diff options
Diffstat (limited to 'lib/sisu/v3/dal_doc_str.rb')
| -rw-r--r-- | lib/sisu/v3/dal_doc_str.rb | 320 |
1 files changed, 164 insertions, 156 deletions
diff --git a/lib/sisu/v3/dal_doc_str.rb b/lib/sisu/v3/dal_doc_str.rb index 7bc36e11..b4a98edd 100644 --- a/lib/sisu/v3/dal_doc_str.rb +++ b/lib/sisu/v3/dal_doc_str.rb @@ -56,7 +56,7 @@ ** Description: document abstraction =end -module SiSU_document_structure_extract +module SiSU_DAL_DocumentStructureExtract class Instantiate < SiSU_Param::Parameters::Instructions @@flag={} #Beware!! def initialize @@ -69,9 +69,9 @@ module SiSU_document_structure_extract @@flag={} #Beware!! def initialize(md,data) @md,@data=md,data - Instantiate.new - @pb=SiSU_document_structure::Object_layout.new.break(Hx[:br_page]) - @pbn=SiSU_document_structure::Object_layout.new.break(Hx[:br_page_new]) + SiSU_DAL_DocumentStructureExtract::Instantiate.new + @pb=SiSU_DAL_DocumentStructure::ObjectLayout.new.break(Hx[:br_page]) + @pbn=SiSU_DAL_DocumentStructure::ObjectLayout.new.break(Hx[:br_page_new]) end def ln_get(lv) ln=case lv @@ -131,11 +131,11 @@ module SiSU_document_structure_extract if str.nil? else if str =~/(?:^|[ ])\*~([a-z0-9._-]+)(?=[ #{Mx[:br_nl]}]|$)/ - str.gsub!(/(^|[ ])\*~([a-z0-9._-]+)(?=[ #{Mx[:br_nl]}]|$)/i, - "\\1#{Mx[:tag_o]}\\2#{Mx[:tag_c]}") - str.gsub!(/ [ ]+/i,' ') + str=str.gsub(/(^|[ ])\*~([a-z0-9._-]+)(?=[ #{Mx[:br_nl]}]|$)/i, + "\\1#{Mx[:tag_o]}\\2#{Mx[:tag_c]}"). + gsub(/ [ ]+/i,' ') tags=str.scan(/#{Mx[:tag_o]}(\S+?)#{Mx[:tag_c]}/).flatten - str.gsub!(/[ ]?#{Mx[:tag_o]}\S+?#{Mx[:tag_c]}[ ]?/,' ') #may be issues with spaces would leave one, but "code" blocks? + str=str.gsub(/[ ]?#{Mx[:tag_o]}\S+?#{Mx[:tag_c]}[ ]?/,' ') #may be issues with spaces would leave one, but "code" blocks? end tags=nametag ? (tags << nametag) : tags end @@ -148,7 +148,7 @@ module SiSU_document_structure_extract @@counter,@verse_count=0,0 @metadata={} @data.each do |t_o| - t_o.gsub!(/(?:\n\s*\n)+/m,"\n") unless @@flag['code'] + t_o=t_o.gsub(/(?:\n\s*\n)+/m,"\n") unless @@flag['code'] if t_o !~/^(?:code|poem|alt|group|block)\{|^\}(?:code|poem|alt|group|block)|^(?:table\{|\{table)[ ~]/ \ and not @@flag['code'] \ and not @@flag['poem'] \ @@ -158,7 +158,7 @@ module SiSU_document_structure_extract and not @@flag['table'] unless t_o =~/^(?:@\S+?:|%+)\s/ # extract book index for paragraph if any idx=if t_o=~/^=\{(.+)\}\s*$\Z/m; m=$1 - t_o.gsub!(/\n=\{.+\}\s*$\Z/m,'') + t_o=t_o.gsub(/\n=\{.+\}\s*$\Z/m,'') m else nil end @@ -173,7 +173,7 @@ module SiSU_document_structure_extract when /^%+\s/ #comment t_o=if t_o=~/^%+\s+(.+)/ h={obj: $1} - SiSU_document_structure::Object_comment.new.comment(h) + SiSU_DAL_DocumentStructure::ObjectComment.new.comment(h) else nil end when /^:?([A-C1-6])\~/ #heading / lv @@ -184,19 +184,19 @@ module SiSU_document_structure_extract note=endnote_test?(obj) obj,tags=extract_tags(obj) h={ lv: lv, ln: ln, obj: obj, idx: idx, tags: tags } - SiSU_document_structure::Object_heading.new.heading(h) + SiSU_DAL_DocumentStructure::ObjectHeading.new.heading(h) elsif t_o=~/^:?[A-C1-6]\~(\S+?)-\s+(.+)/m name,obj=$1,$2 note=endnote_test?(obj) obj,tags=extract_tags(obj) h={ lv: lv, name: name, obj: obj, idx: idx, autonum_: false, tags: tags} - SiSU_document_structure::Object_heading.new.heading(h) + SiSU_DAL_DocumentStructure::ObjectHeading.new.heading(h) elsif t_o=~/^:?[A-C1-6]\~(\S+)\s+(.+)/m name,obj=$1,$2 note=endnote_test?