Class: Puppet::Pops::Parser::Lexer2

Inherits:

Object

Object
Puppet::Pops::Parser::Lexer2

show all

Includes:: EppSupport, HeredocSupport, InterpolationSupport, LexerSupport, SlurpSupport

Defined in:: lib/puppet/pops/parser/lexer2.rb

Constant Summary collapse

TOKEN_LBRACK = ALl tokens have three slots, the token name (a Symbol), the token text (String), and a token text length. All operator and punctuation tokens reuse singleton arrays Tokens that require unique values create a unique array per token. PEFORMANCE NOTES: This construct reduces the amount of object that needs to be created for operators and punctuation. The length is pre-calculated for all singleton tokens. The length is used both to signal the length of the token, and to advance the scanner position (without having to advance it with a scan(regexp)).

[:LBRACK,       '['.freeze,   1].freeze

TOKEN_LISTSTART =

[:LISTSTART,    '['.freeze,   1].freeze

TOKEN_RBRACK =

[:RBRACK,       ']'.freeze,   1].freeze

TOKEN_LBRACE =

[:LBRACE,       '{'.freeze,   1].freeze

TOKEN_RBRACE =

[:RBRACE,       '}'.freeze,   1].freeze

TOKEN_SELBRACE =

[:SELBRACE,     '{'.freeze,   1].freeze

TOKEN_LPAREN =

[:LPAREN,       '('.freeze,   1].freeze

TOKEN_WSLPAREN =

[:WSLPAREN,     '('.freeze,   1].freeze

TOKEN_RPAREN =

[:RPAREN,       ')'.freeze,   1].freeze

TOKEN_EQUALS =

[:EQUALS,       '='.freeze,   1].freeze

TOKEN_APPENDS =

[:APPENDS,      '+='.freeze,  2].freeze

TOKEN_DELETES =

[:DELETES,      '-='.freeze,  2].freeze

TOKEN_ISEQUAL =

[:ISEQUAL,      '=='.freeze,  2].freeze

TOKEN_NOTEQUAL =

[:NOTEQUAL,     '!='.freeze,  2].freeze

TOKEN_MATCH =

[:MATCH,        '=~'.freeze,  2].freeze

TOKEN_NOMATCH =

[:NOMATCH,      '!~'.freeze,  2].freeze

TOKEN_GREATEREQUAL =

[:GREATEREQUAL, '>='.freeze,  2].freeze

TOKEN_GREATERTHAN =

[:GREATERTHAN,  '>'.freeze,   1].freeze

TOKEN_LESSEQUAL =

[:LESSEQUAL,    '<='.freeze,  2].freeze

TOKEN_LESSTHAN =

[:LESSTHAN,     '<'.freeze,   1].freeze

TOKEN_FARROW =

[:FARROW,       '=>'.freeze,  2].freeze

TOKEN_PARROW =

[:PARROW,       '+>'.freeze,  2].freeze

TOKEN_LSHIFT =

[:LSHIFT,       '<<'.freeze,  2].freeze

TOKEN_LLCOLLECT =

[:LLCOLLECT,    '<<|'.freeze, 3].freeze

TOKEN_LCOLLECT =

[:LCOLLECT,     '<|'.freeze,  2].freeze

TOKEN_RSHIFT =

[:RSHIFT,       '>>'.freeze,  2].freeze

TOKEN_RRCOLLECT =

[:RRCOLLECT,    '|>>'.freeze, 3].freeze

TOKEN_RCOLLECT =

[:RCOLLECT,     '|>'.freeze,  2].freeze

TOKEN_PLUS =

[:PLUS,         '+'.freeze,   1].freeze

TOKEN_MINUS =

[:MINUS,        '-'.freeze,   1].freeze

TOKEN_DIV =

[:DIV,          '/'.freeze,   1].freeze

TOKEN_TIMES =

[:TIMES,        '*'.freeze,   1].freeze

TOKEN_MODULO =

[:MODULO,       '%'.freeze,   1].freeze

TOKEN_NOT =

[:NOT,          '!'.freeze,   1].freeze

TOKEN_DOT =

[:DOT,          '.'.freeze,   1].freeze

TOKEN_PIPE =

[:PIPE,         '|'.freeze,   1].freeze

TOKEN_AT =

[:AT ,          '@'.freeze,   1].freeze

TOKEN_ATAT =

[:ATAT ,        '@@'.freeze,  2].freeze

TOKEN_COLON =

[:COLON,        ':'.freeze,   1].freeze

TOKEN_COMMA =

[:COMMA,        ','.freeze,   1].freeze

TOKEN_SEMIC =

[:SEMIC,        ';'.freeze,   1].freeze

TOKEN_QMARK =

[:QMARK,        '?'.freeze,   1].freeze

