ydocx 1.0.3 → 1.0.4

data/History.txt

@@ -1,3 +1,13 @@
+=== 1.0.4 / 02.05.2012
+
+* Fixed bug for action to_xml in command
+* Updated help message
+* Added option handling into command
+* Improved heading/chapter detection
+* Added docx2xml/to_xml simple xml format output
+* Changed module and lib structure
+* Updated executable file
+
 === 1.0.3 / 01.05.2012
 
 * Fixed blank tag without fachinfo format

data/Manifest.txt

@@ -3,11 +3,12 @@ Manifest.txt
 README.txt
 Rakefile
 bin/docx2html
-lib/docx2html/bin/docx2html
-lib/docx2html/lib/docx2html.rb
-lib/docx2html/lib/docx2html/builder.rb
-lib/docx2html/lib/docx2html/document.rb
-lib/docx2html/lib/docx2html/html_methods.rb
-lib/docx2html/lib/docx2html/parser.rb
-lib/fachinfo.rb
+bin/docx2xml
 lib/version.rb
+lib/ydocx.rb
+lib/ydocx/builder.rb
+lib/ydocx/document.rb
+lib/ydocx/markup_method.rb
+lib/ydocx/parser.rb
+lib/ydocx/command.rb
+lib/ydocx/templates/fachinfo.rb

data/bin/docx2html

@@ -4,28 +4,9 @@
 if $0 == __FILE__
   require 'pathname'
   root = Pathname.new(__FILE__).realpath.parent.parent
-  %w(
-    lib/docx2html/lib
-    lib
-  ).each do |lib|
-    $:.unshift root.join(lib)
-  end
-  require 'docx2html'
-else
-  require 'docx2html/lib/docx2html'
+  $:.unshift root.join('lib')
 end
 
-if file = ARGV.first
-  unless File.exist?(file)
-    exit
-  else
-    # TODO args handling
-    if true
-      require 'fachinfo'
-    end
-    options = {
-      :style => :frame
-    }
-    Docx2html::Document.open(file).to_html file, options
-  end
-end
+require 'ydocx/command'
+
+YDocx::Command.run(:to_html)

data/bin/docx2xml

@@ -0,0 +1,12 @@
+#!/usr/bin/env ruby
+# coding: utf-8
+
+if $0 == __FILE__
+  require 'pathname'
+  root = Pathname.new(__FILE__).realpath.parent.parent
+  $:.unshift root.join('lib')
+end
+
+require 'ydocx/command'
+
+YDocx::Command.run(:to_xml)

