normalise-url

#!/usr/bin/env ruby

require "English"

# Whitelisted grammar providers
HOSTS = %w[
  github.com
  gitlab.com
  bitbucket.org
]

# Construct a well-formed URL from the output of `parse_url()`
def build_url(hash)
  path = "#{hash[:user]}/#{hash[:repo]}.git"
  if hash[:protocol] == "ssh"
    "git@#{hash[:host]}:#{path}"
  else
    "#{hash[:protocol]}://#{hash[:host]}/#{path}"
  end
end

# Break a repository URL into its separate components
def parse_url(input)
  hosts = %r[#{HOSTS.map { |x| Regexp.escape x }.join "|"}]i

  # HTTPS/HTTP link pointing to recognised hosts
  case input
  when %r[^
    (?:
      (?<protocol> https?|ssh|git)
      (?:\+(?:git|ssh))? # Allowed, but deprecated
    )? :* /*
    (?:[^.@]*@)?
    (?:www\.)?
    (?<host> #{hosts}) /+
    (?<user> [^:@/]+)  /+
    (?<repo> [^:@/]+?)
    (?:\.git)? (?=$|[/#])
  ]ix; {
    protocol: ($LAST_MATCH_INFO[:protocol] or "https").downcase,
    host:      $LAST_MATCH_INFO[:host].downcase,
    user:      $LAST_MATCH_INFO[:user],
    repo:      $LAST_MATCH_INFO[:repo]
  }

  # SSH
  when %r[^
    git (?:\+(?:ssh|https?))? @
    (?:www\.)?
    (?<host> #{hosts}) :/*
    (?<user> [^:@/]+) /+
    (?<repo> [^:@/]+?)
    (?:\.git)? /*$
  ]ix; {
    protocol: "ssh",
    host: $LAST_MATCH_INFO[:host].downcase,
    user: $LAST_MATCH_INFO[:user],
    repo: $LAST_MATCH_INFO[:repo]
  }

  # provider:user/repo
  when %r[^
    (?<host>
      gh | github |
      gl | gitlab |
      bb | bitbucket
    ) :/*
    (?<user> [^:@/]+) /+
    (?<repo> [^:@/]+?)
    (?:\.git)? /*$
  ]ix; {
    protocol: "https",
    host: (case $LAST_MATCH_INFO[:host].downcase
      when "gh", "github";    "github.com"
      when "gl", "gitlab";    "gitlab.com"
      when "bb", "bitbucket"; "bitbucket.org"
    end),
    user: $LAST_MATCH_INFO[:user],
    repo: $LAST_MATCH_INFO[:repo]
  }

  # user/repo - Common GitHub shorthand
  when %r[^
    /*
    (?<user> [^:@/]+) /+
    (?<repo> [^:@/]+?)
    (?:\.git)? /*$
  ]ix; {
    protocol: "https",
    host: "github.com",
    user: $LAST_MATCH_INFO[:user],
    repo: $LAST_MATCH_INFO[:repo]
  }

  # Not something we recognise
  else
    raise "Unsupported URL: #{input}"
  end
end


require "optparse"
$json     = false
$protocol = nil

OptionParser.new do |opts|
  opts.banner = <<~END
  #{$PROGRAM_NAME}: Resolve a repository URL from various formats

  Usage:
      #{$PROGRAM_NAME} [-p|--protocol name] ...urls
      #{$PROGRAM_NAME} [-j|--json] ...urls
      #{$PROGRAM_NAME} [-h|--hosts]

  Examples:
     $ #{$PROGRAM_NAME} Alhadis/language-etc BB:user/name
    => https://github.com/Alhadis/language-etc.git
       https://bitbucket.org/user/name.git

  Options:
  END
  opts.on("-h", "--hosts", "Print a list of whitelisted grammar hosts, then exit") do
    puts HOSTS.join $RS
    exit
  end
  opts.on("-j", "--json", "Output parsed URLs as an array of JSON objects") do
    $json = true
  end
  opts.on("-pNAME", "--protocol=NAME", "Force URLs to use a protocol, even if different. Ignored for JSON output.") do |name|
    $protocol = name.to_s.downcase
  end
end.parse!

if $json
  require "json"
  puts JSON.pretty_generate ARGV.map { |x| parse_url x }, { indent: "\t" }
else
  ARGV.each do |arg|
    url = parse_url arg
    url[:protocol] = $protocol unless $protocol.nil?
    puts build_url(url)
  end
end