TOKEN_TILDE = lexed but not an operator in Puppet

[:TILDE,        '~'.freeze,   1].freeze

TOKEN_REGEXP =

[:REGEXP,       nil,   0].freeze

TOKEN_IN_EDGE =

[:IN_EDGE,      '->'.freeze,  2].freeze

TOKEN_IN_EDGE_SUB =

[:IN_EDGE_SUB,  '~>'.freeze,  2].freeze

TOKEN_OUT_EDGE =

[:OUT_EDGE,     '<-'.freeze,  2].freeze

TOKEN_OUT_EDGE_SUB =

[:OUT_EDGE_SUB, '<~'.freeze,  2].freeze

TOKEN_STRING = Tokens that are always unique to what has been lexed

[:STRING, nil,          0].freeze

TOKEN_WORD =

[:WORD, nil,            0].freeze

TOKEN_DQPRE =

[:DQPRE,  nil,          0].freeze

TOKEN_DQMID =

[:DQPRE,  nil,          0].freeze

TOKEN_DQPOS =

[:DQPRE,  nil,          0].freeze

TOKEN_NUMBER =

[:NUMBER, nil,          0].freeze

TOKEN_VARIABLE =

[:VARIABLE, nil,        1].freeze

TOKEN_VARIABLE_EMPTY =

[:VARIABLE, ''.freeze,  1].freeze

TOKEN_HEREDOC = HEREDOC has syntax as an argument.

[:HEREDOC, nil, 0].freeze

TOKEN_EPPSTART = EPP_START is currently a marker token, may later get syntax

[:EPP_START, nil, 0].freeze

TOKEN_EPPEND =

[:EPP_END, '%>', 2].freeze

TOKEN_EPPEND_TRIM =

[:EPP_END_TRIM, '-%>', 3].freeze

TOKEN_OTHER = This is used for unrecognized tokens, will always be a single character. This particular instance is not used, but is kept here for documentation purposes.

[:OTHER,  nil,  0]

KEYWORDS = Keywords are all singleton tokens with pre calculated lengths. Booleans are pre-calculated (rather than evaluating the strings “false” “true” repeatedly.

{
  'case'     => [:CASE,     'case',     4],
  'class'    => [:CLASS,    'class',    5],
  'default'  => [:DEFAULT,  'default',  7],
  'define'   => [:DEFINE,   'define',   6],
  'if'       => [:IF,       'if',       2],
  'elsif'    => [:ELSIF,    'elsif',    5],
  'else'     => [:ELSE,     'else',     4],
  'inherits' => [:INHERITS, 'inherits', 8],
  'node'     => [:NODE,     'node',     4],
  'and'      => [:AND,      'and',      3],
  'or'       => [:OR,       'or',       2],
  'undef'    => [:UNDEF,    'undef',    5],
  'false'    => [:BOOLEAN,  false,      5],
  'true'     => [:BOOLEAN,  true,       4],
  'in'       => [:IN,       'in',       2],
  'unless'   => [:UNLESS,   'unless',   6],
  'function' => [:FUNCTION, 'function', 8],
  'type'     => [:TYPE,     'type',     4],
  'attr'     => [:ATTR,     'attr',     4],
  'private'  => [:PRIVATE,  'private',  7],
  'application' => [:APPLICATION, 'application',  11],
  'consumes'    => [:CONSUMES,    'consumes',  8],
  'produces'    => [:PRODUCES,    'produces',  8],
  'site'        => [:SITE,        'site',  4],
}

KEYWORD_NAMES = Reverse lookup of keyword name to string

{}

PATTERN_WS =

%r{[[:blank:]\r]+}

PATTERN_NON_WS =

%r{\w+\b?}

PATTERN_COMMENT = The single line comment includes the line ending.

%r{#.*\r?}

PATTERN_MLCOMMENT =

%r{/\*(.*?)\*/}m

PATTERN_REGEX =

%r{/[^/]*/}

PATTERN_REGEX_END =

%r{/}

PATTERN_REGEX_A = for replacement to “”

%r{\A/}

PATTERN_REGEX_Z = for replacement to “”

%r{/\Z}

PATTERN_REGEX_ESC = for replacement to “/”

%r{\\/}

PATTERN_CLASSREF = The NAME and CLASSREF in 4x are strict. Each segment must start with a letter a-z and may not contain dashes (w includes letters, digits and _).

