ydocx 1.0.4 → 1.0.5

data/History.txt

@@ -1,3 +1,13 @@
+=== 1.0.5 / 03.05.2012
+
+* Added img support
+ - without wmf conversion
+ - create ***_files directory into same directory with html
+* Improved chapter regexp for patinfo
+* Updated command option fachinfo format as default
+* Improved regexp for patinfo
+* Added Patinfo format
+
 === 1.0.4 / 02.05.2012
 
 * Fixed bug for action to_xml in command

data/Manifest.txt

@@ -12,3 +12,4 @@ lib/ydocx/markup_method.rb
 lib/ydocx/parser.rb
 lib/ydocx/command.rb
 lib/ydocx/templates/fachinfo.rb
+lib/ydocx/templates/patinfo.rb

data/README.txt

@@ -11,6 +11,12 @@
 * Umlaute
 * bold, italic, underline
 
+== Usage
+
+* Usage: /usr/local/bin/docx2html file [options]
+    -f, --format    Format of style and chapter {(fi|fachinfo)|(pl|plain)}, default none.
+    -h, --help      Display this help message.
+
 == Using the great libraries
 
 * nokogiri

data/lib/version.rb

@@ -2,5 +2,5 @@
 # encoding: utf-8
 
 module Docx2html
-  VERSION = "1.0.4"
+  VERSION = "1.0.5"
 end

data/lib/ydocx/builder.rb

@@ -2,17 +2,19 @@
 # encoding: utf-8
 
 require 'nokogiri'
+require 'pathname'
 require 'ydocx/markup_method'
 
 module YDocx
   class Builder
     include MarkupMethod
     attr_accessor :contents, :container, :indecies,
-                  :style, :title
+                  :files, :style, :title
     def initialize(contents)
       @contents = contents
       @container = {}
       @indecies = []
+      @files = Pathname.new('.')
       @style = false
       @title = ''
       init
@@ -47,10 +49,10 @@ module YDocx
       builder.to_html.gsub(/\n/, '')
     end
     def build_xml
-      chapters = compile(@contents, :xml) 
+      paragraphs = compile(@contents, :xml)
       builder = Nokogiri::XML::Builder.new do |xml|
         xml.document {
-          xml.chapters { xml << chapters }
+          xml.paragraphs { xml << paragraphs }
         }
       end
       builder.to_xml(:indent => 0, :encoding => 'utf-8').gsub(/\n/, '')
@@ -87,7 +89,7 @@ module YDocx
       if tag == :br and mode != :xml
         return "<br/>"
       elsif content.nil? or content.empty?
-        return ''
+        return '' if attributes.nil? # without img
       end
       _content = ''
       if content.is_a? Array
@@ -109,7 +111,12 @@ module YDocx
       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}"
+          if tag == :img and key == :src
+            src = @files.join value.to_s
+            _attributes << " src=\"#{src}\""
+          else
+            _attributes << " #{key.to_s}=\"#{value.to_s}\""
+          end
         end
       end
       if mode == :xml
@@ -123,7 +130,11 @@ module YDocx
         when :sub     then _tag = 'subscript'   # text
         end
       end
-      return "<#{_tag}#{_attributes}>#{_content}</#{_tag}>"
+      if tag == :img
+        return "<#{_tag}#{_attributes}/>"
+      else
+        return "<#{_tag}#{_attributes}>#{_content}</#{_tag}>"
+      end
     end
     def style
       style = <<-CSS

data/lib/ydocx/command.rb

@@ -19,7 +19,7 @@ module YDocx
       def help
         banner = <<-BANNER
 Usage: #{$0} file [options]
-  -f, --format    Format of style and chapter {fi|fachinfo}, default none.
+  -f, --format    Format of style and chapter {(fi|fachinfo)|(pl|plain)|none}, default fachinfo.
   -h, --help      Display this help message.
         BANNER
         puts banner
@@ -50,10 +50,14 @@ Usage: #{$0} file [options]
                 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
+                  require 'ydocx/templates/patinfo'
+                  options.merge!({:style => :frame}) if action == :to_html
+                when 'pl', 'plain'
+                  options.merge!({:style => true}) if action == :to_html
+                when 'none'
+                  # pass
                 else
                   self.error "#{self.command}: exit with #{option}: Invalid argument"
                 end
@@ -62,6 +66,10 @@ Usage: #{$0} file [options]
               else
                 self.error "#{self.command}: exit with #{option}: Unknown option"
               end
+            else
+              # default fachinfo
+              require 'ydocx/templates/fachinfo'
+              options.merge!({:style => :frame}) if action == :to_html
             end
             YDocx::Document.open(path).send(action, path, options)
             self.report action, path

