checkpoint harmony

[sbp.git] / tests / tibdoc.g

diff --git a/tests/tibdoc.g b/tests/tibdoc.g
index 1ea798d..81061e5 100644 (file)
--- a/tests/tibdoc.g
+++ b/tests/tibdoc.g
@@ -1,3 +1,5 @@
+x::="x"
+// interactions between !=> and &~ mean that I need to rethink the chartage
 // indentation styling...
 // literal blocks [[need to ignore bracing]] double-colon style?
 // definition -- by prior line indentation, like headings in the original structured text
@@ -5,26 +7,10 @@
 // dropcap
 // output formats: latex, contex, ps, pdf, html, man, txt, rfc
 
-// URGENT: why does swapping [a-zA-Z0-9] for alphanum in "item" cause severe breakage?
-// URGENT: why does enabling "url" cause severe breakage? (probably same cause)
-//   ... something to do with unit productions
-
-// output formats: latex, contex, ps, pdf, html, man, txt, rfc
-// bullet list
-// numbered list
-// horizontal rule
-// dropcap
-// smallcap
-// strikethrough
-// link
+// escapification
+// comment
 // math
 // image
-// today's date, last edit date
-// blockquote
-// superscript
-// subscript
-// citations/references
-// typewriter-text
 // figures
 
 // FIXME: these have to go at the top so they have their dropAll bit set before PreSequence.build...
@@ -34,51 +20,72 @@ nw       ::= ~[\r\n\ ]
 
 //////////////////////////////////////////////////////////////////////////////
 
-s                 ::= {Doc} => top
+s                 ::= Doc => top
 
-Doc               ::= Header   Body                       /ws => doc
-Header            ::= "header" { kv */ ws }               /ws => header
+Doc               ::= {Header} Body                   /ws => doc
+Header            ::= "header" { kv */ ws }           /ws => header
 Body              ::= Section*/ws                             => body
-Section           ::= SectionHeader Paragraph*/ws         /ws => section
+Section           ::= { SectionHeader Paragraph* /ws => section }
 SectionHeader     ::= "==" SectionHeaderBody "=="
 SectionHeaderBody ::=  "=" SectionHeaderBody "="
                     >      ws text ws
 
+sp    !::= " "**
+blank  ::= sp "\n" sp "\n" ws
+
 kv         ::= word "=" text /ws => kv1
 
 num !::= [0-9]++
-Paragraph  ::= { "\"\"" ws text }        => "blockquote"
-             > { "*" " " ws text }       => "ul"
-             > { "#" " " ws text }       => "ol"
-             > { num " " ws text }       => "ol"
-             > { "---" "-"* }            => "hr"
-             > { text }                  => "p"
-
-text       ::= item */ ws
+Paragraph  ::= { "\"\"" ws text  }  => "blockquote"
+             > { "*" " " ws text }  => "ul"
+             > { "#" " " ws text }  => "ol"
+             > { num " " ws text }  => "ol"
+             > { "---" "-"*      }  => "hr"
+             > { text }             => "p"
+
+text       ::= Item
+Itemx      ::= ws Item | ()
+Item       ::= blockquote
+             > pre         Itemx => []
+             > structured  Itemx => []
+             > structuredx Itemx => []
+             > styled      Itemx => []
+             > qtext       Itemx => []
+             > alphanum++  Itemx => []
+             > symbol      Itemx => []
+//             > sym++       Itemx => []
+             > Paragraph   Itemx => []
+
+symbol     ::= sym++
+
+blockquote ::= "\"\"" text "\"\""        => "blockquote"
+             | "\"\"" block              => "blockquote"
+             
+qtext      ::= "\"" text "\""            => "quoted"
 pre        ::= "[verbatim]" { ~[]+ } /ws => "verbatim"   // FIXME doesn't work
-item       ::= pre
-             > email
-             > structured
-             > styled
-             > "\"" text "\""    => quoted
-             > [a-zA-Z0-9]++
-             > symbol
-
-symbol     ::= symbolx & sym++
-symbolx    ::= "--"  => emdash
-             | ","
-             | ":"
-             | ";"
-
-styled     ::= "**" text "**" => bold
-             | "__" text "__" => ul
-             | "~~" text "~~" => it     // hard to type
-             | "((" text "))" => footnote
-
-structured ::= glyph
-//             | url
 
-glyph      ::= "(r)" | "(c)" | "(tm)"  // euro symbol?
+styled     ::= "__" text "__"      => ul
+             | "((" text "))"      => footnote
+             | ( "[[" text "]]"      => tt
+               >  "[" word "]"       => citation
+               )
+             | "!!" text "!!"      => strikethrough
+             | "^^" (word|block)   => superscript
+             | ",," (word|block)   => subscript
+             | "\\sc" block        => smallcap
+             | "**" text "**"      => bold
+             | "!" (word|block)    => keyword
+             >  "*" text "*"       => it
+
+block ::= { text }
+structured ::= { text } "->" (url|email) => link
+              //> alphanum++ "->" (url|email) => link
+structuredx ::= glyph
+              > email
+              > url
+
+glyph      ::= "(r)" | "(c)" | "(tm)" | "--"  // euro symbol?
+             | "\\today" -> ~[a-z] => today
 
 
 // URLs //////////////////////////////////////////////////////////////////////////////
@@ -98,11 +105,11 @@ url_login  ::= username (":" password) "@" => "login"
 method     ::= [+\-.a-z0-9]+ 
 port       ::= [0-9]+
 
-domain     ::= part +/ "."
+domain     ::= (part +/ ".") -> ~"."
 part       ::= [a-zA-Z0-9\-]++    // interesting use of boolean grammars
 //            &~ ([\-0-9] ~[]* | ~[]* [\-0-9])
 
-email      ::= username "@" host      => email
+email      ::= username "@" host -> ~[.] => emailaddr
 host       ::= [0-9]+ "." [0-9]+ "." [0-9]+ "." [0-9]+ => "ip"
              | domain
 
@@ -123,8 +130,8 @@ escaped  ::= "\\n" => "\n"
 // Chars ///////////////////////////////////////////////////////////////
 
 alpha    ::= [a-zA-Z]
-num      ::= [0-9]
+//num      ::= [0-9]
 alphanum ::= [a-zA-Z0-9]
-sym      ::= ~[a-zA-Z0-9\ \r\n]
+sym      ::= ~[a-zA-Z0-9\ \r\n=\">]