%r{((::){0,1}[A-Z][\w]*)+}

PATTERN_NAME =

%r{^((::)?[a-z][\w]*)(::[a-z][\w]*)*$}

PATTERN_BARE_WORD =

%r{((?:::){0,1}(?:[a-z_](?:[\w-]*[\w])?))+}

PATTERN_DOLLAR_VAR =

%r{\$(::)?(\w+::)*\w+}

PATTERN_NUMBER =

%r{\b(?:0[xX][0-9A-Fa-f]+|0?\d+(?:\.\d+)?(?:[eE]-?\d+)?)\b}

STRING_BSLASH_SLASH = PERFORMANCE NOTE: Comparison against a frozen string is faster (than unfrozen).

'\/'.freeze

Constants included from EppSupport

EppSupport::TOKEN_RENDER_EXPR, EppSupport::TOKEN_RENDER_STRING

Constants included from SlurpSupport

SlurpSupport::DQ_ESCAPES, SlurpSupport::SLURP_ALL_PATTERN, SlurpSupport::SLURP_DQ_PATTERN, SlurpSupport::SLURP_SQ_PATTERN, SlurpSupport::SLURP_UQNE_PATTERN, SlurpSupport::SLURP_UQ_PATTERN, SlurpSupport::SQ_ESCAPES, SlurpSupport::UQ_ESCAPES

Constants included from LexerSupport

Constants included from InterpolationSupport

InterpolationSupport::PATTERN_VARIABLE

Constants included from HeredocSupport

HeredocSupport::PATTERN_HEREDOC

Instance Attribute Summary collapse

#locator ⇒ Object readonly

Instance Method Summary collapse

#clear ⇒ Object

Clears the lexer state (it is not required to call this as it will be garbage collected and the next lex call (lex_string, lex_file) will reset the internal state..
#emit(token, byte_offset) ⇒ Object

Emits (produces) a token [:tokensymbol, TokenValue] and moves the scanner’s position past the token.
#emit_completed(token, byte_offset) ⇒ Object

Emits the completed token on the form [:tokensymbol, TokenValue. This method does not alter the scanner’s position..
#enqueue(emitted_token) ⇒ Object

Allows subprocessors for heredoc etc to enqueue tokens that are tokenized by a different lexer instance.
#enqueue_completed(token, byte_offset) ⇒ Object

Enqueues a completed token at the given offset.
#escaped_end(value) ⇒ Object

Determine if last char of value is escaped by a backslash.
#file ⇒ Object

TODO: This method should not be used, callers should get the locator since it is most likely required to compute line, position etc given offsets.
#file=(file) ⇒ Object

Convenience method, and for compatibility with older lexer.
#fullscan ⇒ Object

Scans all of the content and returns it in an array Note that the terminating [false, false] token is included in the result.
#initialize ⇒ Lexer2 constructor

A new instance of Lexer2.
#initvars ⇒ Object
#lex_file(file) ⇒ Object

Initializes lexing of the content of the given file.
#lex_string(string, path = nil) ⇒ Object
#lex_token ⇒ Object

This lexes one token at the current position of the scanner.
#lex_unquoted_string(string, locator, escapes, interpolate) ⇒ Object

Lexes an unquoted string.
#regexp_acceptable? ⇒ Boolean

Answers after which tokens it is acceptable to lex a regular expression.
#scan {|[false, false]| ... } ⇒ Object

A block must be passed to scan.
#string=(string) ⇒ Object

Convenience method, and for compatibility with older lexer.

Constructor Details

#initialize ⇒ `Lexer2`

Returns a new instance of Lexer2.

