~ser/rml commit - sourcehut hg

2 files changed, 38 insertions(+), 27 deletions(-)

M rml.rb
M rml.re

M rml.rb +20 -14

@@ 11,29 11,33 @@ SUBSCRIPT = %r%~[-/*+?^_`\w\d].*?~%
 UNDERSCORE  = %r%_[-/*+?^~`\w\d].*?_%
 RAW       = %r%`[-/*+?^~_\w\d].*?`%
 FENCE     = %r%\{\{\{.*?\}\}\}%
-LINE      = %r%(?![ \t][*-][ \t])(.*?|#{EMPHASIS}|#{STRONG}|#{CITATION}|#{DELETED}|#{INSERTED}|#{SUPERSCRIPT}|#{SUBSCRIPT}|#{UNDERSCORE}|#{RAW}|#{FENCE})+%
+INLINE    = %r%(?:.*?|#{EMPHASIS}|#{STRONG}|#{CITATION}|#{DELETED}|#{INSERTED}|#{SUPERSCRIPT}|#{SUBSCRIPT}|#{UNDERSCORE}|#{RAW}|#{FENCE})+%
+OLIST     = '(?:#|[0-9]+|[A-Za-z])\.'
+LINE      = %r%(?:(?![ \t]*(?:[*-]|(?:#{OLIST})|\[[_X]\])[ \t])#{INLINE}\n)%
 
 RULES     = %r%(\-+|~+|\++|\^+|=+)%
-PbHeading = %r%\A([ \t]*#{RULES})\n([ \t]*)#{LINE}\n\1[ \t]*\n%m
-HEAD      = %r%[ \t]*#{LINE}\n#{RULES}[ \t]*%m
+PbHeading = %r%\A([ \t]*#{RULES})\n([ \t]*)#{LINE}\1[ \t]*\n%m
+HEAD      = %r%[ \t]*#{LINE}#{RULES}[ \t]*%m
 NHeading  = %r%\A[ \t]*([0-9]+|#)\.#{HEAD}\n%m
 Heading   = %r%\A#{HEAD}\n%m
 PageBreak = %r%\A([ \t]*#{RULES})\n\1[ \t]*\n%m
 
-BLOCK     = %r%([ \t]+)#{LINE}(\n+\1#{LINE})*%m
-UList     = %r%\A([ \t]*)[*-]#{LINE}(\n\1[ \t]+#{LINE})*\n%m
-OList     = %r%\A[ \t]*([#0-9A-Za-z]+\.)#{BLOCK}\n%m
-Todo      = %r%\A[ \t]*\[[_X]\]#{BLOCK}\n%m
+def mk_list( mark ) 
+        %r%\A([ \t]*)#{mark}[ \t]+#{LINE}(\1[ \t]+#{LINE})*%m
+end
+UList     = mk_list('[*-]')
+OList     = mk_list(OLIST)
+Todo      = mk_list('\[[_X]\]')
 
-PARA      = %r%#{LINE}(\n#{LINE})*\n%m
-Block     = %r%\A#{PARA}\n%m
+PARA      = %r%#{LINE}(#{LINE})*%m
+Block     = %r%\A#{PARA}%m
 Fenced    = %r%\A\{\{\{(#[\w]+)?\n.*?\n\}\}\}\n%m
 
 ATTRIBUTION = %r%.*?:\n%
-Quote     = %r%\A(#{ATTRIBUTION})?>([ \t]+(>[ \t]+)*#{LINE})?\n%
+Quote     = %r%\A(#{ATTRIBUTION})?>([ \t]+(>[ \t]+)*#{LINE})?%
 Comment   = %r%\A#[ \t]+.*\n%
 
-Definition  = %r%\A\S#{LINE}\n([ \t]+)#{LINE}(\n\1#{LINE})*\n%m
+Definition  = %r%\A(?![ \t])#{LINE}([ \t]+)#{LINE}(?:\1#{LINE})*%m
 
 # Justification requires post-processing interpretation
 # Sidebar requires post-processing interpretation

          
@@ 45,8 49,8 @@ Ident     = %r%\A:[\w \t]+: .*\n%
 
 URI       = %r^[a-zA-Z]{3,}://[a-zA-Z0-9\.]+/*[a-zA-Z0-9/\\%_.]*\?*[a-zA-Z0-9/\\%_.=&amp;]*^
 Link      = %r%\{\w[\w \t]+([ \t]#{URI})?\}%
-Anchor    = %r%\[[:alpha:]#{LINE}\]%
-Footnote  = %r%\{(\d|#)+([ \t]+#{LINE})?\}%
+Anchor    = %r%\[[:alpha:]#{INLINE}\]%
+Footnote  = %r%\{(\d|#)+([ \t]+#{INLINE})?\}%
 
 LinkRef   = %r%\A\.\.[ \t][\w \t]+:[ \t](#{URI}|#{PARA})\n%m
 Directive = %r%\A\.\.[ \t]:[\w \t]+:[ \t].*?\n%