data/lib/version.rb

@@ -2,5 +2,5 @@
 # encoding: utf-8
 
 module Docx2html
-  VERSION = "1.0.3"
+  VERSION = "1.0.4"
 end

data/lib/ydocx/builder.rb

@@ -0,0 +1,144 @@
+#!/usr/bin/env ruby
+# encoding: utf-8
+
+require 'nokogiri'
+require 'ydocx/markup_method'
+
+module YDocx
+  class Builder
+    include MarkupMethod
+    attr_accessor :contents, :container, :indecies,
+                  :style, :title
+    def initialize(contents)
+      @contents = contents
+      @container = {}
+      @indecies = []
+      @style = false
+      @title = ''
+      init
+      if block_given?
+        yield self
+      end
+    end
+    def init
+    end
+    def build_html
+      contents = @contents
+      body = compile(contents, :html)
+      if @container.has_key?(:content)
+        body = build_tag(@container[:tag], body, @container[:attributes])
+      end
+      if before = build_before_content
+        body = build_tag(before[:tag], before[:content], before[:attributes]) << body
+      end
+      if after = build_after_content
+        body << build_tag(after[:tag], after[:content], after[:attributes])
+      end
+      builder = Nokogiri::HTML::Builder.new do |doc|
+        doc.html {
+          doc.head {
+            doc.meta :charset => 'utf-8'
+            doc.title @title
+            doc.style { doc << style } if @style
+          }
+          doc.body { doc << body }
+        }
+      end
+      builder.to_html.gsub(/\n/, '')
+    end
+    def build_xml
+      chapters = compile(@contents, :xml) 
+      builder = Nokogiri::XML::Builder.new do |xml|
+        xml.document {
+          xml.chapters { xml << chapters }
+        }
+      end
+      builder.to_xml(:indent => 0, :encoding => 'utf-8').gsub(/\n/, '')
+    end
+    private
+    def compile(contents, mode)
+      if mode == :xml
+        block_tag = :chapter
+      else
+        block_tag = :div
+      end
+      result = ''
+      headings = 0
+      contents.each do |element|
+        if element[:tag].to_s =~ /^h[1-9]$/ # block
+          if headings == 0
+            result << "<#{block_tag}>"
+          else
+            result << "</#{block_tag}><#{block_tag}>"
+          end
+          headings += 1
+        end
+        result << build_tag(element[:tag], element[:content], element[:attributes], mode)
+      end
+      result << "</#{block_tag}>"
+    end
+    def build_after_content
+      nil
+    end
+    def build_before_content
+      nil
+    end
+    def build_tag(tag, content, attributes, mode=:html)
+      if tag == :br and mode != :xml
+        return "<br/>"
+      elsif content.nil? or content.empty?
+        return ''
+      end
+      _content = ''
+      if content.is_a? Array
+        content.each do |c|
+          next if c.nil? or c.empty?
+          if c.is_a? Hash
+            _content << build_tag(c[:tag], c[:content], c[:attributes], mode)
+          elsif c.is_a? String
+            _content << c.chomp.to_s
+          end
+        end
+      elsif content.is_a? Hash
+        _content = build_tag(content[:tag], content[:content], content[:attributes], mode)
+      elsif content.is_a? String
+        _content = content
+      end
+      _tag = tag.to_s
+      _attributes = ''
+      unless attributes.empty?
+        attributes.each_pair do |key, value|
+          next if mode == :xml and key.to_s =~ /(id|style|colspan)/u
+          _attributes << " #{key.to_s}=#{value.to_s}"
+        end
+      end
+      if mode == :xml
+        case _tag.to_sym
+        when :span    then _tag = 'underline'
+        when :strong  then _tag = 'bold'
+        when :em      then _tag = 'italic'
+        when :p       then _tag = 'paragraph'
+        when :h2, :h3 then _tag = 'heading'
+        when :sup     then _tag = 'superscript' # text
+        when :sub     then _tag = 'subscript'   # text
+        end
+      end
+      return "<#{_tag}#{_attributes}>#{_content}</#{_tag}>"
+    end
+    def style
+      style = <<-CSS
+table, tr, td {
+  border-collapse: collapse;
+  border:          1px solid gray;
+}
+table {
+  margin: 5px 0 5px 0;
+}
+td {
+  padding: 5px 10px;
+}
+      CSS
+      style.gsub(/\s\s+|\n/, ' ')
+    end
+  end
+end