# File 'lib/puppet/pops/parser/lexer2.rb', line 186

def initialize()
  @selector = {
    '.' =>  lambda { emit(TOKEN_DOT, @scanner.pos) },
    ',' => lambda {  emit(TOKEN_COMMA, @scanner.pos) },
    '[' => lambda do
      before = @scanner.pos
      if (before == 0 || @scanner.string[locator.char_offset(before)-1,1] =~ /[[:blank:]\r\n]+/)
        emit(TOKEN_LISTSTART, before)
      else
        emit(TOKEN_LBRACK, before)
      end
    end,
    ']' => lambda { emit(TOKEN_RBRACK, @scanner.pos) },
    '(' => lambda do
      before = @scanner.pos
      # If first on a line, or only whitespace between start of line and '('
      # then the token is special to avoid being taken as start of a call.
      line_start = @lexing_context[:line_lexical_start]
      if before == line_start || @scanner.string.byteslice(line_start, before - line_start) =~ /\A[[:blank:]\r]+\Z/
        emit(TOKEN_WSLPAREN, before)
      else
        emit(TOKEN_LPAREN, before)
      end
    end,
    ')' => lambda { emit(TOKEN_RPAREN, @scanner.pos) },
    ';' => lambda { emit(TOKEN_SEMIC, @scanner.pos) },
    '?' => lambda { emit(TOKEN_QMARK, @scanner.pos) },
    '*' => lambda { emit(TOKEN_TIMES, @scanner.pos) },
    '%' => lambda do
      scn = @scanner
      before = scn.pos
      la = scn.peek(2)
      if la[1] == '>' && @lexing_context[:epp_mode]
        scn.pos += 2
        if @lexing_context[:epp_mode] == :expr
          enqueue_completed(TOKEN_EPPEND, before)
        end
        @lexing_context[:epp_mode] = :text
        interpolate_epp
      else
        emit(TOKEN_MODULO, before)
      end
    end,
    '{' => lambda do
      # The lexer needs to help the parser since the technology used cannot deal with
      # lookahead of same token with different precedence. This is solved by making left brace
      # after ? into a separate token.
      #
      @lexing_context[:brace_count] += 1
      emit(if @lexing_context[:after] == :QMARK
             TOKEN_SELBRACE
           else
             TOKEN_LBRACE
           end, @scanner.pos)
    end,
    '}' => lambda do
      @lexing_context[:brace_count] -= 1
      emit(TOKEN_RBRACE, @scanner.pos)
    end,


    # TOKENS @, @@, @(
    '@' => lambda do
      scn = @scanner
      la = scn.peek(2)
      if la[1] == '@'
        emit(TOKEN_ATAT, scn.pos) # TODO; Check if this is good for the grammar
      elsif la[1] == '('
        heredoc
      else
        emit(TOKEN_AT, scn.pos)
      end
    end,

    # TOKENS |, |>, |>>
    '|' => lambda do
      scn = @scanner
      la = scn.peek(3)
      emit(la[1] == '>' ? (la[2] == '>' ? TOKEN_RRCOLLECT : TOKEN_RCOLLECT) : TOKEN_PIPE, scn.pos)
    end,

    # TOKENS =, =>, ==, =~
    '=' => lambda do
      scn = @scanner
      la = scn.peek(2)
      emit(case la[1]
           when '='
             TOKEN_ISEQUAL
           when '>'
             TOKEN_FARROW
           when '~'
             TOKEN_MATCH
           else
             TOKEN_EQUALS
           end, scn.pos)
    end,

    # TOKENS '+', '+=', and '+>'
    '+' => lambda do
      scn = @scanner
      la = scn.peek(2)
      emit(case la[1]
           when '='
             TOKEN_APPENDS
           when '>'
             TOKEN_PARROW
           else
             TOKEN_PLUS
           end, scn.pos)
    end,

    # TOKENS '-', '->', and epp '-%>' (end of interpolation with trim)
    '-' => lambda do
      scn = @scanner
      la = scn.peek(3)
      before = scn.pos
      if @lexing_context[:epp_mode] && la[1] == '%' && la[2] == '>'
        scn.pos += 3
        if @lexing_context[:epp_mode] == :expr
          enqueue_completed(TOKEN_EPPEND_TRIM, before)
        end
        interpolate_epp(:with_trim)
      else
        emit(case la[1]
             when '>'
               TOKEN_IN_EDGE
             when '='
               TOKEN_DELETES
             else
               TOKEN_MINUS
             end, before)
      end
    end,

    # TOKENS !, !=, !~
    '!' => lambda do
      scn = @scanner
      la = scn.peek(2)
      emit(case la[1]
           when '='
             TOKEN_NOTEQUAL
           when '~'
             TOKEN_NOMATCH
           else
             TOKEN_NOT
           end, scn.pos)
    end,

    # TOKENS ~>, ~
    '~' => lambda do
      scn = @scanner
      la = scn.peek(2)
      emit(la[1] == '>' ? TOKEN_IN_EDGE_SUB : TOKEN_TILDE, scn.pos)
    end,

