path: root/scripts/build

#!/usr/bin/env ruby

require 'open-uri'
require 'open3'
require 'yaml'
require 'fileutils'
require 'set'
require 'json'
require 'tsort'

Dir.chdir(File.dirname(__dir__))

BASE_URL = 'https://raw.githubusercontent.com/github/linguist/master'

def read_section(name)
  "\" Please do not edit this file directly, instead modify polyglot.vim or scripts/build\n\n" +
  File.read('polyglot.vim').match(/""" #{Regexp.escape(name)}\n*(.*)\n(?=\n""")/m)[1]
end

def camelize(str)
  str.split(/[-_\.]/).map { |a| a.capitalize }.join("")
end

def except(hash, *keys)
  h = hash.dup
  keys.each { |k| h.delete(k) }
  h
end

def load_data()
  packages = Hash[YAML.load_stream(File.read('packages.yaml'))
    .group_by { |a| a.fetch("name") }
    .map { |a, b| [a, b.first] }]

  deps = Hash.new { |h, k| h[k] = [] }

  each_node = lambda {|&b| packages.keys.each(&b) }
  each_child = lambda {|n, &b| [packages[n]["after"] || []].flatten.each(&b) }

  languages = load_languages

  # Reason can have ocaml as interpreter but let's not depend on it...
  languages["Reason"]["interpreters"] -= ["ocaml"]

  packages = TSort.tsort(each_node, each_child).map { |a| packages[a] }

  for package in packages
    for filetype in package["filetypes"]
      if filetype["linguist"]
        if filetype["extensions"]
          raise "#{package["name"]} #{filetype["name"]}: extensions can't be set when linguist is defined"
        end

        if filetype["filenames"]
          raise "#{package["name"]} #{filetype["name"]}: filenames can't be set when linguist is defined"
        end

        linguist = languages.fetch(filetype["linguist"])

        filetype["extensions"] = (linguist["extensions"] || []).map { |e| e[1..-1] } |
          filetype.fetch("extra_extensions", [])

        extra_ignored_extensions = filetype.fetch("ignored_extensions", []) - filetype["extensions"]
        if extra_ignored_extensions.size > 0
          raise StandardError.new("[#{filetype["name"]}]: Unnecessary ignored extensions: #{extra_ignored_extensions.join(", ")}") 
        end

        filetype["extensions"] -= filetype.fetch("ignored_extensions", [])

        filetype["filenames"] = ((
          (linguist["filenames"] || []) |
          filetype.fetch("extra_filenames", [])
        ) - filetype.fetch("ignored_filenames", [])).uniq
        
        filetype["interpreters"] = ((
          (linguist["interpreters"] || []) |
          filetype.fetch("extra_interpreters", [])
        ) - filetype.fetch("ignored_interpreters", []).uniq)
      else
        filetype["extensions"] ||= []
        filetype["filenames"] ||= []
        filetype["interpreters"] ||= []

        filetype.keys.each do |key|
          if key.start_with?("extra_")
            raise "[#{filetype["name"]}]: #{key} is not allowed if linguist is not used"
          end
        end
      end
    end
  end

  heuristics = YAML.load_stream(File.read('heuristics.yaml'))

  [packages, transform_patterns(heuristics)]
end

def patterns_to_vim_patterns(patterns)
  stdin, stdout, stderr = Open3.popen3('vim', '-V', '--clean', '/dev/stdin', '-es', '-c', "echo expand('%:p:h') | source #{__dir__}/eregex.vim", '-c', "for line in range(0, line('$')) | call setline(line, ExtendedRegex2VimRegex(getline(line))) | endfor", '-c', ':wq! /dev/stdout', chdir: __dir__)
  stdin.write(patterns.join("\n"))
  stdin.close
  stdout.readlines.map(&:chomp).map do |r|
    r.gsub('\b', '\(\<\|\>\)')
  end
end

def each_hash(data, &block)
  if data.is_a?(Hash)
    yield data
    data.each do |key, val|
      each_hash(val, &block)
    end
  elsif data.is_a?(Array)
    data.map { |d| each_hash(d, &block) }
  end
end


def transform_patterns(heuristics)
  patterns = []
  each_hash(heuristics) do |h|
    if h.has_key?("pattern")
      patterns << h["pattern"]
    end
  end
  patterns_mapping = Hash[patterns.zip(patterns_to_vim_patterns(patterns))]
  each_hash(heuristics) do |h|
    if h.has_key?("pattern")
      h["pattern"] = patterns_mapping.fetch(h["pattern"])
    end
  end
  heuristics
end

def load_languages
  file = 'tmp/languages.yml'

  unless File.exist?(file)
    url = "#{BASE_URL}/lib/linguist/languages.yml"
    data = URI.open(url) { |io| io.read }
    File.write(file, data)
  end

