path: root/scripts/import_vim

#!/usr/bin/env ruby

require 'yaml'
require 'json'

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.map { |e| e.gsub('\\,', ',')}
end

def import_autocommands
  cmd = <<~EOF
vim -es -u DEFAULTS -c 'exe("func! Capture() \\n redir => capture \\n silent autocmd filetypedetect BufRead \\n redir END \\n return capture \\n endfunc") | let a = Capture() | put =a | %print | :q!'
  EOF
  filetypes = Hash.new { |h, k| h[k] = [] }
  autocommands = `#{cmd}`
  for pattern, cmd in autocommands.scan(/^    ([^ ]+)\s*(.*)\n/)
    cmd = cmd.strip

    what = {}
    if match = cmd.match(/^setf\s+(.*)$/)
      what[:ft] = match[1]
    elsif match = cmd.match(/^call s:StarSetf\(\'(.*)\'\)$/)
      what[:ft] = match[1]
    elsif match = cmd.match(/^set filetype=(.*)$/)
      what[:ft] = match[1]
    end

    if what != {}
      for p in comma_expanson(pattern.strip)
        filetypes[what[:ft].strip] << p
      end
    end
  end

  for k, v in filetypes
    filetypes[k].uniq!
  end

  filetypes
end

$with_special_char = /#{[".", ":", "{", "}", "[", "]", ",", "&", "*", "#", "?", "|", "-", "<", ">", "=", "!", "%", "@", "\\"].map { |e| Regexp.escape(e) }.join("|")}/
def print_pattern(p)
  if p.match($with_special_char)
    p.gsub("'", "\\'")
  end

  return p
end

def generate_packages_entries(filetypes, comments)
  entries = []

  current_filetypes = YAML.load_stream(File.read('packages.yaml')).flat_map do |p|
    p["filetypes"].map { |a| a["name"] }
  end

  for ft, patterns in filetypes
    if current_filetypes.include?(ft)
      next
    end

    output = {
      "name" => ft,
      "remote" => "vim/vim:runtime",
      "glob" => "**/" + ft + ".vim",
      "filetypes" => [{
        "name" => ft,
        "patterns" => []
      }]
    }
    paths_with_comments = patterns.group_by { |p| (comments[ft] || {})[p] || "" }
    for comment, paths in paths_with_comments
      output["filetypes"][0]["patterns"] << {
        "pattern" => paths.join(","),
      }

      if comment.strip.size > 0
        output["filetypes"][0]["patterns"].last["description"] = comment
      end
    end
    entries << YAML.dump(output)
  end

  return entries
end

def fix_quotes(a)
  a = a.gsub(/\\/) { '\\\\' }
  a.scan(/^.*?"(.+?)"\n/m).each { |p| a[p[0]] = p[0].gsub('"') { '\\"'} }
  a
end

def get_comments
  comments_cmd = <<~'EOF'
  awk '/^"/ { comment = $0; next } match($0, "(\\S+)\\s+setf (\\S*)$", a) { print "- filetype: \"" a[2] "\"\n  description: \"" substr(comment, 3) "\"\n  patterns: \"" a[1] "\"" }' 'tmp/vim/vim/runtime/filetype.vim'
  EOF

  comments = YAML.load(fix_quotes(`#{comments_cmd}`))

  comments.reject! { |c| c["patterns"].strip == "" || c["patterns"].include?('\\') }
  result = {}
  comments = comments.flat_map do |c| 
    result[c["filetype"]] ||= {}

    for p in comma_expanson(c["patterns"])
      result[c["filetype"]][p] = c["description"]
    end
  end
  result
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 brace_expansion(s)
  if !s.include?('{')
    return [s]
  end
  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 generate_tests(autocommands)
  existing = JSON.parse(File.read('tmp/vim/vim/src/testdir/test_filetype.vim')
    .match(/let s:filename_checks = ({.*?\\ })/m)[1]
    .gsub('    \\', ' ').gsub("'", '"')
    .gsub('$VIMRUNTIME .', '').gsub(",\n  }", "}"))

  output = []

  generated = {}
  for ft in autocommands.keys().sort()
    patterns = autocommands[ft]

    files1 = []
    to_delete = []

    patterns.reject! { |p| p.match?(/[\|\\\(]/) }

    patterns = patterns.map do |pattern|
      pattern = pattern.gsub('[0-9]', '1').gsub('[2-3]', '2').gsub('[1-3]', '1').gsub('?', 'x')

      if pattern.match?(/\/\*\.[^\*\/]+$/)
        pattern = pattern.gsub(/\/\*\.([^\*\/]+)$/, '/file.\1')
      end

      if pattern.match?(/[\/\.-]\*$/)
        pattern = pattern[0..-2] + "file"
      end

      pattern = pattern.gsub(/\/\*\//) { '/any/' }

      pattern
    end

    patterns = patterns.flat_map { |p| brace_expansion(p) }

    for pattern in patterns
      if pattern.match(/^\*[.\,][^\*\/]+$/)
        files1 << pattern.gsub('*', 'file')

        to_delete << pattern
      end
      if pattern.match(/^[^\*\/]+\.\*$/)
        files1 << pattern.gsub('*', 'file')
        to_delete << pattern
      end
    end
    files1.sort!
    patterns = patterns - to_delete


    files2 = []
    for pattern in patterns
      if !pattern.match(/\*/)
        files2 << pattern
        to_delete << pattern
      end
    end
    files2.sort!
    patterns = patterns - to_delete

    patterns = patterns.flat_map do |pattern|
      if pattern.match?(/^\*\//)
        [pattern[1..-1], "any" + pattern[1..-1]]
      else
        [pattern]
      end
    end
    
    patterns = patterns.flat_map do |pattern|
      if pattern.match?(/^\*[-\.]/)
        ["some" + pattern[1..-1]]
      elsif pattern.match?(/^\*/)
        [pattern[1..-1], "some-" + pattern[1..-1]]
      else
        [pattern]
      end
    end

    patterns = patterns.flat_map do |pattern|
      if pattern.match?(/[^\/]+\/\*-[^\\]+$/)
        [pattern.gsub('/*-', '/file-')]
      elsif pattern.match?(/[^\/]+\/\*\.[^\\]+$/)
        [pattern.gsub('/*.', '/file.')]
      elsif pattern.match?(/[^\/]+\/\*[^\\]+$/)
        [pattern.gsub('/*', '/'), pattern.gsub('/*', '/some-')]
      else
        [pattern]
      end
    end

    patterns = patterns.flat_map do |pattern|
      if pattern.include?('-*/')
        [pattern.gsub('-*/', '-file/')]
      elsif pattern.include?('.*/')
        [pattern.gsub('.*/', '.file/')]
      elsif pattern.include?('*/')
        [pattern.gsub('*/', '/'), pattern.gsub('*/', '-some/')]
      else
        [pattern]
      end
    end

    patterns = patterns.flat_map do |pattern|
      if pattern.match?(/[^\/]+\.\*\.([^\*\/]+)$/)
        [pattern.gsub('.*.', '.file.')]
      elsif pattern.match?(/[^\/]+-\*\.([^\*\/]+)$/)
        [pattern.gsub('-*.', '-file.')]
      elsif pattern.match?(/[^\/]+\*\.([^\*\/]+)$/)
        [pattern.gsub('*.', '.'), pattern.gsub('*.', '-file.')]
      else
        [pattern]
      end
    end

    patterns = patterns.flat_map do |pattern|
      if pattern.match?(/\*$/)
        [pattern[0..-2], pattern[0..-2] + "-file"]
      else
        [pattern]
      end
    end
    patterns.sort!

    files = [*files1, *files2, *patterns].flat_map { |e| square_expansion(e) }
    generated[ft] = files

  end


  for ft, original in existing
    for file in generated.fetch(ft, [])

      unless original.include?(file) || ['file.DEF', 'file.MOD', 'file.BUILD', 'BUILD'].include?(file) || ft == "help"
        original << file
      end
    end
    generated.delete(ft)
    output << "    \\ '#{ft}': #{JSON.generate(original).gsub('"', "'").gsub("','", "', '")},"
  end

  for ft, paths in generated
    idx = output.find_index { |a| a > "    \\ '" + ft }
    output.insert(idx, "    \\ '#{ft}': #{JSON.generate(paths).gsub('"', "'").gsub("','", "', '")},")
  end


  output << "    \\ }"

  msgs = <<'EOF'
    \ 'messages': ['/log/auth', '/log/cron', '/log/daemon', '/log/debug', '/log/kern', '/log/lpr', '/log/mail', '/log/messages', '/log/news/news', '/log/syslog', '/log/user',
    \     '/log/auth.log', '/log/cron.log', '/log/daemon.log', '/log/debug.log', '/log/kern.log', '/log/lpr.log', '/log/mail.log', '/log/messages.log', '/log/news/news.log', '/log/syslog.log', '/log/user.log',
    \     '/log/auth.err', '/log/cron.err', '/log/daemon.err', '/log/debug.err', '/log/kern.err', '/log/lpr.err', '/log/mail.err', '/log/messages.err', '/log/news/news.err', '/log/syslog.err', '/log/user.err',
    \      '/log/auth.info', '/log/cron.info', '/log/daemon.info', '/log/debug.info', '/log/kern.info', '/log/lpr.info', '/log/mail.info', '/log/messages.info', '/log/news/news.info', '/log/syslog.info', '/log/user.info',
    \      '/log/auth.warn', '/log/cron.warn', '/log/daemon.warn', '/log/debug.warn', '/log/kern.warn', '/log/lpr.warn', '/log/mail.warn', '/log/messages.warn', '/log/news/news.warn', '/log/syslog.warn', '/log/user.warn',
    \      '/log/auth.crit', '/log/cron.crit', '/log/daemon.crit', '/log/debug.crit', '/log/kern.crit', '/log/lpr.crit', '/log/mail.crit', '/log/messages.crit', '/log/news/news.crit', '/log/syslog.crit', '/log/user.crit',
    \      '/log/auth.notice', '/log/cron.notice', '/log/daemon.notice', '/log/debug.notice', '/log/kern.notice', '/log/lpr.notice', '/log/mail.notice', '/log/messages.notice', '/log/news/news.notice', '/log/syslog.notice', '/log/user.notice'],
EOF


  "let s:filename_checks = {\n" + output.join("\n").gsub("'/doc/help.txt'", "$VIMRUNTIME . '/doc/help.txt'").gsub(/    \\ 'messages':.*?\],\n/m) { msgs }


end

#comments = get_comments()
autocommands = import_autocommands()
# entries = generate_packages_entries(autocommands, comments)
# print(entries.join(""))
result = generate_tests(autocommands)
print(result)