v5: digests cleaning

2 files changed, 18 insertions, 142 deletions

diff --git a/data/doc/sisu/CHANGELOG_v5 b/data/doc/sisu/CHANGELOG_v5
index c25d71ee..41771585 100644
--- a/data/doc/sisu/CHANGELOG_v5
+++ b/data/doc/sisu/CHANGELOG_v5
@@ -49,6 +49,8 @@ http://www.jus.uio.no/sisu/pkg/src/sisu_5.3.5.orig.tar.xz
 * metadata, pdf, remove link to manifest page where --no-manifest is used
   (Closes: #744378)
 
+* digests, cleaning
+
 * added sisu.org emacs:evil:org mode notes related to sisu development
 
 %% 5.3.4.orig.tar.xz (2014-02-14:06/5)
diff --git a/lib/sisu/v5/digests.rb b/lib/sisu/v5/digests.rb
index 3acd51eb..b3dd40bd 100644
--- a/lib/sisu/v5/digests.rb
+++ b/lib/sisu/v5/digests.rb
@@ -71,7 +71,7 @@ module SiSU_DigestView
       @fnb=@opt.fnb
       @@endnotes_para=[]
       @@dg=nil
-      @dg=@@dg ||=SiSU_Env::InfoEnv.new.digest.type
+      @dg=@@dg ||=SiSU_Env::InfoEnv.new.digest(opt).type
       @particulars=SiSU_Particulars::CombinedSingleton.instance.get_all(opt)
     end
     def read
@@ -114,8 +114,8 @@ module SiSU_DigestView
         @particulars=particulars
         @data,@env,@md=@particulars.ao_array,@particulars.env,@particulars.md
         SiSU_Env::FileOp.new(@md).mkdir
-        @@dg ||=@env.digest.type
-        @@dl ||=@env.digest.length
+        @@dg ||=@env.digest(@md.opt).type
+        @@dl ||=@env.digest(@md.opt).length
         @dg,@dl=@@dg,@@dl
         l=SiSU_Env::StandardiseLanguage.new(@md.opt.lng).language
         @language=l[:n]
@@ -158,135 +158,6 @@ module SiSU_DigestView
         txt=txt.gsub(/([()])/,"\\\\\\1")
       end
       def message_digest
-        data=@data
-        sys=SiSU_Env::SystemCall.new
-        @p=[]
-        @g,@v,@r='','',''
-        data.each do |para|
-          x=nil
-          y,para_endnotes=[],[]
-          if para =~/#{Mx[:id_o]}~(\d+);(?:(?:\w|[0-6]:)\d+);(?:\w\d+)#{Mx[:id_c]}#{Mx[:id_o]}([0-9a-f]{#{@dl}}):([0-9a-f]{#{@dl}})#{Mx[:id_c]}/
-            ocn,d_clean,d_all=$1,$2,$3
-            @ocn=ocn unless ocn.to_i==0
-            para=para.gsub(/#{Mx[:fa_bold_o]}(.+?)#{Mx[:fa_bold_c]}/,'<b>\1</b>').
-              gsub(/#{Mx[:fa_italics_o]}(.+?)#{Mx[:fa_italics_c]}/,'<i>\1</i>').