  YAML.load(File.read(file))
end

def parse_remote(remote)
  match = remote.match(/(?<repo>[^@:]+)(?:@(?<branch>[^:]+))?(?::(?<path>.*))?/)
  dir = "tmp/" + match[:repo] + (match[:branch] ? "-#{match[:branch]}" : "")
  [match[:repo], match[:branch] || "master", match[:path], dir]
end

def copy_file(package, src, dest)
  FileUtils.mkdir_p(File.dirname(dest))
  name = package.fetch("name")

  if dest.end_with?(".vim")
    header = '" Polyglot metafile'
    if File.exist?(dest)
      meta_dest = dest
      new_dest = dest
      i = 0
      while File.exist?(new_dest)
        i += 1
        new_dest = "#{dest.gsub(/\.vim$/, '')}-#{i}.vim"
      end

      if File.read(dest).include?(header)
        dest = new_dest
      else
        FileUtils.mv(dest, new_dest)
        File.write(meta_dest, "#{header}\n")
        open(meta_dest, "a+") do |output|
          output << "source <sfile>:h/#{File.basename(new_dest)}\n"
        end
        dest = "#{dest.gsub(/\.vim$/, '')}-#{i+1}.vim"
      end
      open(meta_dest, "a+") do |output|
        output << "source <sfile>:h/#{File.basename(dest)}\n"
      end
    end
  end

  open(src, "r") do |input|
    open(dest, "w") do |output|
      if name == "jsx"
        output << "if !exists('g:polyglot_disabled') || (index(g:polyglot_disabled, 'javascript') == -1 && index(g:polyglot_disabled, 'jsx') == -1)\n\n"
      else
        output << "if !exists('g:polyglot_disabled') || index(g:polyglot_disabled, '#{name}') == -1\n\n"
      end
      contents = File.read(input)
      contents.gsub!(' ..= ', ' .= ')
      output << contents
      output << "\nendif\n"
    end
  end
end

def download(packages)
  packages.map { |p| p["remote"] or raise "No remote for: " + p["name"] }.uniq.each_slice(20) do |remotes|
    remotes.map do |remote|
      Thread.new do
        repo, branch, path, dir = parse_remote(remote)
        unless File.exist?(dir)
          FileUtils.mkdir_p(dir)
          url = "https://codeload.github.com/#{repo}/tar.gz/#{branch}"
          `curl --silent -fL #{url} | tar -zx -C "#{dir}" --strip 1`
        end
        progress
      end
    end.map(&:join)
  end
end

$i = 0
LYRICS = "Never gonna give you up.  Never gonna let you down.  " +
         "Never gonna run around and desert you.  " +
         "Never gonna make you cry.  Never gonna say goodbye.  " +
         "Never gonna tell a lie and hurt you."

$mutex = Mutex.new
def progress
  $mutex.synchronize do
    $stdout.write(LYRICS[$i] || ".")
    $i += 1
  end
end

def indent(str, amount)
  str.gsub(/^(?!$)/, " " * amount).gsub(/\s+$/, "").gsub(/^ +\n/, "")
end

def pattern_to_condition(rule)
  if rule.has_key?("or")
    return rule["or"].map { |p| pattern_to_condition(p) }.join(" || ")
  end

  if rule.has_key?("or")
    return rule["and"].map { |p| pattern_to_condition(p) }.join(" && ")
  end

  operator = (rule["negative"] ? "!" : "=") + "~" + (rule["ignore_case"] ? "?" : "#")

  return "line #{operator} '#{rule["pattern"]}'"
end

def rules_to_code(rules)
  output = ""

  vars = []
  each_hash(rules) do |h|
    if h.has_key?("set")
      vars << h["set"]
    end
  end


  if vars.size > 0
    output << vars.uniq.sort.map do |var|
      "let #{var} = 0"
    end.join("\n") + "\n"
  end

  output << rule_to_code(rules)

  output
end

def rule_to_code(rule)
  if rule.has_key?("lines")
    if rule["lines"] == 1
      return <<~EOS
        let line = getline(nextnonblank(1))

