Class: RDoc::Markup::Parser

Relationships & Source Files
Namespace Children
Exceptions: Error, ParseError
Extension / Inclusion / Inheritance Descendants
Subclasses: TomDoc
Super Chains via Extension / Inclusion / Inheritance
Instance Chain: self, ::RDoc::Text
Inherits:	Object
Defined in:	lib/rdoc/markup/parser.rb

Overview

A recursive-descent parser for ::RDoc::RDoc markup.

The parser tokenizes an input string then parses the tokens into a Document. Documents can be converted into output formats by writing a visitor like RDoc::Markup::ToHTML.

The parser only handles the block-level constructs Paragraph, List, ListItem, Heading, Verbatim, BlankLine and Rule. Inline markup such as +blah+ is handled separately by AttributeManager.

To see what markup the Parser implements read ::RDoc::RDoc. To see how to use ::RDoc::RDoc markup to format text in your program read ::RDoc::Markup.

Constant Summary

LIST_TOKENS =

List token types

# File 'lib/rdoc/markup/parser.rb', line 25

[
  :BULLET,
  :LABEL,
  :LALPHA,
  :NOTE,
  :NUMBER,
  :UALPHA,
]

Class Method Summary

.new ⇒ Parser constructor

Creates a new Parser.
.parse(str)

Parses str into a Document.
.tokenize(str)

Returns a token stream for str, for testing.

Instance Attribute Summary

#debug rw

Enables display of debugging information.
#tokens readonly

Token accessor.

Instance Method Summary

#build_heading(level)

Builds a Heading of level
#build_list(margin)

Builds a List flush to margin
#build_paragraph(margin)

Builds a Paragraph that is flush to margin
#build_verbatim(margin)

Builds a Verbatim that is indented from margin.
#char_pos(byte_offset)

The character offset for the input string at the given byte_offset
#get

Pulls the next token from the stream.
#parse(parent, indent = 0)

Parses the tokens into an array of RDoc::Markup::XXX objects, and appends them to the passed parent RDoc::Markup::YYY object.
#peek_token

Returns the next token on the stream without modifying the stream.
#setup_scanner(input)

Creates the StringScanner.
#skip(token_type, error = true)

Skips the next token if its type is token_type.
#token_pos(byte_offset)

Calculates the column (by character) and line of the current token based on byte_offset.
#tokenize(input)

Turns text input into a stream of tokens.
#unget

Returns the current token to the token stream.

::RDoc::Text - Included

#expand_tabs	Expands tab characters in `text` to eight spaces.
#flush_left	Flush `text` left based on the shortest line.
#markup	Convert a string in markup format into HTML.
#normalize_comment	Strips hashes, expands tabs then flushes `text` to the left.
#parse	Normalizes `text` then builds a Document from it.
#snippet	The first `limit` characters of `text` as HTML.
#strip_hashes	Strips leading # characters from `text`
#strip_newlines	Strips leading and trailing n characters from `text`
#strip_stars	Strips /* */ style comments.
#to_html	Converts ampersand, dashes, ellipsis, quotes, copyright and registered trademark symbols in `text` to properly encoded characters.
#wrap	Wraps `txt` to `line_len`

Constructor Details

.new ⇒ `Parser`

Creates a new Parser. See also .parse

[ GitHub ]

# File 'lib/rdoc/markup/parser.rb', line 78


def initialize
  @binary_input   = nil
  @current_token  = nil
  @debug          = false
  @have_encoding  = Object.const_defined? :Encoding
  @have_byteslice = ''.respond_to? :byteslice
  @input          = nil
  @input_encoding = nil
  @line           = 0
  @line_pos       = 0
  @s              = nil
  @tokens         = []
end

Class Method Details

.parse(str)

Parses str into a Document.

Use RDoc::Markup#parse instead of this method.

