cleaning, and make sure in xml   rather than   is used

8 files changed, 20 insertions, 82 deletions

diff --git a/CHANGELOG b/CHANGELOG
index 1fb7284b..732a1076 100644
--- a/CHANGELOG
+++ b/CHANGELOG
@@ -16,6 +16,8 @@ http://www.jus.uio.no/sisu/pkg/src/sisu_0.60.1.orig.tar.gz
   * html, where no ocn selected, keep ocn id, just remove visible reference to
     ocn, as search result links will continue to function
 
+  * xml, make sure   is used instead of  
+
   * sisu_manual, link to markup and book examples updated/corrected
 
 %% sisu_0.60.0.orig.tar.gz (2007-10-03:40/3)
diff --git a/lib/sisu/v0/html_segments.rb b/lib/sisu/v0/html_segments.rb
index fe2966df..18ed11ec 100644
--- a/lib/sisu/v0/html_segments.rb
+++ b/lib/sisu/v0/html_segments.rb
@@ -178,10 +178,10 @@ module SiSU_HTML_seg
               markup(pg)
             end
           end
-          Seg.new(para,@md).txt
+          #Seg.new(para,@md).txt
           if testforartnum[tracking-1] =~/endnote/
             if printed_endnote_seg == 'n'
-              Seg.new(para,@md).endnote
+              #Seg.new(para,@md).endnote
               printed_endnote_seg='y'
             end
           end
@@ -344,15 +344,11 @@ module SiSU_HTML_seg
         end
       end
     end
-    def txt
-    end
-    def endnote
-    end
     def tail
       format_head_seg=SiSU_HTML_Format_type::Head_seg.new(@md)
       if @md.flag_auto_endnotes
         @@seg[:tail] << format_head_seg.endnote_mark
-        @@seg[:tail] << @@seg_endnotes[@@get_hash_fn] #endnotes deposited at end of individual segments||@|EXTRACTION OF ENDNOTES|
+        @@seg[:tail] << @@seg_endnotes[@@get_hash_fn] #endnotes deposited at end of individual segments ||@|EXTRACTION OF ENDNOTES|
       end
       @@seg[:tail] << '<table summary="whitespace"><tr><td>&nbsp;</td></tr></table>'
       ads=SiSU_HTML_promo::Ad.new(@md)
diff --git a/lib/sisu/v0/html_tune.rb b/lib/sisu/v0/html_tune.rb
index 0b861526..b98f19d8 100644
--- a/lib/sisu/v0/html_tune.rb
+++ b/lib/sisu/v0/html_tune.rb
@@ -278,7 +278,7 @@ module SiSU_Tune
         end
         word
       end
-      @words
+      @words=@words.join(' ')
     end
     def url_markup
       data=@data
@@ -306,9 +306,8 @@ module SiSU_Tune
           end
           if para =~/\{.+?\}((?:https?|file|ftp)\S+|image)/
             @word_mode=para.scan(/\{.+?\}(?:(?:https?|file|ftp)\S+|image)|\S+/)
-            word_mode=urls(@word_mode)
-            words=word_mode.join(' ')
-            para.gsub!(/.+/,words)
+            words=urls(@word_mode)
+            para.gsub!(/.+/m,words)
           end
           if (para !~/^0~|^<:code>/)
             para.gsub!(/\\copyright/i,%{<sup>&copy;</sup>})
diff --git a/lib/sisu/v0/shared_html_lite.rb b/lib/sisu/v0/shared_html_lite.rb
index f258f0a9..85368a28 100644
--- a/lib/sisu/v0/shared_html_lite.rb
+++ b/lib/sisu/v0/shared_html_lite.rb
@@ -93,7 +93,7 @@ module SiSU_Format_Shared
           else m,u=/\{(.+?)\}((?:https?|file|ftp)\S+|image)/.match(word).captures
             d=''
           end
-          case m
+          word=case m
           when /\.png|\.jpg|\.gif|c=|\d+x\d+/
             w,h=/(\d+)x(\d+)/.match(m).captures if m =~/\d+x\d+/
             w=%{width="#{w}"} if w
@@ -117,20 +117,18 @@ module SiSU_Format_Shared
             #ins=%{#{link} <a href="#{u}">[link]</a>#{d}}
