Adding docbook support.

Rakefile

@@ -11,7 +11,7 @@ end
 
 desc "Run notroff"
 task :run do |t|
-  sh "ruby -Ilib -rnotroff bin/notroff test.nr out.html"
+  sh "ruby -Ilib -rnotroff bin/notroff -d test.nr test.xml"
 end
 
 

bin/notroff

@@ -7,19 +7,16 @@ options={}
 formatter_class = HtmlFormatter
 
 while true
-  if ARGV.first == '-fn'
-    options[:fn] = true
-    ARGV.shift
-  elsif ARGV.first == '-html'
-    formatter_class = HtmlFormatter
+  if ARGV.first == '-d'
+    formatter_class = DocbookFormatter
     ARGV.shift
   else
     break
   end
 end
 
 unless ARGV.size == 2
-  raise "Usage: notroff <nr file> <output file>"
+  raise "Usage: notroff [-d] <nr file> <output file>"
 end
 
 

lib/notroff.rb

@@ -11,11 +11,8 @@
 require "notroff/processor"
 require "notroff/command_processor"
 require "notroff/composite_processor"
-#require "notroff/processor"
 require "notroff/tokenize"
-#require "notroff/odt_renderer"
 require "notroff/html_renderer"
-#require "notroff/odt_replacer"
-#require "notroff/template_expander"
+require "notroff/docbook_renderer"
 require "notroff/formatter"
 require "notroff/filter"

lib/notroff/composite_processor.rb

@@ -16,12 +16,21 @@ def process(paras=[])
       puts "Applying processor #{processor.class} to #{paras.size} paragraphs"
       paras = processor.process( paras )
       puts "After processor #{processor.class}"
-      puts "Now have #{paras.size} paragraphs" if paras
-      p paras
-      puts "========="
+      dump(paras)
     end
     paras
   end
+
+  def dump(array)
+    if array.nil?
+      puts "Array: #{array}"
+    else
+      puts "======="
+      puts "Array with #{array.size} items:"
+      array.each_with_index {|item, i| puts "[#{i}] - #{item}" }
+      puts "======="
+    end
+  end
 end
 
 

lib/notroff/docbook_renderer.rb

@@ -0,0 +1,155 @@
+require 'rexml/document'
+require 'pp'
+
+class DocbookRenderer < Processor
+  include Tokenize
+  include REXML
+
+  def process( paragraphs )
+    @author = 'John Smith'
+    @title = 'Wealth of Nations'
+    @chapters = []
+    @chapter = nil
+    @section = nil
+
+    paragraphs.each do |paragraph|
+      format( paragraph )
+    end
+    @book = Element.new('book')
+    @book.add_namespace('http://docbook.org/ns/docbook')
+    @book.add_attribute('version', '5.0')
+    @book.add_element element_for('title', @title)
+    @book.add_element element_for('author', @author)
+    @chapters.each {|ch| @book << ch}
+    doc = Document.new
+    decl = XMLDecl.new
+    decl.version = '1.0'
+    doc << decl
+    doc << @book
+    doc
+  end
+
+  def element_for(type, text)
+    result = Element.new(type)
+    result.add_text(text)
+    result
+  end
+
+  def format( p )
+    type = p[:type]
+    text = p.string
+
+    return nil if text.empty? and :type != :code
+
+    case type
+    when :title
+      @title = text
+    when :author
+      @author = text
+    when :chapter
+      puts "adding chapter #{text}"
+      @chapter = Element.new('chapter')
+      @chapters << @chapter
+      @section = nil
+      title_element = Element.new('title')
+      add_body_text(title_element, text)
+      @chapter.add_element(title_element)
+    when :section
+      puts "adding section #{text}"
+      @section = Element.new('section')
+      @chapter.add(@section)
+      title_element = Element.new('title')
+      add_body_text(title_element, text)
+      @section.add_element(title_element)
+    when :body
+      puts "adding body #{text[0..5]}"
+      paragraph = Element.new('para')
+      add_body_text(paragraph, text)
+      add_content_element(paragraph)
+    when :code
+      add_content_element(code_element(type, text))
+    else
+      raise "#{type}???"
+    end
+  end
+
+  def add_content_element(el)
+    if @section
+      @section.add_element(el)
+    elsif @chapter
+      @chapter.add_element(el)
+    else
+      raise "No chapter to add #{el} to"
+    end
+  end
+
+  def text_element(type, text)
+    element = [ tag_for(type) ]
+    add_body_text(element, text)
+    element
+  end
+
+  def tag_for(type)
+    case type
+    when :body
+      'para'
+    when :text
+      'p'
+    when :author
+      'h3'
+    when :section
+      'h3'
+    when :chapter
+      'chapter'
+    when :sec
+      'h3'
+    when :title
+      'h2'
+    else
+        raise "Dont know what to do with #{type}"
+    end
+  end
+
+  def add_body_text( element, text )
+    tokens = tokenize_body_text( text )
+    tokens.each {|token| add_span( token, element ) }
+  end
+
+  def add_span( token, element )
+    puts "Add span: token: #{token} element: #{element}"
+    case token[:type]
+    when :italic
+      element.add( span_for( token.string, "emphasis" ))
+    when :code
+      element.add( span_for( token.string, "code" ))
+    when :bold
+      element.add( span_for( token.string, "emphasis" ))
+    when :normal
+      element.add_text( token.string )
+    when :footnote
+      element.add(footnote_for(token.string))
+    else
+      raise "Dont know what to do with #{token}"
+    end
+  end
+
+  def footnote_for( text )
+    fn = Element.new('footnote')
+    fn.add_element(element_for('para', text))
+    fn
+  end
+
+  def code_element(type, text)
+    element = Element.new('informalexample')
+    prog_element = element_for('programlisting', text)
+    prog_element.add_attribute('xml:space', 'preserve')
+    element.add_element(prog_element)
+    element
+  end
+
+  def span_for( text, style )
+    span = Element.new(style)
+    span.text = remove_escapes(text)
+    span
+  end
+end

