class PuppetLint::Lexer::StringSlurper

Internal: A class for slurping strings from a Puppet manifest.

Constants

END_INTERP_PATTERN
END_STRING_PATTERN
ESC_DQUOTE_PATTERN
LBRACE_PATTERN
START_INTERP_PATTERN
UNENC_VAR_PATTERN

Attributes

interp_stack[RW]

results[RW]

scanner[RW]

Public Class Methods

new(string) click to toggle source

# File lib/puppet-lint/lexer/string_slurper.rb, line 16
def initialize(string)
  @scanner = StringScanner.new(string)
  @results = []
  @interp_stack = []
  @segment = []
end

Public Instance Methods

consumed_chars() click to toggle source

Get the number of characters consumed by the StringSlurper.

StringScanner from Ruby 2.0 onwards supports charpos which returns the number of characters and is multibyte character aware.

Prior to this, Ruby’s multibyte character support in Strings was a bit unusual and neither String#length nor String#split behave as expected, so we use String#scan to split all the consumed text using a UTF-8 aware regex and use the length of the result

# File lib/puppet-lint/lexer/string_slurper.rb, line 105
def consumed_chars
  return scanner.charpos if scanner.respond_to?(:charpos)

  (scanner.pre_match + scanner.matched).scan(%r{.}mu).length
end

end_heredoc(pattern) click to toggle source

# File lib/puppet-lint/lexer/string_slurper.rb, line 153
def end_heredoc(pattern)
  results << [:HEREDOC, @segment.join]
  results << [:HEREDOC_TERM, scanner.scan(pattern)]
end

end_interp() click to toggle source

# File lib/puppet-lint/lexer/string_slurper.rb, line 128
def end_interp
  if interp_stack.empty?
    @segment << scanner.scan(END_INTERP_PATTERN)
    return
  else
    interp_stack.pop
  end

  if interp_stack.empty?
    results << [:INTERP, @segment.join]
    @segment = []
    scanner.skip(END_INTERP_PATTERN)
  else
    @segment << scanner.scan(END_INTERP_PATTERN)
  end
end

end_string() click to toggle source

# File lib/puppet-lint/lexer/string_slurper.rb, line 158
def end_string
  if interp_stack.empty?
    @segment << scanner.scan(END_STRING_PATTERN).gsub!(%r{"\Z}, '')
    results << [@segment_type, @segment.join]
  else
    @segment << scanner.scan(END_STRING_PATTERN)
  end
end

parse() click to toggle source

# File lib/puppet-lint/lexer/string_slurper.rb, line 23
def parse
  @segment_type = :STRING

  until scanner.eos?
    if scanner.match?(START_INTERP_PATTERN)
      start_interp
    elsif !interp_stack.empty? && scanner.match?(LBRACE_PATTERN)
      read_char
    elsif scanner.match?(END_INTERP_PATTERN)
      end_interp
    elsif unenclosed_variable?
      unenclosed_variable
    elsif scanner.match?(END_STRING_PATTERN)
      end_string
      break if interp_stack.empty?
    elsif scanner.match?(ESC_DQUOTE_PATTERN)
      @segment << scanner.scan(ESC_DQUOTE_PATTERN)
    else
      read_char
    end
  end

  raise UnterminatedStringError if results.empty? && scanner.matched?

  results
end

parse_heredoc(heredoc_tag) click to toggle source

# File lib/puppet-lint/lexer/string_slurper.rb, line 56
def parse_heredoc(heredoc_tag)
  heredoc_name = heredoc_tag[%r{\A"?(.+?)"?(:.+?)?#{PuppetLint::Lexer::WHITESPACE_RE}*(/.*)?\Z}o, 1]
  end_heredoc_pattern = %r{^\|?\s*-?\s*#{Regexp.escape(heredoc_name)}$}
  interpolation = heredoc_tag.start_with?('"')

  @segment_type = :HEREDOC

  until scanner.eos?
    if scanner.match?(end_heredoc_pattern)
      end_heredoc(end_heredoc_pattern)
      break if interp_stack.empty?
    elsif interpolation && scanner.match?(START_INTERP_PATTERN)
      start_interp
    elsif interpolation && !interp_stack.empty? && scanner.match?(LBRACE_PATTERN)
      read_char
    elsif interpolation && unenclosed_variable?
      unenclosed_variable
    elsif interpolation && scanner.match?(END_INTERP_PATTERN)
      end_interp
    else
      read_char
    end
  end

  results
end

read_char() click to toggle source

# File lib/puppet-lint/lexer/string_slurper.rb, line 83
def read_char
  @segment << scanner.getch

  return if interp_stack.empty?

  case @segment.last
  when '{'
    interp_stack.push(true)
  when '}'
    interp_stack.pop
  end
end

start_interp() click to toggle source

# File lib/puppet-lint/lexer/string_slurper.rb, line 111
def start_interp
  if @segment.last && @segment.last == '\\'
    read_char
    return
  end

  if interp_stack.empty?
    scanner.skip(START_INTERP_PATTERN)
    results << [@segment_type, @segment.join]
    @segment = []
  else
    @segment << scanner.scan(START_INTERP_PATTERN)
  end

  interp_stack.push(true)
end

unenclosed_variable() click to toggle source

# File lib/puppet-lint/lexer/string_slurper.rb, line 145
def unenclosed_variable
  read_char if scanner.match?(%r{.\$})

  results << [@segment_type, @segment.join]
  results << [:UNENC_VAR, scanner.scan(UNENC_VAR_PATTERN)]
  @segment = []
end

unenclosed_variable?() click to toggle source

# File lib/puppet-lint/lexer/string_slurper.rb, line 50
def unenclosed_variable?
  interp_stack.empty? &&
    scanner.match?(UNENC_VAR_PATTERN) &&
    (@segment.last.nil? ? true : !@segment.last.end_with?('\\'))
end