          
@@ 128,8 132,10 @@ def parse( input )
 			rest = $'
 			puts "#"*60 + " EMPTY"
 			#puts "Rest:\n#{rest[0,200]}"
+                when ""
+			break
 		else
-			puts "No match for:\n#{rest[0,200]}"
+			puts "No match for:\n#{rest[0,400].inspect}"
 			exit
 		end
 	end

M rml.re +18 -13

@@ 9,28 9,33 @@ SUBSCRIPT = %r%~[-/*+?^_`\w\d].*?~%
 UNDERSCORE  = %r%_[-/*+?^~`\w\d].*?_%
 RAW       = %r%`[-/*+?^~_\w\d].*?`%
 FENCE     = %r%\{\{\{.*?\}\}\}%
-LINE      = %r%(?![ \t][*-][ \t])(.*?|#{EMPHASIS}|#{STRONG}|#{CITATION}|#{DELETED}|#{INSERTED}|#{SUPERSCRIPT}|#{SUBSCRIPT}|#{UNDERSCORE}|#{RAW}|#{FENCE})+%
+INLINE    = %r%(?:.*?|#{EMPHASIS}|#{STRONG}|#{CITATION}|#{DELETED}|#{INSERTED}|#{SUPERSCRIPT}|#{SUBSCRIPT}|#{UNDERSCORE}|#{RAW}|#{FENCE})+%
+OLIST     = '(?:#|[0-9]+|[A-Za-z])\.'
+LINE      = %r%(?:(?![ \t]*(?:[*-]|(?:#{OLIST})|\[[_X]\])[ \t])#{INLINE}\n)%
 
 RULES     = %r%(\-+|~+|\++|\^+|=+)%
-PbHeading = %r%\A([ \t]*#{RULES})\n([ \t]*)#{LINE}\n\1[ \t]*\n%m
-HEAD      = %r%[ \t]*#{LINE}\n#{RULES}[ \t]*%m
+PbHeading = %r%\A([ \t]*#{RULES})\n([ \t]*)#{LINE}\1[ \t]*\n%m
+HEAD      = %r%[ \t]*#{LINE}#{RULES}[ \t]*%m
 NHeading  = %r%\A[ \t]*([0-9]+|#)\.#{HEAD}\n%m
 Heading   = %r%\A#{HEAD}\n%m
 PageBreak = %r%\A([ \t]*#{RULES})\n\1[ \t]*\n%m
 
-BLOCK     = %r%([ \t]+)#{LINE}(\n+\1#{LINE})*%m
-UList     = %r%\A([ \t]*)[*-]#{LINE}(\n\1[ \t]+#{LINE})*\n%m
-OList     = %r%\A[ \t]*([#0-9A-Za-z]+\.)#{BLOCK}\n%m
-Todo      = %r%\A[ \t]*\[[_X]\]#{BLOCK}\n%m
+def mk_list( mark ) 
+        %r%\A([ \t]*)#{mark}[ \t]+#{LINE}(\1[ \t]+#{LINE})*%m
+end
+UList     = mk_list('[*-]')
+OList     = mk_list(OLIST)
+Todo      = mk_list('\[[_X]\]')
 
-PARA      = %r%#{LINE}(\n#{LINE})*\n%m
-Block     = %r%\A#{PARA}\n%m
+PARA      = %r%#{LINE}(#{LINE})*%m
+Block     = %r%\A#{PARA}%m
 Fenced    = %r%\A\{\{\{(#[\w]+)?\n.*?\n\}\}\}\n%m
 
-Quote     = %r%\A>[ \t]+(>[ \t]+#{LINE})*#{LINE}\n%
+ATTRIBUTION = %r%.*?:\n%
+Quote     = %r%\A(#{ATTRIBUTION})?>([ \t]+(>[ \t]+)*#{LINE})?%
 Comment   = %r%\A#[ \t]+.*\n%
 
-Definition = %r%\A\w#{LINE}\n([ \t]+)#{LINE}(\n\1#{LINE})*\n%m
+Definition  = %r%\A(?![ \t])#{LINE}([ \t]+)#{LINE}(?:\1#{LINE})*%m
 
 # Justification requires post-processing interpretation
 # Sidebar requires post-processing interpretation

          
@@ 42,8 47,8 @@ Ident     = %r%\A:[\w \t]+: .*\n%
 
 URI       = %r^[a-zA-Z]{3,}://[a-zA-Z0-9\.]+/*[a-zA-Z0-9/\\%_.]*\?*[a-zA-Z0-9/\\%_.=&amp;]*^
 Link      = %r%\{\w[\w \t]+([ \t]#{URI})?\}%
-Anchor    = %r%\[[:alpha:]#{LINE}\]%
-Footnote  = %r%\{(\d|#)+([ \t]+#{LINE})?\}%
+Anchor    = %r%\[[:alpha:]#{INLINE}\]%
+Footnote  = %r%\{(\d|#)+([ \t]+#{INLINE})?\}%
 
 LinkRef   = %r%\A\.\.[ \t][\w \t]+:[ \t](#{URI}|#{PARA})\n%m
 Directive = %r%\A\.\.[ \t]:[\w \t]+:[ \t].*?\n%