[ GitHub ]

# File 'lib/rdoc/markup/parser.rb', line 59


def self.parse str
  parser = new
  parser.tokenize str
  doc = RDoc::Markup::Document.new
  parser.parse doc
end

.tokenize(str)

Returns a token stream for str, for testing

[ GitHub ]

# File 'lib/rdoc/markup/parser.rb', line 69


def self.tokenize str
  parser = new
  parser.tokenize str
  parser.tokens
end

Instance Attribute Details

#debug (rw)

Enables display of debugging information

[ GitHub ]

# File 'lib/rdoc/markup/parser.rb', line 47


attr_accessor :debug

#tokens (readonly)

Token accessor

[ GitHub ]

# File 'lib/rdoc/markup/parser.rb', line 52


attr_reader :tokens

Instance Method Details

#build_heading(level)

Builds a Heading of level

[ GitHub ]

# File 'lib/rdoc/markup/parser.rb', line 95


def build_heading level
  type, text, = get

  text = case type
         when :TEXT then
           skip :NEWLINE
           text
         else
           unget
           ''
         end

  RDoc::Markup::Heading.new level, text
end

#build_list(margin)

Builds a List flush to margin

[ GitHub ]

# File 'lib/rdoc/markup/parser.rb', line 113


def build_list margin
  p :list_start => margin if @debug

  list = RDoc::Markup::List.new
  label = nil

  until @tokens.empty? do
    type, data, column, = get

    case type
    when *LIST_TOKENS then
      if column < margin || (list.type && list.type != type) then
        unget
        break
      end

      list.type = type
      peek_type, _, column, = peek_token

      case type
      when :NOTE, :LABEL then
        label = [] unless label

        if peek_type == :NEWLINE then
          # description not on the same line as LABEL/NOTE
          # skip the trailing newline & any blank lines below
          while peek_type == :NEWLINE
            get
            peek_type, _, column, = peek_token
          end

          # we may be:
          #   - at end of stream
          #   - at a column < margin:
          #         [text]
          #       blah blah blah
          #   - at the same column, but with a different type of list item
          #       [text]
          #       * blah blah
          #   - at the same column, with the same type of list item
          #       [one]
          #       [two]
          # In all cases, we have an empty description.
          # In the last case only, we continue.
          if peek_type.nil? || column < margin then
            empty = true
          elsif column == margin then
            case peek_type
            when type
              empty = :continue
            when *LIST_TOKENS
              empty = true
            else
              empty = false
            end
          else
            empty = false
          end

          if empty then
            label << data
            next if empty == :continue
            break
          end
        end
      else
        data = nil
      end

      if label then
        data = label << data
        label = nil
      end

      list_item = RDoc::Markup::ListItem.new data
      parse list_item, column
      list << list_item

    else
      unget
      break
    end
  end

  p :list_end => margin if @debug

  if list.empty? then
    return nil unless label
    return nil unless [:LABEL, :NOTE].include? list.type

    list_item = RDoc::Markup::ListItem.new label, RDoc::Markup::BlankLine.new
    list << list_item
  end

  list
end

#build_paragraph(margin)

Builds a Paragraph that is flush to margin

[ GitHub ]

# File 'lib/rdoc/markup/parser.rb', line 213


def build_paragraph margin
  p :paragraph_start => margin if @debug

  paragraph = RDoc::Markup::Paragraph.new

  until @tokens.empty? do
    type, data, column, = get

    if type == :TEXT and column == margin then
      paragraph << data

      break if peek_token.first == :BREAK

      data << ' ' if skip :NEWLINE
    else
      unget
      break
    end
  end

  paragraph.parts.last.sub!(/ \z/, '') # cleanup

  p :paragraph_end => margin if @debug

  paragraph
end

#build_verbatim(margin)

Builds a Verbatim that is indented from margin.

The verbatim block is shifted left (the least indented lines start in column 0). Each part of the verbatim is one line of text, always terminated by a newline. Blank lines always consist of a single newline character, and there is never a single newline at the end of the verbatim.

[ GitHub ]

# File 'lib/rdoc/markup/parser.rb', line 248


def build_verbatim margin
  p :verbatim_begin => margin if @debug
  verbatim = RDoc::Markup::Verbatim.new

  min_indent = nil
  generate_leading_spaces = true
  line = ''

  until @tokens.empty? do
    type, data, column, = get

    if type == :NEWLINE then
      line << data
      verbatim << line
      line = ''
      generate_leading_spaces = true
      next
    end

    if column <= margin
      unget
      break
    end

    if generate_leading_spaces then
      indent = column - margin
      line << ' ' * indent
      min_indent = indent if min_indent.nil? || indent < min_indent
      generate_leading_spaces = false
    end

    case type
    when :HEADER then
      line << '=' * data
      _, _, peek_column, = peek_token
      peek_column ||= column + data
      indent = peek_column - column - data
      line << ' ' * indent
    when :RULE then
      width = 2 + data
      line << '-' * width
      _, _, peek_column, = peek_token
      peek_column ||= column + width
      indent = peek_column - column - width
      line << ' ' * indent
    when :BREAK, :TEXT then
      line << data
    else # *LIST_TOKENS
      list_marker = case type
                    when :BULLET then data
                    when :LABEL  then "[#{data}]"
                    when :NOTE   then "#{data}::"
                    else # :LALPHA, :NUMBER, :UALPHA
                      "#{data}."
                    end
      line << list_marker
      peek_type, _, peek_column = peek_token
      unless peek_type == :NEWLINE then
        peek_column ||= column + list_marker.length
        indent = peek_column - column - list_marker.length
        line << ' ' * indent
      end
    end

  end

  verbatim << line << "\n" unless line.empty?
  verbatim.parts.each { |p| p.slice!(0, min_indent) unless p == "\n" } if min_indent > 0
  verbatim.normalize

  p :verbatim_end => margin if @debug

  verbatim
end

#char_pos(byte_offset)

The character offset for the input string at the given byte_offset

[ GitHub ]

# File 'lib/rdoc/markup/parser.rb', line 326


def char_pos byte_offset
  if @have_byteslice then
    @input.byteslice(0, byte_offset).length
  elsif @have_encoding then
    matched = @binary_input[0, byte_offset]
    matched.force_encoding @input_encoding
    matched.length
  else
    byte_offset
  end
end

#get

Pulls the next token from the stream.

[ GitHub ]

# File 'lib/rdoc/markup/parser.rb', line 341


def get
  @current_token = @tokens.shift
  p :get => @current_token if @debug
  @current_token
end

#parse(parent, indent = 0)

Parses the tokens into an array of RDoc::Markup::XXX objects, and appends them to the passed parent RDoc::Markup::YYY object.

Exits at the end of the token stream, or when it encounters a token in a column less than indent (unless it is a NEWLINE).

Returns parent.

[ GitHub ]

# File 'lib/rdoc/markup/parser.rb', line 356


def parse parent, indent = 0
  p :parse_start => indent if @debug

  until @tokens.empty? do
    type, data, column, = get

    case type
    when :BREAK then
      parent << RDoc::Markup::BlankLine.new
      skip :NEWLINE, false
      next
    when :NEWLINE then
      # trailing newlines are skipped below, so this is a blank line
      parent << RDoc::Markup::BlankLine.new
      skip :NEWLINE, false
      next
    end

    # indentation change: break or verbatim
    if column < indent then
      unget
      break
    elsif column > indent then
      unget
      parent << build_verbatim(indent)
      next
    end

    # indentation is the same
    case type
    when :HEADER then
      parent << build_heading(data)
    when :RULE then
      parent << RDoc::Markup::Rule.new(data)
      skip :NEWLINE
    when :TEXT then
      unget
      parse_text parent, indent
    when *LIST_TOKENS then
      unget
      parent << build_list(indent)
    else
      type, data, column, line = @current_token
      raise ParseError, "Unhandled token #{type} (#{data.inspect}) at #{line}:#{column}"
    end
  end

  p :parse_end => indent if @debug

  parent

end

#peek_token

Returns the next token on the stream without modifying the stream

[ GitHub ]

# File 'lib/rdoc/markup/parser.rb', line 419


def peek_token
  token = @tokens.first || []
  p :peek => token if @debug
  token
end

#setup_scanner(input)

Creates the StringScanner

[ GitHub ]

# File 'lib/rdoc/markup/parser.rb', line 428


def setup_scanner input
  @line     = 0
  @line_pos = 0
  @input    = input.dup

  if @have_encoding and not @have_byteslice then
    @input_encoding = @input.encoding
    @binary_input   = @input.force_encoding Encoding::BINARY
  end

  @s = StringScanner.new input
end

#skip(token_type, error = true)

Skips the next token if its type is token_type.

Optionally raises an error if the next token is not of the expected type.

Raises:

(ParseError)

[ GitHub ]

# File 'lib/rdoc/markup/parser.rb', line 446


def skip token_type, error = true
  type, = get
  return unless type # end of stream
  return @current_token if token_type == type
  unget
  raise ParseError, "expected #{token_type} got #{@current_token.inspect}" if error
end

#token_pos(byte_offset)

Calculates the column (by character) and line of the current token based on byte_offset.

[ GitHub ]

# File 'lib/rdoc/markup/parser.rb', line 542


def token_pos byte_offset
  offset = char_pos byte_offset

  [offset - @line_pos, @line]
end

#tokenize(input)

Turns text input into a stream of tokens

[ GitHub ]

# File 'lib/rdoc/markup/parser.rb', line 457


def tokenize input
  setup_scanner input

  until @s.eos? do
    pos = @s.pos

    # leading spaces will be reflected by the column of the next token
    # the only thing we loose are trailing spaces at the end of the file
    next if @s.scan(/ +/)

    # note: after BULLET, LABEL, etc.,
    # indent will be the column of the next non-newline token

    @tokens << case
               # [CR]LF => :NEWLINE
               when @s.scan(/\r?\n/) then
                 token = [:NEWLINE, @s.matched, *token_pos(pos)]
                 @line_pos = char_pos @s.pos
                 @line += 1
                 token
               # === text => :HEADER then :TEXT
               when @s.scan(/(=+)(\s*)/) then
                 level = @s[1].length
                 header = [:HEADER, level, *token_pos(pos)]

                 if @s[2] =~ /^\r?\n/ then
                   @s.pos -= @s[2].length
                   header
                 else
                   pos = @s.pos
                   @s.scan(/.*/)
                   @tokens << header
                   [:TEXT, @s.matched.sub(/\r$/, ''), *token_pos(pos)]
                 end
               # --- (at least 3) and nothing else on the line => :RULE
               when @s.scan(/(-{3,}) *\r?$/) then
                 [:RULE, @s[1].length - 2, *token_pos(pos)]
               # * or - followed by white space and text => :BULLET
               when @s.scan(/([*-]) +(\S)/) then
                 @s.pos -= @s[2].bytesize # unget \S
                 [:BULLET, @s[1], *token_pos(pos)]
               # A. text, a. text, 12. text => :UALPHA, :LALPHA, :NUMBER
               when @s.scan(/([a-z]|\d+)\. +(\S)/i) then
                 # FIXME if tab(s), the column will be wrong
                 # either support tabs everywhere by first expanding them to
                 # spaces, or assume that they will have been replaced
                 # before (and provide a check for that at least in debug
                 # mode)
                 list_label = @s[1]
                 @s.pos -= @s[2].bytesize # unget \S
                 list_type =
                   case list_label
                   when /[a-z]/ then :LALPHA
                   when /[A-Z]/ then :UALPHA
                   when /\d/    then :NUMBER
                   else
                     raise ParseError, "BUG token #{list_label}"
                   end
                 [list_type, list_label, *token_pos(pos)]
               # [text] followed by spaces or end of line => :LABEL
               when @s.scan(/\[(.*?)\]( +|\r?$)/) then
                 [:LABEL, @s[1], *token_pos(pos)]
               # text:: followed by spaces or end of line => :NOTE
               when @s.scan(/(.*?)::( +|\r?$)/) then
                 [:NOTE, @s[1], *token_pos(pos)]
               # anything else: :TEXT
               else @s.scan(/(.*?)(  )?\r?$/)
                 token = [:TEXT, @s[1], *token_pos(pos)]

                 if @s[2] then
                   @tokens << token
                   [:BREAK, @s[2], *token_pos(pos + @s[1].length)]
                 else
                   token
                 end
               end
  end

  self
end

#unget

Returns the current token to the token stream

Raises:

(Error)

[ GitHub ]

# File 'lib/rdoc/markup/parser.rb', line 551


def unget
  token = @current_token
  p :unget => token if @debug
  raise Error, 'too many #ungets' if token == @tokens.first
  @tokens.unshift token if token
end

Class: RDoc::Markup::Parser

Overview

Constant Summary

::RDoc::Text - Included

Class Method Summary

Instance Attribute Summary

Instance Method Summary

::RDoc::Text - Included

Constructor Details

.new ⇒ Parser

Class Method Details

.parse(str)

.tokenize(str)

Instance Attribute Details

#debug (rw)

#tokens (readonly)

Instance Method Details

#build_heading(level)

#build_list(margin)

#build_paragraph(margin)

#build_verbatim(margin)

#char_pos(byte_offset)

#get

#parse(parent, indent = 0)

#peek_token

#setup_scanner(input)

#skip(token_type, error = true)

#token_pos(byte_offset)

#tokenize(input)

#unget

.new ⇒ `Parser`