mte90 / gh-license

Scan your online repo, check if it is missing the license (or install the Git Hooks in your machine)

GitHosting License

This script scans every repo of a user (on GitHub/Bitbucket/GitLab or other providers available) for:

• Check if include a license file
• Download a license to the project folder (automatically execute a git commit)
• Adds a license badge in the readme (automatically execute a git commit)
• Include git hooks to check if your project has one.

Remember, without a license file your project is proprietary even if it is online!

Install

pip3 install gh-license

In case of a git clone:

pip3 install -e ./folder-of-the-repo

Info: This overwrites the init.templatedir global git config after creating a backup of the current value in ~/.gh-license/config.ini with the key templatedir in section main.
The installed git template contains a hook that will be installed every time you run git init. The hook reminds you to install a license if you ever forget, and disables itself after adding one.

Example

gh-license --scan Mte90

With this command you will get a report in a file called Mte90-github-license-report

gh-license --scan Mte90 --provider bitbucket

With this command you will get a report in a file called Mte90-bitbucket-license-report

gh-license --scan Mte90 --report my-report

With this command you will get a report in a file called my-report

gh-license --license-list

With this command will be showed the licenses avalaible

gh-license --license GPLv3

With this command, a GPLv3 license will be downloaded, a shields will be added in the readme and if Git is available a commit will be added and the changes will be pushed to the repo.

gh-license --license GPLv3 -- origin upstream

With this command the commit will be pushed on the upstream origin

Example of output

gh-license's Issues

Migrate to a multithread check

In this way is more fast...

Remember the last used license

Actually when you use the tool you have to remember the last license or what you use mostly with the codename used in the package.
It will be cool to remind what are the most used on creating when the license is not specified with a prompt to ask to the user what of the license most used want to use.

Search license also in other branches

As we can see

GH-License/ghlicense/repobase.py

Line 19 in 817ba6c

default_branch -- The branch to check (default "master").

we set the default branch.
But as today repository can have different as default, like for a fork and so on. We should find for all the repository behind services the "main" branch for a better check.

Revamp the report file

[Very shiny feature, not urgent at all, but may be useful someway]

Currently, the report file is "only" a text file listing all repos with their licensing characteristics.

What could be great is to generate a visual file, HTML or Markdown (to be discussed) with minimal formating that would give a shinier view of what's going on in your organization/user account. This option could be available by giving a --visual-report option with html|markdown flag

Raising IndexError on empty README file

touch README.md
python gh-license.py --license GPLv2

Traceback (most recent call last):
File "gh-license.py", line 69, in
downloadLicense("http://www.gnu.org/licenses/gpl-2.0.txt", args.args[0], '(https://img.shields.io/badge/License-GPL%20v2-blue.svg)](https://img.shields.io/badge/License-GPL%20v2-blue.svg)')
File "gh-license.py", line 26, in downloadLicense
if text[0][0] == '#':
IndexError: list index out of range

Add a `--no-license` or `--show` with options to filter repos on their licensing state

When launching the CLI on a specific user/organization, it'll be great to be able to pass a --no-license that would print and report only faulty reports.

Another way of implementing would be to add a --show option with 3 available flags all|licensed|unlicensed and filter followingly

Undetected license file for unusual repos

Some repos have license files named somewhat strangely : License-Agplv3.txt, etc. (cf. github.com/decidim/decidim as instance).

It would be great, instead of covering this by having a license_files list, to work followingly :

1. For GH, using the API, which itself relies on Licensee, a ruby gem that do the job pretty well : https://docs.github.com/en/rest/licenses#get-the-license-for-a-repository
2. For Bitbucket, same API endpoint does exist : https://developer.atlassian.com/server/bitbucket/rest/v804/api-group-project/#api-api-latest-projects-projectkey-repos-repositoryslug-license-get

Scan export as file

Will be useful to have the scan report as file with all the links of the repos.

Installation error?

