ydocx 1.2.1 → 1.2.6

checksums.yaml

@@ -0,0 +1,7 @@
+---
+SHA256:
+  metadata.gz: d110c0230d26d291bbc973e022379d669a7fa963dc0baf27fa3b0ccaaa193bab
+  data.tar.gz: f21bfa758230376fadeab73e75645dd06862957c0d8c7d120bd489d7ffeb9b1b
+SHA512:
+  metadata.gz: 75d8e7bed0a19192c43c4b75be52b9c5f9da1ef265052420d299b810189faf3b880cc3e5f207621695a6cd8d156c477aed00c87b6e82366c7d8dedc012411e90
+  data.tar.gz: 84bb26bab8bc03ea116f90f1df1708d4f646f0b0094a6e685723f2e7448e409a5db8c46845c54d44ee723fd9dd0cf866946d49e8b15f6a0a56d63c01610925cc

data/History.txt

@@ -1,3 +1,26 @@
+=== 1.2.6 / 19.01.2021
+
+* Updated to use for ruby 3.0.0. RMagick->rmagic
+* Replaced hoe by bundler/gem_tasks
+
+=== 1.2.5 / 07.07.2014
+
+* Updated to use rubyzip >= 1.0.0
+
+=== 1.2.4 / 26.02.2014
+
+* Many changes by Niklaus Giger to get ready for Pseudo-Fachinformation.
+
+=== 1.2.3 / 12.07.2012
+
+* Updated french char handling, lang in filename
+
+=== 1.2.2 / 12.07.2012
+
+* Fixed typo in french
+* Updated help output
+* Updated help message about --lang option
+
 === 1.2.1 / 11.07.2012
 
 * Updated template, option handling for extension

data/README.txt

@@ -1,4 +1,4 @@
-== ydocx - © ywesee GmbH
+== ydocx by ywesee GmbH
 
 * https://github.com/zdavatz/ydocx
 * Parsing docx files with Ruby and output them as HTML and XML.
@@ -18,17 +18,29 @@
 
 == Usage
 
-* Usage: bin/docx2html file [options]
+* Usage: docx2html file [options]
     -f, --format    Format of style and chapter {(fi|fachinfo)|(pi|patinfo)|(pl|plain)|none}, default fachinfo.
     -h, --help      Display this help message.
+    -l, --lang      Language option for templates {de|fr}
     -v, --version   Show version.
 
 == Using the great libraries
 
-* rubyzip
+* rubyzip (< 1.0.0)
 * nokogiri
 * htmlentities
-* RMagick
+* rmagick
+
+Install them using bundle install.
+
+Requires Ruby >= 1.9.2.
+
+== TODO
+
+For unknown reason the gem fails to  parse properly the Présentation for sinovial_fr.
+Failing spec is: ydocx should convert sinovial_FR to xml
+     Failure/Error: doc.xpath('//chapters/chapter[contains(heading, "Présentation")]').size.should > 0
+     
 
 == License GPLv3.0
 

data/lib/ydocx.rb

@@ -1,8 +1,4 @@
 #!/usr/bin/env ruby
-# encoding: utf-8
 
 require 'ydocx/document'
 