lib/notroff/embedded.rb

@@ -1,5 +1,5 @@
 class BetweenFilter
-  def initialize(re1, re2)
+  def initialize(re1, re2=re1)
     @re1 = re1
     @re2 = re2
   end
@@ -10,6 +10,7 @@ def process(paras)
       if state == :before_first and @re1 =~ para
         para[:included] = true
         state = :after_first
+        break if para =~ @re2
       elsif state == :after_first
         para[:included] = true
         break if para =~ @re2
@@ -57,6 +58,12 @@ def initialize(class_name, include_body)
   end
 end
 
+class ModuleFilter < DefinitionFilter
+  def initialize(module_name, include_body)
+    super /^ *module +#{module_name}(\(|$| )/, include_body
+  end
+end
+
 class EmbeddedRubyProcessor
   def process(paragraphs)
     new_paragraphs = []
@@ -80,6 +87,7 @@ def process_command(ruby_expression)
   def embed(*filters, &block)
     paras = block.call.map {|line| line.rstrip}
     paras.map! {|p| Text.new(p, :type => :code)}
+    puts "EMBED: #{paras}"
     unless filters.empty?
       filters.each {|f| paras = f.process(paras)}
       paras = paras.find_all {|p| p[:included]}
@@ -95,7 +103,7 @@ def run(command, *filters)
     embed(*filters) {File.popen(command).readlines}
   end
 
-  def between(re1, re2)
+  def matches(re1, re2=re1)
     BetweenFilter.new(re1, re2)
   end
 
@@ -107,6 +115,10 @@ def clazz(name, include_body=true)
     ClassFilter.new(name, include_body)
   end
 
+  def mod(name, include_body=true)
+    ModuleFilter.new(name, include_body)
+  end
+
   def indent(delta_indent, paragraphs)
     paragraphs.map do |p| 
       if delta_indent > 0

lib/notroff/formatter.rb

@@ -19,6 +19,16 @@ def initialize(input, output)
   end
 end
 
+class DocbookFormatter < Formatter
+  def initialize(input, output)
+    super()
+    prepend_processor FileReader.new(input)
+    add_processor CodeParagraphJoiner.new
+    add_processor DocbookRenderer.new
+    add_processor FileWriter.new(output)
+  end
+end
+
 
 
 

lib/notroff/paragraph_joiner.rb

@@ -42,7 +42,7 @@ def join?(paragraph)
   end
 
   def skip?(paragraph)
-    paragraph.empty?
+    paragraph[:type] == :body && paragraph.empty?
   end
 end
 

test.nr

@@ -0,0 +1,41 @@
+.title The Roots of Ruby
+.author Fogus and Russ
+.chapter The Way of the World
+AAAAA In more
+
+this is ~~italics~~ and
+
+this is a footnote {{foot notes}} and
+
+this is @@code@@ and
+
+this is  !!bold!! is book.
+All DocBook elements are in an XML Namespace, so the root element has an xmlns attribute to set the current namespace. Also, the root element of a DocBook document must have a version that specifies the version of the format that the document is built on.
+
+.section A section
+In more detail, the root element of the document is book. All DocBook elements are in an XML Namespace, so the root element has an xmlns attribute to set the current namespace. Also, the root element of a DocBook document must have a version that specifies the version of the format that the document is built on.
+Some code
+.code
+
+# A comment surrouded by white space
+
+
+# Another comment
+
+
+
+.body
+body line
+.chapter A second chapter
+More intro text
+.section Section 1
+section 1 text
+more body
+text
+.code
+foo = bar
+bar = foo
+.body
+still more body text
+.section Two
+section 2 text