diff options
Diffstat (limited to 'lib/sisu/v0/shared_xml.rb')
-rw-r--r-- | lib/sisu/v0/shared_xml.rb | 89 |
1 files changed, 84 insertions, 5 deletions
diff --git a/lib/sisu/v0/shared_xml.rb b/lib/sisu/v0/shared_xml.rb index fd27c664..abc6cc1a 100644 --- a/lib/sisu/v0/shared_xml.rb +++ b/lib/sisu/v0/shared_xml.rb @@ -1,3 +1,4 @@ +# coding: utf-8 =begin * Name: SiSU @@ -154,10 +155,44 @@ module SiSU_XML_munge class Trans require "#{SiSU_lib}/defaults" def initialize(md) + @md=md @sys=SiSU_Env::System_call.new - @dir=SiSU_Env::Info_env.new(md.fns) + @dir=SiSU_Env::Info_env.new(@md.fns) @dp=SiSU_Env::Info_env.new.digest.pattern @url_brace=SiSU_Viz::Skin.new.url_decoration + if @md.sem_tag + @ab ||=semantic_tags.default + end + end + def semantic_tags + def default + { + :pub => 'publication', + :ref => 'reference', + :desc => 'description', + :conv => 'convention', + :vol => 'volume', + :pg => 'page', + :ct => 'cite', + :cty => 'city', + :org => 'organization', + :d => 'date', + :t => 'title', + :a => 'author', + :n => 'name', + :fn => 'firstname', + :f => 'firstname', + :mn => 'middlename', + :m => 'middlename', + :ln => 'lastname', + :l => 'lastname', + :i => 'initials', + :q => 'quote', + :y => 'year', + :ab => 'abreviation', + } + end + self end def char_enc #character encode def utf8(para='') @@ -348,10 +383,8 @@ module SiSU_XML_munge end end def markup(para='') - #if para !~/^<:code>/ - wordlist=para.scan(/\S+|\n/) #\n needed for tables, check though added 2005w17 - para=tidywords(wordlist).join(' ').strip - #end + wordlist=para.scan(/\S+|\n/) #\n needed for tables, check though added 2005w17 + para=tidywords(wordlist).join(' ').strip para.gsub!(/(^|\s+)<\s+/,'\1< '); para.gsub!(/\s+>(\s+|$)/,' >\1') para.gsub!(/<:pb>\s*/,'') para.gsub!(/<+[-~]#>+/,'') @@ -426,6 +459,52 @@ module SiSU_XML_munge para.gsub!(/<(\/?en)>/,'<\1>') para end + def xml_sem_block_paired(matched) # colon depth: many, recurs + matched.gsub!(/\b(a):\{(.+?)\}:\1\b/m, %{<sem:#{@ab[:a]} depth="many">\\2</sem:#{@ab[:a]}>}) # sem : + matched.gsub!(/\b(vol):\{(.+?)\}:\1\b/m, %{<sem:#{@ab[:vol]} depth="many">\\2</sem:#{@ab[:vol]}>}) # sem : + matched.gsub!(/\b(pub):\{(.+?)\}:\1\b/m, %{<sem:#{@ab[:pub]} depth="many">\\2</sem:#{@ab[:pub]}>}) # sem : + matched.gsub!(/\b(ref):\{(.+?)\}:\1\b/m, %{<sem:#{@ab[:ref]} depth="many">\\2</sem:#{@ab[:ref]}>}) # sem : + matched.gsub!(/\b(desc):\{(.+?)\}:\1\b/m,%{<sem:#{@ab[:desc]} depth="many">\\2</sem:#{@ab[:desc]}>}) # sem : + matched.gsub!(/\b(conv):\{(.+?)\}:\1\b/m,%{<sem:#{@ab[:conv]} depth="many">\\2</sem:#{@ab[:conv]}>}) # sem : + matched.gsub!(/\b(ct):\{(.+?)\}:\1\b/m, %{<sem:#{@ab[:ct]} depth="many">\\2</sem:#{@ab[:ct]}>}) # sem : + matched.gsub!(/\b(cty):\{(.+?)\}:\1\b/m, %{<sem:#{@ab[:cty]} depth="many">\\2</sem:#{@ab[:cty]}>}) # sem : + matched.gsub!(/\b(org):\{(.+?)\}:\1\b/m, %{<sem:#{@ab[:org]} depth="many">\\2</sem:#{@ab[:org]}>}) # sem : + matched.gsub!(/\b(d):\{(.+?)