#!/usr/bin/env ruby

require "bundler/setup"
require "linguist"
require "json"
require "yaml"

class GrammarList

  ROOT = File.expand_path "../../", __FILE__

  def initialize
    @submodules     = load_submodules()
    @sources        = load_sources()
    @language_names = load_languages()
  end

  # Load .gitmodules
  def load_submodules
    submodules = {}
    submodule_file = File.read("#{ROOT}/.gitmodules")
    pattern = /^\[submodule\s*"([^"]+)"\]$\n((?:^(?!\[).+(?:\n|$))+)/is
    submodule_file.scan(pattern) do |id, attr|
      submod = {}
      submod[:path]  = $1 if attr =~ /^\s*path\s*=\s*(.+)$/
      submod[:url]   = $1 if attr =~ /^\s*url\s*=\s*(.+)$/
      submod[:url].gsub!(/\.git$/, "")
      submod[:short] = shorten(submod[:url])
      submodules["#{id}"] = submod
    end
    submodules
  end

  # Grab the name of each language, sorted case-insensitively
  def load_languages
    Linguist::Language.all.map(&:name).sort do |a, b|
      a.downcase() <=> b.downcase()
    end
  end

  # Load grammars.yml
  def load_sources
    sources = {}
    grammars = YAML.load_file("#{ROOT}/grammars.yml")
    grammars.each do |path, scopes|
      scopes.each { |scope| sources[scope] = path }
    end
    sources
  end

  # Shorten a repository URL
  def shorten(url)
    if url =~ /^https?:\/\/(?:www\.)?github\.com\/([^\/]+\/[^\/]+)/i
      $1
    elsif url =~ /^https?:\/\/(?:www\.)?(bitbucket|gitlab)\.(?:com|org)\/([^\/]+\/[^\/]+)/i
      "#{$1.downcase()}:#{$2}"
    else
      url.replace(/^https?:\/\/(?:www\.)?/i, "")
    end
  end

  # Markdown: Generate grammar list
  def to_markdown
    markdown = ""
    @language_names.each do |item|
      lang  = Linguist::Language["#{item}"]
      scope = lang.tm_scope
      next if scope == "none"
      path  = @sources[scope] || scope
      case path
      when "https://bitbucket.org/Clams/sublimesystemverilog/get/default.tar.gz"
        short_url = "bitbucket:Clams/sublimesystemverilog"
        long_url  = "https://bitbucket.org/Clams/sublimesystemverilog"
      when "https://svn.edgewall.org/repos/genshi/contrib/textmate/Genshi.tmbundle/Syntaxes/Markup%20Template%20%28XML%29.tmLanguage"
        short_url = "genshi.edgewall.org/query"
        long_url  = "https://genshi.edgewall.org/query"
      when "vendor/grammars/oz-tmbundle/Syntaxes/Oz.tmLanguage"
        short_url = "eregon/oz-tmbundle"
        long_url  = "https://github.com/eregon/oz-tmbundle"
      else
        submodule = @submodules[@sources[scope].chomp("/")]
        next unless submodule
        short_url = submodule[:short]
        long_url  = submodule[:url]
      end
      markdown += "- **#{item}:** [#{short_url}](#{long_url})\n"
    end

    markdown
  end

  # Update the file displaying the reader-friendly list of grammar repos
  def update_readme
    readme = "#{ROOT}/vendor/README.md"
    preamble = File.read(readme).match(/\A.+?<!--.+?-->\n/ms)
    list = self.to_markdown
    File.write(readme, preamble.to_s + list)
  end
end

list = GrammarList.new
if ARGV.include? "--print"
  puts list.to_markdown
else
  list.update_readme
end