Merge branch 'upstream' into debian/sid

3 files changed, 43 insertions, 16 deletions

diff --git a/CHANGELOG b/CHANGELOG
index 5edad220..d6676e53 100644
--- a/CHANGELOG
+++ b/CHANGELOG
@@ -9,11 +9,20 @@ Reverse Chronological:
 
 %% STABLE MANIFEST
 
+%% sisu_0.65.0.orig.tar.gz (2008-02-16:06/6)
+http://www.jus.uio.no/sisu/pkg/src/sisu_0.65.0.orig.tar.gz
+  sisu_0.65.0.orig.tar.gz
+  sisu_0.65.0-1.dsc
+  sisu_0.65.0-1.diff.gz
+
+  * semantic tag, match refinement related, (consider a possible extension)
+    [version bump for tags]
+
 %% sisu_0.64.4.orig.tar.gz (2008-02-14:06/4)
 http://www.jus.uio.no/sisu/pkg/src/sisu_0.64.4.orig.tar.gz
-  sisu_0.64.4.orig.tar.gz
-  sisu_0.64.4-1.dsc
-  sisu_0.64.4-1.diff.gz
+  2ab102057702aaf35ec96b9a082fb2d9 1488644 sisu_0.64.4.orig.tar.gz
+  62604c78e2574f80a4e13cb38e1374cd 629 sisu_0.64.4-1.dsc
+  b797bf7f165be762ff97d50914a5b102 146562 sisu_0.64.4-1.diff.gz
 