\}:\1\b/m, %{<sem:#{@ab[:d]} depth="many">\\2</sem:#{@ab[:d]}>}) # sem : + matched.gsub!(/\b(n):\{(.+?)\}:\1\b/m, %{<sem:#{@ab[:n]} depth="many">\\2</sem:#{@ab[:n]}>}) # sem : + matched.gsub!(/([a-z]+(?:[_:.][a-z]+)*)(?::\{(.+?)\}:\1)/m,'<sem:\1 depth="many">\2</sem:\1>') # sem : + end + def xml_semantic_tags(para) + if @md.sem_tag + para.gsub!(/([a-z]+(?:[_:.][a-z]+)*)(?::\{(.+?)\}:\1)/m) {|c| xml_sem_block_paired(c) } # sem : + para.gsub!(/([a-z]+(?:[_:.][a-z]+)*)(?::\{(.+?)\}:\1)/m) {|c| xml_sem_block_paired(c) } # sem : + para.gsub!(/([a-z]+(?:[_:.][a-z]+)*)(?::\{(.+?)\}:\1)/m) {|c| xml_sem_block_paired(c) } # sem : + #colon one / single / flat / shallow + para.gsub!(/:\{(.+?)\}:a\b/m, %{<sem:#{@ab[:a]} depth="one">\\1</sem:#{@ab[:a]}>}) # sem : + para.gsub!(/:\{(.+?)\}:n\b/m, %{<sem:#{@ab[:n]} depth="one">\\1</sem:#{@ab[:n]}>}) # sem : + para.gsub!(/:\{(.+?)\}:t\b/m, %{<sem:#{@ab[:t]} depth="one">\\1</sem:#{@ab[:t]}>}) # sem : + para.gsub!(/:\{(.+?)\}:ref\b/m, %{<sem:#{@ab[:ref]} depth="one">\\1</sem:#{@ab[:ref]}>}) # sem : + para.gsub!(/:\{(.+?)\}:desc\b/m, %{<sem:#{@ab[:desc]} depth="one">\\1</sem:#{@ab[:desc]}>}) # sem : + para.gsub!(/:\{(.+?)\}:cty\b/m, %{<sem:#{@ab[:cty]} depth="one">\\1</sem:#{@ab[:cty]}>}) # sem : + para.gsub!(/:\{(.+?)\}:org\b/m, %{<sem:#{@ab[:org]} depth="one">\\1</sem:#{@ab[:org]}>}) # sem : + para.gsub!(/:\{(.+?)\}:([a-z]+(?:[_:.][a-z]+)*)/m,'<sem:\2 depth="one">\1</sem:\2>') # sem : + #semicolon zero / none + para.gsub!(/;\{([^}]+(?![;]))\};t\b/m, %{<sem:#{@ab[:t]} depth="zero">\\1</sem:#{@ab[:t]}>}) # sem ; + para.gsub!(/;\{([^}]+(?![;]))\};q\b/m, %{<sem:#{@ab[:q]} depth="zero">\\1</sem:#{@ab[:q]}>}) # sem ; + para.gsub!(/;\{([^}]+(?![;]))\};ref\b/m, %{<sem:#{@ab[:ref]} depth="zero">\\1</sem:#{@ab[:ref]}>}) # sem ; + para.gsub!(/;\{([^}]+(?![;]))\};desc\b/m,%{<sem:#{@ab[:desc]} depth="zero">\\1</sem:#{@ab[:desc]}>}) # sem ; + para.gsub!(/;\{([^}]+(?![;]))\};y\b/m, %{<sem:#{@ab[:y]} depth="zero">\\1</sem:#{@ab[:y]}>}) # sem ; + para.gsub!(/;\{([^}]+(?![;]))\};ab\b/m, %{<sem:#{@ab[:ab]} depth="zero">\\1</sem:#{@ab[:ab]}>}) # sem ; + para.gsub!(/;\{([^}]+(?![;]))\};pg\b/m, %{<sem:#{@ab[:pg]} depth="zero">\\1</sem:#{@ab[:pg]}>}) # sem ; + para.gsub!(/;\{([^}]+(?![;]))\};fn?\b/m, %{<sem:#{@ab[:fn]} depth="zero">\\1</sem:#{@ab[:fn]}>}) # sem ; + para.gsub!(/;\{([^}]+(?![;]))\};mn?\b/m, %{<sem:#{@ab[:mn]} depth="zero">\\1</sem:#{@ab[:mn]}>}) # sem ; + para.gsub!(/;\{([^}]+(?![;]))\};ln?\b/m, %{<sem:#{@ab[:ln]} depth="zero">\\1</sem:#{@ab[:ln]}>}) # sem ; + para.gsub!(/;\{([^}]+(?![;]))\};i\b/m, %{<sem:#{@ab[:i]} depth="zero">\\1</sem:#{@ab[:i]}>}) # sem ; + para.gsub!(/;\{([^}]+(?![;]))\};org\b/m, %{<sem:#{@ab[:org]} depth="zero">\\1</sem:#{@ab[:org]}>}) # sem ; + para.gsub!(/;\{([^}]+(?![;]))\};cty\b/m, %{<sem:#{@ab[:cty]} depth="zero">\\1</sem:#{@ab[:cty]}>}) # sem ; + para.gsub!(/;\{([^}]+(?![;]))\};([a-z]+(?:[_:.][a-z]+)*)/m,'<sem:\2 depth="zero">\1</sem:\2>') # sem ; + end + para + end end end module SiSU_XML_tags #Format |