        #{indent(rule_to_code(except(rule, "lines")), 0)}
      EOS
    else
      return <<~EOS
        for lnum in range(1, min([line("$"), #{rule["lines"]}]))
          let line = getline(lnum)

        #{indent(rule_to_code(except(rule, "lines")), 2)}
        endfor
      EOS
    end
  end

  if rule.has_key?("pattern") || rule.has_key?("or") || rule.has_key?("and")
    return <<~EOS
      if #{pattern_to_condition(rule)}
      #{indent(rule_to_code(except(rule, "pattern", "or", "and", "ignore_case", "negative")), 2)}
      endif
    EOS
  end

  if rule.has_key?("rules")
    return rule["rules"].map do |r|
      indent(rule_to_code({ "extensions" => rule["extensions"] }.merge(r)), 0)
    end.join("\n")
  end

  if rule.has_key?("if_set")
    return <<~EOS
      if #{rule["negative"] ? "!" : ""}#{rule["if_set"]}
      #{indent(rule_to_code(except(rule, "if_set", "negative")), 2)}
      endif
    EOS
  end

  if rule.has_key?("if_exists")
    return <<~EOS
      if #{rule["negative"] ? "!" : ""}exists("#{rule["if_exists"]}")
      #{indent(rule_to_code(except(rule, "if_exists", "negative")), 2)}
      endif
    EOS
  end

  if rule.has_key?("set")
    return <<~EOS
      let #{rule["set"]} = 1
      #{indent(rule_to_code(except(rule, "set")), 0)}
    EOS
  end

  if rule.has_key?("filetype")
    if rule.has_key?("fallback")
      return <<~EOS
        set ft=#{rule["filetype"]} | au! BufWritePost <buffer> ++once call polyglot#Detect#{camelize(rule["extensions"].first)}Filetype()
        return
      EOS
    end

    return <<~EOS
      set ft=#{rule["filetype"]} | return
      #{indent(rule_to_code(except(rule, "filetype")), 0)}
    EOS
  end

  if rule.has_key?("override")
    variable = "g:filetype_#{rule["extensions"].first}"
    return <<~EOS
      if exists("#{variable}")
        let &ft = #{variable} | return
      endif
    EOS
  end

  if (rule.keys - ["extensions"]).size > 0
    raise "Unknown rule: #{JSON.generate(rule)}"
  end
  
  return ""
end

def extract(packages)
  all_dirs = %w(syntax indent doc compiler autoload ftplugin ctags extras after)

  default_dirs = %w(
    syntax indent doc compiler autoload ftplugin ctags extras
    after/syntax after/indent after/ftplugin
  )

  FileUtils.rm_rf(all_dirs)

  output = []
  packages.map do |package|
    repo, branch, path, dir = parse_remote(package["remote"])
    dirs = package.fetch("dirs", default_dirs)
    ignored_dirs = package.fetch("ignored_dirs", [])
    if ignored_dirs.size > 0
      dirs = dirs.reject { |d| ignored_dirs.any? { |id| d.start_with?(id) } }
    end
    dirs |= package.fetch("extra_dirs", [])
    for subdir in dirs
      subtree = "#{dir}/#{path ? path + "/" : ""}"
      subpath = "#{subtree}#{subdir}"
      if FileTest.directory?(subpath)
        if repo == "vim/vim" && (["glob", "globs"] & package.keys).size == 0
          raise "Package from vim/vim should define glob or globs: #{package["name"]}"
        end
        globs = [package.fetch("glob", package.fetch('globs', '**/*.{vim,ctags,vital,txt}'))].flatten
        globs.each do |glob|
          Dir.glob("#{subdir}/#{glob}", base: subtree).each do |p|
            next unless File.file?("#{subtree}/#{p}")
            if p.include?("samba")
              raise package["name"]
            end
            copy_file(package, "#{subtree}/#{p}", p)
          end
        end
      elsif File.exist?(subpath)
        copy_file(package, subpath, subdir)
      end
    end

