v5 v6: ao, set document tags

4 files changed, 50 insertions, 28 deletions

diff --git a/lib/sisu/v5/ao_doc_str.rb b/lib/sisu/v5/ao_doc_str.rb
index f7168c26..dd7f32f3 100644
--- a/lib/sisu/v5/ao_doc_str.rb
+++ b/lib/sisu/v5/ao_doc_str.rb
@@ -163,10 +163,13 @@ module SiSU_AO_DocumentStructureExtract
           str=str.gsub(/(^|[ ])\*~([a-z0-9._-]+)(?=[ #{Mx[:br_nl]}]|$)/i,
               "\\1#{Mx[:tag_o]}\\2#{Mx[:tag_c]}").
             gsub(/ [ ]+/i,' ')
-          tags=str.scan(/#{Mx[:tag_o]}(\S+?)#{Mx[:tag_c]}/).flatten
+          tags=str.scan(/#{Mx[:tag_o]}(\S+?)#{Mx[:tag_c]}/).flatten.uniq
           str=str.gsub(/[ ]?#{Mx[:tag_o]}\S+?#{Mx[:tag_c]}[ ]?/,' ') #may be issues with spaces would leave one, but "code" blocks?
         end
         tags=nametag ? (tags << nametag) : tags
+        tags.each do |t|
+          t.gsub!(/[^a-z0-9._-]/,'')
+        end
       end
       [str,tags]
     end
diff --git a/lib/sisu/v5/ao_numbering.rb b/lib/sisu/v5/ao_numbering.rb
index 8caa5ccc..984675cc 100644
--- a/lib/sisu/v5/ao_numbering.rb
+++ b/lib/sisu/v5/ao_numbering.rb
@@ -78,6 +78,14 @@ module SiSU_AO_Numbering
       data=set_heading_top(data) unless @md.set_heading_top
       [data,tags_map,ocn_html_seg_map]
     end
+    def set_tags(tags,tag)
+      tags=if not tag.empty? \
+      and tag !~/^\d+$/
+        tag=tag.gsub(/[^a-z0-9._-]/,'')
+        [tag,tags].flatten
+      else tags
+      end
+    end
     def number_plaintext_para(data)
       @tuned_file=[]
       data.each do |dob|
@@ -168,10 +176,10 @@ module SiSU_AO_Numbering
               and not @md.seg_names.include?(title_no)
                 if dob.ln==no1
                   dob.name="#{title_no}" if not dob.name
-                  dob.tags=[title_no,dob.tags].flatten if title_no !~/^\d+$/  #check whether will work across file types with stop signs
+                  dob.tags=set_tags(dob.tags,title_no)
                   tag=dob.obj.gsub(/(Article|Clause|Section|Chapter)\s+/,"\\1_#{title_no}").downcase
                   tag=heading_tag_clean(tag)
-                  dob.tags=[tag,dob.tags].flatten if tag !~/^\d+$/ #check whether will work across file types with stop signs
+                  dob.tags=set_tags(dob.tags,tag)
                   dob.obj=(dob.obj =~/(Article|Clause|Section)\s+/) \
                   ? (dob.obj.gsub(/(Article|Clause|Section)\s+/,"\\1 #{title_no} "))
                   : (dob.obj.gsub(/^/,"#{title_no}. ")) #fix stop later
@@ -179,7 +187,7 @@ module SiSU_AO_Numbering
                 if dob.ln !=no1 \
                 and dob.obj =~/^[\d.]+\s/ #fix -> if the title starts with a numbering scheme, do not auto-number, review
                   dob.name ="#{title_no}" if not dob.name
-                  dob.tags=[title_no,dob.tags].flatten if title_no !~/^\d+$/ #check whether will work across file types with stop signs
+                  dob.tags=set_tags(dob.tags,title_no)
                   dob.obj=dob.obj.gsub(/^/,"#{title_no}. ")
                 end
                 @md.seg_names << title_no
@@ -187,28 +195,28 @@ module SiSU_AO_Numbering
               if dob.ln!=no1 \
               and dob.name!~/^[a-z_\.]+$/ \
               and dob.obj !~/[A-Z]\.?\s/ #bug -> tmp fix, excludes A. B. C. lettering, but not roman numerals, is arbitrary, review required # not fixed, work on
-                dob.tags=[title_no,dob.tags].flatten if title_no !~/^\d+$/ #check whether will work across file types with stop signs
+                dob.tags=set_tags(dob.tags,title_no)
                 dob.obj=dob.obj.gsub(/^/i,"#{title_no}. ")
               end
             end
             if dob.ln==no1         #watch because here you change dob.name
-              dob.tags=["h#{title_no}",dob.tags].flatten #check whether will work across file types with stop signs