(obj) obj,tags=extract_tags(obj,name) h={ lv: lv, name: name, obj: obj, idx: idx, tags: tags } - SiSU_document_structure::Object_heading.new.heading(h) + SiSU_DAL_DocumentStructure::ObjectHeading.new.heading(h) else nil end when /^_(?:[1-9]!?|[1-9]?\*)\s+/ #indented and/or bullet paragraph @@ -213,7 +213,7 @@ module SiSU_document_structure_extract obj,tags=extract_tags(obj) unless obj=~/\A\s*\Z/m h={ bullet_: bullet, hang: hang, indent: indent, obj: obj, idx: idx, note_: note, image_: image, tags: tags } - SiSU_document_structure::Object_para.new.paragraph(h) + SiSU_DAL_DocumentStructure::ObjectPara.new.paragraph(h) end else nil end @@ -230,23 +230,23 @@ module SiSU_document_structure_extract obj,tags=extract_tags(obj) unless obj=~/\A\s*\Z/m h={ hang: hang, indent: indent, obj: obj, idx: idx, note_: note, image_: image, tags: tags } - SiSU_document_structure::Object_para.new.paragraph(h) + SiSU_DAL_DocumentStructure::ObjectPara.new.paragraph(h) end else nil end when /^[<\[](?:br)?:(?:pa?r|o(?:bj|---)?)[>\]]\s*$/ #[br:par] #[br:obj] - SiSU_document_structure::Object_layout.new.break(Hx[:br_obj]) + SiSU_DAL_DocumentStructure::ObjectLayout.new.break(Hx[:br_obj]) when /^(?:[<\[](?:br)?:pg[>\]]|<?:pb>?)\s*$/ #[br:pg] - SiSU_document_structure::Object_layout.new.break(Hx[:br_page]) + SiSU_DAL_DocumentStructure::ObjectLayout.new.break(Hx[:br_page]) when /^[<\[](?:br)?:pg?n[>\]]\s*$/ #[br:pgn] - SiSU_document_structure::Object_layout.new.break(Hx[:br_page_new]) + SiSU_DAL_DocumentStructure::ObjectLayout.new.break(Hx[:br_page_new]) else #paragraph image=image_test(t_o) note=endnote_test?(t_o) obj,tags=extract_tags(t_o) unless obj=~/\A\s*\Z/m h={ bullet_: false, indent: 0, hang: 0, obj: obj, idx: idx, note_: note, image_: image, tags: tags } - SiSU_document_structure::Object_para.new.paragraph(h) + SiSU_DAL_DocumentStructure::ObjectPara.new.paragraph(h) end end elsif not @@flag['code'] @@ -255,30 +255,30 @@ module SiSU_document_structure_extract @@counter=1 @codeblock_numbered=(t_o =~/^code\{#/) ? true : false h={ obj: 'code block start' } #introduce a counter - t_o=SiSU_document_structure::Object_comment.new.comment(h) #t_o=SiSU_document_structure::Object_layout.new.insert(h) + t_o=SiSU_DAL_DocumentStructure::ObjectComment.new.comment(h) #t_o=SiSU_DAL_DocumentStructure::ObjectLayout.new.insert(h) elsif t_o =~/^poem\{/ @@flag['poem']=true h={ obj: 'poem start' } #introduce a counter - t_o=SiSU_document_structure::Object_comment.new.comment(h) #t_o=SiSU_document_structure::Object_layout.new.insert(h) + t_o=SiSU_DAL_DocumentStructure::ObjectComment.new.comment(h) #t_o=SiSU_DAL_DocumentStructure::ObjectLayout.new.insert(h) tuned_file << t_o elsif t_o =~/^group\{/ @@flag['group']=true h={ obj: 'group text start' } #introduce a counter - t_o=SiSU_document_structure::Object_comment.new.comment(h) #t_o=SiSU_document_structure::Object_layout.new.insert(h) + t_o=SiSU_DAL_DocumentStructure::ObjectComment.new.comment(h) #t_o=SiSU_DAL_DocumentStructure::ObjectLayout.new.insert(h) tuned_file << t_o elsif t_o =~/^block\{/ @@flag['block']=true h={ obj: 'block text start' } #introduce a counter - t_o=SiSU_document_structure::Object_comment.new.comment(h) #t_o=SiSU_document_structure::Object_layout.new.insert(h) + t_o=SiSU_DAL_DocumentStructure::ObjectComment.new.comment(h) #t_o=SiSU_DAL_DocumentStructure::ObjectLayout.new.insert(h) tuned_file << t_o elsif t_o =~/^alt\{/ @@flag['alt']=true h={ obj: 'alt text start' } #introduce a counter - t_o=SiSU_document_structure::Object_comment.new.comment(h) #t_o=SiSU_document_structure::Object_layout.new.insert(h) + t_o=SiSU_DAL_DocumentStructure::ObjectComment.new.comment(h) #t_o=SiSU_DAL_DocumentStructure::ObjectLayout.new.insert(h) tuned_file << t_o elsif t_o =~/^(?:table\{|\{table)[ ~]/ h={ obj: 'table start' } #introduce a counter - ins=SiSU_document_structure::Object_comment.new.comment(h) #ins=SiSU_document_structure::Object_layout.new.insert(h) + ins=SiSU_DAL_DocumentStructure::ObjectComment.new.comment(h) #ins=SiSU_DAL_DocumentStructure::ObjectLayout.new.insert(h) tuned_file << ins if t_o=~/^table\{(?:~h)?