-module YDocx
-  VERSION = '1.2.1'
-end

data/lib/ydocx/command.rb

@@ -6,8 +6,9 @@ require 'ydocx'
 module YDocx
   class Command
     class << self
-      @@help    = /^\-(h|\-help)$/u
       @@format  = /^\-(f|\-format)$/u
+      @@help    = /^\-(h|\-help)$/u
+      @@lang    = /^\-(l|\-lang)$/u
       @@version = /^\-(v|\-version)$/u
       def error(message='')
         puts message
@@ -25,11 +26,57 @@ module YDocx
 Usage: #{self.command} file [options]
   -f, --format    Format of style and chapter {(fi|fachinfo)|(pi|patinfo)|(pl|plain)|none}, default fachinfo.
   -h, --help      Display this help message.
+  -l, --lang      Language option for templates {de|fr}
   -v, --version   Show version.
         BANNER
         puts banner
         exit
       end
+      def parse(action, argv)
+        if argv.length.odd?
+          self.error "#{self.command}: exit with: Invalid option"
+        else
+          args = Hash[*argv]
+        end
+        options = {}
+        args.each_pair do |option, value|
+          if option =~ @@format
+            case value
+            when 'fi', 'fachinfo'
+              require 'ydocx/templates/fachinfo'
+              options.merge!({:style => :frame}) if action == :to_html
+            when 'pi', 'patinfo'
+              require 'ydocx/templates/patinfo'
+              options.merge!({:style => :frame}) if action == :to_html
+            when 'pl', 'plain'
+              options.merge!({:style => true}) if action == :to_html
+            when 'none'
+              # pass
+            else
+              self.error "#{self.command}: exit with #{option}: Invalid argument"
+            end
+          elsif option =~ @@help
+            self.help
+          elsif option =~ @@lang
+            options.merge!({:lang => value})
+          elsif option.downcase =~ /\.(jpeg|jpg|png|gif)$/u and action == :to_html
+            # allow as default
+            # TODO
+            # refactor as normal option
+            # currently, support fachinfo/patinfo format only
+            require 'ydocx/templates/fachinfo'
+            options.merge!({:style => :frame})
+          else
+            self.error "#{self.command}: exit with #{option}: Unknown option"
+          end
+        end
+        if !args.include?('-f') and !args.include?('--format')
+          # default fachinfo
+          require 'ydocx/templates/fachinfo'
+          options.merge!({:style => :frame}) if action == :to_html
+        end
+        options
+      end
       def report(action, path)
         puts "#{self.command}: generated #{File.expand_path(path)}"
         exit
@@ -48,41 +95,8 @@ Usage: #{self.command} file [options]
           elsif !File.extname(path).match(/^\.docx$/)
             self.error "#{self.command}: cannot open #{file}: Not a docx file"
           else
-            options = {}
-            if option = argv.shift
-              if option =~ @@format
-                case argv[0]
-                when 'fi', 'fachinfo'
-                  require 'ydocx/templates/fachinfo'
-                  options.merge!({:style => :frame}) if action == :to_html
-                when 'pi', 'patinfo'
-                  require 'ydocx/templates/patinfo'
-                  options.merge!({:style => :frame}) if action == :to_html
-                when 'pl', 'plain'
-                  options.merge!({:style => true}) if action == :to_html
-                when 'none'
-                  # pass
-                else
-                  self.error "#{self.command}: exit with #{option}: Invalid argument"
-                end
-              elsif option =~ @@help
-                self.help
-              elsif option.downcase =~ /\.(jpeg|jpg|png|gif)$/u and action == :to_html
-                # allow as default
-                # TODO
-                # refactor as normal option
-                # currently, support fachinfo/patinfo format only
-                require 'ydocx/templates/fachinfo'
-                options.merge!({:style => :frame})
-              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
-            doc = YDocx::Document.open(path)
+            options = self.parse(action, argv)
+            doc = YDocx::Document.open(path, options)
             doc.send(action, path, options)
             ext = self.extname(action)
             self.report action, doc.output_file(ext[1..-1])

data/lib/ydocx/document.rb

@@ -2,11 +2,11 @@
 # encoding: utf-8
 
 require 'pathname'
-require 'zip/zip'
+require 'zip'
 begin
-  require 'RMagick'
+  require 'rmagick'
 rescue LoadError
-  warn "Couldn't load RMagick: .wmf conversion off"
+  warn "Couldn't load rmagick: .wmf conversion off"
 end
 require 'ydocx/parser'
 require 'ydocx/builder'
@@ -83,7 +83,7 @@ module YDocx
     def create_files
       files_dir = output_directory
       mkdir Pathname.new(files_dir) unless files_dir.exist?
-      @zip = Zip::ZipFile.open(@path.realpath)
+      @zip = Zip::File.open(@path.realpath)
       @images.each do |image|
         origin_path = Pathname.new image[:origin] # media/filename.ext
         source_path = Pathname.new image[:source] # images/filename.ext
@@ -118,10 +118,11 @@ module YDocx
     end
     def read(file)
       @path = Pathname.new file
-      @zip = Zip::ZipFile.open(@path.realpath)
+      @zip = Zip::File.open(@path.realpath)
       doc = @zip.find_entry('word/document.xml').get_input_stream
       rel = @zip.find_entry('word/_rels/document.xml.rels').get_input_stream
       @parser = Parser.new(doc, rel) do |parser|
+        parser.lang = @options[:lang] if @options[:lang]
         @contents = parser.parse
         @indecies = parser.indecies
         @images = parser.images