OS: Manjaro

Collecting gh-license
  Downloading gh-license-0.2.4.tar.gz
Collecting pygithub>=1.5.0 (from gh-license)
  Downloading PyGithub-1.35.tar.gz (2.6MB)
    100% |████████████████████████████████| 2.6MB 387kB/s 
Collecting bitbucket-api>=0.5.0 (from gh-license)
  Downloading bitbucket-api-0.5.0.tar.gz
Collecting pyjwt (from pygithub>=1.5.0->gh-license)
  Downloading PyJWT-1.5.3-py2.py3-none-any.whl
Requirement already satisfied: requests in /usr/lib/python3.6/site-packages (from bitbucket-api>=0.5.0->gh-license)
Collecting sh (from bitbucket-api>=0.5.0->gh-license)
  Downloading sh-1.12.14-py2.py3-none-any.whl
Collecting requests-oauthlib (from bitbucket-api>=0.5.0->gh-license)
  Downloading requests_oauthlib-0.8.0-py2.py3-none-any.whl
Requirement already satisfied: chardet<3.1.0,>=3.0.2 in /usr/lib/python3.6/site-packages (from requests->bitbucket-api>=0.5.0->gh-license)
Requirement already satisfied: idna<2.7,>=2.5 in /usr/lib/python3.6/site-packages (from requests->bitbucket-api>=0.5.0->gh-license)
Requirement already satisfied: urllib3<1.23,>=1.21.1 in /usr/lib/python3.6/site-packages (from requests->bitbucket-api>=0.5.0->gh-license)
Collecting oauthlib>=0.6.2 (from requests-oauthlib->bitbucket-api>=0.5.0->gh-license)
  Downloading oauthlib-2.0.6.tar.gz (127kB)
    100% |████████████████████████████████| 133kB 721kB/s 
Installing collected packages: pyjwt, pygithub, sh, oauthlib, requests-oauthlib, bitbucket-api, gh-license
  Running setup.py install for pygithub ... done
  Running setup.py install for oauthlib ... done
  Running setup.py install for bitbucket-api ... done
  Running setup.py install for gh-license ... done
Successfully installed bitbucket-api-0.5.0 gh-license-0.2.4 oauthlib-2.0.6 pygithub-1.35 pyjwt-1.5.3 requests-oauthlib-0.8.0 sh-1.12.14

$ gh-license --scan italia
ERROR: Provider 'github' does not exist!

Add GitLab provider

In requirements.txt there are traces of a tentative to implement that support but is missing.

Add new licenses

Improve the script with new opensource licenses.

Code refactoring

The code on https://github.com/Mte90/GH-License/blob/master/ghlicense/cmd.py is too long and probably is better to split in various files.

If no license is passed there is an error

With gh-license --license the output is License True not found!

Improve the code organization

Divide the code in more file or more functions with comments can improve the code quality

Force a commit after the license download

As you can see in the readme I suggest to do a commit but will be useful a check if the folder have github and done automatically with a parameter for the command itself.

Git hooks

It will be helpful to have a git hooks installed globally when is executed git init to remember to download a license.

Add URL link for unlicensed repositories in analysis phase

As a developer, it would be awesome to have a https link to the repository in which there is no license.

Currently, link is here for already licensed projects and in the output file, but I think it could be helpful to ease the process of License adding if you provide a link also for unlicensed repos, since in few clicks I can solve this afterward.

Fix broken character

This line:

GH-License/ghlicense/cmd.py

Line 280 in 1290856

print(' ☐ Is a fork, check the original or create a PR!')

has a character which can't be displayed for me on macOS High Sierra, neither in terminal, nor the browser:

Is this happening only for me? What character should it be?

Detect actual git branch before pushing

Right now push automatically to master but there could be different branches, like main or trunk

GH-License/ghlicense/cmd.py

Line 116 in 817ba6c

os.system(f"git push {ARGS.origin} master")

Found by @fatualux

Fix lint errors

• https://www.codacy.com/app/mte90/GH-License/issues

Integrate with ScanCode-Toolkit

FYI, I maintain https://github.com/nexB/scancode-toolkit ... It could help for the license detection side. It is also written in Python

Add EUPL 1.2

https://it.wikipedia.org/wiki/European_Union_Public_Licence

--scan should default to Sources

Currently --scan processes forks, mirrors, etc. While that is useful, it is usually not what users would expect as the default.

If license not exist crash

Example command: gh-license --license gplv3

Add GitLab source

The wrapper available here seems to be functionable : https://github.com/python-gitlab/python-gitlab

Add stats in the report file

Will be useful get in the end of the report the stats about the repo with license and without and how many of them are fork with the same problem.

Can there ever be any provider in the "bad" list?

GH-License/ghlicense/repobase.py

Lines 44 to 49 in 70cb31f

	for name in providers:
	if providers[name]:
	good.append(name)
	else:
	bad.append(name)
	return good, bad

With the above logic,
under what circumstances will a provider be added to the "bad" list?...

Add unit tests

The package is working but unit tests can transform the tool in a swiss knife for every developer if we have unit tests.

Improve License suggestion help

Actually this is the output:
First parameter is missing: The license: GPLv2, GPLv3, LGPLv3, AGPLv3, FDLv1.3, Apachev2, CC-BY

Add a list with all the licenses with an explanation will be useful to choice.

Support for GitLab, BitBucket

The script born for GitHub but support other services is important.
The script check a profile for all the repos and search in all of them for a license file.
Is hardcoded the support for GitHub but probably with only a switch of the domain works also with other services.

Remove bitbucket-api support

That package create issue with the license because is not GPL friendly.