Updated french char handling, lang in filename

lib/ydocx/parser.rb

@@ -55,7 +55,7 @@ def apply_fonts(rpr, text)
       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 @@ def apply_align(rpr, text)
       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 @@ def optional_replace(code)
         #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
@@ -230,7 +233,7 @@ def parse_paragraph(node)
             end
             unless r.xpath('w:sym').empty?
               code = r.xpath('w:sym').first['char'].downcase # w:char
-              content << optional_replace(code)
+              content << character_replace(code)
               pos += 1
             end
             if !r.xpath('w:pict').empty? or !r.xpath('w:drawing').empty?
@@ -276,6 +279,7 @@ def parse_table(node)
     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')

lib/ydocx/templates/fachinfo.rb

@@ -6,6 +6,7 @@
 module YDocx
   class Parser
     attr_accessor :code, :lang
+    @@figure_pattern = /‘|’|'|´/
     def init
       @image_path = 'image'
       @code = nil
@@ -93,7 +94,7 @@ def escape_id(text)
                  gsub(/\s*\/\s*|\s+|\/|\-/, '_').gsub(/\./, '').downcase)
     end
     def parse_code(text) # swissmedic number
-      if text.gsub(/‘|’|'|´/, '') =~
+      if text.gsub(@@figure_pattern, '') =~
          /^\s*(\d{5})(.*|\s*)\s*\(\s*Swiss\s*medic\s*\)(\s*|.)$/iu
         @code = "%5d" % $1
       else
@@ -116,21 +117,20 @@ def parse_title(node, text)
     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
           # next if !node.previous.inner_text.empty? and !node.next.inner_text.empty?
           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
@@ -256,9 +256,10 @@ def output_directory
       @files
     end
     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