data/lib/ydocx/command.rb

@@ -0,0 +1,73 @@
+#!/usr/bin/env ruby
+# encoding: utf-8
+
+require 'ydocx/document'
+
+module YDocx
+  class Command
+    class << self
+      @@help = /^\-(h|\-help)$/u
+      @@format = /^\-(f|\-format)$/u
+      def error(message='')
+        puts message
+        puts "see `#{self.command} --help`"
+        exit
+      end
+      def command
+        File.basename $0
+      end
+      def help
+        banner = <<-BANNER
+Usage: #{$0} file [options]
+  -f, --format    Format of style and chapter {fi|fachinfo}, default none.
+  -h, --help      Display this help message.
+        BANNER
+        puts banner
+        exit
+      end
+      def report(action, path)
+        dir = File.dirname path
+        base = File.basename path, '.docx'
+        ext = (action == :to_xml) ? '.xml' : '.html'
+        puts "#{self.command}: generated #{dir}/#{base}#{ext}"
+        exit
+      end
+      def run(action=:to_html)
+        argv = ARGV.dup
+        if argv.empty? or argv[0] =~ @@help
+          self.help
+        else
+          file = argv.shift
+          path = File.expand_path(file)
+          if !File.exist?(path)
+            self.error "#{self.command}: cannot open #{file}: No such file"
+          elsif !File.extname(path).match(/^\.docx$/)
+            self.error "#{self.command}: cannot open #{file}: Not a docx file"
+          else
+            options = {}
+            if option = argv.shift
+              if option =~ @@format
+                case argv[0]
+                when 'fi', 'fachinfo'
+                  require 'ydocx/templates/fachinfo'
+                  # TODO style option?
+                  options.merge!({:style => :frame}) if action == :to_html
+                when 'pi', 'patinfo'
+                  # pending
+                else
+                  self.error "#{self.command}: exit with #{option}: Invalid argument"
+                end
+              elsif option =~ @@help
+                self.help
+              else
+                self.error "#{self.command}: exit with #{option}: Unknown option"
+              end
+            end
+            YDocx::Document.open(path).send(action, path, options)
+            self.report action, path
+          end
+        end
+      end
+    end
+  end
+end

data/lib/{docx2html/lib/docx2html → ydocx}/document.rb

@@ -3,12 +3,12 @@
 
 require 'pathname'
 require 'zip/zip'
-require 'docx2html/lib/docx2html/parser'
-require 'docx2html/lib/docx2html/builder'
+require 'ydocx/parser'
+require 'ydocx/builder'
 
-module Docx2html
+module YDocx
   class Document
-    attr_reader :contents
+    attr_reader :contents, :indecies
     def self.open(file)
       self.new(file)
     end
@@ -25,7 +25,7 @@ module Docx2html
         if @indecies
           builder.indecies = @indecies
         end
-        html = builder.build
+        html = builder.build_html
       end
       unless file.empty?
         path = Pathname.new(file).realpath.sub_ext('.html')
@@ -36,6 +36,20 @@ module Docx2html
         html
       end
     end
+    def to_xml(file='', options={})
+      xml = ''
+      Builder.new(@contents) do |builder|
+        xml = builder.build_xml
+      end
+      unless file.empty?
+        path = Pathname.new(file).realpath.sub_ext('.xml')
+        File.open(path, 'w:utf-8') do |f|
+          f.puts xml
+        end
+      else
+        xml
+      end
+    end
     private
     def read(file)
       @path = File.expand_path(file)

data/lib/{docx2html/lib/docx2html/html_methods.rb → ydocx/markup_method.rb}

@@ -1,9 +1,9 @@
 #!/usr/bin/env ruby
 # encoding: utf-8
 
-module Docx2html
-  module HtmlMethods
-    def tag(tag, content = [], attributes = {})
+module YDocx
+  module MarkupMethod
+    def markup(tag, content = [], attributes = {})
       tag_hash = {
         :tag        => tag,
         :content    => content,

data/lib/{docx2html/lib/docx2html → ydocx}/parser.rb

@@ -3,11 +3,11 @@
 
 require 'nokogiri'
 require 'htmlentities'
-require 'docx2html/lib/docx2html/html_methods'
+require 'ydocx/markup_method'
 