\s+/ @@flag['table']=true @@ -304,7 +304,7 @@ module SiSU_document_structure_extract cols=nil rws.each do |r| cols=(cols ? cols : (r.scan('|').length) +1) - r.gsub!(/\s*\|\s*/m,"#{Mx[:tc_p]}") #r.gsub!(/\|/m,"#{Mx[:tc_p]}") + r=r.gsub(/\s*\|\s*/m,"#{Mx[:tc_p]}") #r.gsub!(/\|/m,"#{Mx[:tc_p]}") rows += r + Mx[:tc_c] end col=[] @@ -318,10 +318,10 @@ module SiSU_document_structure_extract cols.times { col << width } end h={ head_: hd, cols: cols, widths: col, obj: rows, idx: idx, tags: tags } - t_o=SiSU_document_structure::Object_table.new.table(h) unless h.nil? + t_o=SiSU_DAL_DocumentStructure::ObjectTable.new.table(h) unless h.nil? tuned_file << t_o h={ obj: 'table end' } #introduce a counter - t_o=SiSU_document_structure::Object_comment.new.comment(h) + t_o=SiSU_DAL_DocumentStructure::ObjectComment.new.comment(h) t_o elsif t_o=~/^\{table(?:~h)?\s+/ m1,m2,hd=nil,nil,nil @@ -337,14 +337,14 @@ module SiSU_document_structure_extract rws=tbl.split(/\n/) rows='' rws.each do |r| - r.gsub!(/\s*\|\s*/m,"#{Mx[:tc_p]}") #r.gsub!(/\|/m,"#{Mx[:tc_p]}") + r=r.gsub(/\s*\|\s*/m,"#{Mx[:tc_p]}") #r.gsub!(/\|/m,"#{Mx[:tc_p]}") rows += r + Mx[:tc_c] end h={ head_: hd, cols: col.length, widths: col, obj: rows, idx: idx, tags: tags } - t_o=SiSU_document_structure::Object_table.new.table(h) unless h.nil? + t_o=SiSU_DAL_DocumentStructure::ObjectTable.new.table(h) unless h.nil? tuned_file << t_o h={ obj: 'table end' } #introduce a counter - t_o=SiSU_document_structure::Object_comment.new.comment(h) + t_o=SiSU_DAL_DocumentStructure::ObjectComment.new.comment(h) t_o end end @@ -356,18 +356,18 @@ module SiSU_document_structure_extract @@flag['table']=false headings,columns,widths,idx=@h[:head_],@h[:cols],@h[:widths],@h[:idx] @h={ head_: headings, cols: columns, widths: widths, idx: idx, obj: @rows } - t_o=SiSU_document_structure::Object_table.new.table(@h) + t_o=SiSU_DAL_DocumentStructure::ObjectTable.new.table(@h) tuned_file << t_o @h,@rows=nil,'' t_o h={ obj: 'table end' } #introduce a counter - t_o=SiSU_document_structure::Object_comment.new.comment(h) #t_o=SiSU_document_structure::Object_layout.new.insert(h) + t_o=SiSU_DAL_DocumentStructure::ObjectComment.new.comment(h) #t_o=SiSU_DAL_DocumentStructure::ObjectLayout.new.insert(h) t_o else if t_o !~/^table\{/ \ and not t_o.nil? - t_o.gsub!(/^\n+/m,'') #check added for ruby 1.9.2 not needed in 1.8 series (tested in v2) - t_o.gsub!(/\n+/m,"#{Mx[:tc_p]}") + t_o=t_o.gsub(/^\n+/m,''). #check added for ruby 1.9.2 not needed in 1.8 series (tested in v2) + gsub(/\n+/m,"#{Mx[:tc_p]}") @rows += t_o + Mx[:tc_c] end t_o=nil @@ -379,11 +379,11 @@ module SiSU_document_structure_extract obj=@tuned_code.join("\n") tags=[] h={ obj: obj, tags: tags, number_: @codeblock_numbered } - t_o=SiSU_document_structure::Object_block_txt.new.code(h) + t_o=SiSU_DAL_DocumentStructure::ObjectBlockTxt.new.code(h) @tuned_code=[] tuned_file << t_o h={ obj: 'code block end' } #introduce a counter - t_o=SiSU_document_structure::Object_comment.new.comment(h) #t_o=SiSU_document_structure::Object_layout.new.insert(h) + t_o=SiSU_DAL_DocumentStructure::ObjectComment.new.comment(h) #t_o=SiSU_DAL_DocumentStructure::ObjectLayout.new.insert(h) end if @@flag['code'] \ and t_o.class==String \ @@ -392,7 +392,7 @@ module SiSU_document_structure_extract @line_mode=sub_array.scan(/.