notroff 0.2.13 → 0.3.0

data/bin/yesroff

@@ -0,0 +1,30 @@
+#!/usr/bin/env ruby
+require 'pp'
+require "yesroff"
+
+unless ARGV.size == 1
+  raise "Usage: yesroff <<odt_file>>"
+end
+
+
+parser = OdtParser.new(ARGV[0])
+parser.parse
+parser.render
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+

data/lib/notroff/type_assigner.rb

@@ -21,7 +21,7 @@ class TypeAssigner
         processed_paragraphs << paragraph
       end
 
-      current_type = :body if [ :section, :title, :code1 ].include?(type)
+      current_type = :body if [ :section, :sec, :c1, :subsec, :title, :code1 ].include?(type)
     end
     processed_paragraphs
   end

data/lib/yesroff.rb

@@ -0,0 +1,3 @@
+require 'yesroff/nr_writer'
+require 'yesroff/odt_parser'
+require 'yesroff/text'

data/lib/yesroff/nr_writer.rb

@@ -0,0 +1,51 @@
+class NRWriter
+  attr_reader :para_style
+
+  def initialize
+    @para_style = :body
+    @single_line = false
+  end
+
+  def switch_para_style(new_style, single_line)
+    #end_paragraph
+    return if new_style == @para_style
+    print ".#{new_style}"
+    if single_line
+      print ' '
+    else
+      print "\n"
+    end
+    @para_style = new_style
+    @single_line = single_line
+  end
+
+  def end_paragraph
+    puts
+    if @single_line
+      @para_style = :body
+      @single_line = false
+    elsif @para_style != :code && @para_style != :listing
+      puts
+    end
+  end
+
+  def indent(n)
+    print (' ' * n)
+    
+  end
+  def toggle_bold
+    print "!!"
+  end
+
+  def toggle_italic
+    print "~~"
+  end
+
+  def toggle_code
+    print '@@'
+  end
+
+  def text(t)
+    print t
+  end
+end

data/lib/yesroff/odt_parser.rb

@@ -0,0 +1,154 @@
+require 'rexml/document'
+require 'pp'
+require 'zip/zipfilesystem'
+
+class OdtParser
+  def initialize(odt_path)
+    Zip::ZipFile.open(odt_path ) do |zipfile|
+      zipfile.file.open("content.xml") do |content|
+        @doc = REXML::Document.new(content.read)
+      end
+    end
+
+    @writer = NRWriter.new
+    @paras = []
+    @text_styles = default_text_styles
+    @para_styles = default_para_styles
+  end
+
+  def default_text_styles
+    cd1 = TextStyle.new("CD1")
+    cd1.code = true
+
+    styles = [
+      cd1,
+      TextStyle.new("Default"),
+      TextStyle.new("C1"),
+      TextStyle.new("C1_20_HD"),
+      TextStyle.new("FN"),
+      TextStyle.new("Base_20_Font"),
+      TextStyle.new("Chapter_20_Word")
+    ]
+    hash = {}
+    styles.each {|s| hash[s.name] = s}
+    hash
+  end
+
+  def default_para_styles
+    styles = [
+    ParagraphStyle.new('FT', nil, :body, false),
+    ParagraphStyle.new('IT', nil, :body, false),
+    ParagraphStyle.new('Quotation', nil, :quote, true),
+    ParagraphStyle.new('CDT1', nil, :code, false),
+    ParagraphStyle.new('CDT', nil, :code, false),
+    ParagraphStyle.new('HA', nil, :title, true),
+    ParagraphStyle.new('HB', nil, :subtitle, true),
+    ParagraphStyle.new('HC', nil, :sec, true),
+    ParagraphStyle.new('HD', nil, :subsec, true),
+    ParagraphStyle.new('LH', nil, :ltitle, true),
+    ParagraphStyle.new('LC', nil, :listing, false),
+    ParagraphStyle.new('LC2', nil, :listing, false),
+    ParagraphStyle.new('LX', nil, :listing, false),
+    ParagraphStyle.new('C1', nil, :c1, true),
+    ParagraphStyle.new('BL1', nil, :bullet, true),
+    ParagraphStyle.new('BL', nil, :bullet, true),
+    ParagraphStyle.new('BX', nil, :bullet, true),
+    ParagraphStyle.new('Quotation_20_Attribution', nil, :attribution, true),
+    ParagraphStyle.new('CDTX', nil, :code, false)
+    ]
+
+    hash = {}
+    styles.each {|s| hash[s.name] = s}
+    hash
+  end
+
+  def parse
+    parse_text_styles
+    parse_paragraph_styles
+    @paras = parse_paragraphs
+  end
+
+  def render
+    @paras.each {|p| p.render(@writer)}
+  end
+
+  def parse_text_styles
+    styles = REXML::XPath.match(@doc, "//style:style[@style:family='text']")
+    styles.each do |s|
+      attrs = s.attributes
+      style = TextStyle.new(attrs['style:name'])
+      props = REXML::XPath.first(s, "./style:text-properties")
+      if props
+        style.bold = (props.attributes['fo:font-weight'] == 'bold')
+        style.italic = (/italic/i =~ props.attributes['style:font-name']) ||
+          (props.attributes['fo:font-style'] == 'italic')
+      end
+      @text_styles[style.name] = style
+    end
+  end
+
+  def parse_paragraph_styles
+    styles = REXML::XPath.match(@doc, "//style:style[@style:family='paragraph']")
+    styles.each do |s|
+      attrs = s.attributes
+      style = ParagraphStyle.new(attrs['style:name'])
+      style.parent = lookup_para_style(attrs['parent-style-name'])
+      @para_styles[style.name] = style
+    end
+  end
+
+  def parse_paragraphs
+    results = []
+    paras = REXML::XPath.match(@doc, '//text:p')
+    paras.each do |p|
+      results << parse_paragraph(p)
+    end
+    results
+  end
+
+  def lookup_para_style(name)
+    s = @para_styles[name]
+    raise "No such para style #{name}" unless s
+    s
+  end
+
+  def lookup_text_style(name)
+    name = "Default" if name.nil? or name.empty?
+    s = @text_styles[name]
+    raise "No such text style [[#{name}]]" unless s
+    s
+  end
+
+  def parse_paragraph(p)
+    attrs = p.attributes
+#    puts "Parsing paragraph, attrs #{attrs}"
+#    puts "==> style-name: [[#{attrs['text:style-name']}]]"
+    style = lookup_para_style(attrs['text:style-name'])
+
+    para = Paragraph.new(style)
+    para.contents = parse_contents(para, p)
+    para
+  end
+
+  def parse_span(el)
+    attrs = el.attributes
+    style = lookup_text_style(attrs['text:style-name'])
+    indent = attrs['text:c'] ? attrs['text:c'].to_i : 0
+    span = Span.new(style)
+    span.indent = indent
+    span.contents = parse_contents(span, el)
+    span
+  end
+
+  def parse_contents(contents, el)
+    results = []
+    el.each_child do |kid|
+      if REXML::Text === kid
+        results << Text.new(REXML::Text.unnormalize(kid.value))
+      else
+        results << parse_span(kid)
+      end
+    end
+    results
+  end
+end