data/lib/ydocx/document.rb

@@ -8,19 +8,25 @@ require 'ydocx/builder'
 
 module YDocx
   class Document
-    attr_reader :contents, :indecies
+    attr_reader :contents, :indecies, :pictures
     def self.open(file)
       self.new(file)
     end
     def initialize(file)
       @contents = nil
       @indecies = nil
+      @pictures = []
+      @path = nil
+      @files = nil
+      @zip = nil
       read(file)
     end
     def to_html(file='', options={})
       html = ''
+      @files = @path.dirname.join(@path.basename('.docx').to_s + '_files')
       Builder.new(@contents) do |builder|
-        builder.title = @path
+        builder.title = @path.basename
+        builder.files = @files
         builder.style = options[:style] if options.has_key?(:style)
         if @indecies
           builder.indecies = @indecies
@@ -28,8 +34,9 @@ module YDocx
         html = builder.build_html
       end
       unless file.empty?
-        path = Pathname.new(file).realpath.sub_ext('.html')
-        File.open(path, 'w:utf-8') do |f|
+        create_files if has_picture?
+        html_file = @path.sub_ext('.html')
+        File.open(html_file, 'w:utf-8') do |f|
           f.puts html
         end
       else
@@ -42,8 +49,8 @@ module YDocx
         xml = builder.build_xml
       end
       unless file.empty?
-        path = Pathname.new(file).realpath.sub_ext('.xml')
-        File.open(path, 'w:utf-8') do |f|
+        xml_file = @path.sub_ext('.xml')
+        File.open(xml_file, 'w:utf-8') do |f|
           f.puts xml
         end
       else
@@ -51,13 +58,32 @@ module YDocx
       end
     end
     private
+    def has_picture?
+      !@pictures.empty?
+    end
+    def create_files
+      FileUtils.mkdir @files unless @files.exist?
+      @zip = Zip::ZipFile.open(@path.realpath)
+      @pictures.each do |pic|
+        pic_path = Pathname.new pic # id/filename.ext
+        pic_dir = @files.join pic_path.dirname
+        FileUtils.mkdir pic_dir unless pic_dir.exist?
+        binary = @zip.find_entry("word/media/#{pic_path.basename}").get_input_stream
+        @files.join(pic_path).open('w') do |f|
+          f.puts binary.read
+        end
+      end
+      @zip.close
+    end
     def read(file)
-      @path = File.expand_path(file)
-      @zip = Zip::ZipFile.open(@path)
-      stream = @zip.find_entry('word/document.xml').get_input_stream
-      Parser.new(stream) do |parser|
+      @path = Pathname.new file
+      @zip = Zip::ZipFile.open(@path.realpath)
+      doc = @zip.find_entry('word/document.xml').get_input_stream
+      ref = @zip.find_entry('word/_rels/document.xml.rels').get_input_stream
+      Parser.new(doc, ref) do |parser|
         @contents = parser.parse
         @indecies = parser.indecies
+        @pictures = parser.pictures
       end
       @zip.close
     end