+              dob.tags=set_tags(dob.tags,"h#{title_no}")
             end
             if dob.ln==no2         #watch because here you change dob.name
               t_no2+=1; t_no3=0
               title_no="#{t_no1}.#{t_no2}"
-              dob.tags=["h#{title_no}",dob.tags].flatten #check whether will work across file types with stop signs
+              dob.tags=set_tags(dob.tags,"h#{title_no}")
               dob=number_sub_heading(dob,no2,title_no)
             end
             if dob.ln==no3         #watch because here you change dob.name
               t_no3+=1
               title_no="#{t_no1}.#{t_no2}.#{t_no3}"
-              dob.tags=["h#{title_no}",dob.tags].flatten #check whether will work across file types with stop signs
+              dob.tags=set_tags(dob.tags,"h#{title_no}")
               dob=number_sub_heading(dob,no3,title_no)
             end
           elsif dob.ln.to_s =~/^[0-6]/ \
           and dob.name =~ /^[\w-]+-/ # endnotes, watch2005# endnotes, watch2005
-            dob.tags=[dob.name,dob.tags].flatten if dob.name !~/^\d+$/  #check whether will work across file types with stop signs
+            dob.tags=set_tags(dob.tags,dob.name)
             dob.name.gsub(/^([a-z_\.]+)-$/,'\1')
           end
         elsif dob.is ==:heading \
@@ -219,13 +227,13 @@ module SiSU_AO_Numbering
           and dob.ln.to_s =~/^[0-9]/ \
           and dob.obj =~ /^([\d\.]+)/ #risky (must be unique) consider output to 4~~\d instead of 4~\d
             dob.name=$1
-            dob.tags=[dob.name,dob.tags].flatten if dob.name !~/^\d+$/ #check whether will work across file types with stop signs
+            dob.tags=set_tags(dob.tags,dob.name)
           end
           if @md.toc_lev_limit
           end
         elsif defined? dob.name \
         and  dob.name
-          dob.tags=[dob.name,dob.tags].flatten if dob.name !~/^\d+$/ #check whether will work across file types with stop signs
+          dob.tags=set_tags(dob.tags,dob.name)
         end
         dob.tags=dob.tags.uniq if defined? dob.tags
         @tuned_file << dob
@@ -312,7 +320,7 @@ module SiSU_AO_Numbering
             if @md.seg_names.is_a?(Array) \
             and not @md.seg_names.include?(possible_seg_name)
               dob.name=possible_seg_name
-              dob.tags=[dob.name,dob.tags].flatten if dob.name !~/^\d+$/
+              dob.tags=set_tags(dob.tags,dob.name)
               @md.seg_names << possible_seg_name
             elsif (@md.opt.act[:verbose_plus][:set]==:on \
             or @md.opt.act[:maintenance][:set]==:on)
@@ -323,7 +331,7 @@ module SiSU_AO_Numbering
           and dob.name                                     #extract segment name from embedded document structure info
             if @md.seg_names.is_a?(Array) \
             and not @md.seg_names.include?(dob.name)
-              dob.tags=[dob.name,dob.tags].flatten if dob.name !~/^\d+$/
+              dob.tags=set_tags(dob.tags,dob.name)
               @md.seg_names << dob.name
             end
           end
@@ -334,7 +342,7 @@ module SiSU_AO_Numbering
             if @md.seg_names.is_a?(Array) \
             and not @md.seg_names.include?(segn_auto)
              dob.name=segn_auto
-             dob.tags=[dob.name,dob.tags].flatten if dob.name !~/^\d+$/ #check whether will work across file types with stop signs
+             dob.tags=set_tags(dob.tags,dob.name)
               @md.seg_names << segn_auto
             else puts 'segment name (numbering) error'
             end
diff --git a/lib/sisu/v6/ao_doc_str.rb b/lib/sisu/v6/ao_doc_str.rb
index 0cdd553b..d8c012fd 100644
--- a/lib/sisu/v6/ao_doc_str.rb
+++ b/lib/sisu/v6/ao_doc_str.rb
@@ -163,10 +163,13 @@ module SiSU_AO_DocumentStructureExtract
           str=str.gsub(/(^|[ ])\*~([a-z0-9._-]+)(?=[ #{Mx[:br_nl]}]|$)/i,
               "\\1#{Mx[:tag_o]}\\2#{Mx[:tag_c]}").
             gsub(/ [ ]+/i,' ')
-          tags=str.scan(/#{Mx[:tag_o]}(\S+?)#{Mx[:tag_c]}/).flatten
+          tags=str.scan(/#{Mx[:tag_o]}(\S+?)#{Mx[:tag_c]}/).flatten.uniq
           str=str.gsub(/[ ]?#{Mx[:tag_o]}\S+?#{Mx[:tag_c]}[ ]?/,' ') #may be issues with spaces would leave one, but "code" blocks?
         end
         tags=nametag ? (tags << nametag) : tags
+        tags.each do |t|
+          t.gsub!(/[^a-z0-9._-]/,'')
+        end
       end
       [str,tags]
     end
diff --git a/lib/sisu/v6/ao_numbering.rb b/lib/sisu/v6/ao_numbering.rb
index 23d9fd73..acb2351a 100644
--- a/lib/sisu/v6/ao_numbering.rb
+++ b/lib/sisu/v6/ao_numbering.rb
@@ -78,6 +78,14 @@ module SiSU_AO_Numbering
       data=set_heading_top(data) unless @md.set_heading_top
       [data,tags_map,ocn_html_seg_map]
     end
+    def set_tags(tags,tag)
+      tags=if not tag.empty? \
+      and tag !~/^\d+$/
+        tag=tag.gsub(/[^a-z0-9._-]/,'')
+        [tag,tags].flatten
+      else tags
+      end
+    end
     def number_plaintext_para(data)
       @tuned_file=[]
       data.each do |dob|
@@ -168,10 +176,10 @@ module SiSU_AO_Numbering
               and not @md.seg_names.include?(title_no)
                 if dob.ln==no1
                   dob.name="#{title_no}" if not dob.name
-                  dob.tags=[title_no,dob.tags].flatten if title_no !~/^\d+$/  #check whether will work across file types with stop signs
+                  dob.tags=set_tags(dob.tags,title_no)
                   tag=dob.obj.gsub(/(Article|Clause|Section|Chapter)\s+/,"\\1_#{title_no}").downcase
                   tag=heading_tag_clean(tag)
-                  dob.tags=[tag,dob.tags].flatten if tag !~/^\d+$/ #check whether will work across file types with stop signs
+                  dob.tags=set_tags(dob.tags,tag)
                   dob.obj=(dob.obj =~/(Article|Clause|Section)\s+/) \
                   ? (dob.obj.gsub(/(Article|Clause|Section)\s+/,"\\1 #{title_no} "))
                   : (dob.obj.gsub(/^/,"#{title_no}. ")) #fix stop later
@@ -179,7 +187,7 @@ module SiSU_AO_Numbering
                 if dob.ln !=no1 \
                 and dob.obj =~/^[\d.]+\s/ #fix -> if the title starts with a numbering scheme, do not auto-number, review
                   dob.name ="#{title_no}" if not dob.name
-                  dob.tags=[title_no,dob.tags].flatten if title_no !~/^\d+$/ #check whether will work across file types with stop signs
+                  dob.tags=set_tags(dob.tags,title_no)
                   dob.obj=dob.obj.gsub(/^/,"#{title_no}. ")
                 end
                 @md.seg_names << title_no
@@ -187,28 +195,28 @@ module SiSU_AO_Numbering
               if dob.ln!=no1 \
               and dob.name!~/^[a-z_\.]+$/ \
               and dob.obj !~/[A-Z]\.?\s/ #bug -> tmp fix, excludes A. B. C. lettering, but not roman numerals, is arbitrary, review required # not fixed, work on
-                dob.tags=[title_no,dob.tags].flatten if title_no !~/^\d+$/ #check whether will work across file types with stop signs
+                dob.tags=set_tags(dob.tags,title_no)
                 dob.obj=dob.obj.gsub(/^/i,"#{title_no}. ")
               end
             end
             if dob.ln==no1         #watch because here you change dob.name
-              dob.tags=["h#{title_no}",dob.tags].flatten #check whether will work across file types with stop signs
+              dob.tags=set_tags(dob.tags,"h#{title_no}")
