I took some time to create an activity diagram in descriptive PlantUML syntax which shows how the current header fixing mechanism works. I think this could simplify future discussions about the header fixing mechanism and its technical requirements.

My original idea was to add this to PR #40 which is merged already, therefore I create this new PR. Feel free to refuse this PR, if the diagram or the SVG file is not desired (in this repository) or if you think this does not belong here.

Solution: copy, using mvn dependency:copy-dependencies, all dependencies and transitive dependencies, and then resolve license from the jar files.

Known disadvantages:

• depends on maven
• replication of the jar files needs to read and write the disk

Example config snippet:

header:
  license:
    #spdx-id: Apache-2.0 # the spdx id of the license, it's convenient when your license is standard SPDX license.
    copyright-owner: Example Ltd # the copyright owner to replace the [owner] in the `spdx-id` template.

    content: |
      Licensed Materials - Property of Example Ltd.
      (C) Copyright Example Ltd 2021-2022. All Rights Reserved.

    pattern: |
      Licensed Materials - Property of Example Ltd\.
      \(C\) Copyright Example Ltd [0-9-]+\. All Rights Reserved\.

  paths:
    - '**/*.go'
    - '**/*.sh'
    - '**/Dockerfile'

Although we try to use ignore pattern present by a .gitignore file, it's possible that users have a global git ignore settings or manually ignore a file.

We can use git command to list files when runs under a git worksapce, and if there's no git workspace, we don't have to see .gitignore. Normally, it's via directly download and all ignored files already excluded. If users want to respect to the ignore pattern, they can simply git init.

cc @kezhenxu94 @wu-sheng @fgksgf @spacewander

Reference:

• https://github.com/apache/skywalking/issues/7543
• https://github.com/apache/skywalking/issues/7538

This allows for multiple license configurations to be contained within the same .licenserc.yaml file. The intended use is for a project that has multiple packages or modules with differing licenses.

To do this we introduce a ConfigV2 config format since it is not valid YAML for a key to be either a dictionary or a sequence.

We try to parse as V2 first, and fall back to V1.

Recently, I use this project to fix myself project license header, It is a GoLand project.

I find that the header comment style config at the languages.yaml and styles.yaml, and it doesn't support config.

The Go file license header comment style is //, like:

// Licensed to the Apache Software Foundation (ASF) under one
// ... ...

But I want the license header comment style is *, like:

/*
 * Licensed to the Apache Software Foundation (ASF) under one
 * ... ...
 */

So, I fix code and submit this PR.

I add the config languages at the .licenserc.yaml, and the license header comment style use this file config preferentially, if this file doesn't have this config, it use the default config at the languages.yaml.

This patch enhances the NPM dependency resolver to resolve all the dependent packages. First, it runs the npm command npm ls --all --parseable to list all the packages' absolute paths. Then, each package's name is inferred from its relative path from the node_modules dir. Finally, walk through each package's root path to resolve licenses from the package.json file or the license file.

Why do we need this fix?

As we are using license-eyes to check license header in our open-sourced projects, we notice that it can help detect unexpected changes in the license header, except the copyright line. In my opinion, it's also necessary to protect the copyright line from unintentional changes.

How was this issue introduced?

https://github.com/apache/skywalking-eyes/pull/46 introduced this line processor to ignore copyright line, mostly for the purpose of dependency resolution. However, in https://github.com/apache/skywalking-eyes/pull/107 starts using Google's license check for this purpose, and this line processor only works to ignore copyright line when checking / fixing license header.

Reviewer

@kezhenxu94

1. Parse the project package.json file to gather the required packages
2. Run command 'npm install' to install or update the required node packages (can skip)
3. Walk through each package's root directory to resolve licenses
   • STEP 1: Try to find and parse the package.json file to capture the license field
   • STEP 2: Try to find the license file to identify the license

This addition allows users of the pre-commit Git hook management tool to use Skywalking Eyes' license-eyes header fix command to automate fixing their license headers at regular points in the development lifecycle.

This is a missing but important feature that helps to organize the distribution package especially in ASF projects, it resolves and saves all LICENSE files of the dependencies.