data/lib/ydocx/parser.rb

@@ -8,11 +8,13 @@ require 'ydocx/markup_method'
 module YDocx
   class Parser
     include MarkupMethod
-    attr_accessor :indecies, :result, :space
-    def initialize(stream)
-      @xml = Nokogiri::XML.parse(stream)
+    attr_accessor :indecies, :pictures, :result, :space
+    def initialize(doc, rel)
+      @doc = Nokogiri::XML.parse(doc)
+      @rel = Nokogiri::XML.parse(rel)
       @coder = HTMLEntities.new
       @indecies = []
+      @pictures = []
       @result = []
       @space = ' '
       init
@@ -23,13 +25,13 @@ module YDocx
     def init
     end
     def parse
-      @xml.xpath('//w:document//w:body').children.map do |node|
+      @doc.xpath('//w:document//w:body').children.map do |node|
         case node.node_name
         when 'text'
           @result << parse_paragraph(node)
         when 'tbl'
           @result << parse_table(node)
-        when 'image'
+        when 'pict'
           # pending
         when 'p'
           @result << parse_paragraph(node)
@@ -82,7 +84,6 @@ module YDocx
       nil # default no block element
     end
     def optional_escape(text)
-      return text = @space if text.empty?
       text.force_encoding('utf-8')
       # NOTE
       # :named only for escape at Builder
@@ -149,8 +150,22 @@ module YDocx
         #p "char : " + @coder.decode("&#%s;" % code.hex.to_s)
       end
     end
-    def parse_image
-      # pending
+    def parse_image(r)
+      if pict = r.xpath('w:pict') and
+         shape = pict.xpath('v:shape') and
+         image = shape.xpath('v:imagedata')
+          id = image.first['id'] # r:id
+          @rel.xpath('/').children.each do |element|
+            element.children.each do |rel|
+              if rel['Id'] == id and rel['Target']
+                src = id.downcase + '/' + File.basename(rel['Target'])
+                @pictures << src
+                return markup :img, [], {:src => src}
+              end
+            end
+          end
+      end
+      nil
     end
     def parse_paragraph(node)
       content = []
@@ -158,12 +173,12 @@ module YDocx
       pos = 0
       node.xpath('w:r').each do |r|
         unless r.xpath('w:t').empty?
-          content << parse_text(r)
+          content << parse_text(r, (pos == 0)) # rm indent
           pos += 1
         else
           unless r.xpath('w:tab').empty?
             if content.last != @space and pos != 0 # ignore tab at line head
-              content << optional_escape('')
+              content << @space
               pos += 1
             end
           end
@@ -172,6 +187,9 @@ module YDocx
             content << optional_replace(code)
             pos += 1
           end
+          unless r.xpath('w:pict').empty?
+            content << parse_image(r)
+          end
         end
       end
       content.compact!
@@ -209,7 +227,7 @@ module YDocx
       end
       table
     end
-    def parse_text(r)
+    def parse_text(r, lstrip=false)
       text = r.xpath('w:t').map(&:text).join('')
       text = optional_escape(text)
       if rpr = r.xpath('w:rPr')
@@ -218,7 +236,7 @@ module YDocx
           block
         else
           # inline tag
-          text = text.strip
+          text = text.lstrip if lstrip
           text = apply_align(rpr, text)
           unless rpr.xpath('w:u').empty?
             text = markup(:span, text, {:style => "text-decoration:underline;"})

data/lib/ydocx/templates/fachinfo.rb

@@ -6,6 +6,9 @@ require 'cgi'
 module YDocx
   class Parser
     private
+    def escape_as_id(text)
+      CGI.escape(text.gsub(/&(.)uml;/, '\1').gsub(/\s*\/\s*|\/|\s+/, '_').gsub(/(\?|_$)/, '').downcase)
+    end
     def parse_as_block(r, text)
       text = text.strip
       # TODO
@@ -17,7 +20,7 @@ module YDocx
         'Ind./Anw.mögl.' => /^Indikationen(\s+|\s*(\/|und)\s*)Anwendungsmöglichkeiten$|^Indications/u, # 4
         'Interakt.'           => /^Interaktionen$|^Interactions/u, # 8
         'Kontraind.'          => /^Kontraindikationen($|\s*\(\s*absolute\s+Kontraindikationen\s*\)$)/u, # 6
-        'Name'                => /^Name\s+des\s+Präparates$/, # 1
+        'Name'                => /^Name\s+des\s+Präparates$/u, # 1
         'Packungen'           => /^Packungen($|\s*\(\s*mit\s+Angabe\s+der\s+Abgabekategorie\s*\)$)/u, # 18
         'Präklin.'       => /^Präklinische\s+Daten$/u, # 15
         'Pharm.kinetik'       => /^Pharmakokinetik($|\s*\((Absorption,\s*Distribution,\s*Metabolisms,\s*Elimination\s|Kinetik\s+spezieller\s+Patientengruppen)*\)$)|^Pharmacocin.tique?/iu, # 14
@@ -35,7 +38,7 @@ module YDocx
         if text =~ regexp
           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)
+          id = escape_as_id(text)
           @indecies << {:text => chapter, :id => id}
           return markup(:h3, text, {:id => id})
         end
@@ -52,6 +55,15 @@ module YDocx
     def init
       @container = markup(:div, [], {:id => 'container'})
     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 build_before_content
       if @indecies