    if branch != "master" || path
      if path
        output << "- [#{package["name"]}](https://github.com/#{repo}/tree/#{branch}/#{path})"
      else
        output << "- [#{package["name"]}](https://github.com/#{repo}/tree/#{branch})"
      end
    else
      output << "- [#{package["name"]}](https://github.com/#{repo})"
    end
    progress
  end

  readme = File.read('README.md')

  readme.gsub!(
    %r{(?<=<!--Package Count-->).*?(?=<!--/Package Count-->)},
    output.size.to_s
  )

  readme.gsub!(
    %r{(?<=<!--Language Packs-->).*?(?=<!--/Language Packs-->)}m,
    "\n" + output.sort.join("\n") + "\n"
  )

  File.write('README.md', readme)
end

def generate_ftdetect(packages, heuristics)
  output = "\n"

  extensions = Hash.new { |h, k| h[k] = [] }

  for package in packages
    for filetype in package["filetypes"]
      for ext in filetype["extensions"]
        extensions[ext] << filetype["name"]
      end
    end
  end

  ambiguous_extensions = extensions
    .select { |a, b| b.uniq.size > 1 }.keys.sort

  all_filetypes = packages.flat_map { |f| f["filetypes"] || [] }
  filetype_names = Set.new(all_filetypes.map { |f| f["name"] })

  native_filetypes = detect_filetypes('tmp/vim/vim/runtime/filetype.vim')
  expected_filetypes = detect_filetypes('tmp/**/ftdetect/*.vim') + native_filetypes

  expected_filetypes = expected_filetypes.select { |e| filetype_names.include?(e["name"]) }

  native_extensions = Set.new(native_filetypes.flat_map { |f| f["extensions"] || [] })
  native_filenames = Set.new(native_filetypes.flat_map { |f| f["filenames"] || [] })

  for package in packages
    name = package.fetch("name")

    to_disable = []

    for filetype in package["filetypes"]
      for extension in filetype["extensions"]
        if native_extensions.include?(extension)
          to_disable << "*." + extension
        end
      end

      for filename in filetype["filenames"]
        if native_filenames.include?(filename)
          to_disable << filename
        end
      end
    end

    if to_disable.size > 0
      output << "if !has_key(s:disabled_packages, '#{name}')\n"
      output << "  au! BufRead,BufNewFile #{to_disable.join(",")}\n"
      output << "endif\n\n"
    end
  end

  for package in packages
    filetypes = package["filetypes"] or raise "Unknown filetype for: #{package["name"]}"

    package_heuristics = []

    autocommands = ""
    for filetype in filetypes
      name = filetype.fetch("name")
      syntax = filetype["syntax"] ? " | set syntax=#{filetype["syntax"]}" : ""

      set_command = "set ft=#{name}"

      if filetype["syntax"]
        set_command = "set ft=#{name} syntax=#{filetype["syntax"]}"
      end

      if filetype["custom_set"]
        set_command = filetype["custom_set"]
      end

      extensions = filetype["extensions"]
      filenames = filetype["filenames"]

      for extension in extensions.sort
        outer_filetype = filetype["outer_filetype"]
        if outer_filetype
          autocommands << "  au BufNewFile *.*.#{extension} execute \"do BufNewFile filetypedetect \" . expand(\"<afile>:r\") | #{outer_filetype}\n"
          autocommands << "  au BufReadPre *.*.#{extension} execute \"do BufRead filetypedetect \" . expand(\"<afile>:r\") | #{outer_filetype}\n"
        end

        heuristic = heuristics.find { |h| h["extensions"].include?(extension) }
        if heuristic
          package_heuristics << heuristic
        else
          # if ambiguous_extensions.include?(extension)
          #   puts "Ambiguous extension without heuristic: #{extension} => #{filetype["name"]}"
          # end
          #
          autocommands << "  au BufNewFile,BufRead *.#{extension} #{set_command}\n"
        end
      end