+/) @line_mode=[] sub_array.scan(/.+/) {|w| @line_mode << w if w =~/[\S]+/} - t_o=SiSU_document_structure_extract::Build.new(@md,@line_mode).build_lines('code').join + t_o=SiSU_DAL_DocumentStructureExtract::Build.new(@md,@line_mode).build_lines('code').join @tuned_code << t_o t_o=nil end @@ -404,37 +404,37 @@ module SiSU_document_structure_extract and t_o =~/^\}poem/ @@flag['poem']=false h={ obj: 'poem end' } #introduce a counter - t_o=SiSU_document_structure::Object_comment.new.comment(h) #t_o=SiSU_document_structure::Object_layout.new.insert(h) + t_o=SiSU_DAL_DocumentStructure::ObjectComment.new.comment(h) #t_o=SiSU_DAL_DocumentStructure::ObjectLayout.new.insert(h) elsif ( @@flag['group'] \ and t_o =~/^\}group/ ) @@flag['group']=false obj,tags=extract_tags(@tuned_block.join("\n")) h={ obj: obj, tags: tags } @tuned_block=[] - t_o=SiSU_document_structure::Object_block_txt.new.group(h) + t_o=SiSU_DAL_DocumentStructure::ObjectBlockTxt.new.group(h) tuned_file << t_o h={ obj: 'group text end' } #introduce a counter - t_o=SiSU_document_structure::Object_comment.new.comment(h) #t_o=SiSU_document_structure::Object_layout.new.insert(h) + t_o=SiSU_DAL_DocumentStructure::ObjectComment.new.comment(h) #t_o=SiSU_DAL_DocumentStructure::ObjectLayout.new.insert(h) elsif ( @@flag['block'] \ and t_o =~/^\}block/ ) @@flag['block']=false obj,tags=extract_tags(@tuned_block.join("\n")) h={ obj: obj, tags: tags } @tuned_block=[] - t_o=SiSU_document_structure::Object_block_txt.new.block(h) + t_o=SiSU_DAL_DocumentStructure::ObjectBlockTxt.new.block(h) tuned_file << t_o h={ obj: 'block text end' } #introduce a counter - t_o=SiSU_document_structure::Object_comment.new.comment(h) #t_o=SiSU_document_structure::Object_layout.new.insert(h) + t_o=SiSU_DAL_DocumentStructure::ObjectComment.new.comment(h) #t_o=SiSU_DAL_DocumentStructure::ObjectLayout.new.insert(h) elsif ( @@flag['alt'] \ and t_o =~/^\}alt/ ) @@flag['alt']=false obj,tags=extract_tags(@tuned_block.join("\n")) h={ obj: obj, tags: tags } - t_o=SiSU_document_structure::Object_block_txt.new.alt(h) + t_o=SiSU_DAL_DocumentStructure::ObjectBlockTxt.new.alt(h) @tuned_block=[] tuned_file << t_o h={ obj: 'alt text end' } #introduce a counter - t_o=SiSU_document_structure::Object_comment.new.comment(h) #t_o=SiSU_document_structure::Object_layout.new.insert(h) + t_o=SiSU_DAL_DocumentStructure::ObjectComment.new.comment(h) #t_o=SiSU_DAL_DocumentStructure::ObjectLayout.new.insert(h) end if @@flag['poem'] \ or @@flag['group'] \ @@ -444,13 +444,13 @@ module SiSU_document_structure_extract sub_array=t_o.dup @line_mode=sub_array.scan(/.+/) type=if @@flag['poem']; 'poem' - t_o=SiSU_document_structure_extract::Build.new(@md,@line_mode).build_lines(type).join + t_o=SiSU_DAL_DocumentStructureExtract::Build.new(@md,@line_mode).build_lines(type).join poem=t_o.split(/\n\n/) poem.each do |v| - v.gsub!(/\n/m,"#{Mx[:br_nl]}\n") + v=v.gsub(/\n/m,"#{Mx[:br_nl]}\n") obj,tags=extract_tags(v) h={ obj: obj, tags: tags } - t_o=SiSU_document_structure::Object_block_txt.new.verse(h) + t_o=SiSU_DAL_DocumentStructure::ObjectBlockTxt.new.verse(h) tuned_file << t_o end else 'group' @@ -463,17 +463,17 @@ module SiSU_document_structure_extract or @@flag['group'] \ or @@flag['alt'] if t_o.class==String - t_o.gsub!