-              gsub(/#{Mx[:fa_underscore_o]}(.+?)#{Mx[:fa_underscore_c]}/,'<u>\1</u>')
-            if para=~/#{Mx[:en_a_o]}[\d*+]+.+?#{Mx[:id_o]}[0-9a-f]{#{@dl}}#{Mx[:id_c]}#{Mx[:en_a_c]}/
-              para_endnotes << para.scan(/(?:#{Mx[:en_a_o]}|#{Mx[:en_b_o]})([\d*+]+).+?#{Mx[:id_o]}([0-9a-f]{#{@dl}})#{Mx[:id_c]}(?:#{Mx[:en_a_c]}|#{Mx[:en_b_c]})/)
-            end
-            images=[]
-            if para !~/^%+\s/ \
-            and para =~/(?:^|[^_\\])#{Mx[:lnk_o]}(\S+\.(png|jpg|gif))\s.+?#{Mx[:lnk_c]}(?:#{Mx[:url_o]}\S+?#{Mx[:url_c]}|image)/
-              images=para.scan(/(?:^|[^_\\])#{Mx[:lnk_o]}(\S+\.(?:png|jpg|gif))\s.+?#{Mx[:lnk_c]}(?:#{Mx[:url_o]}\S+?#{Mx[:url_c]}|image)/).flatten
-            else nil
-            end
-            x=case para
-            when /^#{Mx[:meta_o]}title#{Mx[:meta_c]}/
-              "\n" << '  '*0 << '@' << '  '*9
-            when /^#{Mx[:meta_o]}subtitle#{Mx[:meta_c]}/
-              "\n" << '  '*1 << '@' << '  '*8
-            when /^#{Mx[:lv_o]}1:/ #fix Mx[:lv_o]
-              "\n" << '  '*2 << ':A ' << '  '*6 << '- ' << ocn << ' '*(10-ocn.length) << d_clean << ' ' << d_all
-            when /^#{Mx[:lv_o]}2:/ #fix Mx[:lv_o]
-              "\n" << '  '*3 << ':B ' << '  '*5 << '- ' << ocn << ' '*(10-ocn.length) << d_clean << ' ' << d_all
-            when /^#{Mx[:lv_o]}3:/ #fix Mx[:lv_o]
-              "\n" << '  '*4 << ':C ' << '  '*4 << '- ' << ocn << ' '*(10-ocn.length) << d_clean << ' ' << d_all
-            when /^#{Mx[:lv_o]}4:/ #fix Mx[:lv_o]
-              "\n" << '  '*5 << '1' << '  '*4 << '- ' << ocn << ' '*(10-ocn.length) << d_clean << ' ' << d_all
-            when /^#{Mx[:lv_o]}5:/ #fix Mx[:lv_o]
-              "\n" << '  '*6 << '2' << '  '*3 << '- ' << ocn << ' '*(10-ocn.length) << d_clean << ' ' << d_all
-            when /^#{Mx[:lv_o]}6:/ #fix Mx[:lv_o]
-              "\n" << '  '*7 << '3' << '  '*2 << '- ' << ocn << ' '*(10-ocn.length) << d_clean << ' ' << d_all
-            else
-              if para =~/MD5\(\S+?\.sst\)=\s*<u>([0-9a-f]{#{@dl}})<\/u>/                                        #watch
-                @n,@s=/MD5\((\S+?\.sst)\)=\s*<u>([0-9a-f]{#{@dl}})<\/u>/.match(para)[1,2]
-              end
-              x=unless ocn =~ /^0$/
-                if images \
-                and images.length > 0 # then get path of image & produce digest
-                  @image_name,@image_dgst,@img=[],[],[]
-                  images.each do |i|
-                    image_source=if FileTest.file?("#{@env.path.image_source_include_local}/#{i}")
-                      @env.path.image_source_include_local
-                    elsif FileTest.file?("#{@env.path.image_source_include_remote}/#{i}")
-                      @env.path.image_source_include_remote
-                    elsif FileTest.file?("#{@env.path.image_source_include}/#{i}")
-                      @env.path.image_source_include
-                    else
-                      SiSU_Screen::Ansi.new(@md.opt.act[:color_state][:set],"ERROR - image:", %{"#{i}" missing}, "search locations: #{@env.path.image_source_include_local}, #{@env.path.image_source_include_remote} and #{@env.path.image_source_include}").error2 unless @md.opt.act[:quiet][:set]==:on
-                      nil
-                    end
-                    @img << /\S+\.(png|jpg|gif)/.match(i)[1]