data/lib/ydocx/parser.rb

@@ -8,7 +8,7 @@ require 'ydocx/markup_method'
 module YDocx
   class Parser
     include MarkupMethod
-    attr_accessor :indecies, :images, :result, :space
+    attr_accessor :indecies, :images, :result, :space, :lang
     def initialize(doc, rel)
       @doc = Nokogiri::XML.parse(doc)
       @rel = Nokogiri::XML.parse(rel)
@@ -55,7 +55,7 @@ module YDocx
       if symbol
         _text = ''
         text.unpack('U*').each do |char|
-          _text << optional_replace(char.to_s(16))
+          _text << character_replace(char.to_s(16))
         end
         text = _text
       end
@@ -76,14 +76,14 @@ module YDocx
       end
       text
     end
-    def optional_escape(text)
+    def character_encode(text)
       text.force_encoding('utf-8')
       # NOTE
       # :named only for escape at Builder
       text = @coder.encode(text, :named)
       text
     end
-    def optional_replace(code)
+    def character_replace(code)
       code = '0x' + code
       # NOTE
       # replace with rsemble html character ref
@@ -143,6 +143,9 @@ module YDocx
         #p "char : " + @coder.decode("&#%s;" % code.hex.to_s)
       end
     end
+    def optional_escape(text)
+      text
+    end
     def parse_block(node)
       nil # default no block element
     end
@@ -155,19 +158,19 @@ module YDocx
       ns = r.namespaces.merge additional_namespaces
       [
         { # old type shape
-          :attr => 'id',
+          :attr => 'r:id',
           :path => 'w:pict//v:shape//v:imagedata',
           :wrap => 'w:pict//v:shape//w10:wrap',
           :type => '',
         },
         { # in anchor
-          :attr => 'embed',
+          :attr => 'r:embed',
           :path => 'w:drawing//wp:anchor//a:graphic//a:graphicData//pic:pic//pic:blipFill//a:blip',
           :wrap => 'w:drawing//wp:anchor//wp:wrapTight',
           :type => 'wrapText',
         },
         { # stand alone
-          :attr => 'embed',
+          :attr => 'r:embed',
           :path => 'w:drawing//a:graphic//a:graphicData//pic:pic//pic:blipFill//a:blip',
           :wrap => 'w:drawing//wp:wrapTight',
           :type => 'wrapText',
@@ -229,8 +232,8 @@ module YDocx
               end
             end
             unless r.xpath('w:sym').empty?
-              code = r.xpath('w:sym').first['char'].downcase # w:char
-              content << optional_replace(code)
+              code = r.xpath('w:sym').first['w:char'].downcase # w:char
+              content << character_replace(code)
               pos += 1
             end
             if !r.xpath('w:pict').empty? or !r.xpath('w:drawing').empty?
@@ -276,6 +279,7 @@ module YDocx
     end
     def parse_text(r, lstrip=false)
       text = r.xpath('w:t').map(&:text).join('')
+      text = character_encode(text)
       text = optional_escape(text)
       text = text.lstrip if lstrip
       if rpr = r.xpath('w:rPr')