      for filename in filenames.sort
        if filename[0] == "."
          filename = "{.,}" + filename[1..-1]
        end
        if filename[-1] == "*"
          autocommands << "  au BufNewFile,BufRead #{filename} call s:StarSetf('#{name}')\n"
        else
          autocommands << "  au BufNewFile,BufRead #{filename} #{set_command}\n"
        end
      end
    end

    for heuristic in package_heuristics.uniq
      extensions = heuristic["extensions"].map { |e| "*.#{e}" }
      autocommands << "  au! BufNewFile,BufRead #{extensions.join(",")} call polyglot#Detect#{camelize(heuristic["extensions"].first)}Filetype()\n"
    end

    if autocommands != ""
      output << "if !has_key(s:disabled_packages, '#{package["name"]}')\n"
      output << autocommands
      output << "endif\n\n"
    end
  end
 
  defined_extensions = all_filetypes.flat_map { |f| expand_all(f["extensions"] || []).map { |e| [f["name"], e] } }
  expected_extensions = expected_filetypes.flat_map { |f| expand_all(f["extensions"] || []).map { |e| [f["name"], e] } }
  ignored_extensions = all_filetypes.flat_map { |f| expand_all(f.fetch("ignored_extensions", [])).map { |e| [f["name"], e] } }

  defined_filenames = all_filetypes.flat_map { |f| expand_all(f["filenames"] || []).map { |e| [f["name"], e] } }
  expected_filenames = expected_filetypes.flat_map { |f| expand_all(f["filenames"] || []).map { |e| [f["name"], e] } }
  ignored_filenames = all_filetypes.flat_map { |f| expand_all(f.fetch("ignored_filenames", [])).map { |e| [f["name"], e] } }

  ignored_warnings = all_filetypes.flat_map { |f| expand_all(f.fetch("ignored_warnings", [])).map { |e| [f["name"], e] } + [f, "*"] }

  for name, e in expected_extensions - defined_extensions - ignored_extensions - ignored_warnings
    puts "Missing extension for #{name}: #{e}"
  end

  for name, e in expected_filenames - defined_filenames - ignored_filenames - ignored_warnings
    puts "Missing filename for #{name}: #{e}"
  end

  ftdetect = read_section('ftdetect/polyglot.vim')

  File.write(
    'ftdetect/polyglot.vim',
     ftdetect.gsub('" scripts/build inserts here filetype detection autocommands') { output }
  )

  output = <<~EOS
    let s:interpreters = {
    EOS

    for filetype in packages.flat_map { |p| p.fetch("filetypes", []) }.sort_by { |a| a["name"] }
      for interpreter in filetype["interpreters"]
        output << "  \\ '#{interpreter}': '#{filetype["name"]}',\n"
      end
    end

  output << <<~EOS
      \\ }

  EOS

  for heuristic in heuristics
    output << <<~EOS
      func! polyglot#Detect#{camelize(heuristic["extensions"].first)}Filetype()
      #{indent(rules_to_code(heuristic), 2)}
      endfunc

    EOS
  end

  autoload_script = read_section('autoload/polyglot.vim')
  autoload_script["\" scripts/build generates heuristics functions here\n"] = output

  File.write('autoload/polyglot.vim', autoload_script)
end

def generate_tests(packages)
  output = <<~EOS
    function! TestFiletype(filetype)
      call Log('Loading ' . a:filetype . ' filetype...')

      enew
      exec 'set ft=' . a:filetype
    endfunction

  EOS

  for package in packages
    for filetype in package.fetch("filetypes", [])
      output << "call TestFiletype('#{filetype["name"]}')\n"
    end
  end

  File.write('tests/filetypes.vim', output)


end

def brace_expansion(s)
  r=1                                       # Dummy value to forward-declare the parse function `r`
  t=->x{                                    # Function to parse a bracket block
    x=x[0].gsub(/^{(.*)}$/){$1}             # Remove outer brackets if both are present
                                            # x[0] is required because of quirks in the `scan` function
    x=x.scan(/(({(\g<1>|,)*}|[^,{}]|(?<=,|^)(?=,|$))+)/)
                                            # Regex black magic: collect elements of outer bracket
    x.map{|i|i=i[0];i[?{]?r[i]:i}.flatten   # For each element with brackets, run parse function
  }
  r=->x{                                    # Function to parse bracket expansions a{b,c}{d,e}
    i=x.scan(/({(\g<1>)*}|[^{} ]+)/)        # Regex black magic: scan for adjacent sets of brackets
    i=i.map(&t)                             # Map all elements against the bracket parser function `t`
    i.shift.product(*i).map &:join          # Combine the adjacent sets with cartesian product and join them together
  }
  s.split.map(&r).flatten
end