(/\n/m,"#{Mx[:br_nl]}") - t_o.gsub!(/[ ][ ]/m,"#{Mx[:nbsp]*2}") - t_o.gsub!(/#{Mx[:nbsp]}\s/,"#{Mx[:nbsp]*2}") + t_o=t_o.gsub(/\n/m,"#{Mx[:br_nl]}"). + gsub(/[ ][ ]/m,"#{Mx[:nbsp]*2}"). + gsub(/#{Mx[:nbsp]}\s/,"#{Mx[:nbsp]*2}") t_o=t_o + Mx[:br_nl] if t_o =~/\S+/ - elsif t_o.is=='group' \ - or t_o.is=='block' \ - or t_o.is=='alt' \ - or t_o.is=='verse' - t_o.obj.gsub!(/\n/m,"#{Mx[:br_nl]}") - t_o.obj.gsub!(/[ ][ ]/m,"#{Mx[:nbsp]*2}") - t_o.obj.gsub!(/#{Mx[:nbsp]}\s/,"#{Mx[:nbsp]*2}") + elsif t_o.is==:group \ + || t_o.is==:block \ + || t_o.is==:alt \ + || t_o.is==:verse + t_o.obj=t_o.obj.gsub(/\n/m,"#{Mx[:br_nl]}"). + gsub(/[ ][ ]/m,"#{Mx[:nbsp]*2}"). + gsub(/#{Mx[:nbsp]}\s/,"#{Mx[:nbsp]*2}") end @tuned_block << t_o if t_o =~/\S+/ else tuned_file << t_o @@ -484,32 +484,27 @@ module SiSU_document_structure_extract if @md.flag_endnotes tuned_file << @pb h={ ln: 2, obj: 'Endnotes', autonum_: false } - tuned_file << SiSU_document_structure::Object_heading.new.heading_insert(h) + tuned_file << SiSU_DAL_DocumentStructure::ObjectHeading.new.heading_insert(h) h={ ln: 4, obj: 'Endnotes', name: 'endnotes', autonum_: false } - tuned_file << SiSU_document_structure::Object_heading.new.heading_insert(h) + tuned_file << SiSU_DAL_DocumentStructure::ObjectHeading.new.heading_insert(h) h={ obj: 'Endnotes' } end if @md.book_idx tuned_file << @pb h={ ln: 2, obj: 'Index', autonum_: false } - tuned_file << SiSU_document_structure::Object_heading.new.heading_insert(h) + tuned_file << SiSU_DAL_DocumentStructure::ObjectHeading.new.heading_insert(h) h={ ln: 4, obj: 'Index', name: 'book_index', autonum_: false } - tuned_file << SiSU_document_structure::Object_heading.new.heading_insert(h) + tuned_file << SiSU_DAL_DocumentStructure::ObjectHeading.new.heading_insert(h) h={ obj: 'Index' } end tuned_file << @pb h={ ln: 2, obj: 'Metadata', autonum_: false, ocn_: false } - tuned_file << SiSU_document_structure::Object_heading.new.heading_insert(h) + tuned_file << SiSU_DAL_DocumentStructure::ObjectHeading.new.heading_insert(h) h={ ln: 4, obj: 'SiSU Metadata, document information', name: 'metadata', autonum_: false, ocn_: false } - tuned_file << SiSU_document_structure::Object_heading.new.heading_insert(h) - tuned_file << @pb - h={ ln: 2, obj: 'Manifest', autonum_: false, ocn_: false } - tuned_file << SiSU_document_structure::Object_heading.new.heading_insert(h) - h={ ln: 4, obj: 'SiSU Manifest, alternative outputs etc.', name: 'sisu_manifest', autonum_: false, ocn_: false } - tuned_file << SiSU_document_structure::Object_heading.new.heading_insert(h) + tuned_file << SiSU_DAL_DocumentStructure::ObjectHeading.new.heading_insert(h) tuned_file h={ obj: 'eof' } - meta=SiSU_document_structure::Object_metadata.new.metadata(@metadata) + meta=SiSU_DAL_DocumentStructure::ObjectMetadata.new.metadata(@metadata) [tuned_file,meta] end def table_rows_and_columns_array(table_str) @@ -522,33 +517,33 @@ module SiSU_document_structure_extract end def meta_heading(h) h={ lv: h[:lv], ln: h[:ln], name: h[:name], obj: h[:obj], ocn: '0' } - SiSU_document_structure::Object_heading.new.heading(h) + SiSU_DAL_DocumentStructure::ObjectHeading.new.heading(h) end def meta_para(str) h={ obj: str, ocn_: false } - SiSU_document_structure::Object_para.new.paragraph(h) + SiSU_DAL_DocumentStructure::ObjectPara.new.paragraph(h) end def build_lines(type='') - data=@data - data.each do |line| - if line =~/\S/ \ + lines,lines_new=@data,[] + lines.each do |line| + line=if line =~/\S/ \ and line !