-module Docx2html
+module YDocx
   class Parser
-    include HtmlMethods
+    include MarkupMethod
     attr_accessor :indecies, :result, :space
     def initialize(stream)
       @xml = Nokogiri::XML.parse(stream)
@@ -63,9 +63,17 @@ module Docx2html
       unless rpr.xpath('w:vertAlign').empty?
         script = rpr.xpath('w:vertAlign').first['val'].to_sym
         if script == :subscript
-          text = tag(:sub, text)
+          if text =~ /^[0-9]$/
+            text = "&sub" + text + ";"
+          else
+            text = markup(:sub, text)
+          end
         elsif script == :superscript
-          text = tag(:sup, text)
+          if text =~ /^[0-9]$/
+            text = "&sup" + text + ";"
+          else
+            text = markup(:sup, text)
+          end
         end
       end
       text
@@ -138,8 +146,7 @@ module Docx2html
       else
         #p "code : " + ("&#%s;" % code)
         #p "hex  : " + code.hex.to_s
-        #p "char : " + @coder.decode("&#%s;" % code)
-        #@coder.decode("&#%s;" % code.hex.to_s)
+        #p "char : " + @coder.decode("&#%s;" % code.hex.to_s)
       end
     end
     def parse_image
@@ -173,7 +180,7 @@ module Docx2html
           c.is_a?(Hash) and c[:tag].to_s =~ /^h[1-9]/u
         end.empty?
         if paragraph
-          tag :p, content
+          markup :p, content
         else
           content.first
         end
@@ -182,9 +189,9 @@ module Docx2html
       end
     end
     def parse_table(node)
-      table = tag :table
+      table = markup :table
       node.xpath('w:tr').each do |tr|
-        cells = tag :tr
+        cells = markup :tr
         tr.xpath('w:tc').each do |tc|
           attributes = {}
           tc.xpath('w:tcPr').each do |tcpr|
@@ -192,7 +199,7 @@ module Docx2html
               attributes[:colspan] = span.first['val'] # w:val
             end
           end
-          cell = tag :td, [], attributes
+          cell = markup :td, [], attributes
           tc.xpath('w:p').each do |p|
             cell[:content] << parse_paragraph(p)
           end
@@ -214,13 +221,13 @@ module Docx2html
           text = text.strip
           text = apply_align(rpr, text)
           unless rpr.xpath('w:u').empty?
-            text = tag(:span, text, {:style => "text-decoration:underline;"})
+            text = markup(:span, text, {:style => "text-decoration:underline;"})
           end
           unless rpr.xpath('w:i').empty?
-            text = tag(:em, text) 
+            text = markup(:em, text) 
           end
           unless rpr.xpath('w:b').empty?
-            text = tag(:strong, text)
+            text = markup(:strong, text)
           end
           text
         end

data/lib/{fachinfo.rb → ydocx/templates/fachinfo.rb}

@@ -3,19 +3,10 @@
 
 require 'cgi'
 
-module Docx2html
+module YDocx
   class Parser
     private
-    def escape(text)
-      CGI.escape(text.gsub(/&(.)uml;/, '\1').gsub(/\s*\/\s*|\/|\s+/, '_').downcase)
-    end
     def parse_as_block(r, text)
-      if r.parent.previous.nil? and @indecies.empty?
-        # The first line as package name
-        id = escape('titel')
-        @indecies << {:text => 'Titel', :id => id}
-        return tag(:h2, text, {:id => id})
-      end
       text = text.strip
       # TODO
       # Franzoesisch
@@ -42,18 +33,24 @@ module Docx2html
         'Zusammens.'          => /^Zusammensetzung($|\s*\/\s*(Wirkstoffe|Hilsstoffe)$)/u, # 2
       }.each_pair do |chapter, regexp|
         if text =~ regexp
-          next unless r.next.nil? # without line break
-          id = escape(text)
+          next if !r.next.nil? and # skip matches in paragraph
+                  r.next.name.downcase != 'bookmarkend'
+          id = CGI.escape(text.gsub(/&(.)uml;/, '\1').gsub(/\s*\/\s*|\/|\s+/, '_').downcase)
           @indecies << {:text => chapter, :id => id}
-          return tag(:h3, text, {:id => id})
+          return markup(:h3, text, {:id => id})
         end
       end
-      nil
+      if r.parent.previous.nil? and @indecies.empty?
+        # The first line as package name
+        @indecies << {:text => 'Titel', :id => 'titel'}
+        return markup(:h2, text, {:id => 'titel'})
+      end
+      return nil
     end
   end
   class Builder
     def init
-      @container = tag(:div, [], {:id => 'container'})
+      @container = markup(:div, [], {:id => 'container'})
     end
     private
     def build_before_content
@@ -61,11 +58,11 @@ module Docx2html
         indices = []
         @indecies.each do |index|
           if index.has_key?(:id)
-            link = tag(:a, index[:text], {:href => "#" + index[:id]})
-            indices << tag(:li, link)
+            link = markup(:a, index[:text], {:href => "#" + index[:id]})
+            indices << markup(:li, link)
           end
         end
-        tag(:div, tag(:ul, indices), {:id => 'indecies'})
+        markup(:div, markup(:ul, indices), {:id => 'indecies'})
       end
     end
     def style

data/lib/ydocx.rb

@@ -0,0 +1,7 @@
+#!/usr/bin/env ruby
+# encoding: utf-8
+
+require 'ydocx/document'
+
+module YDocx
+end

metadata

@@ -1,91 +1,88 @@
---- !ruby/object:Gem::Specification 
+--- !ruby/object:Gem::Specification
 name: ydocx
-version: !ruby/object:Gem::Version 
-  prerelease: false
-  segments: 
-  - 1
-  - 0
-  - 3
-  version: 1.0.3
+version: !ruby/object:Gem::Version
+  version: 1.0.4
+  prerelease: 
 platform: ruby
-authors: 
+authors:
 - Yasuhiro Asaka, Zeno R.R. Davatz
 autorequire: 
 bindir: bin
 cert_chain: []
-
-date: 2012-05-01 00:00:00 +02:00
-default_executable: 
-dependencies: 
-- !ruby/object:Gem::Dependency 
-  name: hoe
+date: 2012-05-02 00:00:00.000000000 Z
+dependencies:
+- !ruby/object:Gem::Dependency
+  name: rdoc
+  requirement: &19114300 !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ~>
+      - !ruby/object:Gem::Version
+        version: '3.10'
+  type: :development
   prerelease: false
-  requirement: &id001 !ruby/object:Gem::Requirement 
-    requirements: 
-    - - ">="
-      - !ruby/object:Gem::Version 
-        segments: 
-        - 2
-        - 9
-        - 1
-        version: 2.9.1
+  version_requirements: *19114300
+- !ruby/object:Gem::Dependency
+  name: hoe
+  requirement: &19113780 !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ~>
+      - !ruby/object:Gem::Version
+        version: '2.13'
   type: :development
-  version_requirements: *id001
-description: ""
-email: 
+  prerelease: false
+  version_requirements: *19113780
+description: ''
+email:
 - yasaka@ywesee.com,  zdavatz@ywesee.com
-executables: 
+executables:
 - docx2html
+- docx2xml
 extensions: []
-
-extra_rdoc_files: 
+extra_rdoc_files:
 - History.txt
 - Manifest.txt
 - README.txt
-files: 
+files:
 - History.txt
 - Manifest.txt
 - README.txt
 - Rakefile
 - bin/docx2html
-- lib/docx2html/bin/docx2html
-- lib/docx2html/lib/docx2html.rb
-- lib/docx2html/lib/docx2html/builder.rb
-- lib/docx2html/lib/docx2html/document.rb
-- lib/docx2html/lib/docx2html/html_methods.rb
-- lib/docx2html/lib/docx2html/parser.rb
-- lib/fachinfo.rb
+- bin/docx2xml
 - lib/version.rb
-has_rdoc: true
+- lib/ydocx.rb
+- lib/ydocx/builder.rb
+- lib/ydocx/document.rb
+- lib/ydocx/markup_method.rb
+- lib/ydocx/parser.rb
+- lib/ydocx/command.rb
+- lib/ydocx/templates/fachinfo.rb
 homepage: https://github.com/zdavatz/ydocx
 licenses: []
-
 post_install_message: 
-rdoc_options: 
+rdoc_options:
 - --main
 - README.txt
-require_paths: 
+require_paths:
 - lib
-required_ruby_version: !ruby/object:Gem::Requirement 
-  requirements: 
-  - - ">="
-    - !ruby/object:Gem::Version 
-      segments: 
-      - 0
-      version: "0"
-required_rubygems_version: !ruby/object:Gem::Requirement 
-  requirements: 
-  - - ">="
-    - !ruby/object:Gem::Version 
-      segments: 
-      - 0
-      version: "0"
+required_ruby_version: !ruby/object:Gem::Requirement
+  none: false
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+required_rubygems_version: !ruby/object:Gem::Requirement
+  none: false
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
 requirements: []
-
 rubyforge_project: ydocx
-rubygems_version: 1.3.6
+rubygems_version: 1.8.15
 signing_key: 
 specification_version: 3
-summary: ""
+summary: ''
 test_files: []
-

