Support license header with shebang

[irrandon]

Hi,
for the license header step to work for linux script files correct handling of shebang lines is required.
Currently it is not handled correctly.

[nedtwigg]

Can you provide an example?

spotless {...}

# EXAMPLE SCRIPT:

# ACTUAL OUTPUT:

# DESIRED OUTPUT:

[vlsi]

@irrandon , you might want to give https://github.com/autostyle/autostyle a try

It supports Shell (e.g. #!/), Batch (e.g. @echo off), XML (e.g. <?..?>), Java (including package-info.java and module-info.java), and Gradle (Kotlin / Gradle scripts) automatically.

Samples: apache/jmeter#549, apache/calcite#1682

[irrandon]

We have a temporary solution with a python script but a support via spotless would be nice :)

Here an example:

spotless{
  format('python') {
    licenseHeaderFile 'com/company/gradle/hash.license', '(import |""")'
    target '**/*.py'
  }
}

(If you add the shebang to the delimiter regex, it push the shebang under the license header, what would be invalid.)

hash.license:

#==========================================================================
#  Copyright (c) Awesome Corp., 1980-2020.
#
#  All rights reserved. All hardware and software names used are registered
#  trade names and/or registered trademarks of the respective manufacturers.
#
#  The user of this computer program acknowledges that the above copyright
#  notice, which constitutes the Universal Copyright Convention, will be
#  attached at the position in the function of the computer program which the
#  author has deemed to sufficiently express the reservation of copyright.
#  It is prohibited for customers, users and/or third parties to remove,
#  modify or move this copyright notice.
#==========================================================================

EXAMPLE SCRIPT:

#!/usr/bin/env python3
import sys
print("Hello world")

ACTUAL OUTPUT:

#==========================================================================
#  Copyright (c) Awesome Corp., 1980-2020.
#
#  All rights reserved. All hardware and software names used are registered
#  trade names and/or registered trademarks of the respective manufacturers.
#
#  The user of this computer program acknowledges that the above copyright
#  notice, which constitutes the Universal Copyright Convention, will be
#  attached at the position in the function of the computer program which the
#  author has deemed to sufficiently express the reservation of copyright.
#  It is prohibited for customers, users and/or third parties to remove,
#  modify or move this copyright notice.
#==========================================================================
import sys
print("Hello world")

DESIRED OUTPUT:

#!/usr/bin/env python3
#==========================================================================
#  Copyright (c) Awesome Corp., 1980-2020.
#
#  All rights reserved. All hardware and software names used are registered
#  trade names and/or registered trademarks of the respective manufacturers.
#
#  The user of this computer program acknowledges that the above copyright
#  notice, which constitutes the Universal Copyright Convention, will be
#  attached at the position in the function of the computer program which the
#  author has deemed to sufficiently express the reservation of copyright.
#  It is prohibited for customers, users and/or third parties to remove,
#  modify or move this copyright notice.
#==========================================================================
import sys
print("Hello world")

[nedtwigg]

An easy workaround is to put the #!/usr/bin ... inside the license that you give to Spotless. Any problems with that?

[irrandon]

The problem is that this only works, when you have the same shebang for all python files.
But we also have #!/usr/bin/env python2. The same problem is there with linux script files.(#!/usr/bin/env sh | #!/usr/bin/env bash)

[nedtwigg]

(comment deletions just to remove off-topic discussion and condense on the feature request)

Aha! This is a good feature request. It's tricky - the license step has so far had a very simple model: find the "top" of the file with regex Y, and then add header X. Solving this issue will require complicating the model somewhat.

A simple way to do this would be to allow something like $FIRST_LINE within the replacement text. So your hash.license could be

$FIRST_LINE
#==========================================================================
#  Copyright (c) Awesome Corp., 1980-2020.
#
#  All rights reserved. All hardware and software names used are registered
#  trade names and/or registered trademarks of the respective manufacturers.

The other option would be to make the license step generally more powerful and context-sensitive, as @vlsi has done in his fork, linked above. I would definitely merge a PR to add support for $FIRST_LINE, since it is so simple. I'm also open to a more powerful context-sensitive option, but backwards compatibility and long-term maintenance burden would be an important challenge to keep in mind.

[abelk2]

I think this is supported now with #1441.

[nedtwigg]

Thanks @abelk2! Relevant example:

format "licenseHeaderSh", {
  target "*.sh"
  licenseHeaderFile(file("$rootDir/config/license/sh.txt"), "^(?!##).+")
    .skipLinesMatching("^#!.+?\$")
  // license header MUST start with ##
}