~/^code\{|^\}code/ \ and line.class != Hash - line.gsub!(/\s\s/,"#{Mx[:nbsp]*2}") - line.gsub!(/#{Mx[:nbsp]}\s/,"#{Mx[:nbsp]*2}") - line.gsub!(/^/,"#{Mx[:gr_o]}codeline#{Mx[:gr_c]}") if type=='code' # REMOVE try sort for texpdf special case - if line =~/(?:https?|file|ftp):\/\/\S+$/ - line.gsub!(/\s*$/," #{Mx[:br_nl]}") - else line.gsub!(/\s*$/,"#{Mx[:br_nl]}") #unless type=='code' - end - if @@flag['code']; @@counter+=1 - else + @@counter+=1 if @@flag['code'] + line=line.gsub(/\s\s/,"#{Mx[:nbsp]*2}"). + gsub(/#{Mx[:nbsp]}\s/,"#{Mx[:nbsp]*2}") + line=line.gsub(/^/,"#{Mx[:gr_o]}codeline#{Mx[:gr_c]}") if type=='code' # REMOVE try sort for texpdf special case + line=if line =~/(?:https?|file|ftp):\/\/\S+$/ + line.gsub(/\s*$/," #{Mx[:br_nl]}") + else line.gsub(/\s*$/,"#{Mx[:br_nl]}") #unless type=='code' end elsif line =~/^\s*$/ - line.gsub!(/\s*$/,"#{Mx[:br_nl]}") + line.gsub(/\s*$/,"#{Mx[:br_nl]}") + else line end + lines_new << line end - data + lines_new end end class Structure # this must happen early @@ -561,29 +556,29 @@ module SiSU_document_structure_extract @dob end def structure_markup #build structure where structure provided only in meta header - @dob=if @dob.is =~/para/ \ - and ((@dob.hang !~/[1-9]/ and @dob.indent !~/[1-9]/) \ - or (@dob.hang != @dob.indent)) \ + @dob=if @dob.is ==:para \ + && (((@dob.hang !~/[1-9]/) && (@dob.indent !~/[1-9]/)) \ + || (@dob.hang != @dob.indent)) \ and not @dob.bullet_ @dob=case @dob.obj when /^#{@md.lv1}/ h={ lv: 'A', ln: 1 } - SiSU_document_structure::Object_heading.new.heading(h,@dob) + SiSU_DAL_DocumentStructure::ObjectHeading.new.heading(h,@dob) when /^#{@md.lv2}/ h={ lv: 'B', ln: 2 } - SiSU_document_structure::Object_heading.new.heading(h,@dob) + SiSU_DAL_DocumentStructure::ObjectHeading.new.heading(h,@dob) when /^#{@md.lv3}/ h={ lv: 'C', ln: 3 } - SiSU_document_structure::Object_heading.new.heading(h,@dob) + SiSU_DAL_DocumentStructure::ObjectHeading.new.heading(h,@dob) when /^#{@md.lv4}/ h={ lv: '1', ln: 4 } - SiSU_document_structure::Object_heading.new.heading(h,@dob) + SiSU_DAL_DocumentStructure::ObjectHeading.new.heading(h,@dob) when /^#{@md.lv5}/ h={ lv: '2', ln: 5 } - SiSU_document_structure::Object_heading.new.heading(h,@dob) + SiSU_DAL_DocumentStructure::ObjectHeading.new.heading(h,@dob) when /^#{@md.lv6}/ h={ lv: '3', ln: 6 } - SiSU_document_structure::Object_heading.new.heading(h,@dob) + SiSU_DAL_DocumentStructure::ObjectHeading.new.heading(h,@dob) else @dob end else @dob @@ -592,26 +587,26 @@ module SiSU_document_structure_extract end def structure_markup_normalize #needs a bit of thinking dob=if @md.markup_version.determined < 0.38 #%convert internal representation, consider making 0.38 structure default ([A-C1-6] instead of [1-9]), requires downstream changes - @dob.gsub!(/^[456]~/,'!_') - @dob.gsub!(/^3~(\S+)/,"#{Mx[:lv_o]}6:\\1#{Mx[:lv_c]}") - @dob.gsub!(/^3~\s+/,"#{Mx[:lv_o]}6:#{Mx[:lv_c]}") - @dob.gsub!(/^2~(\S+)/,"#{Mx[:lv_o]}5:\\1#{Mx[:lv_c]}") - @dob.gsub!(/^2~\s+/,"#{Mx[:lv_o]}5:#{Mx[:lv_c]}") - @dob.gsub!(/^1~(\S+)/,"#{Mx[:lv_o]}4:\\1#{Mx[:lv_c]}") - @dob.gsub!(/^1~\s+/,"#{Mx[:lv_o]}4:#{Mx[:lv_c]}") - @dob.gsub!(/^:?C~(\S+)/,"#{Mx[:lv_o]}3:\\1#{Mx[:lv_c]}") - @dob.gsub!(/^:?C~\s+/,"#{Mx[:lv_o]}3:#{Mx[:lv_c]}") - @dob.gsub!(/^:?B~(\S+)/,"#{Mx[:lv_o]}2:\\1#{Mx[:lv_c]}") - @dob.gsub!(/^:?B~\s+/,"#{Mx[:lv_o]}2:#{Mx[:lv_c]}") - @dob.gsub!(/^:?A~(\S+)/,"#{Mx[:lv_o]}1:\\1#{Mx[:lv_c]}") - @dob.gsub!(/^:?A~\s+/,"#{Mx[:lv_o]}1:#{Mx[:lv_c]}") + @dob=@dob.gsub(/^[456]~/,'!