data/lib/docx2html/bin/docx2html

@@ -1,16 +0,0 @@
-#!/usr/bin/env ruby
-# coding: utf-8
-
-require 'pathname'
-
-root = Pathname.new(__FILE__).realpath.parent.parent
-$:.unshift root.join('lib/docx2html') if $0 == __FILE__
-require 'docx2html'
-
-if file = ARGV.first
-  unless File.exist?(file)
-    exit
-  else
-    Docx2html::Document.open(file).to_html file, :style => true
-  end
-end

data/lib/docx2html/lib/docx2html/builder.rb

@@ -1,105 +0,0 @@
-#!/usr/bin/env ruby
-# encoding: utf-8
-
-require 'nokogiri'
-require 'docx2html/lib/docx2html/html_methods'
-
-module Docx2html
-  class Builder
-    include HtmlMethods
-    attr_accessor :body, :indecies, :title,
-                  :frame, :style
-    def initialize(body)
-      @body = body
-      @container = {}
-      @indecies = []
-      @style = false
-      @title = ''
-      init
-      if block_given?
-        yield self
-      end
-    end
-    def init
-    end
-    def build
-      if @container.has_key?(:content)
-        @container[:content] = @body
-        @body = [@container]
-      end
-      if before_content = build_before_content
-        @body.unshift before_content
-      end
-      if after_content = build_after_content
-        @body.push after_content
-      end
-      body = ''
-      @body.each do |e|
-        body << build_tag(e[:tag], e[:content], e[:attributes])
-      end
-      builder = Nokogiri::HTML::Builder.new do |doc|
-        doc.html {
-          doc.head {
-            doc.meta :charset => 'utf-8'
-            doc.title @title
-            doc.style { doc << style } if @style
-          }
-          doc.body { doc << body }
-        }
-      end
-      builder.to_html.gsub(/\n/, '')
-    end
-    private
-    def build_after_content
-      nil
-    end
-    def build_before_content
-      nil
-    end
-    def build_tag(tag, content, attributes)
-      if tag == :br
-        return "<br/>"
-      elsif content.nil? or content.empty?
-        return ''
-      end
-      _content = ''
-      if content.is_a? Array
-        content.each do |c|
-          next if c.nil? or c.empty?
-          if c.is_a? Hash
-            _content << build_tag(c[:tag], c[:content], c[:attributes])
-          elsif c.is_a? String
-            _content << c.chomp.to_s
-          end
-        end
-      elsif content.is_a? Hash
-        _content = build_tag(content[:tag], content[:content], content[:attributes])
-      elsif content.is_a? String
-        _content = content
-      end
-      _tag = tag.to_s
-      _attributes = ''
-      unless attributes.empty?
-        attributes.each_pair do |k, v|
-          _attributes << " #{k.to_s}=#{v.to_s}"
-        end
-      end
-      return "<#{_tag}#{_attributes}>#{_content}</#{_tag}>"
-    end
-    def style
-      style = <<-CSS
-table, tr, td {
-  border-collapse: collapse;
-  border:          1px solid gray;
-}
-table {
-  margin: 5px 0 5px 0;
-}
-td {
-  padding: 5px 10px;
-}
-      CSS
-      style.gsub(/\s\s+|\n/, ' ')
-    end
-  end
-end

data/lib/docx2html/lib/docx2html.rb

@@ -1,7 +0,0 @@
-#!/usr/bin/env ruby
-# encoding: utf-8
-
-require 'docx2html/lib/docx2html/document'
-
-module Docx2html
-end