v5 v6: html, (output) cleaning

6 files changed, 40 insertions, 10 deletions

diff --git a/lib/sisu/v5/html.rb b/lib/sisu/v5/html.rb
index ef62b9bd..4a911d2c 100644
--- a/lib/sisu/v5/html.rb
+++ b/lib/sisu/v5/html.rb
@@ -623,7 +623,8 @@ WOK
         toc_shared << format_txt_obj.center_bold
         @segtoc << format_txt_obj.center_bold
         if defined? @md.creator.author
-          creator_endnote=@md.creator.author.gsub(/(\*+)/,
+          creator=SiSU_HTML_Tune::CleanHTML.new(@md.creator.author).clean_for_html
+          creator_endnote=creator.gsub(/(\*+)/,
             %{&nbsp;<sup><a href="#notes">\\1</a></sup>})
           tmp_head=creator_endnote + "\n"
           txt_obj={ txt: tmp_head }
diff --git a/lib/sisu/v5/html_format.rb b/lib/sisu/v5/html_format.rb
index a933d091..e244bbb8 100644
--- a/lib/sisu/v5/html_format.rb
+++ b/lib/sisu/v5/html_format.rb
@@ -502,7 +502,8 @@ WOK
     end
     def rights
       def all
-        rights=@md.rights.all.gsub(/^\s*Copyright\s+\(C\)/,'Copyright <sup>&copy;</sup>&nbsp;')
+        rights=SiSU_HTML_Tune::CleanHTML.new(@md.rights.all).clean_for_html
+        rights=rights.gsub(/^\s*Copyright\s+\(C\)/,'Copyright <sup>&copy;</sup>&nbsp;')
         %{<p class="small_left">Rights: #{rights}</p>
 <p />}
       end
@@ -977,7 +978,9 @@ WOK
     end
     def para_form_css(tag,attrib)                                                    # regular paragraphs shaped here
       ul=ulc=''
-      ul,ulc="<ul>\n  ","\n  </ul>" if @tag =~/li/
+      if tag =~/li/
+        ul,ulc="<ul>\n  ","\n  </ul>"
+      end
       %{
 <div class="substance">
   #{@p_num.ocn_display}
@@ -1250,7 +1253,7 @@ WOK
         note=$1
         note=note.gsub(/[\s]+/m,' ')
         txt=txt.gsub(/(?:#{Mx[:en_a_o]}.+?#{Mx[:en_a_c]}|#{Mx[:en_b_o]}.+?#{Mx[:en_b_c]})\s*/m,' ').
-          gsub(/<a[\n\s]+name="-\d+"[\n\s]+href="#_\d+">&nbsp;<sup>\d+<\/sup>&nbsp;/m,'')
+          gsub(/<a[\s]+name="-\d+"[\s]+href="#_\d+">&nbsp;<sup>\d+<\/sup>&nbsp;/m,'')
       end
       %{<#{tag} class="#{attrib}">
     <a href="##{@ocn_html_identifier}#{@ocn}"><i>#{txt}</i></a> #{note}
diff --git a/lib/sisu/v5/html_tune.rb b/lib/sisu/v5/html_tune.rb
index 84336861..39cbb708 100644
--- a/lib/sisu/v5/html_tune.rb
+++ b/lib/sisu/v5/html_tune.rb
@@ -104,6 +104,16 @@ module SiSU_HTML_Tune
     def initialize(html='')
       @html=html
     end
+    def clean_for_html
+      html=@html
+      str=if html.is_a?(String)
+        html
+      else html.obj
+      end
+      str=str.gsub(/#{Mx[:gl_o]}(#[0-9]{3})#{Mx[:gl_c]}/u,'&\1;').
+        gsub(/#{Mx[:gl_o]}#([a-z]{2,4})#{Mx[:gl_c]}/u,'&\1;').
+        gsub(/[<]/m,'&lt;').gsub(/[>]/m,'&gt;')
+    end
     def clean
       html=@html
       str=if html.is_a?(String)
@@ -111,7 +121,8 @@ module SiSU_HTML_Tune
       else html.obj
       end
       str=str.gsub(/#{Mx[:gl_o]}(#[0-9]{3})#{Mx[:gl_c]}/u,'&\1;').
-        gsub(/#{Mx[:gl_o]}#([a-z]{2,4})#{Mx[:gl_c]}/u,'&\1;')
+        gsub(/#{Mx[:gl_o]}#([a-z]{2,4})#{Mx[:gl_c]}/u,'&\1;').
+        gsub(/[\\]{2}/m,'<br />')
     end
   end
   class Tune
diff --git a/lib/sisu/v6/html.rb b/lib/sisu/v6/html.rb
index f1651326..8c9869e2 100644
--- a/lib/sisu/v6/html.rb
+++ b/lib/sisu/v6/html.rb
@@ -623,7 +623,8 @@ WOK
         toc_shared << format_txt_obj.center_bold
         @segtoc << format_txt_obj.center_bold
         if defined? @md.creator.author
-          creator_endnote=@md.creator.author.gsub(/(\*+)/,
+          creator=SiSU_HTML_Tune::CleanHTML.new(@md.creator.author).clean_for_html
+          creator_endnote=creator.gsub(/(\*+)/,
             %{&nbsp;<sup><a href="#notes">\\1</a></sup>})
           tmp_head=creator_endnote + "\n"
           txt_obj={ txt: tmp_head }
diff --git a/lib/sisu/v6/html_format.rb b/lib/sisu/v6/html_format.rb
index 63271d95..de5a7144 100644
--- a/lib/sisu/v6/html_format.rb
+++ b/lib/sisu/v6/html_format.rb
@@ -502,7 +502,8 @@ WOK
     end
     def rights
       def all
-        rights=@md.rights.all.gsub(/^\s*Copyright\s+\(C\)/,'Copyright <sup>&copy;</sup>&nbsp;')
+        rights=SiSU_HTML_Tune::CleanHTML.new(@md.rights.all).clean_for_html
+        rights=rights.gsub(/^\s*Copyright\s+\(C\)/,'Copyright <sup>&copy;</sup>&nbsp;')
         %{<p class="small_left">Rights: #{rights}</p>
 <p />}
       end
@@ -977,7 +978,9 @@ WOK
     end
     def para_form_css(tag,attrib)                                                    # regular paragraphs shaped here
       ul=ulc=''
-      ul,ulc="<ul>\n  ","\n  </ul>" if @tag =~/li/
+      if tag =~/li/
+        ul,ulc="<ul>\n  ","\n  </ul>"
+      end
       %{
 <div class="substance">
   #{@p_num.ocn_display}
@@ -1250,7 +1253,7 @@ WOK
         note=$1
         note=note.gsub(/[\s]+/m,' ')
         txt=txt.gsub(/(?:#{Mx[:en_a_o]}.+?#{Mx[:en_a_c]}|#{Mx[:en_b_o]}.+?#{Mx[:en_b_c]})\s*/m,' ').
-          gsub(/<a[\n\s]+name="-\d+"[\n\s]+href="#_\d+">&nbsp;<sup>\d+<\/sup>&nbsp;/m,'')
+          gsub(/<a[\s]+name="-\d+"[\s]+href="#_\d+">&nbsp;<sup>\d+<\/sup>&nbsp;/m,'')
       end
       %{<#{tag} class="#{attrib}">
     <a href="##{@ocn_html_identifier}#{@ocn}"><i>#{txt}</i></a> #{note}
diff --git a/lib/sisu/v6/html_tune.rb b/lib/sisu/v6/html_tune.rb
index 516deaa5..42568a36 100644
--- a/lib/sisu/v6/html_tune.rb
+++ b/lib/sisu/v6/html_tune.rb
@@ -104,6 +104,16 @@ module SiSU_HTML_Tune
     def initialize(html='')
       @html=html
     end
+    def clean_for_html
+      html=@html
+      str=if html.is_a?(String)
+        html
+      else html.obj
+      end
+      str=str.gsub(/#{Mx[:gl_o]}(#[0-9]{3})#{Mx[:gl_c]}/u,'&\1;').
+        gsub(/#{Mx[:gl_o]}#([a-z]{2,4})#{Mx[:gl_c]}/u,'&\1;').
+        gsub(/[<]/m,'&lt;').gsub(/[>]/m,'&gt;')
+    end
     def clean
       html=@html
       str=if html.is_a?(String)
@@ -111,7 +121,8 @@ module SiSU_HTML_Tune
       else html.obj
       end
       str=str.gsub(/#{Mx[:gl_o]}(#[0-9]{3})#{Mx[:gl_c]}/u,'&\1;').
-        gsub(/#{Mx[:gl_o]}#([a-z]{2,4})#{Mx[:gl_c]}/u,'&\1;')
+        gsub(/#{Mx[:gl_o]}#([a-z]{2,4})#{Mx[:gl_c]}/u,'&\1;').
+        gsub(/[\\]{2}/m,'<br />')
     end
   end
   class Tune