_'). + gsub(/^3~(\S+)/,"#{Mx[:lv_o]}6:\\1#{Mx[:lv_c]}"). + gsub(/^3~\s+/,"#{Mx[:lv_o]}6:#{Mx[:lv_c]}"). + gsub(/^2~(\S+)/,"#{Mx[:lv_o]}5:\\1#{Mx[:lv_c]}"). + gsub(/^2~\s+/,"#{Mx[:lv_o]}5:#{Mx[:lv_c]}"). + gsub(/^1~(\S+)/,"#{Mx[:lv_o]}4:\\1#{Mx[:lv_c]}"). + gsub(/^1~\s+/,"#{Mx[:lv_o]}4:#{Mx[:lv_c]}"). + gsub(/^:?C~(\S+)/,"#{Mx[:lv_o]}3:\\1#{Mx[:lv_c]}"). + gsub(/^:?C~\s+/,"#{Mx[:lv_o]}3:#{Mx[:lv_c]}"). + gsub(/^:?B~(\S+)/,"#{Mx[:lv_o]}2:\\1#{Mx[:lv_c]}"). + gsub(/^:?B~\s+/,"#{Mx[:lv_o]}2:#{Mx[:lv_c]}"). + gsub(/^:?A~(\S+)/,"#{Mx[:lv_o]}1:\\1#{Mx[:lv_c]}"). + gsub(/^:?A~\s+/,"#{Mx[:lv_o]}1:#{Mx[:lv_c]}") @dob=if @dob =~/^@(?:level|markup):\s/ - @dob.gsub!(/3/,'6') - @dob.gsub!(/2/,'5') - @dob.gsub!(/1/,'4') - @dob.gsub!(/:?C/,'3') - @dob.gsub!(/:?B/,'2') - @dob.gsub!(/:?A/,'1') + @dob=@dob.gsub(/3/,'6'). + gsub(/2/,'5'). + gsub(/1/,'4'). + gsub(/:?C/,'3'). + gsub(/:?B/,'2'). + gsub(/:?A/,'1') @dob else @dob end @@ -620,19 +615,19 @@ module SiSU_document_structure_extract end def structure_marks t_o=if @md.markup_version.determined < 0.38 - @t_o.gsub!(/^1~(\S+)/,"#{Mx[:lv_o]}1:\\1#{Mx[:lv_c]}") - @t_o.gsub!(/^1~\s+/,"#{Mx[:lv_o]}1:#{Mx[:lv_c]}") - @t_o.gsub!(/^2~(\S+)/,"#{Mx[:lv_o]}2:\\1#{Mx[:lv_c]}") - @t_o.gsub!(/^2~\s+/,"#{Mx[:lv_o]}2:#{Mx[:lv_c]}") - @t_o.gsub!(/^3~(\S+)/,"#{Mx[:lv_o]}3:\\1#{Mx[:lv_c]}") - @t_o.gsub!(/^3~\s+/,"#{Mx[:lv_o]}3:#{Mx[:lv_c]}") - @t_o.gsub!(/^4~(\S+)/,"#{Mx[:lv_o]}4:\\1#{Mx[:lv_c]}") - @t_o.gsub!(/^4~\s+/,"#{Mx[:lv_o]}4:#{Mx[:lv_c]}") - @t_o.gsub!(/^5~(\S+)/,"#{Mx[:lv_o]}5:\\1#{Mx[:lv_c]}") - @t_o.gsub!(/^5~\s+/,"#{Mx[:lv_o]}5:#{Mx[:lv_c]}") - @t_o.gsub!(/^6~(\S+)/,"#{Mx[:lv_o]}6:\\1#{Mx[:lv_c]}") - @t_o.gsub!(/^6~\s+/,"#{Mx[:lv_o]}6:#{Mx[:lv_c]}") - @t_o.gsub!(/^[789]~/,'!_') + @t_o=@t_o.gsub(/^1~(\S+)/,"#{Mx[:lv_o]}1:\\1#{Mx[:lv_c]}"). + gsub(/^1~\s+/,"#{Mx[:lv_o]}1:#{Mx[:lv_c]}"). + gsub(/^2~(\S+)/,"#{Mx[:lv_o]}2:\\1#{Mx[:lv_c]}"). + gsub(/^2~\s+/,"#{Mx[:lv_o]}2:#{Mx[:lv_c]}"). + gsub(/^3~(\S+)/,"#{Mx[:lv_o]}3:\\1#{Mx[:lv_c]}"). + gsub(/^3~\s+/,"#{Mx[:lv_o]}3:#{Mx[:lv_c]}"). + gsub(/^4~(\S+)/,"#{Mx[:lv_o]}4:\\1#{Mx[:lv_c]}"). + gsub(/^4~\s+/,"#{Mx[:lv_o]}4:#{Mx[:lv_c]}"). + gsub(/^5~(\S+)/,"#{Mx[:lv_o]}5:\\1#{Mx[:lv_c]}"). + gsub(/^5~\s+/,"#{Mx[:lv_o]}5:#{Mx[:lv_c]}"). + gsub(/^6~(\S+)/,"#{Mx[:lv_o]}6:\\1#{Mx[:lv_c]}"). + gsub(/^6~\s+/,"#{Mx[:lv_o]}6:#{Mx[:lv_c]}"). + gsub(/^[789]~/,'!_') @t_o else @t_o end @@ -651,12 +646,14 @@ module SiSU_document_structure_extract parent=node1=node2=node3=node4=node5=node6=nil data.each do |dob| h={} - if (dob.obj !~ regex_exclude_ocn_and_node || dob.is =='code') \ - and dob.of !~/(?:comment|layout|meta)/ \ - and dob.obj !~/#{Mx[:pa_non_object_no_heading]}|#{Mx[:pa_non_object_dummy_heading]}/ \ - and dob.ocn_ + if (dob.obj !~ regex_exclude_ocn_and_node || dob.is ==:code) \ + && (dob.of !=:comment \ + && dob.of !=:layout \ + && dob.of !=:meta) \ + && dob.obj !~/#{Mx[:pa_non_object_no_heading]}|#{Mx[:pa_non_object_dummy_heading]}/ \ + && dob.ocn_ #dob.ln now is determined, and set earlier, check how best to remove this --> - if dob.is=='heading' + if dob.is==:heading ln=case dob.lv when 'A'; 1 when 'B'; 2 @@ -672,7 +669,7 @@ module SiSU_document_structure_extract if not dob.obj =~/<:#>|~#|-#/ \ or not dob.toc_ # fix this no longer in dob.obj ocn+=1 - if dob.is=='heading' \ + if dob.is==:heading \ and (ln.to_s =~/^[1-9]/ \ or ln.to_s =~@md.lv1 \ or ln.to_s =~@md.lv2 \ @@ -708,56 +705,67 @@ module SiSU_document_structure_extract end else ocno+=1 - if dob.is=='table' + if dob.is==:table ocnt+=1 ocn_sp,parent="t#{ocnt}",node - elsif dob.is=='code' + elsif dob.is==:code ocnc+=1 ocn_sp,parent="c#{ocnc}",node - elsif dob.is=~/^(?:group|block|alt|verse)/ + elsif dob.is==:group \ + || dob.is==:block \ + || dob.is==:alt \ + || dob.is==:verse ocng+=1 #group, poem ocn_sp,parent="g#{ocng}",node - elsif dob.is=~/image|#{Mx[:lnk_o]}\S+?