data/lib/yesroff/text.rb

@@ -0,0 +1,110 @@
+
+class Style
+  attr_accessor :name, :parent
+
+
+  def initialize(name, parent = nil)
+    @name = name
+    @parent = parent
+  end  
+
+  def child_of?(name)
+    return true if name == @name
+    return false unless @parent
+    @parent.child_of?(name)
+  end
+
+  def base_style
+    return self unless @parent
+    @parent.base_style
+  end
+end
+
+class TextStyle < Style
+  attr_accessor :bold, :italic, :code
+
+  def to_s
+    result = "[Style: #{name}"
+    result += 'B' if @bold
+    result += 'I' if @italic
+    result += 'C' if @code
+    result += ']'
+  end
+
+  def render(w)
+    w.toggle_bold if @bold
+    w.toggle_code if @code
+    w.toggle_italic if @italic
+  end
+end
+
+
+class ParagraphStyle < Style
+  attr_accessor :nr_name, :single_line
+
+  def initialize(name, parent=nil, nr_name='', single_line=true)
+    super(name, parent)
+    @nr_name = nr_name
+    @single_line = single_line
+  end
+
+  def render(w)
+    base = self.base_style
+    w.switch_para_style(base.nr_name, base.single_line)
+  end
+end
+
+class Container
+  attr_accessor :style, :parent, :contents
+
+  def initialize(style, parent=nil, contents=[])
+    @style = style
+    @contents = contents
+  end
+
+  def <<(content)
+    @contents << content
+  end
+
+  def render(w)
+    #puts "========= rendering #{self.class} size: #{contents.size}"
+    #pp contents
+    #puts "Rendering:"
+    contents.each {|c|  c.render(w)}
+  end
+end
+
+class Span < Container
+  attr_accessor :indent
+
+  def render(w)
+    style.render(w)
+    w.indent(@indent) if @indent
+    super(w)
+    style.render(w)
+  end
+end
+
+class Paragraph < Container
+  def render(w)
+    style.render(w)
+    super(w)
+    w.end_paragraph
+  end
+end
+
+class Text
+  attr_accessor :text
+
+  def initialize(t)
+    @text = t
+  end
+
+  def render(w)
+    w.text(@text)
+  end
+
+  def to_s
+    "{{Text: #{text.class} : #{text}}}"
+  end
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: notroff
 version: !ruby/object:Gem::Version
-  version: 0.2.13
+  version: 0.3.0
   prerelease: 
 platform: ruby
 authors:
@@ -9,12 +9,13 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-01-15 00:00:00.000000000Z
+date: 2012-09-03 00:00:00.000000000Z
 dependencies: []
 description: NotRoff A simple text to openoffice filter
 email: russ@russolsen.com
 executables:
 - notroff
+- yesroff
 extensions: []
 extra_rdoc_files: []
 files:
@@ -58,7 +59,12 @@ files:
 - lib/notroff/type_assigner.rb
 - lib/notroff/type_refiner.rb
 - lib/notroff.rb
+- lib/yesroff/nr_writer.rb
+- lib/yesroff/odt_parser.rb
+- lib/yesroff/text.rb
+- lib/yesroff.rb
 - bin/notroff
+- bin/yesroff
 homepage: http://www.russolsen.com
 licenses: []
 post_install_message: