class RDoc::Parser

A parser is simple a class that subclasses RDoc::Parser and implements scan to fill in an RDoc::TopLevel with parsed data.

The initialize method takes an RDoc::TopLevel to fill with parsed content, the name of the file to be parsed, the content of the file, an RDoc::Options object and an RDoc::Stats object to inform the user of parsed items. The scan method is then called to parse the file and must return the RDoc::TopLevel object. By calling super these items will be set for you.

In order to be used by RDoc the parser needs to register the file extensions it can parse. Use ::parse_files_matching to register extensions.

require 'rdoc'

class RDoc::Parser::Xyz < RDoc::Parser
  parse_files_matching /\.xyz$/

  def initialize top_level, file_name, content, options, stats
    super

    # extra initialization if needed
  end

  def scan
    # parse file and fill in @top_level
  end
end

Attributes

parsers [R]

An Array of arrays that maps file extension (or name) regular expressions to parser classes that will parse matching filenames.

Use parse_files_matching to register a parser’s file extensions.

file_name [R]

The name of the file being parsed

Public Class Methods

alias_extension (old_ext, new_ext)

Alias an extension to another extension. After this call, files ending “new_ext” will be parsed using the same parser as “old_ext”

# File lib/rdoc/parser.rb, line 58
def self.alias_extension(old_ext, new_ext)
  old_ext = old_ext.sub(/^\.(.*)/, '\1')
  new_ext = new_ext.sub(/^\.(.*)/, '\1')

  parser = can_parse_by_name "xxx.#{old_ext}"
  return false unless parser

  Parser.parsers.unshift [/\.#{new_ext}$/, parser]

  true
end

binary? (file)

Determines if the file is a “binary” file which basically means it has content that an RDoc parser shouldn’t try to consume.

# File lib/rdoc/parser.rb, line 74
def self.binary?(file)
  return false if file =~ /\.(rdoc|txt)$/

  s = File.read(file, 1024) or return false

  return true if s[0, 2] == Marshal.dump('')[0, 2] or s.index("\x00")

  mode = 'r:utf-8' # default source encoding has been changed to utf-8
  s.sub!(/\A#!.*\n/, '')     # assume shebang line isn't longer than 1024.
  encoding = s[/^\s*\#\s*(?:-\*-\s*)?(?:en)?coding:\s*([^\s;]+?)(?:-\*-|[\s;])/, 1]
  mode = "rb:#{encoding}" if encoding
  s = File.open(file, mode) {|f| f.gets(nil, 1024)}

  not s.valid_encoding?
end

can_parse (file_name)

Return a parser that can handle a particular extension

# File lib/rdoc/parser.rb, line 107
def self.can_parse(file_name)
  parser = can_parse_by_name file_name

  # HACK Selenium hides a jar file using a .txt extension
  return if parser == Parser::Simple and zip? file_name

  parser
end

can_parse_by_name (file_name)

Returns a parser that can handle the extension for file_name. This does not depend upon the file being readable.

# File lib/rdoc/parser.rb, line 120
def self.can_parse_by_name(file_name)
  _, parser = Parser.parsers.find { |regexp,| regexp =~ file_name }

  # The default parser must not parse binary files
  ext_name = File.extname file_name
  return parser if ext_name.empty?

  if parser == Parser::Simple and ext_name !~ /txt|rdoc/
    case mode = check_modeline(file_name)
    when nil, 'rdoc' # continue
    else
      Parser.parsers.find { |_, p| return p if mode.casecmp?(p.name[/\w+\z/]) }
      return nil
    end
  end

  parser
rescue Errno::EACCES
end

check_modeline (file_name)

Returns the file type from the modeline in file_name

# File lib/rdoc/parser.rb, line 143
def self.check_modeline(file_name)
  line = File.open file_name do |io|
    io.gets
  end

  /-\*-\s*(.*?\S)\s*-\*-/ =~ line

  return nil unless type = $1

  if /;/ =~ type
    return nil unless /(?:\s|\A)mode:\s*([^\s;]+)/i =~ type
    type = $1
  end

  return nil if /coding:/i =~ type

  type.downcase
rescue ArgumentError
rescue ::Encoding::InvalidByteSequenceError # invalid byte sequence

end

for (top_level, content, options, stats)

Finds and instantiates the correct parser for the given file_name and content.

