, and then to exclude the ones from the test1/ directory.

Also, spec.match_tree() and spec.match_files() now return an iterable instead of a set, so you'll have to convert it to a list or set manually. E.g.,

list(spec.match_tree('d:\dev\eclipse_workspace\test_scan'))

The changes are only reflected on Github right now. If my changes work for you, let me know and I'll update the version on PyPI.

[wangxuepeng]

Thanks, works like charm now.

So the patterns are working like filters from up to down? or just the excludes need be put in the last?

[cpburnz]

Yes, the filters work from up to down. The order is what really matters. This is so that you can include (or exclude) a large amount of files, but then further down you really want to include (or exclude) a specific directory or smaller set of files.

[wangxuepeng]

That make sense, thanks very much for the good library :)