data/lib/ydocx/templates/patinfo.rb

@@ -0,0 +1,47 @@
+#!/usr/bin/env ruby
+# encoding: utf-8
+
+require 'ydocx/templates/fachinfo'
+
+module YDocx
+  class Parser
+    private
+    def parse_as_block(r, text)
+      text = text.strip
+      # TODO
+      # Franzoesisch
+      chapters = {
+        'Abänderung'     => /^Was\s+sollte\s+dazu\s+beachtet\s+werden\s*\??$/u, # 4
+        'Dos./Anw.'           => /^Wie\s+verwenden\s+Sie\s+\w+\s*\??$/u, # 8
+        'Eigensch.'           => /^Was\s+ist\s+\w+\s+und\s+wann\s+wird\s+es\s+angewendet\s*\??$/u, # 3
+        'Gewöhnliche H.' => /^Was\s+ist\s+ferner\s+zu\s+beachten\s*\??$/u, # 10
+        'Hersteller'          => /^Herstellerin$/u, # 15
+        'Information'         => /^Information\s+für\sPatientinnen\s+und\s+Patienten$/u, # 1
+        'Kontraind.'          => /^Wann\s+darf\s+\w+\s+nicht\s+(eingenommen\s*\/\s*angewendet|eingenommen|angewendet)\s*werden\s*\??$/u, # 5
+        'Name'                => /^Name\s+des\s+Präparates$/u, # 2
+        'Packungen'           => /^Wo\s+erhalten\s+Sie\s+\w+\s*\?\s*Welche\s+Packungen\s+sind\s+erhältlich\s*\??$/u, # 13
+        'Schwanderschaft'     => /^Darf\s+\w+\s+während\s+einer\s+Schwangerschaft\s+oder\s+in\s+der\s+Stillzeit\s+(eingenommen\s*\/\s*angewendet|eingenommen|angewendet)\s*werden\s*\??$/u, # 7
+        'Stand d. Info.'      => /^Diese\sPackungsbeilage\s+wurde\s+im\s+[\.A-z\s0-9]+(\s+|\s*\/\s*\w+\s+\(Monat\s*\/\s*Jahr\)\s*)letztmals\s+durch\s+die\s+Arzneimittelbehörde\s*\(\s*Swissmedic\s*\)\s*geprüft.?$/u, # 16
+        'Swissmedic-Nr.'      => /^Zulassungsnummer$/u, # 12
+        'Unerw.Wirkungen'     => /^Welche\s+Nebenwirkungen\s+kann\s+\w+\s+haben\s*\??$/u, # 9
+        'Verteiler'           => /^Zulassungsinhaberin$/u, # 14
+        'Vorbeugung'          => /^Wann\s+ist\s+bei\s+der\s+(Einnahme\s*\/\s*Anwendung|Einnahme|Anwendung)\s*von\s+\w+\s+Vorsicht\s+geboten\s*\??$/u, # 6
+        'Zusammens.'          => /^Was\s+ist\s+in\s+\w+\s+enthalten\s*\??$/u, # 11
+      }.each_pair do |chapter, regexp|
+        if text =~ regexp
+          next if !r.next.nil? and # skip matches in paragraph
+                  r.next.name.downcase != 'bookmarkend'
+          id = escape_as_id(text)
+          @indecies << {:text => chapter, :id => id}
+          return markup(:h3, text, {:id => id})
+        end
+      end
+      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
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: ydocx
 version: !ruby/object:Gem::Version
-  version: 1.0.4
+  version: 1.0.5
   prerelease: 
 platform: ruby
 authors:
@@ -9,11 +9,11 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-05-02 00:00:00.000000000 Z
+date: 2012-05-03 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rdoc
-  requirement: &19114300 !ruby/object:Gem::Requirement
+  requirement: &9555940 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -21,10 +21,10 @@ dependencies:
         version: '3.10'
   type: :development
   prerelease: false
-  version_requirements: *19114300
+  version_requirements: *9555940
 - !ruby/object:Gem::Dependency
   name: hoe
-  requirement: &19113780 !ruby/object:Gem::Requirement
+  requirement: &9555520 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -32,7 +32,7 @@ dependencies:
         version: '2.13'
   type: :development
   prerelease: false
-  version_requirements: *19113780
+  version_requirements: *9555520
 description: ''
 email:
 - yasaka@ywesee.com,  zdavatz@ywesee.com
@@ -59,6 +59,7 @@ files:
 - lib/ydocx/parser.rb
 - lib/ydocx/command.rb
 - lib/ydocx/templates/fachinfo.rb
+- lib/ydocx/templates/patinfo.rb
 homepage: https://github.com/zdavatz/ydocx
 licenses: []
 post_install_message: