faster_url_parser.cr

# to_unsafe AND avoid creating Chars

class URL
  property scheme, path, host
  property non_relative_flag
end

class Parser
  property url

  SPECIAL_SCHEME = Set{"ftp", "file", "gopher", "http", "https", "ws", "wss"}

  # https://url.spec.whatwg.org/
  def initialize(input)
    @url = URL.new
    @input = input.to_unsafe
    @state = :scheme_start
    # @buffer = MemoryIO.new
    @at_flag = false
    @bracket_flag = false
    @ptr = 0
  end

  def c
    @input[@ptr]
  end

  def run
    parse_scheme_start
  end

  def special_scheme?
    SPECIAL_SCHEME.includes? url.scheme
  end

  def parse_scheme_start
    if alpha?
      parse_scheme
    else
      # Nothing
    end
  end

  def alpha?
    ('a'.ord <= c && c <= 'z'.ord) ||
      ('A'.ord <= c && c <= 'Z'.ord)
  end

  def parse_scheme
    start = @ptr
    while true
      if alpha? || c === '-' || c === '.' || c === '+'
        # @buffer << c.chr
        @ptr += 1
      elsif c === ':'
        @url.scheme = String.new(@input + start, @ptr - start)
        # todo file and other special cases
        if @input[@ptr + 1] === '/'
          # @state = :path_or_authority
          @ptr += 1
          parse_path_or_authority
          break
        else
          @url.non_relative_flag = true
          @url.path = ""
          parse_non_relative_path
          # @state = :non_relative_path
        end
      else
        # Nothing
        break
        # @state = :no_scheme
        # @ptr = 0
        # break
      end
    end
  end

  def parse_path_or_authority
    if c === '/'
      @ptr += 1
      parse_authority
    else
      parse_path
    end
  end

  def parse_authority
    while true
      if c === '@'
        # todo
      elsif c === '\0' || c === '/' || c === '?' || c === '#' || (special_scheme? && c === '\\')
        @ptr += 1
        parse_host
        break
      end
      @ptr += 1
    end
  end

  def parse_host
    bracket_flag = false
    start = @ptr
    while true
      if c === ':' && !bracket_flag
        # todo if url is special and buffer empty fail
        # url.host = @buffer.to_s
        # reset_buffer
        parse_port
        break
      elsif c === '\0' || c === '/' || c === '?' || c === '#' || (special_scheme? && c === '\\')
        @url.host = String.new(@input + start, @ptr - start)
        @ptr -= 1
        # todo if url is special and buffer empty fail
        # todo host parsing buffer
        parse_path
        break
      else
        bracket_flag = true if c == '['.ord
        bracket_flag = false if c == ']'.ord
      end
      @ptr += 1
    end
  end

  def parse_port
  end

  def parse_path
  end

  def parse_non_relative_path
  end
end

require "benchmark"
require "uri"

par = Parser.new("http://bitfission.com")
par.run
puts [par.url.scheme, par.url.host]

uri = URI.parse("http://bitfission.com")
puts [uri.scheme, uri.host]

Benchmark.ips do |x|
  x.report("new") { Parser.new("http://bitfission.com").run }
  x.report("old") { URI.parse("http://bitfission.com") }
end