diff options
Diffstat (limited to 'lib/sisu/v0/plaintext.rb')
| -rw-r--r-- | lib/sisu/v0/plaintext.rb | 142 |
1 files changed, 79 insertions, 63 deletions
diff --git a/lib/sisu/v0/plaintext.rb b/lib/sisu/v0/plaintext.rb index ddcf3a42..886e50a2 100644 --- a/lib/sisu/v0/plaintext.rb +++ b/lib/sisu/v0/plaintext.rb @@ -70,12 +70,12 @@ module SiSU_Plaintext class Source def initialize(opt) @opt=opt - if @opt.fns =~/(.+?)\.[_-]?sst$/ - case @opt.cmd - when /[af]/; @@dostype='unix footnotes' - when /e/; @@dostype='unix endnotes' - when /[AF]/; @@dostype='msdos footnotes' - when /E/; @@dostype='msdos endnotes' + @@dostype=if @opt.fns =~/(.+?)\.[_-]?sst$/ + if @opt.mod.inspect =~ /--footnote/ and @opt.mod.inspect =~ /--dos/; 'msdos footnotes' + elsif @opt.mod.inspect =~ /--endnote/ and @opt.mod.inspect =~ /--dos/; 'msdos endnotes' + elsif @opt.mod.inspect =~ /--footnote/; 'unix footnotes' + elsif @opt.mod.inspect =~ /--endnote/; 'unix endnotes' + else 'unix footnotes' end else puts "#{sf} not a processed file type" end @@ -153,36 +153,36 @@ module SiSU_Plaintext require "#{SiSU_lib}/defaults" require "#{SiSU_lib}/shared_txt" include SiSU_text_utils - @@endnotes_para=[] - @@plaintext={ :body=>[],:open=>[],:close=>[],:head=>[],:metadata=>[],:tail=>[],:endnotes=>[] } + @@endnotes={ :para=>[],:end=>[] } @@dp=nil def initialize(data,md) @data,@md=data,md @url_brace=SiSU_Viz::Skin.new.url_decoration @vz=SiSU_Env::Get_init.instance.skin @dp=@@dp ||=SiSU_Env::Info_env.new.digest.pattern - @regx=/^(?:(?:<:p[bn]>\s*)?\d~(?:(\S+))?\s+)?(.+?)\s*<~(\d+);(?:\w|[0-6]:)\d+;\w\d+><#@dp:#@dp>$/ #m # 2004w18 pb pn removal added + @regx=/^(?:(?:<:p[bn]>\s*)?\d~(?:(\S+))?\s+)?(.+?)\s*<~(\d+);(?:\w|[0-6]:)\d+;\w\d+><#@dp:#@dp>$/m # 2004w18 pb pn removal added @tab="\t" - @br=case md.cmd - when /[af]/ - @@dostype='unix footnotes' - "\n" - when /e/ - @@dostype='unix endnotes' - "\n" - when /[AF]/ + @br=if md.mod.inspect =~ /--footnote/ and md.mod.inspect =~ /--dos/ @@dostype='msdos footnotes' "\r\n" - when /E/ + elsif md.mod.inspect =~ /--endnote/ and md.mod.inspect =~ /--dos/ @@dostype='msdos endnotes' "\r\n" - else "\n" + elsif md.mod.inspect =~ /--footnote/ + @@dostype='unix footnotes' + "\n" + elsif md.mod.inspect =~ /--endnote/ + @@dostype='unix endnotes' + "\n" + else + @@dostype='unix footnotes' + "\n" end + @plaintext={ :body=>[],:open=>[],:close=>[],:head=>[],:metadata=>[],:tail=>[] } end def songsheet - @data=markup(@data) - publish - #@data.each { |x| puts x.inspect if x =~/\[table/ } + plaintext=markup(@data) + publish(plaintext) end # Used for extraction of endnotes from paragraphs def extract_endnotes(para='') @@ -201,13 +201,12 @@ module SiSU_Plaintext end notes=@n.flatten notes.each do |e| - util=if e.to_s =~/^\[[\d*+]+\]:/; SiSU_text_utils::Wrap.new(e.to_s,70,4,1) - else SiSU_text_utils::Wrap.new(e.to_s,70,1,1) + util=if e.to_s =~/^\[[\d*+]+\]:/; SiSU_text_utils::Wrap.new(e.to_s,78,4,1) + else SiSU_text_utils::Wrap.new(e.to_s,78,1,1) end wrap=util.line_wrap if wrap =~ /^\s*[\d*+]+\s+.+?\s*\Z/m wrap.gsub!(/^(\s*)([\d*+]+)\s+(.+?)