# File lib/rdoc/parser.rb, line 169
def self.for(top_level, content, options, stats)
  file_name = top_level.absolute_name
  return if binary? file_name

  parser = use_markup content

  unless parser
    parse_name = file_name

    # If no extension, look for shebang
    if file_name !~ /\.\w+$/ && content =~ %r{\A#!(.+)}
      shebang = $1
      case shebang
      when %r{env\s+ruby}, %r{/ruby}
        parse_name = 'dummy.rb'
      end
    end

    parser = can_parse parse_name
  end

  return unless parser

  content = remove_modeline content

  parser.new top_level, content, options, stats
rescue SystemCallError
  nil
end

new (top_level, content, options, stats)

Creates a new Parser storing top_level, file_name, content, options and stats in instance variables. In +@preprocess+ an RDoc::Markup::PreProcess object is created which allows processing of directives.

# File lib/rdoc/parser.rb, line 261
def initialize(top_level, content, options, stats)
  @top_level = top_level
  @top_level.parser = self.class
  @store = @top_level.store

  @file_name = top_level.absolute_name
  @content = content
  @options = options
  @stats = stats

  @preprocess = Markup::PreProcess.new @file_name, @options.rdoc_include
  @preprocess.options = @options
end

parse_files_matching (regexp)

Record which file types this parser can understand.

It is ok to call this multiple times.

# File lib/rdoc/parser.rb, line 204
def self.parse_files_matching(regexp)
  Parser.parsers.unshift [regexp, self]
end

remove_modeline (content)

Removes an emacs-style modeline from the first line of the document

# File lib/rdoc/parser.rb, line 211
def self.remove_modeline(content)
  content.sub(/\A.*-\*-\s*(.*?\S)\s*-\*-.*\r?\n/, '')
end

use_markup (content)

If there is a markup: parser_name comment at the front of the file, use it to determine the parser. For example:

# markup: rdoc
# Class comment can go here

class C
end

The comment should appear as the first line of the content.

If the content contains a shebang or editor modeline the comment may appear on the second or third line.

Any comment style may be used to hide the markup comment.

The tomdoc and markdown markups name comment formats rather than parsers, so no parser is selected for them and RDoc::Parser.for picks one from the file name instead.

# File lib/rdoc/parser.rb, line 236
def self.use_markup(content)
  markup = content.lines.first(3).grep(/markup:\s+(\w+)/) { $1 }.first

  return unless markup

  # tomdoc and markdown name a comment format, not a parser.  Skipping them
  # keeps the search below from matching RDoc::Parser::Markdown for a file
  # that is not markdown.
  return if %w[tomdoc markdown].include? markup

  markup = Regexp.escape markup

  _, selected = Parser.parsers.find do |_, parser|
    /^#{markup}$/i =~ parser.name.sub(/.*:/, '')
  end

  selected
end

zip? (file)

Checks if file is a zip file in disguise. Signatures from www.garykessler.net/library/file_sigs.html

# File lib/rdoc/parser.rb, line 94
def self.zip?(file)
  zip_signature = File.read file, 4

  zip_signature == "PK\x03\x04" or
    zip_signature == "PK\x05\x06" or
    zip_signature == "PK\x07\x08"
rescue
  false
end

Public Instance Methods

handle_tab_width (body)

Normalizes tabs in body

# File lib/rdoc/parser.rb, line 280
def handle_tab_width(body)
  if /\t/ =~ body
    tab_width = @options.tab_width
    body.split(/\n/).map do |line|
      1 while line.gsub!(/\t+/) do
        b, e = $~.offset(0)
        ' ' * (tab_width * (e-b) - b % tab_width)
      end
      line
    end.join "\n"
  else
    body
  end
end