   * markup tag, semantic tag added
 
diff --git a/data/sisu/conf/editor-syntax-etc/vim/syntax/sisu.vim b/data/sisu/conf/editor-syntax-etc/vim/syntax/sisu.vim
index 9a351d88..1de9a34d 100644
--- a/data/sisu/conf/editor-syntax-etc/vim/syntax/sisu.vim
+++ b/data/sisu/conf/editor-syntax-etc/vim/syntax/sisu.vim
@@ -40,9 +40,9 @@ if !exists("sisu_no_identifiers")
   syn match   sisu_sem_ex_marker_block               "\]:[a-z]\+"
   syn match   sisu_error_sem_marker                  "[a-z]\+[:;]{"
   syn match   sisu_error_sem_marker                  "}[:;][a-z]\+"
-  syn match   sisu_sem_block contains=sisu_error,sisu_error_wspace,sisu_content_alt,sisu_link,sisu_linked,sisu_break,sisu_sem_block,sisu_sem_content,sisu_sem_marker_block,sisu_sem_marker,sisu_sem_ex_marker_block,sisu_sem_ex_marker,sisu_mark_endnote,sisu_content_endnote "\([a-z]\+\):{[^}].\{-}}:\1"he=e-1
+  syn match   sisu_sem_block contains=sisu_error,sisu_error_wspace,sisu_content_alt,sisu_link,sisu_linked,sisu_break,sisu_sem_block,sisu_sem_content,sisu_sem_marker_block,sisu_sem_marker,sisu_sem_ex_marker_block,sisu_sem_ex_marker,sisu_mark_endnote,sisu_content_endnote "\([a-z]\+\):{[^}].\{-}}:\1"
   syn match   sisu_sem_content contains=sisu_error,sisu_error_wspace,sisu_content_alt,sisu_link,sisu_linked,sisu_break,sisu_sem_marker_block,sisu_sem_marker,sisu_sem_ex_marker_block,sisu_sem_ex_marker "\([a-z]\+\);{[^}].\{-}};\1"
-  syn match   sisu_sem_ex_block contains=sisu_error,sisu_error_wspace,sisu_content_alt,sisu_link,sisu_linked,sisu_break,sisu_sem_block,sisu_sem_content,sisu_sem_marker_block,sisu_sem_marker,sisu_sem_ex_marker_block,sisu_sem_ex_marker,sisu_mark_endnote,sisu_content_endnote "\([a-z]\+\):\[[^}].\{-}\]:\1"he=e-1
+  syn match   sisu_sem_ex_block contains=sisu_error,sisu_error_wspace,sisu_content_alt,sisu_link,sisu_linked,sisu_break,sisu_sem_block,sisu_sem_content,sisu_sem_marker_block,sisu_sem_marker,sisu_sem_ex_marker_block,sisu_sem_ex_marker,sisu_mark_endnote,sisu_content_endnote "\([a-z]\+\):\[[^}].\{-}\]:\1"
   syn match   sisu_sem_ex_content contains=sisu_error,sisu_error_wspace,sisu_content_alt,sisu_link,sisu_linked,sisu_break,sisu_sem_marker_block,sisu_sem_marker,sisu_sem_ex_marker_block,sisu_sem_ex_marker "\([a-z]\+\);\[[^}].\{-}\];\1"
 endif
 "% 9 URLs Numbers: and ASCII Codes
diff --git a/lib/sisu/v0/shared_sem.rb b/lib/sisu/v0/shared_sem.rb
index da989999..f6303ee0 100644
--- a/lib/sisu/v0/shared_sem.rb
+++ b/lib/sisu/v0/shared_sem.rb
@@ -65,38 +65,56 @@ module SiSU_sem
       @para=para
     end
     def rgx
+      def exclude
+        /^<:code>/
+      end
       def each_csc
-        /[a-z]+[:;]\{|\}[:;][a-z]+/
+        /\b[a-z]+[:;]\{|\}[:;][a-z]+\b/m
+        #/\b(?:[a-z]+(?:[_.][a-z]+)+|[a-z]+)[:;]\{|\}[:;](?:[a-z]+(?:[_.][a-z]+)+|[a-z]+)\b/m
       end
       def each_c
-        /[a-z]+:\{|\}:[a-z]+/
+        /\b[a-z]+:\{|\}:[a-z]+\b/m
+        #/\b(?:[a-z]+(?:[_.][a-z]+)+|[a-z]+):\{|\}:(?:[a-z]+(?:[_.][a-z]+)+|[a-z]+)\b/m
       end
       def each_sc
-        /[a-z]+;\{|\};[a-z]+/
+        /\b[a-z]+;\{|\};[a-z]+\b/m
+        #/\b(?:[a-z]+(?:[_.][a-z]+)+|[a-z]+);\{|\};(?:[a-z]+(?:[_.][a-z]+)+|[a-z]+)\b/m
       end
       def pair_csc
-        /(([a-z]+)(?::\{(.+?)\}:\2|;\{(.+?)\};\2))/
-        #/(([a-z]+);\{(.+?)\};\2)/
-        #/(([a-z]+)([:;])\{(.+?)\}\3\2)/
+        /(([a-z]+)(?::\{(.+?)\}:\2|;\{(.+?)\};\2))/m
+        #/(([a-z]+(?:[_.][a-z]+)+|[a-z]+)(?::\{(.+?)\}:\2|;\{(.+?)\};\2))/m
       end
       def pair_c
-        /(([a-z]+):\{(.+?)\}:\2)/
+        /(([a-z]+):\{(.+?)\}:\2)/m
+        #/(([a-z]+(?:[_.][a-z]+)+|[a-z]+):\{(.+?)\}:\2)/m
       end
       def pair_sc
-        /(([a-z]+);\{.+?\};\2)/
+        /(([a-z]+);\{.+?\};\2)/m
+        #/(([a-z]+(?:[_.][a-z]+)+|[a-z]+);\{.+?\};\2)/m
       end
       def whole_csc_ae
-        /(([a-z]+)(?::\[(.+?)\]:\2|;\[(.+?)\];\2))/
+        /(([a-z]+)(?::\[(.+?)\]:\2|;\[(.+?)\];\2))/m
+        #/(([a-z]+(?:[_.][a-z]+)+|[a-z]+)(?::\[(.+?)\]:\2|;\[(.+?)\];\2))/m
+      end
+      def each_csc_ae
+        /\b[a-z]+[:;]\[|\][:;][a-z]+\b/m
+        #/\b(?:[a-z]+(?:[_.][a-z]+)+|[a-z]+)[:;]\[|\][:;](?:[a-z]+(?:[_.][a-z]+)+|[a-z]+)\b/m
       end
       self
     end
     def rm
       def sem_marker_parts
-        @para.gsub!(rgx.each_csc,'')
+        unless @para =~ rgx.exclude
+          @para.gsub!(rgx.each_csc,'')
+        end
         @para
       end
       def sem_marker_added_extra_parts
-        @para.gsub!(rgx.whole_csc_ae,'')
+        unless @para =~ rgx.exclude
+          @para.gsub!(rgx.whole_csc_ae,'')
+          if para =~rgx.each_csc_ae; puts "WARNING semantic taggin error: #@para"
+          end
+        end
         @para
       end
       def all