data/lib/ydocx/templates/fachinfo.rb

@@ -6,45 +6,100 @@ require 'cgi'
 module YDocx
   class Parser
     attr_accessor :code, :lang
+    @@figure_pattern = /‘|’|'|´/
     def init
       @image_path = 'image'
       @code = nil
-      @lang = 'DE'
+      @lang ||= 'de'
     end
-    private
+    ###
+    # Fachinfo Chapters
+    #  1. name
+    #  2. composition
+    #  3. galenic form
+    #  4. indications
+    #  5. usage
+    #  6. contra_indications
+    #  7. restrictions
+    #  8. interactions
+    #  9. pregnancy
+    # 10. driving_ability
+    # 11. unwanted_effects
+    # 12. overdose
+    # 13. effects
+    # 14. kinetic
+    # 15. preclinic
+    # 16. other_advice
+    # 17. iksnr
+    # 19. packages
+    # 19. registration_owner
+    # 20. date
     def chapters
-      # TODO
-      # Franzoesisch
       chapters = {
-        'Name'                => /^Name\s+des\s+Präparates$/u, # 1
-        'Zusammens.'          => /^Zusammensetzung($|\s*\/\s*(Wirkstoffe|Hilsstoffe)$)/u, # 2
-        'Galen.Form'          => /^Galenische\s+Form\s*(und|\/)\s*Wirkstoffmenge\s+pro\s+Einheit$|^Forme\s*gal.nique/iu, # 3
-        'Ind./Anw.mögl.' => /^Indikationen(\s+|\s*(\/|und)\s*)Anwendungsmöglichkeiten$|^Indications/u, # 4
-        'Dos./Anw.'           => /^Dosierung\s*(\/|und)\s*Anwendung/u, # 5
-        'Kontraind.'          => /^Kontraindikationen($|\s*\(\s*absolute\s+Kontraindikationen\s*\)$)/u, # 6
-        'Warn.hinw.'          => /^Warnhinweise\s+und\s+Vorsichtsmassnahmen($|\s*\/\s*(relative\s+Kontraindikationen|Warnhinweise\s*und\s*Vorsichtsmassnahmen)$)/u, # 7
-        'Interakt.'           => /^Interaktionen$|^Interactions/u, # 8
-        'Schwangerschaft'     => /^Schwangerschaft(,\s*|\s*\/\s*|\s+und\s+)Stillzeit$/u, # 9
-        'Fahrtücht.'     => /^Wirkung\s+auf\s+die\s+Fahrtüchtigkeit\s+und\s+auf\s+das\s+Bedienen\s+von\s+Maschinen$/u, # 10
-        'Unerw.Wirkungen'     => /^Unerwünschte\s+Wirkungen$/u, # 11
-        'Überdos.'       => /^Überdosierung$|^Surdosage$/u, # 12
-        'Eigensch.'           => /^Eigenschaften\s*\/\s*Wirkungen($|\s*\(\s*(ATC\-Code|Wirkungsmechanismus|Pharmakodyamik|Klinische\s+Wirksamkeit)\s*\)\s*$)|^Propri.t.s/iu, # 13
-        'Pharm.kinetik'       => /^Pharmakokinetik($|\s*\((Absorption,\s*Distribution,\s*Metabolisms,\s*Elimination\s|Kinetik\s+spezieller\s+Patientengruppen)*\)$)|^Pharmacocin.tique?/iu, # 14
-        'Präklin.'       => /^Präklinische\s+Daten$/u, # 15
-        'Sonstige H.'         => /^Sonstige\s*Hinweise($|\s*\(\s*(Inkompatibilitäten|Beeinflussung\s*diagnostischer\s*Methoden|Haltbarkeit|Besondere\s*Lagerungshinweise|Hinweise\s+für\s+die\s+Handhabung)\s*\)$)|^Remarques/u, # 16
-        'Swissmedic-Nr.'      => /^Zulassungsnummer(n|:|$|\s*\(\s*Swissmedic\s*\)$)/u, # 17
-        'Packungen'           => /^Packungen($|\s*\(\s*mit\s+Angabe\s+der\s+Abgabekategorie\s*\)$)/u, # 18
-        'Reg.Inhaber'         => /^Zulassungsinhaberin($|\s*\(\s*Firma\s+und\s+Sitz\s+gemäss\s*Handelsregisterauszug\s*\))/u, # 19
-        'Stand d. Info.'      => /^Stand\s+der\s+Information$|^Mise\s+.\s+jour$/iu, # 20
+        :de => {
+          'name'                 => /^Name\s+des\s+Präparates$/u, # 1
+          'composition'          => /^Zusammensetzung|Wirkstoffe|Hilsstoffe/u, # 2
+          'galenic_form'         => /^Galenische\s+Form\s*(und|\/)\s*Wirkstoffmenge\s+pro\s+Einheit$/iu, # 3