             word.gsub!(/\{.+?\}(?:https?|file|ftp)\S+/,ins)
           end
-          word
         else word
         end
         word
       end
-      @words
+      @words=@words.join(' ')
     end
     def markup(para)
       if para !~/^<:code>/
         if para =~/\{.+?\}((?:https?|file|ftp)\S+|image)/
           wm=para.scan(/\{.+?\}(?:(?:https?|file|ftp)\S+|image)|\S+/)
-          word_mode=urls(wm)
-          words=word_mode.join(' ')
-          para.gsub!(/.+/,words)
+          words=urls(wm)
+          para.gsub!(/.+/m,words)
         end
         para.gsub!(/\b[_\\]((?:https?|file|ftp):\/\/\S+?\.[^'"><\s]+?)([;.,]?(?:\s|$))/,'<a href="\1" target="_top">\1</a>\2') #http ftp matches escaped, no decoration
         para.gsub!(/((?:^|\s)[}])((?:https?|file|ftp):\/\/\S+?\.[^'"><\s]+?)([;.,]?(?:\s|$))/,'\1<a href="\2" target="_top">\2</a>\3') #special case \{ e.g. \}http://url
diff --git a/lib/sisu/v0/shared_xml.rb b/lib/sisu/v0/shared_xml.rb
index 135f368e..4b9ccb12 100644
--- a/lib/sisu/v0/shared_xml.rb
+++ b/lib/sisu/v0/shared_xml.rb
@@ -355,22 +355,18 @@ module SiSU_XML_munge
       para.gsub!(/<0;\w\d+;[um]\d+><#@dp:#@dp>/,'')
       if para !~/^<:code>/
         #embeds a red-bullet image -->
-        #para.gsub!(/^(<:i[1-9]>\s*_\*)\s+/,'\1 <image xmlns:xlink="http://www.w3.org/1999/xlink" xlink:type="simple" xlink:actuate="onLoad" xlink:show="embed" xlink:href="/usr/share/sisu/image/bullet_09.png" width="12" height="12" alt="*" /> ')
-        #para.gsub!(/^_\*\s+/,'<image xmlns:xlink="http://www.w3.org/1999/xlink" xlink:type="simple" xlink:actuate="onLoad" xlink:show="embed" xlink:href="/usr/share/sisu/image/bullet_09.png" width="12" height="12" alt="*" /> ')
         para.gsub!(/(^|\s)\{\s*(\S+?\.(?:jpg|png|gif))\s+(\d+)x(\d+)(\s+[^}]+)?\}(https?:\/\/\S+)/,%{\\1<image xmlns:xlink="http://www.w3.org/1999/xlink" xlink:type="simple" xlink:actuate="onLoad" xlink:show="embed" xlink:href="#{@dir.url.images_local}/\\2" width="\\3" height="\\4" />[\\2] \\5})
         para.gsub!(/(^|\s)\{\s*(\S+?\.(?:jpg|png|gif))(\s+[^}]+)?\}(https?:\/\/\S+)/,%{\\1<image xmlns:xlink="http://www.w3.org/1999/xlink" xlink:type="simple" xlink:actuate="onLoad" xlink:show="embed" xlink:href="#{@dir.url.images_local}/\\2"/>\\2})
         para.gsub!(/(^|\s)\{([^}]+)\}(https?:\/\/[^"><]+?)([,.:;"><]?(?=\s|$))/,
           '\1<link xmlns:xlink="http://www.w3.org/1999/xlink" xlink:type="simple" xlink:href="\3">\2</link>\4') #watch, compare html_tune
-        #para.gsub!(/\B\{([^}]+)\}(https?:\/\/[^"><]+?)([,.:;"><]?(?:\s|$))/,
-        #  '<link xmlns:xlink="http://www.w3.org/1999/xlink" xlink:type="simple" xlink:href="\2">\1</link>\3') #watch, compare html_tune
         para.gsub!(/(^|\s)((?:https?|file|ftp):\/\/\S+?\.[^'"><\s]+?)([;.,]?(?=\s|$))/,
           %{\\1#{@url_brace.xml_open}<link xmlns:xlink="http://www.w3.org/1999/xlink" xlink:type="simple" xlink:href="\\2">\\2</link>#{@url_brace.xml_close}\\3})
-        #para.gsub!(/\b((?:https?|file|ftp):\/\/\S+?\.[^'"><\s]+?)([.,]?(?:\s|$))/, #also works
-          #%{#{@url_brace.xml_open}<link xmlns:xlink="http://www.w3.org/1999/xlink" xlink:type="simple" xlink:href="\\1">\\1</link>#{@url_brace.xml_close}\\2})
         para.gsub!(/\b[_\\]((?:https?|file|ftp):\/\/\S+?\.[^'"><\s]+?)([;.,]?(?:\s|$))/,'<link xmlns:xlink="http://www.w3.org/1999/xlink" xlink:type="simple" xlink:href="\1">\1</link>\2') #escaped urls not linked, deal with later
-        #para.gsub!(/(^|\s)[_\\]((?:https?|ftp):\/\/\S+?\.[^'"><\s]+?)([.,]?(?:\s|$))/,'\1<link xmlns:xlink="http://www.w3.org/1999/xlink" xlink:type="simple" xlink:href="\2">\2</link>\3') #escaped urls not linked, deal with later
-        para.gsub!(/&nbsp;/,' ') #clean
-      else para.gsub!(/_</m,'&lt;'); para.gsub!(/_>/m,'&gt;') #code-block: angle brackets special characters
+        para.gsub!(/&nbsp;/,'&#160;')
+        #para.gsub!(/&nbsp;/,' ') #clean
+      else
+        para.gsub!(/_</m,'&lt;'); para.gsub!(/_>/m,'&gt;') #code-block: angle brackets special characters
+        para.gsub!(/&nbsp;/,'&#160;')
       end
       para
     end
@@ -386,7 +382,8 @@ module SiSU_XML_munge
       para.gsub!(/&([^;]{1,5})/,'&amp;\1') #sort, rough estimate, revisit #WATCH found in node not sax
       para.gsub!(/\{(\S+?\.(?:png|jpg|gif)) .+?\}(?:(?:https?|file|ftp):\/\/\S+|image)/,
         "<image.path>#{@dir.url.images_local}\/\\1</image.path>")
-      para.gsub!(/&nbsp;/,' ')
+      para.gsub!(/&nbsp;/,'&#160;')
+      #para.gsub!(/&nbsp;/,' ') #clean
       wordlist=para.scan(/\S+|\n/) #\n needed for tables, check though added 2005w17
       para=tidywords(wordlist).join(' ').strip
       para
@@ -412,15 +409,6 @@ module SiSU_XML_tags #Format
       if @md.dc_title                                                            # DublinCore 1 - title
         @rdf_title=%{    dc.title="#{seg_name}#{@md.dc_title}"\n}
         @dc_title=%{  <meta name="dc.title" content="#{seg_name}#{@md.dc_title}" />\n}
-        #if @md.subtitle
-        #  @rdf_title=%{    dc.title="#{seg_name}#{@md.title} - #{@md.subtitle}"\n}
-        #  @title=%{<meta name="dc.title" } +
-        #    %{content="#{seg_name}#{@md.title} - #{@md.subtitle}" />\n}
-        #else
-        #  @rdf_title=%{    dc.title="#{seg_name}#{@md.title}"\n}
-        #  @title=%{<meta name="dc.title" } +
-        #    %{content="#{seg_name}#{@md.title}" />\n}
-        #end
       end
       if @md.dc_creator                                                          # DublinCore 2 - creator/author (author)
         @rdf_creator=%{    dc.creator="#{@md.dc_creator}"\n}
diff --git a/lib/sisu/v0/texpdf_format.rb b/lib/sisu/v0/texpdf_format.rb
index 55f1e19a..51681ce7 100644
--- a/lib/sisu/v0/texpdf_format.rb
+++ b/lib/sisu/v0/texpdf_format.rb
@@ -798,9 +798,6 @@ WOK
           if word =~/\\\{(.+?)\\\}((?:https?|file|ftp):\/\/\S+?\.[^'"><\s]+?)([;.,]?(?:\s|$))/
             r=%r/\\\{(.+?)\\?\}((?:https?|file|ftp):\/\/\S+?\.[^'"><\s]+?)(?:[;.,]?(?:\s|$)|(?:\s|$))/
             d=/\\\{.+?\\?\}(?:https?|file|ftp):\/\/\S+?\.[^'"><\s]+?([;.,]?(?:\s|$))/.match(word).captures.to_s
-#          if word =~/\\\{(.+?)\\\}((?:https?|file|ftp)\S+?)([;.,]?(?:\s|$))/
-#            r=%r/\\\{(.+?)\\?\}((?:https?|file|ftp):\S+?)(?:[;.,]?(?:\s|$)|(?:\s|$))/
-#            d=/\\\{.+?\\?\}(?:https?|file|ftp):\S+?([;.,]?(?:\s|$))/.match(word).captures.to_s
           else
             r=%r/\\\{(.+?)\\?\}((?:https?|file|ftp):\S+)/
             d=''
@@ -820,8 +817,6 @@ WOK
             when /a5/;     @tx.a5
             else           @tx.a4
             end
-            ##dm=SiSU_Env::Get_init.instance.tex(@md.papersize).dimensions #test!