\.(?:png|jpg|gif)\s+/m + elsif dob.is==:image #check ocni+=1 ocn_sp,parent="i#{ocni}",node else ocnp+=1 #paragraph ocn_sp,parent="p#{ocnp}",node end end - if dob.is=='heading' + if dob.is==:heading dob.ln,dob.node,dob.ocn,dob.odv,dob.osp,dob.parent=ln,node,ocn,ocn_dv,ocn_sp,parent else - unless dob.of=~/meta|comment|layout/ + if dob.of !=:meta \ + && dob.of !=:comment \ + && dob.of !=:layout dob.ocn,dob.odv,dob.osp,dob.parent=ocn,ocn_dv,ocn_sp,parent end end else ocnu+=1 - dob.obj.gsub!(/#{Mx[:fa_o]}~##{Mx[:fa_c]}/,'') if dob.obj + dob.obj=dob.obj.gsub(/#{Mx[:fa_o]}~##{Mx[:fa_c]}/,'') if dob.obj ocn_dv,ocn_sp="u#{ocnu}","u#{ocnu}" dob.ocn,dob.odv,dob.osp=ocn,ocn_dv,ocn_sp end h elsif dob.obj=~/#{Mx[:pa_non_object_no_heading]}/ - dob.obj.gsub!(/#{Mx[:pa_non_object_no_heading]}/,'') - if dob.is=='para' + dob.obj=dob.obj.gsub(/#{Mx[:pa_non_object_no_heading]}/,'') + if dob.is==:para h={ obj: dob.obj, ocn_: false, ocn: nil } - dob=SiSU_document_structure::Object_para.new.paragraph(h,dob) - elsif dob.is=='heading' + dob=SiSU_DAL_DocumentStructure::ObjectPara.new.paragraph(h,dob) + elsif dob.is==:heading h={ obj: dob.obj, ocn_: false, ocn: nil, toc_: true } - dob=SiSU_document_structure::Object_heading.new.heading(h,dob) + dob=SiSU_DAL_DocumentStructure::ObjectHeading.new.heading(h,dob) end elsif dob.obj=~/#{Mx[:pa_non_object_dummy_heading]}/ - dob.obj.gsub!(/#{Mx[:pa_non_object_dummy_heading]}/,'') - if dob.is=='para' + dob.obj=dob.obj.gsub(/#{Mx[:pa_non_object_dummy_heading]}/,'') + if dob.is==:para h={ obj: dob.obj, ocn_: false, ocn: nil } - dob=SiSU_document_structure::Object_para.new.paragraph(h,dob) - elsif dob.is=='heading' + dob=SiSU_DAL_DocumentStructure::ObjectPara.new.paragraph(h,dob) + elsif dob.is==:heading h={ obj: dob.obj, ocn_: false, ocn: nil, toc_: false } - dob=SiSU_document_structure::Object_heading.new.heading(h,dob) + dob=SiSU_DAL_DocumentStructure::ObjectHeading.new.heading(h,dob) end else dob end - dob.obj.gsub!(/\n\n/,"\n") if dob.is =~/(?:code|verse|alt|group|block)/ #newlines taken out + if dob.is==:code \ + || dob.is==:verse \ + || dob.is==:alt \ + || dob.is==:group \ + || dob.is==:block + dob.obj=dob.obj.gsub(/\n\n/,"\n") #newlines taken out + end @o_array << dob end @o_array @@ -791,7 +799,8 @@ module SiSU_document_structure_extract puts "<#{@s[0]}>" end data.each_with_index do |o,i| - if o.is =~/^heading/ + if o.is ==:heading \ + || o.is ==:heading_insert case o.ln when 1 tuned_file << tag_close(o.ln,hs) @@ -847,8 +856,7 @@ module SiSU_document_structure_extract end puts_tag_close(0,hs) if @md.opt.cmd =~/V/ tuned_file << tag_close(0,hs) - tuned_file.flatten! - tuned_file + tuned_file=tuned_file.flatten end def tags(o) tag=if o[:status]=='open' @@ -867,7 +875,7 @@ module SiSU_document_structure_extract when '6'; 9 end h={ tag: tag, node: o[:node], lv: o[:lv], ln: ln, status: o[:status] } - SiSU_document_structure::Object_structure.new.xml_dom(h) #downstream code utilise else ignore like comments + SiSU_DAL_DocumentStructure::ObjectStructure.new.xml_dom(h) #downstream code utilise else ignore like comments end def tag_open(o,tag) t={ lv: tag[o.ln], node: o.node, status: 'open' } |