\s*\Z/m, <<GSUB - \\1[\\2]: \\3 GSUB ) @@ -217,14 +216,15 @@ GSUB GSUB ) end - @@plaintext[:endnotes] << wrap - @@endnotes_para << wrap + @@endnotes[:para] << "-#{wrap}" + @@endnotes[:end] << wrap end + @@endnotes end def plaintext_metadata(meta) - util=SiSU_text_utils::Wrap.new(meta.text,70,15,1) + util=SiSU_text_utils::Wrap.new(meta.text,78,15,1) txt=util.line_wrap - @@plaintext[:metadata] <<= if meta.type == 'meta' + @plaintext[:metadata] <<= if meta.type == 'meta' <<WOK #{@tab}#{meta.el}: #{txt} @@ -242,7 +242,7 @@ WOK "Source file: #{@md.sc_filename}#{@br}Version number: #{@md.sc_number}#{@br}Version date: #{@md.sc_date}#{@br}" else '' end - @@plaintext[:tail] <<<<WOK + @plaintext[:tail] <<<<WOK #@br Other versions of this document: #@br manifest: @@ -269,37 +269,35 @@ WOK n=lv - 1 n3=lv + 2 lv=nil if lv == 0 - extract_endnotes(para) - para.gsub!(/~[{\[]([\d*+]+)\s+(?:.+?)[}\]]~/,'[^\1]') # endnote marker marked up wrapped=if para[@regx] paragraph=para[@regx,2] if paragraph =~/<:i([1-9])>/ m=$1.to_i paragraph.gsub!(/<:i#{m}>/,'') - util=SiSU_text_utils::Wrap.new(paragraph,70,m*2) - else util=SiSU_text_utils::Wrap.new(paragraph,70,0) + util=SiSU_text_utils::Wrap.new(paragraph,78,m*2) + else util=SiSU_text_utils::Wrap.new(paragraph,78,0) end util.line_wrap end if lv times=wrapped.length - times=70 if times > 70 - @@plaintext[:body] << case lv + times=78 if times > 78 + @plaintext[:body] << case lv when 1; wrapped.upcase << @br << '*'*times << @br when 2..3; wrapped.upcase << @br << '='*times << @br when 4; wrapped.upcase << @br << '-'*times << @br when 5..6; wrapped.upcase << @br << '.'*times << @br end else - @@plaintext[:body] << wrapped << @br # main text, contents, body KEEP + @plaintext[:body] << wrapped << @br # main text, contents, body KEEP end - if @@endnotes_para and @@dostype =~/footnote/ #edit out to switch off endnotes following paragraph to which they belong - @@plaintext[:body] << @br - @@endnotes_para.each {|e| @@plaintext[:body] << e << @br} - elsif @@endnotes_para and @@dostype =~/endnote/ - @@plaintext[:body] << @br*2 + if @@endnotes[:para] and @@dostype =~/footnote/ #edit out to switch off endnotes following paragraph to which they belong + @plaintext[:body] << @br + @@endnotes[:para].each {|e| @plaintext[:body] << e << @br} + elsif @@endnotes[:para] and @@dostype =~/endnote/ + @plaintext[:body] << @br*2 end - @@endnotes_para=[] + @@endnotes[:para]=[] end def markup(data) # Used for major markup instructions dir=SiSU_Env::Info_env.new(@md.fns) @@ -313,29 +311,44 @@ WOK para.gsub!(/<!Th?¡.+/m,"#@br#{table_message}") para.gsub!(/.+?<-#>/,'') # remove dummy headings (used by html) #check para.gsub!(/_\*\s+/,'* ') # bullet markup, marked down - #para.gsub!(/<br(?: \/)?>/,"\n") # introduces a bug - para.gsub!(/©/,'©') # bullet markup, marked down - para.gsub!(/&/,'&') # bullet markup, marked down para.gsub!(/<sup>(.+?)<\/sup>/,'^\1^') para.gsub!(/<sub>(.+?)<\/sub>/,'[\1]') para.gsub!(/<i>(.+?)<\/i>/,'/\1/') para.gsub!(/<b>(.+?)<\/b>/,'*\1*') para.gsub!(/<u>(.+?)<\/u>/,'_\1_') + unless para =~/<:code>/ + para.gsub!(/\{(.+?)\}((?:https?|file|ftp):\/\/\S+|image)/,'\1 [link:] \2') + para.gsub!(/(^|\s)((?:https?|file|ftp):\/\/\S+?\.[^'"><\s]+?)([.,]?(?:\s|$))/,"\\1#{@url_brace.txt_open}\\2#{@url_brace.txt_close}\\3") + para.gsub!(/_((?:https?|file|ftp):\/\/\S+)/,'\1') + extract_endnotes(para) + para.gsub!(/~[{\[]([\d*+]+)\s+(?:.+?)[}\]]~/,'[^\1]') # endnote marker marked up + para.gsub!(/&/,'&') + para.gsub!(/!/,'!') + para.gsub!(/#/,'#') + para.gsub!(/*/,'*') + para.gsub!(/-/,'-') + para.gsub!(///,'/') + para.gsub!(/_/,'_') + para.gsub!(/{/,'{') + para.gsub!(/}/,'}') + para.gsub!(/~/,'~') + para.gsub!(/©/,'©') + end if para =~/<:(?:group|verse|alt|code)(?:-end)?>(?:\s+<~(\d+);(?:\w|[0-6]:)\d+;\w\d+><#@dp:#@dp>)?/ if para =~/<:code>/ para.gsub!