-            #dm=SiSU_Viz::TeX.new(@md.papersize).dimensions
             width=if orientation =~/portrait/ #value is not currently passed
               width=if width.to_i > dm.portrait.img_px; dm.portrait.img_px
               else                                      width
@@ -875,7 +870,7 @@ WOK
       @words.join
     end
     def http(orientation)
-      wm=@string.dup.scan(/\\\{.+?\\\}(?:(?:https?|file|ftp):\S+|image)|\w+\s*|./m) #first match of interest others passed through
+      wm=@string.dup.scan(/\\\{.+?\\\}(?:(?:https?|file|ftp):\S+|image)|\w+\s*|./m)
       @string=SiSU_TeX_Pdf::Format_text_object.new(@md,wm).http_word_mode(orientation)
     end
     def language
diff --git a/lib/sisu/v0/xhtml.rb b/lib/sisu/v0/xhtml.rb
index 1d7d63d1..6b91f34f 100644
--- a/lib/sisu/v0/xhtml.rb
+++ b/lib/sisu/v0/xhtml.rb
@@ -190,25 +190,6 @@ WOK
         else ''
         end
       end
-      def xml_structure_ee(para='',lv='',ocn='',hname='') #embedded endnotes
-        lv=lv.to_i
-        n=lv - 1
-        n3=lv + 2
-        lv=nil if lv == 0
-        embedded_endnotes(para)
-        if para[@regx]
-          paragraph="#{para[@regx,2]}"
-          util=SiSU_text_utils::Wrap.new(paragraph,70)
-          wrapped=util.line_wrap
-        end
-        @@xml[:body] << "#{@tab*0}<object>" << "\n" if para[@regx]
-        @@xml[:body] << "#{@tab*1}<ocn>#{para[@regx,3]}</ocn>" << "\n" if para[@regx,3]
-        @@xml[:body] << if lv; %{#{@tab*1}<text class="h#{lv}">#{wrapped}</text>\n} << "\n"
-        else                   %{#{@tab*1}<text class="norm">#{wrapped}</text>\n} # main text, contents, body KEEP
-        end
-        ##@@xml[:body] << "#{@tab*1}<text>#{para[@regx,2]}</text>\n" if para[@regx,2] # old unwrapped main text, contents, body KEEP
-        @@xml[:body] << "#{@tab*0}</object>" << "\n" if para[@regx]
-      end
       def xml_structure(para='',lv='',ocn='',hname='',type='norm') #extracted endnotes
         lv=lv.to_i
         n=lv - 1
diff --git a/lib/sisu/v0/xml.rb b/lib/sisu/v0/xml.rb
index 2dd25491..b230ec9f 100644
--- a/lib/sisu/v0/xml.rb
+++ b/lib/sisu/v0/xml.rb
@@ -214,26 +214,6 @@ WOK
         end
         @@xml[:sc]=sc
       end
-      def xml_structure_ee(para='',lv='',ocn='',hname='') #embedded endnotes
-        lv=lv.to_i
-        n=lv - 1
-        n3=lv + 2
-        lv=nil if lv == 0
-        embedded_endnotes(para)
-        if para[@regx]
-          paragraph="#{para[@regx,2]}"
-          util=SiSU_text_utils::Wrap.new(paragraph,70)
-          wrapped=util.line_wrap
-        end
-        @@xml[:body] << "#{@tab*0}<object>" << "\n" if para[@regx]
-        @@xml[:body] << "#{@tab*1}<ocn>#{para[@regx,3]}</ocn>" << "\n" if para[@regx,3]
-        @@xml[:body] << if lv; %{#{@tab*1}<text class="h#{lv}">#{wrapped}</text>\n} << "\n"
-        else                   %{#{@tab*1}<text class="norm">#{wrapped}</text>\n} # main text, contents, body KEEP
-        end
-        #@@xml[:body] << "#{@tab*1}<text>#{wrapped}</text>\n" # main text, contents, body KEEP
-        ##@@xml[:body] << "#{@tab*1}<text>#{para[@regx,2]}</text>\n" if para[@regx,2] # old unwrapped main text, contents, body KEEP
-        @@xml[:body] << "#{@tab*0}</object>" << "\n" if para[@regx]
-      end
       def xml_structure(para='',lv='',ocn='',hname='',type='norm') #extracted endnotes
         lv=lv.to_i
         n=lv - 1
@@ -249,7 +229,6 @@ WOK
         @@xml[:body] << if para[@regx,3]; %{#{@tab*0}<object id="#{para[@regx,3]}">} << "\n"
         else                              "#{@tab*0}<object>" << "\n" if para[@regx]
         end
-        #@@xml[:body] << "#{@tab*1}<heading>#{lv}</heading>" << "\n" if lv
         @@xml[:body] << "#{@tab*1}<ocn>#{para[@regx,3]}</ocn>" << "\n" if para[@regx,3]
         @@xml[:body] << if lv; %{#{@tab*1}<text class="h#{lv}">\n#{@tab*2}#{wrapped}\n#{@tab*1}</text>\n} << "\n"
         else                   %{#{@tab*1}<text class="#{type}">\n#{@tab*2}#{wrapped}\n#{@tab*1}</text>\n} # main text, contents, body KEEP