    '#' => lambda { @scanner.skip(PATTERN_COMMENT); nil },

    # TOKENS '/', '/*' and '/ regexp /'
    '/' => lambda do
      scn = @scanner
      la = scn.peek(2)
      if la[1] == '*'
        lex_error(Issues::UNCLOSED_MLCOMMENT) if scn.skip(PATTERN_MLCOMMENT).nil?
        nil
      else
        before = scn.pos
        # regexp position is a regexp, else a div
        if regexp_acceptable? && value = scn.scan(PATTERN_REGEX)
          # Ensure an escaped / was not matched
          while escaped_end(value)
            more = scn.scan_until(PATTERN_REGEX_END)
            return emit(TOKEN_DIV, before) unless more
            value << more
          end
          regex = value.sub(PATTERN_REGEX_A, '').sub(PATTERN_REGEX_Z, '').gsub(PATTERN_REGEX_ESC, '/')
          emit_completed([:REGEX, Regexp.new(regex), scn.pos-before], before)
        else
          emit(TOKEN_DIV, before)
        end
      end
    end,

    # TOKENS <, <=, <|, <<|, <<, <-, <~
    '<' => lambda do
      scn = @scanner
      la = scn.peek(3)
      emit(case la[1]
           when '<'
             if la[2] == '|'
               TOKEN_LLCOLLECT
             else
               TOKEN_LSHIFT
             end
           when '='
             TOKEN_LESSEQUAL
           when '|'
             TOKEN_LCOLLECT
           when '-'
             TOKEN_OUT_EDGE
           when '~'
             TOKEN_OUT_EDGE_SUB
           else
             TOKEN_LESSTHAN
           end, scn.pos)
    end,

    # TOKENS >, >=, >>
    '>' => lambda do
      scn = @scanner
      la = scn.peek(2)
      emit(case la[1]
           when '>'
             TOKEN_RSHIFT
           when '='
             TOKEN_GREATEREQUAL
           else
             TOKEN_GREATERTHAN
           end, scn.pos)
    end,

    # TOKENS :, ::CLASSREF, ::NAME
    ':' => lambda do
      scn = @scanner
      la = scn.peek(3)
      before = scn.pos
      if la[1] == ':'
        # PERFORMANCE NOTE: This could potentially be speeded up by using a case/when listing all
        # upper case letters. Alternatively, the 'A', and 'Z' comparisons may be faster if they are
        # frozen.
        #
        la2 = la[2]
        if la2 >= 'A' && la2 <= 'Z'
          # CLASSREF or error
          value = scn.scan(PATTERN_CLASSREF)
          if value && scn.peek(2) != '::'
            after = scn.pos
            emit_completed([:CLASSREF, value.freeze, after-before], before)
          else
            # move to faulty position ('::<uc-letter>' was ok)
            scn.pos = scn.pos + 3
            lex_error(Issues::ILLEGAL_FULLY_QUALIFIED_CLASS_REFERENCE)
          end
        else
          value = scn.scan(PATTERN_BARE_WORD)
          if value
            if value =~ PATTERN_NAME
              emit_completed([:NAME, value.freeze, scn.pos - before], before)
            else
              emit_completed([:WORD, value.freeze, scn.pos - before], before)
            end
          else
            # move to faulty position ('::' was ok)
            scn.pos = scn.pos + 2
            lex_error(Issues::ILLEGAL_FULLY_QUALIFIED_NAME)
          end
        end
      else
        emit(TOKEN_COLON, before)
      end
    end,

    '$' => lambda do
      scn = @scanner
      before = scn.pos
      if value = scn.scan(PATTERN_DOLLAR_VAR)
        emit_completed([:VARIABLE, value[1..-1].freeze, scn.pos - before], before)
      else
        # consume the $ and let higher layer complain about the error instead of getting a syntax error
        emit(TOKEN_VARIABLE_EMPTY, before)
      end
    end,

    '"' => lambda do
      # Recursive string interpolation, 'interpolate' either returns a STRING token, or
      # a DQPRE with the rest of the string's tokens placed in the @token_queue
      interpolate_dq
    end,

    "'" => lambda do
      scn = @scanner
      before = scn.pos
      emit_completed([:STRING, slurp_sqstring.freeze, scn.pos - before], before)
    end,