+          'indications'          => /^Indikationen(\s+|\s*(\/|und)\s*)Anwendungsmöglichkeiten$/u, # 4
+          'usage'                => /^Dosierung\s*(\/|und)\s*Anwendung/u, # 5
+          'contra_indications'   => /^Kontraindikationen($|\s*\(\s*absolute\s+Kontraindikationen\s*\)$)/u, # 6
+          'restrictions'         => /^Warnhinweise\s+und\s+Vorsichtsmassnahmen($|\s*\/\s*(relative\s+Kontraindikationen|Warnhinweise\s*und\s*Vorsichtsmassnahmen)$)/u, # 7
+          'interactions'         => /^Interaktionen$/u, # 8
+          'pregnancy'            => /^Schwangerschaft(,\s*|\s*\/\s*|\s+und\s+)Stillzeit$/u, # 9
+          'driving_ability'      => /^Wirkung\s+auf\s+die\s+Fahrtüchtigkeit\s+und\s+auf\s+das\s+Bedienen\s+von\s+Maschinen$/u, # 10
+          'unwanted_effects'     => /^Unerwünschte\s+Wirkungen$/u, # 11
+          'overdose'             => /^Überdosierung$/u, # 12
+          'effects'              => /^Eigenschaften\s*\/\s*Wirkungen($|\s*\(\s*(ATC\-Code|Wirkungsmechanismus|Pharmakodyamik|Klinische\s+Wirksamkeit)\s*\)\s*$)/iu, # 13
+          'kinetic'              => /^Pharmakokinetik($|\s*\((Absorption,\s*Distribution,\s*Metabolisms,\s*Elimination\s|Kinetik\s+spezieller\s+Patientengruppen)*\)$)/iu, # 14
+          'preclinic'            => /^Präklinische\s+Daten$/u, # 15
+          'other_advice'         => /^Sonstige\s*Hinweise($|\s*\(\s*(Inkompatibilitäten|Beeinflussung\s*diagnostischer\s*Methoden|Haltbarkeit|Besondere\s*Lagerungshinweise|Hinweise\s+für\s+die\s+Handhabung)\s*\)$)|^Remarques/u, # 16
+          'iksnrs'               => /^Zulassungsnummer(n|:|$|\s*\(\s*Swissmedic\s*\)$)/u, # 17
+          'packages'             => /^Packungen($|\s*\(\s*mit\s+Angabe\s+der\s+Abgabekategorie\s*\)$)/u, # 18
+          'registration_owner'   => /^Zulassungsinhaberin($|\s*\(\s*Firma\s+und\s+Sitz\s+gemäss\s*Handelsregisterauszug\s*\))/u, # 19
+          'date'                 => /^Stand\s+der\s+Information$/iu, # 20
+          'fabrication'          => /^Herstellerin/u,
+          'company'              => /^Vertriebsfirma/u,
+        },
+        :fr => {
+          'name'                => /^Nom$/u, # 1
+          'composition'         => /^Composition$/u, # 2
+          'galenic_form'        => /^Forme\s+galénique\s+et\s+quantité\s+de\s+principe\s+actif\s+par\s+unité|^Forme\s*gal.nique/iu, # 3
+          'indications'         => /^Indications/u, # 4
+          'usage'               => /^Posologiei/u, # 5
+          'contra_indications'  => /^Contre\-indications/iu, # 6
+          'restrictions'        => /^Mises/u, # 7
+          'interactions'        => /^Interactions/u, # 8
+          'pregnancy'           => /^Grossesse\s*\/\s*Allaitement/u, # 9
+          'driving_ability'     => /^Effet\s+sur\s+l'aptitude\s+à;\s+la\s+conduite\s+et\s+l'utilisation\s+de\s+machines/u, # 10
+          'unwanted_effects'    => /^Effets/u, # 11
+          'overdose'            => /^Surdosage$/u, # 12
+          'effects'             => /^Propriétés/iu, # 13
+          'kinetic'             => /^Pharmacocinétique$/iu, # 14
+          'preclinic'           => /^Données\s+précliniques$/u, # 15
+          'other_advice'        => /^Remarques/u, # 16
+          'iksnrs'              => /^Numéro\s+d'autorisation$/u, # 17
+          'packages'            => /^Présentation/iu, # 18
