Fix shebang handling for languages with non-# comments #434
Conversation
Codecov Report
@@ Coverage Diff @@
## master #434 +/- ##
==========================================
+ Coverage 98.94% 98.94% +<.01%
==========================================
Files 75 75
Lines 7646 7651 +5
==========================================
+ Hits 7565 7570 +5
Misses 81 81
Continue to review full report at Codecov.
|
|
Oh that's interesting! Thank you @jonasbb for pointing this out. |
| @@ -144,7 +144,7 @@ def header_to_metadata_and_cell(lines, header_prefix, ext=None): | |||
| encoding_re = re.compile(r'^[ \t\f]*{}.*?coding[:=][ \t]*([-_.a-zA-Z0-9]+)'.format(comment)) | |||
There was a problem hiding this comment.
Regarding the encoding, do you know it it should use the language line comment, or #?
There was a problem hiding this comment.
I only know the encoding line from Python 2 programs, so I am not sure if other languages use this feature as well.
tests/test_read_simple_rust.py
Outdated
|
|
||
| # remove version information | ||
| def remove_version_info(text): | ||
| return '\n'.join([line for line in text.splitlines() if 'version' not in line]) |
There was a problem hiding this comment.
For this, typically I use the fixture no_jupytext_version_number (you just need to add that argument to the test function, and remove the version information in the text)... If that's not clear I can also do it from here.
There was a problem hiding this comment.
Thanks, I didn't know this feature exists. I push a new version to use it.
FYI: I copied this test, which is where I found this remove_version_info function.
There was a problem hiding this comment.
Thanks! Oh interesting, later on I will update that other test.
The shebang always starts with the sequence "#!" independend of the comment symbol of the programming language. Add a test with Rust, which uses "//" as the comment symbol.
The shebang always starts with the sequence
#!independent of the comment symbol of the programming language. Add a test with Rust, which uses//as the comment symbol.The problem with the existing code is that it uses the comment symbol in the shebang line. The shebang can only start with
#!though. This leads to two problems.When parsing the shebang, when converting from text to notebook format, it uses a fixed offset of two characters to take the executable path (
line[2:]), which fails if the comment is not only one character, like the two characters of//.Similarly, when converting from notebook to text, it uses the comment symbol again and writes an invalid shebang.
The new test is mostly copied from the Python version with slight adaptions.