    "\n" => lambda do
      # If heredoc_cont is in effect there are heredoc text lines to skip over
      # otherwise just skip the newline.
      #
      ctx = @lexing_context
      if ctx[:newline_jump]
        @scanner.pos = ctx[:newline_jump]
        ctx[:newline_jump] = nil
      else
        @scanner.pos += 1
      end
      ctx[:line_lexical_start] = @scanner.pos
      nil
    end,
    '' => lambda { nil } # when the peek(1) returns empty
  }

  [ ' ', "\t", "\r" ].each { |c| @selector[c] = lambda { @scanner.skip(PATTERN_WS); nil } }

  [ '0', '1', '2', '3', '4', '5', '6', '7', '8', '9'].each do |c|
    @selector[c] = lambda do
      scn = @scanner
      before = scn.pos
      value = scn.scan(PATTERN_NUMBER)
      if value
        length = scn.pos - before
        assert_numeric(value, before)
        emit_completed([:NUMBER, value.freeze, length], before)
      else
        invalid_number = scn.scan_until(PATTERN_NON_WS)
        if before > 1
          after = scn.pos
          scn.pos = before - 1
          if scn.peek(1) == '.'
            # preceded by a dot. Is this a bad decimal number then?
            scn.pos = before - 2
            while scn.peek(1) =~ /^\d$/
              invalid_number = nil
              before = scn.pos
              break if before == 0
              scn.pos = scn.pos - 1
            end
          end
          scn.pos = before
          invalid_number = scn.peek(after - before) unless invalid_number
        end
        length = scn.pos - before
        assert_numeric(invalid_number, before)
        scn.pos = before + 1
        lex_error(Issues::ILLEGAL_NUMBER, {:value => invalid_number})
      end
    end
  end
  ['a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm',
    'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', '_'].each do |c|
    @selector[c] = lambda do

      scn = @scanner
      before = scn.pos
      value = scn.scan(PATTERN_BARE_WORD)
      if value && value =~ PATTERN_NAME
        emit_completed(KEYWORDS[value] || [:NAME, value.freeze, scn.pos - before], before)
      elsif value
        emit_completed([:WORD, value.freeze, scn.pos - before], before)
      else
        # move to faulty position ([a-z_] was ok)
        scn.pos = scn.pos + 1
        fully_qualified = scn.match?(/::/)
        if fully_qualified
          lex_error(Issues::ILLEGAL_FULLY_QUALIFIED_NAME)
        else
          lex_error(Issues::ILLEGAL_NAME_OR_BARE_WORD)
        end
      end
    end
  end

  ['A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M',
    'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z'].each do |c|
    @selector[c] = lambda do
      scn = @scanner
      before = scn.pos
      value = @scanner.scan(PATTERN_CLASSREF)
      if value && @scanner.peek(2) != '::'
        emit_completed([:CLASSREF, value.freeze, scn.pos - before], before)
      else
        # move to faulty position ([A-Z] was ok)
        scn.pos = scn.pos + 1
        lex_error(Issues::ILLEGAL_CLASS_REFERENCE)
      end
    end
  end

  @selector.default = lambda do
    # In case of unicode spaces of various kinds that are captured by a regexp, but not by the
    # simpler case expression above (not worth handling those special cases with better performance).
    scn = @scanner
    if scn.skip(PATTERN_WS)
      nil
    else
      # "unrecognized char"
      emit([:OTHER, scn.peek(0), 1], scn.pos)
    end
  end
  @selector.each { |k,v| k.freeze }
  @selector.freeze
end

Instance Attribute Details

#locator ⇒ `Object` (readonly)



184
185
186

# File 'lib/puppet/pops/parser/lexer2.rb', line 184

def locator
  @locator
end

Instance Method Details

#clear ⇒ `Object`

Clears the lexer state (it is not required to call this as it will be garbage collected and the next lex call (lex_string, lex_file) will reset the internal state.

# File 'lib/puppet/pops/parser/lexer2.rb', line 596

def clear()
  # not really needed, but if someone wants to ensure garbage is collected as early as possible
  @scanner = nil
  @locator = nil
  @lexing_context = nil
end

#emit(token, byte_offset) ⇒ `Object`

Emits (produces) a token [:tokensymbol, TokenValue] and moves the scanner’s position past the token

# File 'lib/puppet/pops/parser/lexer2.rb', line 720

def emit(token, byte_offset)
  @scanner.pos = byte_offset + token[2]
  [token[0], TokenValue.new(token, byte_offset, @locator)]
end

#emit_completed(token, byte_offset) ⇒ `Object`

Emits the completed token on the form [:tokensymbol, TokenValue. This method does not alter the scanner’s position.