+          'registration_owner'  => /^Titulaire\s+de\s+l'autorisation$/u, # 19
+          'date'                => /^Mise à jour/iu, # 20
+          'fabrication'         => /^Fabricant$/u,
+          'company'             => /^Distributeur/u,
+        }
       }
+      if @lang == 'fr' || @lang == :fr
+        chapters[:fr]
+      else
+        chapters[:de]
+      end
     end
+    private
     def escape_id(text)
-      CGI.escape(text.gsub(/&(.)uml;/, '\1e').gsub(/\s*\/\s*|\s+|\/|\-/, '_').gsub(/\./, '').downcase)
+      CGI.escape(text.
+                 gsub(/&(.)uml;/, '\1e').gsub(/'/, '').gsub(/&(eacute|agrave);/, 'e').
+                 gsub(/\s*\/\s*|\s+|\/|\-/, '_').gsub(/\./, '').downcase)
     end
-    def parse_code(text) # swissmedic nummer
-      if text =~ /^\s*(\d{2})(‘|’|'|.|\s*)(\d{3})\s*\(\s*Swiss\s*medic\s*\)(\s*|.)$/iu
-        @code = "%5d" % ($1 + $3)
-      elsif text =~ /^\s*(\d{5})(.*|\s*)\s*\(\s*Swiss\s*medic\s*\)(\s*|.)$/iu
+    def parse_code(text) # swissmedic number
+      if text.gsub(@@figure_pattern, '') =~
+         /^\s*(\d{5})(.*|\s*)\s*\(\s*Swiss\s*medic\s*\)(\s*|.)$/iu
         @code = "%5d" % $1
       else
         nil
@@ -54,18 +109,19 @@ module YDocx
       return markup(:h2, text, {:id => id})
     end
     def parse_title(node, text)
-      if @indecies.empty? and !text.empty? and
-         (node.previous.inner_text.strip.empty? or node.parent.previous.nil?)
+      if @indecies.empty? and !text.empty? and node.previous and
+         (node.parent.previous.nil? or node.previous.inner_text.strip.empty?)
         # The first line as package name
-        @indecies << {:text => 'Titel', :id => 'titel'}
-        return markup(:h1, text, {:id => 'titel'})
+        title = (@lang == 'fr' ? 'Titre' : 'Titel')
+        @indecies << {:text => title, :id => title.downcase}
+        return markup(:h1, text, {:id => title.downcase})
       else
         return nil
       end
     end
     def parse_block(node)
       text = node.inner_text.strip
-      text = optional_escape text
+      text = character_encode(text)
       chapters.each_pair do |chapter, regexp|
         if text =~ regexp
           # allow without line break
@@ -73,13 +129,12 @@ module YDocx
           id = escape_id(chapter)
           @indecies << {:text => chapter, :id => id}
           return parse_heading(text, id)
-        elsif parse_code(text)
-          return nil
         end
       end
       if title = parse_title(node, text)
         return title
       end
+      parse_code(text)
       return nil
     end
   end
@@ -177,11 +232,13 @@ div#container {
       end
       style.gsub(/\s\s+|\n/, ' ')
     end
-    def resolve_path(path)
+    def resolve_path(path) # image src
       if reference = @references.shift
         File.dirname(path) + '/' + reference.basename.to_s
-      else
+      elsif @files.to_s =~ /\d{5}/
         path
+      else
+        @files.join path
       end
     end
   end
@@ -202,10 +259,11 @@ div#container {
       end
       @files
     end
-    def output_file(ext)
+    def output_file(ext) # html
+      lang = (@parser.lang.downcase == 'fr' ? 'fr' : 'de')
       if @parser.code
         filename = @parser.code
-        output_directory.join "#{filename}.#{ext.to_s}"
+        output_directory.join "#{lang}_#{filename}.#{ext.to_s}"
       else # default
         @path.sub_ext(".#{ext.to_s}")
       end