             end
             if dob.ln==no2         #watch because here you change dob.name
               t_no2+=1; t_no3=0
               title_no="#{t_no1}.#{t_no2}"
-              dob.tags=["h#{title_no}",dob.tags].flatten #check whether will work across file types with stop signs
+              dob.tags=set_tags(dob.tags,"h#{title_no}")
               dob=number_sub_heading(dob,no2,title_no)
             end
             if dob.ln==no3         #watch because here you change dob.name
               t_no3+=1
               title_no="#{t_no1}.#{t_no2}.#{t_no3}"
-              dob.tags=["h#{title_no}",dob.tags].flatten #check whether will work across file types with stop signs
+              dob.tags=set_tags(dob.tags,"h#{title_no}")
               dob=number_sub_heading(dob,no3,title_no)
             end
           elsif dob.ln.to_s =~/^[0-6]/ \
           and dob.name =~ /^[\w-]+-/ # endnotes, watch2005# endnotes, watch2005
-            dob.tags=[dob.name,dob.tags].flatten if dob.name !~/^\d+$/  #check whether will work across file types with stop signs
+            dob.tags=set_tags(dob.tags,dob.name)
             dob.name.gsub(/^([a-z_\.]+)-$/,'\1')
           end
         elsif dob.is ==:heading \
@@ -219,13 +227,13 @@ module SiSU_AO_Numbering
           and dob.ln.to_s =~/^[0-9]/ \
           and dob.obj =~ /^([\d\.]+)/ #risky (must be unique) consider output to 4~~\d instead of 4~\d
             dob.name=$1
-            dob.tags=[dob.name,dob.tags].flatten if dob.name !~/^\d+$/ #check whether will work across file types with stop signs
+            dob.tags=set_tags(dob.tags,dob.name)
           end
           if @md.toc_lev_limit
           end
         elsif defined? dob.name \
         and  dob.name
-          dob.tags=[dob.name,dob.tags].flatten if dob.name !~/^\d+$/ #check whether will work across file types with stop signs
+          dob.tags=set_tags(dob.tags,dob.name)
         end
         dob.tags=dob.tags.uniq if defined? dob.tags
         @tuned_file << dob
@@ -312,7 +320,7 @@ module SiSU_AO_Numbering
             if @md.seg_names.is_a?(Array) \
             and not @md.seg_names.include?(possible_seg_name)
               dob.name=possible_seg_name
-              dob.tags=[dob.name,dob.tags].flatten if dob.name !~/^\d+$/
+              dob.tags=set_tags(dob.tags,dob.name)
               @md.seg_names << possible_seg_name
             elsif (@md.opt.act[:verbose_plus][:set]==:on \
             or @md.opt.act[:maintenance][:set]==:on)
@@ -323,7 +331,7 @@ module SiSU_AO_Numbering
           and dob.name                                     #extract segment name from embedded document structure info
             if @md.seg_names.is_a?(Array) \
             and not @md.seg_names.include?(dob.name)
-              dob.tags=[dob.name,dob.tags].flatten if dob.name !~/^\d+$/
+              dob.tags=set_tags(dob.tags,dob.name)
               @md.seg_names << dob.name
             end
           end
@@ -334,7 +342,7 @@ module SiSU_AO_Numbering
             if @md.seg_names.is_a?(Array) \
             and not @md.seg_names.include?(segn_auto)
              dob.name=segn_auto
-             dob.tags=[dob.name,dob.tags].flatten if dob.name !~/^\d+$/ #check whether will work across file types with stop signs
+             dob.tags=set_tags(dob.tags,dob.name)
               @md.seg_names << segn_auto
             else puts 'segment name (numbering) error'
             end