(/_</,'<'); para.gsub!(/_>/m,'>') #code-block: angle brackets special characters end para.gsub!(/<br(?: \/)?>/,"\n") # watch para.gsub!(/<:(?:group|verse|alt|code)(?:-end)?>(?:\s+<~(\d+);(?:\w|[0-6]:)\d+;\w\d+><#@dp:#@dp>)?/,'') + else para.gsub!(/<br(?: \/)?>/,"\n\n") # watch introduces a bug end para.gsub!(/<:p[bn]>/,'') # remove page breaks para.gsub!(/^\s*<~\d+;(?:\w|[0-6]:)\d+;\w\d+><#@dp:#@dp>$/,'') # remove empty lines - check - para.gsub!(/(^|\s)((?:https?|ftp):\/\/\S+?\.[^'"><\s]+?)([.,]?(?:\s|$))/,"\\1#{@url_brace.txt_open}\\2#{@url_brace.txt_close}\\3") - para.gsub!(/(^|\s)[_\\]((?:https?|ftp):\/\/\S+?\.[^'"><\s]+?)([.,]?(?:\s|$))/,'\1\2\3') + para.gsub!(/(^|\s)[_\\]((?:https?|file|ftp):\/\/\S+?\.[^'"><\s]+?)([.,]?(?:\s|$))/,'\1\2\3') para.gsub!(/<a href=".+?">(.+?)<\/a>/m,'\1') para.gsub!(/<:name#\S+?>/,'') # remove name links para.gsub!(/ /,' ') # decide on - para.gsub!(/\{(\S+?\.(?:png|jpg|gif)) .+?\}(?:(?:https?|ftp):\/\/\S+|image)/,' [ \1 ]') #"[ #{dir.url.images_local}\/\\1 ]") + para.gsub!(/\{(\S+?\.(?:png|jpg|gif)) .+?\}(?:(?:https?|file|ftp):\/\/\S+|image)/,' [ \1 ]') #"[ #{dir.url.images_local}\/\\1 ]") para.gsub!(/^\{\S+?\.(?:png|jpg|gif)\s+.+?"(.*?)"\s*\}\S+/,'[image: "\1"]') wordlist=para.scan(/\S+/) if para =~/^0~(\S+)\s+(.+?)\Z/m # for headers @@ -395,14 +408,14 @@ WOK para end elsif para =~/#{table_message}/ - @@plaintext[:body] << para << @br + @plaintext[:body] << para << @br elsif para =~/(Note|Endnotes?)/ and para !~/<~\d+;(?:\w|[0-6]:)\d+;\w\d+><#@dp:#@dp>$/ elsif para =~/(MetaData)/ and para =~/<~(\d+);[um]\d+;\w\d+><#@dp:#@dp>$/ #debug 2003w46 add rc info ####suspect visit #formatMono=MonoSiSU.new('<br /><a name="metadata">MetaData</a>') #para=formatMono.bold_para elsif para.include? 'Owner Details' and para !~/<~(\d+);(?:[oh]|[0-6]:)\d+;\w\d+><#@dp:#@dp>$/ #formatMono=MonoSiSU.new('<br /><a name="owner.details">Owner Details</a>') - #@@plaintext[:owner_details]=formatMono.bold_para + #@plaintext[:owner_details]=formatMono.bold_para #para='' elsif para =~/(¡|<!Th?)/ #tables ! elsif para =~/(.*)<!#!>(.*)/ @@ -430,21 +443,22 @@ WOK para.gsub!(/<:\S+>/,' ') if para ## Clean Prepared Text end end + @plaintext end - def publish - divider="=" + def publish(plaintext) + divider='=' content=[] - content << @@plaintext[:open] - content << @@plaintext[:head] - content << @@plaintext[:body] - content << @@plaintext[:endnotes] if @@dostype =~/endnotes/ - content << "#@br#{divider*70}#@br" - content << @@plaintext[:metadata] - content << "#@br#{divider*70}#@br" if @md.stmp =~/\w+/ #not used? - content << @@plaintext[:owner_details] if @md.stmp =~/\w+/ #not used? - content << @@plaintext[:tail] + content << plaintext[:open] + content << plaintext[:head] + content << plaintext[:body] + content << @@endnotes[:end] if @@dostype =~/endnotes/ + content << "#@br#{divider*78}#@br" + content << plaintext[:metadata] + content << "#@br#{divider*78}#@br" if @md.stmp =~/\w+/ #not used? + content << plaintext[:owner_details] if @md.stmp =~/\w+/ #not used? + content << plaintext[:tail] Output.new(content,@md).plaintext - @@plaintext[:head],@@plaintext[:body],@@plaintext[:tail],@@plaintext[:metadata]=[],[],[],[] + @@endnotes[:para],@@endnotes[:end]=[],[] end end class Output <Source @@ -466,8 +480,10 @@ WOK else filename_plaintext.puts para #unix plaintext # /^([*=-]|\.){5}/ end end + filename_plaintext.close end end end end __END__ +!\|#\|&*\|-\|/\|_\|{\|}\|~\|&# |