data/lib/ydocx/version.rb

@@ -0,0 +1,5 @@
+#!/usr/bin/env ruby
+
+module YDocx
+   VERSION = '1.2.6'
+end

data/spec/lib/sinovial_spec.rb

@@ -0,0 +1,87 @@
+#!/usr/bin/env ruby
+# encoding: utf-8
+require 'spec_helper'
+
+RSpec.configure do |config|
+  config.expect_with(:rspec) { |c| c.syntax = :should }
+end
+
+describe "ydocx" do
+ 
+  before :all do
+  end
+  
+  before :each do
+    Dir.glob("#{YDcoxHelper::DataDir}/*.xml").each { |file| FileUtils.rm_f(file, :verbose => $VERBOSE) }
+    Dir.glob("#{YDcoxHelper::DataDir}/*.html").each { |file| FileUtils.rm_f(file, :verbose => $VERBOSE) }
+  end
+  
+  after :each do
+  end
+  
+  after :all do
+    Dir.glob("#{YDcoxHelper::DataDir}/*.xml").each { |file| FileUtils.rm_f(file, :verbose => $VERBOSE) }
+    Dir.glob("#{YDcoxHelper::DataDir}/*.html").each { |file| FileUtils.rm_f(file, :verbose => $VERBOSE) }
+  end
+
+  it "should convert sinovial_FR to xml" do
+    require 'ydocx/templates/fachinfo'
+    sinovial_FR = File.join(YDcoxHelper::DataDir, 'Sinovial_FR.docx')
+    File.exists?(sinovial_FR).should be true
+    doc = YDocx::Document.open(sinovial_FR, { :lang => :fr})
+    sinovial_FR_xml = sinovial_FR.sub('.docx', '.xml')
+    doc.to_xml(sinovial_FR_xml, {:format => :fachinfo})
+    out = doc.output_file('xml')    
+    File.exists?(sinovial_FR_xml).should be true
+    doc.parser.lang.to_s.should == 'fr'
+    doc = Nokogiri::XML(open(sinovial_FR_xml))
+    doc.xpath('//chapters/chapter[contains(heading, "Fabricant")]').size.should > 0
+    doc.xpath('//chapters/chapter[contains(heading, "Distributeur")]').size.should > 0
+    doc.xpath('//chapters/chapter[contains(heading, "Remarques particulières")]').size.should > 0
+  end
+
+  it "should convert sinovial_DE to xml" do
+    sinovial_DE = File.join(YDcoxHelper::DataDir, 'Sinovial_DE.docx')
+    File.exists?(sinovial_DE).should be true
+    doc = YDocx::Document.open(sinovial_DE)
+    sinovial_DE_xml = sinovial_DE.sub('.docx', '.xml')
+    doc.to_xml(sinovial_DE_xml, {:format => :fachinfo})
+    out = doc.output_file('xml')
+    File.exists?(sinovial_DE_xml).should be true
+    doc.parser.lang.to_s.should == 'de'
+    doc = Nokogiri::XML(open(sinovial_DE_xml))
+    doc.xpath('//chapters/chapter[contains(heading, "Packung")]').size.should > 0
+    doc.xpath('//chapters/chapter[contains(heading, "Hersteller")]').size.should > 0
+    doc.xpath('//chapters/chapter[contains(heading, "Vertriebsfirma")]').size.should > 0
+  end
+
+  it "should convert sinovial_DE to html" do
+    sinovial_DE = File.join(YDcoxHelper::DataDir, 'Sinovial_DE.docx')
+    File.exists?(sinovial_DE).should be true
+    doc = YDocx::Document.open(sinovial_DE)
+    sinovial_DE_html = sinovial_DE.sub('.docx', '.html')
+    doc.to_html(sinovial_DE_html, {:format => :fachinfo})
+    out = doc.output_file('html')
+    File.exists?(sinovial_DE_html).should be true
+  end
+
+  it "should convert various pseudo fachinfo to xml" do
+            require 'ydocx/templates/fachinfo'
+    files = [ 'Sinovial_0.8_DE.docx', 'Sinovial_0.8_FR.docx',
+              'Sinovial_DE.docx',     'Sinovial_FR.docx',
+            ]
+    files.each {
+      |file|
+        file_name = File.join(YDcoxHelper::DataDir, file)
+        File.exists?(file_name).should be true
+        file.match('_DE') ? lang = 'de' : lang = 'fr'
+        doc = YDocx::Document.open(file_name, { :lang => lang} )
+        file_name_xml = file_name.sub('.docx', '.xml')
+        doc.to_xml(file_name_xml, {:format => :fachinfo})
+        out = doc.output_file('xml')
+        File.exists?(file_name_xml).should be true
+        doc.parser.lang.should == lang
+        doc = Nokogiri::XML(open(file_name_xml))
+    }
+  end
+end

data/spec/spec_helper.rb

@@ -0,0 +1,17 @@
+#!/usr/bin/env ruby
+# encoding: utf-8
+
+require 'simplecov'
+if ENV['COVERAGE']
+	SimpleCov.start
+end
+
+require 'fileutils'
+require 'pp'
+require 'ydocx'
+
+module YDcoxHelper
+  DataDir = File.join(File.dirname(__FILE__), '..', 'spec', 'data')
+end
+require 'rspec'
+require 'ydocx'

metadata

@@ -1,88 +1,64 @@
 --- !ruby/object:Gem::Specification
 name: ydocx
 version: !ruby/object:Gem::Version
-  version: 1.2.1
-  prerelease: 
+  version: 1.2.6
 platform: ruby
 authors:
-- Yasuhiro Asaka, Zeno R.R. Davatz
+- "'Yasuhiro Asak, Zeno R.R. Davatz"
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-07-11 00:00:00.000000000 Z
-dependencies:
-- !ruby/object:Gem::Dependency
-  name: rdoc
-  requirement: &6667440 !ruby/object:Gem::Requirement
-    none: false
-    requirements:
-    - - ~>
-      - !ruby/object:Gem::Version
-        version: '3.10'
-  type: :development
-  prerelease: false
-  version_requirements: *6667440
-- !ruby/object:Gem::Dependency
-  name: hoe
-  requirement: &6667000 !ruby/object:Gem::Requirement
-    none: false
-    requirements:
-    - - ~>
-      - !ruby/object:Gem::Version
-        version: '2.13'
-  type: :development
-  prerelease: false
-  version_requirements: *6667000
+date: 2021-01-20 00:00:00.000000000 Z
+dependencies: []
 description: ''
-email:
-- yasaka@ywesee.com,  zdavatz@ywesee.com
+email: yasaka@ywesee.com, zdavatz@ywesee.com
 executables:
-- docx2html
 - docx2xml
+- docx2html
 extensions: []
-extra_rdoc_files:
-- History.txt
-- Manifest.txt
-- README.txt
+extra_rdoc_files: []
 files:
 - History.txt
 - Manifest.txt
 - README.txt
-- Rakefile
 - bin/docx2html
 - bin/docx2xml
 - lib/ydocx.rb
 - lib/ydocx/builder.rb
+- lib/ydocx/command.rb
 - lib/ydocx/document.rb
 - lib/ydocx/markup_method.rb
 - 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: []
+- lib/ydocx/version.rb
+- spec/data/Sinovial_0.8_DE.docx
+- spec/data/Sinovial_0.8_FR.docx
+- spec/data/Sinovial_DE.docx
+- spec/data/Sinovial_FR.docx
+- spec/lib/sinovial_spec.rb
+- spec/spec_helper.rb
+homepage: https://github.com/zdavatz/ydocx/
+licenses:
+- GPL-3.0
+metadata: {}
 post_install_message: 
-rdoc_options:
-- --main
-- README.txt
+rdoc_options: []
 require_paths:
 - lib
 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: 1.8.15
+rubygems_version: 3.1.2
 signing_key: 
-specification_version: 3
-summary: ''
+specification_version: 4
+summary: Parsing docx files with Ruby and output them as HTML and XML
 test_files: []

data/Rakefile

@@ -1,17 +0,0 @@
-# -*- ruby -*-
-
-require 'rubygems'
-require 'hoe'
-
-# Hoe.plugin :compiler
-# Hoe.plugin :gem_prelude_sucks
-# Hoe.plugin :inline
-# Hoe.plugin :minitest
-# Hoe.plugin :racc
-# Hoe.plugin :rubyforge
-
-Hoe.spec 'ydocx' do
-
-  developer('Yasuhiro Asaka, Zeno R.R. Davatz', 'yasaka@ywesee.com,  zdavatz@ywesee.com')
-
-end