def square_expansion(s)
  return [s] unless s.include?('[')
  s.scan(/(\[[^\]]+\]|[^\[]+)/).map { |x| x[0] }
    .map { |x| x[0] == "[" ? x[1..-2].split("") : [x] }
    .reduce(&:product).map(&:flatten).map(&:join)
end

def comma_expanson(s)
  s.scan(/{[^{]+}|[^{]+/).map { |a| a[0] == "{" ? a : a.split(",", -1) }.reduce([]) do |a, b|
    a.size > 0 ?
      (b.is_a?(String) ?
         a[0..-2] + [a[-1] + b] :
         a[0..-2] + [a[-1] + b[0]] + b[1..-1]) :
      [b].flatten
  end
end

def expand_all(pattern)
  if pattern.is_a?(Array)
    return pattern.flat_map { |p| expand_all(p) }
  end

  comma_expanson(pattern).flat_map do |e|
    brace_expansion(e).flat_map do |e2|
      square_expansion(e2)
    end
  end
end

def detect_filetypes(glob)
  filetypes = Dir[glob].flat_map do |file|
    contents = File.read(file).gsub(/^\s*au(tocmd)?!?\s*$/, '')
    results = contents.scan(/^\s*(?:au!|au|au[^g][^ ]*) +(?:\S+)\s+(\S+)[\s\\]+([^\n]+)/)
    results = results.map do |a, b|
      [
        a,
        b.gsub(/call (?:s:setf|s:StarSetf)\('([^']+)'\)/i, 'setf \1')
          .gsub(/set(?:local)?\s+(?:ft|filetype)=(\S+)/, 'setf \1')
        .gsub(/setf\S*/, 'setf')
        .gsub(/.*setf\s+(\S+).*/, 'setf \1')
      ]
    end.select { |a, b| b.match(/setf \S+/) }.map { |a, b| [a, b.split(" ")[1]] }
    results
  end

  filetypes.flat_map do |ext, filetype|
    expand_all(ext).map { |e| [filetype, e] }
  end.group_by { |a, b| a }.map { |a, b| [a, b.map { |c, d| d }] }.map { |a, b|
    {
      "name" => a,
      "extensions" => b.select { |x| x.match(/^\*\.[^\/]+$/) }.map { |a| a.strip[2..-1] },
      "filenames" => expand_all(b.select { |x| !x.match(/^\*\.[^\/]+$/) })
    }
  }
end

def generate_plugins(packages)
  output = "let s:globs = {\n"

  patterns = Hash.new { |h, k| h[k] = [] }

  for package in packages
    for filetype in package["filetypes"]
      extensions = (filetype["extensions"] || []).map { |e| "*.#{e}" }
      files = (filetype["filenames"] || []).reject { |e| e.match(/\*\*|\//) || e[0] == "." }
      patterns[filetype["name"]].concat(extensions)
      patterns[filetype["name"]].concat(files)
    end
  end

  for filetype in patterns.keys.sort
    output << "  \\ '#{filetype}': '#{patterns[filetype].uniq.join(",")}',\n"
  end
  output << "  \\}\n\n"


  output << <<~EOS
    func! sleuth#GlobForFiletype(type)
      return get(s:globs, a:type, '')
    endfunc
  EOS


  File.write('autoload/sleuth.vim', output)
end


if __FILE__ == $0
  if !ENV["DEV"]
    FileUtils.rm_rf("tmp")
  end

  Dir.mkdir('tmp') unless File.exists?('tmp')

  packages, heuristics = load_data()
  download(packages)
  extract(packages)
  generate_ftdetect(packages, heuristics)
  generate_plugins(packages)
  generate_tests(packages)
  puts(" Bye! Have a wonderful time!")

  if !ENV["DEV"]
    FileUtils.rm_rf("tmp")
  end
end