-                    not_found_msg='image not found'
-                    if image_source
-                      para_image = image_source + '/' + i
-                      @image_name << i
-                      @image_dgst << (@dg =~/^sha(?:2|256)$/) \
-                      ? sys.sha256(para_image)
-                      : sys.md5(para_image)
-                    else
-                      @image_name << '  '*16 + i + ' [image missing]'
-                      @image_dgst << ''
-                      @image_dgst[1]=not_found_msg + ' '*(32-not_found_msg.length)
-                    end
-                  end
-                  line= "\n" + '  '*9 + ' - ' + ocn + ' '*(10-ocn.length) + d_clean + ' ' + d_all + "\n"
-                  line_image=[]
-                  c=0
-                  @image_name.each do |ok|
-                    line_image << %{                     #{@img[c]}       #{@image_dgst[c][1]}                                  #{@image_name[c]}}
-                    c +=1
-                  end
-                  line=line + line_image.join("\n")
-                else "\n" + '  '*9 + ' - ' + ocn + ' '*(10-ocn.length) + d_clean + ' ' + d_all
-                end
-              else
-                prefix=''
-                metad=[@tr.full_title,@tr.author,@tr.translator,@tr.illustrator,@tr.prepared_by,@tr.digitized_by,@tr.description,@tr.subject,@tr.abstract,@tr.publisher,@tr.contributor,@tr.date_created,@tr.date_issued,@tr.date_available,@tr.date_modified,@tr.date_valid,@tr.date,@tr.type,@tr.format,@tr.rights,@tr.identifier,@tr.source,@tr.language,@tr.language_original,@tr.relation,@tr.coverage,@tr.keywords,@tr.comments,@tr.cls_loc,@tr.cls_dewey,@tr.cls_gutenberg,@tr.cls_isbn,@tr.prefix_a,@tr.prefix_b,@tr.sourcefile,@tr.sourcefile_digest,@tr.last_generated,@tr.sisu_version,@tr.ruby_version,@tr.sc_number,@tr.sc_date,'Generated by: ','Ruby version: ']
-                metad.each do |n|
-                  m=rgx_txt(n)
-                  if m=~/\S+/ \
-                  and para=~/^#{m}:/
-                    x,o=0,18
-                    while x < 2; o = o + 2
-                      x=o - n.length
-                    end
-                    space=' '*x
-                    prefix="#{n.downcase}#{space}"
-                    break
-                  else prefix='  '*9
-                  end
-                end
-                m_title=rgx_txt(@tr.full_title)
-                m_author=rgx_txt(@tr.author)
-                m_sourcefile_digest=rgx_txt(@tr.sourcefile_digest)
-                m_sisu_version=rgx_txt(@tr.sisu_version)
-                m_last_generated=rgx_txt(@tr.last_generated)
-                m_ruby_version=rgx_txt(@tr.ruby_version)
-                case para
-                when /#{m_title}: /
-                  @t=/#{m_title}: (.+?)#{Mx[:id_o]}~\d;(?:\w|[0-6]:)\d+;\w\d+#{Mx[:id_c]}#{Mx[:id_o]}[0-9a-f]{#{@dl}}:[0-9a-f]{#{@dl}}#{Mx[:id_c]}/.match(para)[1].gsub(/<\/?u>/,'').strip
-                when /#{m_author}: /
-                  @c=/#{m_author}: (.+?)#{Mx[:id_o]}~\d;(?:\w|[0-6]:)\d+;\w\d+#{Mx[:id_c]}#{Mx[:id_o]}[0-9a-f]{#{@dl}}:[0-9a-f]{#{@dl}}#{Mx[:id_c]}/.match(para)[1].gsub(/<\/?u>/,'').strip
-                when /#{m_sourcefile_digest}.+?<u>/                                        #watch
-                  dgst_extra="\n" << ' '*21 << 'source' << ' '*4 << @md.dgst[1] << ' '*34 << @md.fns
-                when /#{m_sisu_version}: /
-                  @v=/#{m_sisu_version}: (.+?)#{Mx[:id_o]}~\d;(?:\w|[0-6]:)\d+;\w\d+#{Mx[:id_c]}#{Mx[:id_o]}[0-9a-f]{#{@dl}}:[0-9a-f]{#{@dl}}#{Mx[:id_c]}/.match(para)[1].gsub(/<\/?u>/,'').strip
-                when /#{m_last_generated}: /
-                  @g=/#{m_last_generated}: (.+?)#{Mx[:id_o]}~\d;(?:\w|[0-6]:)\d+;\w\d+#{Mx[:id_c]}#{Mx[:id_o]}[0-9a-f]{#{@dl}}:[0-9a-f]{#{@dl}}#{Mx[:id_c]}/.match(para)[1].gsub(/<\/?u>/,'').strip
-                when /#{m_ruby_version}: /
-                  @r=/#{m_ruby_version}: (.+?)#{Mx[:id_o]}~\d;(?:\w|[0-6]:)\d+;\w\d+#{Mx[:id_c]}#{Mx[:id_o]}[0-9a-f]{#{@dl}}:[0-9a-f]{#{@dl}}#{Mx[:id_c]}/.match(para)[1].gsub(/<\/?u>/,'').strip
-                end
-                dgst_extra ||=''
-                "\n" << prefix << ' - ' << ocn << ' '*(10-ocn.length) << d_clean << ' ' << d_all << dgst_extra << "\n"
-              end
-            end
-            para_endnotes[0].each { |e| y << "\n" + ' '*(28-e[0].length) + "[#{e[0].to_s}] #{e[1].to_s}" } if para_endnotes[0]
-            if y; digests(x,y)
-            else  digests(x)
-            end
-          end
-        end
         manifest="#{@env.url.root}/#{@md.fnb}/sisu_manifest.html"
         a=%{ocn      digest clean (no markup/notes),#{@sp*33}digest all (includes markup & endnotes)\n}
         description("#{@md.title.full}\n")
@@ -340,18 +211,20 @@ module SiSU_DigestView
         data.each do |t_o|
           if t_o.is==:heading
             x=case t_o.ln
-            when 1 then l[1] +=1 #fix Mx[:lv_o]
+            when 1 then l[0] +=1 #fix Mx[:lv_o]
               '  '*0 << ':A'
-            when 2 then l[2] +=1 #fix Mx[:lv_o]
+            when 1 then l[1] +=1 #fix Mx[:lv_o]
               '  '*1 << ':B'
-            when 3 then l[3] +=1 #fix Mx[:lv_o]
+            when 2 then l[2] +=1 #fix Mx[:lv_o]
               '  '*2 << ':C'
+            when 3 then l[3] +=1 #fix Mx[:lv_o]
+              '  '*3 << ':D'
             when 4 then l[4] +=1 #fix Mx[:lv_o]
-              '  '*3 << '1'
+              '  '*4 << '1'
             when 5 then l[5] +=1 #fix Mx[:lv_o]
-              '  '*4 << '2'
+              '  '*5 << '2'
             when 6 then l[6] +=1 #fix Mx[:lv_o]
-              '  '*5 << '3'
+              '  '*6 << '3'
             else nil
             end
           end
@@ -363,9 +236,10 @@ module SiSU_DigestView
         ao_structure_summary("document structure[*]\n")
         [0,1,2,3,4,5,6].each do |y|
           v=case y
-          when 1 then ':A'
-          when 2 then ':B'
-          when 3 then ':C'
+          when 0 then ':A'
+          when 1 then ':B'
+          when 2 then ':C'
+          when 3 then ':D'
           when 4 then '1 '
           when 5 then '2 '
           when 6 then '3 '
@@ -374,7 +248,7 @@ module SiSU_DigestView
         end
         ao_structure_summary("objects (ocn) = #{ocn}\n")
         ao_structure_summary("endnotes      = #{endnotes}\n")
-        ao_structure_summary("  [*] number of headers (@) and of each heading level (:A to :C and 1 to 3)\n")
+        ao_structure_summary("  [*] number of headers (@) and of each heading level (:A to :D and 1 to 3)\n")
       end
       def supplementary
         if defined? @md.sc_number \