728
729
730

# File 'lib/puppet/pops/parser/lexer2.rb', line 728

def emit_completed(token, byte_offset)
  [token[0], TokenValue.new(token, byte_offset, @locator)]
end

#enqueue(emitted_token) ⇒ `Object`

Allows subprocessors for heredoc etc to enqueue tokens that are tokenized by a different lexer instance



739
740
741

# File 'lib/puppet/pops/parser/lexer2.rb', line 739

def enqueue(emitted_token)
  @token_queue << emitted_token
end

#enqueue_completed(token, byte_offset) ⇒ `Object`

Enqueues a completed token at the given offset



733
734
735

# File 'lib/puppet/pops/parser/lexer2.rb', line 733

def enqueue_completed(token, byte_offset)
  @token_queue << emit_completed(token, byte_offset)
end

#escaped_end(value) ⇒ `Object`

Determine if last char of value is escaped by a backslash

# File 'lib/puppet/pops/parser/lexer2.rb', line 579

def escaped_end(value)
  escaped = false
  if value.end_with?(STRING_BSLASH_SLASH)
    value[1...-1].each_codepoint do |cp|
      if cp == 0x5c # backslash
        escaped = !escaped
      else
        escaped = false
      end
    end
  end
  escaped
end

#file ⇒ `Object`

TODO: This method should not be used, callers should get the locator since it is most likely required to compute line, position etc given offsets.



644
645
646

# File 'lib/puppet/pops/parser/lexer2.rb', line 644

def file
  @locator ? @locator.file : nil
end

#file=(file) ⇒ `Object`

Convenience method, and for compatibility with older lexer. Use the lex_file instead. (Bad form to use overloading of assignment operator for something that is not really an assignment).



637
638
639

# File 'lib/puppet/pops/parser/lexer2.rb', line 637

def file=(file)
  lex_file(file)
end

#fullscan ⇒ `Object`

Scans all of the content and returns it in an array Note that the terminating [false, false] token is included in the result.

# File 'lib/puppet/pops/parser/lexer2.rb', line 671

def fullscan
  result = []
  scan {|token| result.push(token) }
  result
end

#initvars ⇒ `Object`

# File 'lib/puppet/pops/parser/lexer2.rb', line 658

def initvars
  @token_queue = []
  # NOTE: additional keys are used; :escapes, :uq_slurp_pattern, :newline_jump, :epp_*
  @lexing_context = {
    :brace_count => 0,
    :after => nil,
    :line_lexical_start => 0
  }
end

#lex_file(file) ⇒ `Object`

Initializes lexing of the content of the given file. An empty string is used if the file does not exist.

# File 'lib/puppet/pops/parser/lexer2.rb', line 650

def lex_file(file)
  initvars
  contents = Puppet::FileSystem.exist?(file) ? Puppet::FileSystem.read(file, :encoding => 'utf-8') : ''
  assert_not_bom(contents)
  @scanner = StringScanner.new(contents.freeze)
  @locator = Locator.locator(contents, file)
end

#lex_string(string, path = nil) ⇒ `Object`

# File 'lib/puppet/pops/parser/lexer2.rb', line 612

def lex_string(string, path=nil)
  initvars
  assert_not_bom(string)
  @scanner = StringScanner.new(string)
  @locator = Locator.locator(string, path)
end

#lex_token ⇒ `Object`

This lexes one token at the current position of the scanner. PERFORMANCE NOTE: Any change to this logic should be performance measured.



714
715
716

# File 'lib/puppet/pops/parser/lexer2.rb', line 714

def lex_token
  @selector[@scanner.peek(1)].call
end

#lex_unquoted_string(string, locator, escapes, interpolate) ⇒ `Object`

Lexes an unquoted string.

Parameters:

string (String) —

the string to lex
locator (Locator) —

the locator to use (a default is used if nil is given)
escapes (Array<String>) —

array of character strings representing the escape sequences to transform
interpolate (Boolean) —

whether interpolation of expressions should be made or not.

# File 'lib/puppet/pops/parser/lexer2.rb', line 625

def lex_unquoted_string(string, locator, escapes, interpolate)
  initvars
  assert_not_bom(string)
  @scanner = StringScanner.new(string)
  @locator = locator || Locator.locator(string, '')
  @lexing_context[:escapes] = escapes || UQ_ESCAPES
  @lexing_context[:uq_slurp_pattern] = interpolate ? (escapes.include?('$') ? SLURP_UQ_PATTERN : SLURP_UQNE_PATTERN) : SLURP_ALL_PATTERN
end

#regexp_acceptable? ⇒ `Boolean`

Answers after which tokens it is acceptable to lex a regular expression. PERFORMANCE NOTE: It may be beneficial to turn this into a hash with default value of true for missing entries. A case expression with literal values will however create a hash internally. Since a reference is always needed to the hash, this access is almost as costly as a method call.

Returns:

(Boolean)

# File 'lib/puppet/pops/parser/lexer2.rb', line 749

def regexp_acceptable?
  case @lexing_context[:after]

  # Ends of (potential) R-value generating expressions
  when :RPAREN, :RBRACK, :RRCOLLECT, :RCOLLECT
    false

  # End of (potential) R-value - but must be allowed because of case expressions
  # Called out here to not be mistaken for a bug.
  when :RBRACE
    true

  # Operands (that can be followed by DIV (even if illegal in grammar)
  when :NAME, :CLASSREF, :NUMBER, :STRING, :BOOLEAN, :DQPRE, :DQMID, :DQPOST, :HEREDOC, :REGEX, :VARIABLE, :WORD
    false

  else
    true
  end
end

#scan {|[false, false]| ... } ⇒ `Object`

A block must be passed to scan. It will be called with two arguments, a symbol for the token, and an instance of LexerSupport::TokenValue PERFORMANCE NOTE: The TokenValue is designed to reduce the amount of garbage / temporary data and to only convert the lexer’s internal tokens on demand. It is slightly more costly to create an instance of a class defined in Ruby than an Array or Hash, but the gain is much bigger since transformation logic is avoided for many of its members (most are never used (e.g. line/pos information which is only of value in general for error messages, and for some expressions (which the lexer does not know about).

Yields:

([false, false])

# File 'lib/puppet/pops/parser/lexer2.rb', line 685

def scan
  # PERFORMANCE note: it is faster to access local variables than instance variables.
  # This makes a small but notable difference since instance member access is avoided for
  # every token in the lexed content.
  #
  scn   = @scanner
  lex_error_without_pos(Issues::NO_INPUT_TO_LEXER) unless scn

  ctx   = @lexing_context
  queue = @token_queue
  selector = @selector

  scn.skip(PATTERN_WS)

  # This is the lexer's main loop
  until queue.empty? && scn.eos? do
    if token = queue.shift || selector[scn.peek(1)].call
      ctx[:after] = token[0]
      yield token
    end
  end

  # Signals end of input
  yield [false, false]
end

#string=(string) ⇒ `Object`

Convenience method, and for compatibility with older lexer. Use the lex_string instead which allows passing the path to use without first having to call file= (which reads the file if it exists). (Bad form to use overloading of assignment operator for something that is not really an assignment. Also, overloading of = does not allow passing more than one argument).



608
609
610

# File 'lib/puppet/pops/parser/lexer2.rb', line 608

def string=(string)
  lex_string(string, nil)
end

Class: Puppet::Pops::Parser::Lexer2

Constant Summary collapse

Constants included from EppSupport

Constants included from SlurpSupport

Constants included from LexerSupport

Constants included from InterpolationSupport

Constants included from HeredocSupport

Instance Attribute Summary collapse

Instance Method Summary collapse

Methods included from EppSupport

Methods included from SlurpSupport

Methods included from LexerSupport

Methods included from InterpolationSupport

Methods included from HeredocSupport

Constructor Details

#initialize ⇒ Lexer2

Instance Attribute Details

#locator ⇒ Object (readonly)

Instance Method Details

#clear ⇒ Object

#emit(token, byte_offset) ⇒ Object

#emit_completed(token, byte_offset) ⇒ Object

#enqueue(emitted_token) ⇒ Object

#enqueue_completed(token, byte_offset) ⇒ Object

#escaped_end(value) ⇒ Object

#file ⇒ Object

#file=(file) ⇒ Object

#fullscan ⇒ Object

#initvars ⇒ Object

#lex_file(file) ⇒ Object

#lex_string(string, path = nil) ⇒ Object

#lex_token ⇒ Object

#lex_unquoted_string(string, locator, escapes, interpolate) ⇒ Object

#regexp_acceptable? ⇒ Boolean

#scan {|[false, false]| ... } ⇒ Object

#string=(string) ⇒ Object