tests/tibdoc.g

   1 // interactions between !=> and &~ mean that I need to rethink the chartage
   2 // indentation styling...
   3 // literal blocks [[need to ignore bracing]] double-colon style?
   4 // definition -- by prior line indentation, like headings in the original structured text
   5 // tables
   6 // dropcap
   7 // output formats: latex, contex, ps, pdf, html, man, txt, rfc
   8
   9 // escapification
  10 // comment
  11 // math
  12 // image
  13 // figures
  14
  15 // "reference-style" links
  16 //
  17 //   this[1] is fun
  18 //
  19 //   [1] http://...
  20 //
  21
  22 // nonbreaking text?
  23 // degree: 15^o
  24 // Arrows: <- -> => <= <->
  25
  26 // textblocks:
  27 //  - attention, caution, danger, error, hint, important, note, tip, warning
  28 // definition
  29 // sidebar
  30 // figure-with-caption
  31 // epigraph (end-of-chapter note)
  32 // compound paragraph (??)
  33 // csv-table?
  34 // table of contents
  35 // header, footer
  36 // #include
  37
  38 // simple macros (#define) (\define)
  39
  40 // table representation
  41 //
  42 //  \table
  43 //     a bbb c
  44 //     ddd   e
  45 //
  46 //     [a] ...
  47 //     [b] ...
  48 //     [c] ...
  49
  50 // FIXME: these have to go at the top so they have their dropAll bit set before PreSequence.build...
  51 ws         = w**
  52 w          =  [\r\n\ ]
  53 nw         = ~[\r\n\ ]
  54
  55 //////////////////////////////////////////////////////////////////////////////
  56
  57 s                   = Doc
  58
  59 Doc                 = Doc:: head:{Header} body:Body   /ws
  60 Header              = Header:: "header" attrs:{ kv */ ws }     /ws
  61 Body                = Body:: sections:(Section*/ws)
  62 Section             = { Section:: header:SectionHeader paragraphs:Paragraph* /ws }
  63 SectionHeader       = "==" SectionHeaderBody "=="
  64 SectionHeaderBody   =  "=" SectionHeaderBody "="
  65                     >      !ws (Chars:: text:alphanum++) !ws
  66
  67 sp       = " "**
  68 blank    = !sp "\n" !sp "\n" !ws
  69
  70
  71 kv           = kv:: key:word "=" val:text /ws
  72 wp           = w++
  73 num          = [0-9]++
  74 Paragraph    = Blockquote:: { "\"\" "    text }
  75              > HR::         { "---" "-"*      }
  76              >              { P:: t:text }
  77
  78 onums        = nums !(". "|") ")
  79 any          = ~[]*
  80
  81 uli          =  "* "         (!ws text &~ any (oli|uli))
  82 oli          = !("# "|onums) (!ws text &~ any (oli|uli))
  83
  84 text         = Item
  85 Itemx        = !ws Item
  86              | ()
  87 Item         = blockquote
  88              > "[]":: { UL:: uli+/ws }          Itemx
  89              | "[]":: { OL:: oli+/ws }          Itemx
  90              > "[]":: pre                       Itemx
  91              > "[]":: link                      Itemx
  92              > "[]":: structured                Itemx
  93              > "[]":: styled                    Itemx
  94              > "[]":: (Chars:: text:alphanum++)      Itemx
  95              > "[]":: "\"" text "\""            Itemx
  96              > "[]":: symbol                    Itemx
  97              > "[]":: (Symbol:: sym++)          Itemx
  98              > "[]":: Paragraph                 Itemx
  99
 100 blockquote   = Blockquote:: "\"\"" text "\"\""
 101              | Blockquote:: "\"\"" block
 102
 103 pre          = Verbatim:: "[verbatim]" { ~[]+ } /ws   // FIXME doesn't work
 104
 105 styled       = Underline::     "__" text "__"
 106              | Footnote::      "((" text "))"
 107              | TT::            "[[" text "]]"
 108              | Citation::       "[" word "]"
 109              | Strikethrough:: "!!" text "!!"
 110              | Superscript::   "^^" (word|block)
 111              | Subscript::     ",," (word|block)
 112              | Smallcap::      "\\sc" block
 113              | Bold::           "++" text "++"
 114              | Keyword::         "!" (word|block)
 115              | Italic::         "**" text "**"
 116
 117 block         = { text }
 118
 119 link          = Link:: text:({ text })     "->" href:(url|email)
 120               > Link:: text:alphanum++ !ws "->" href:(url|email)
 121
 122 structured    = command & "\\" [a-zA-Z0-9]++ block?
 123               > glyph
 124               > email
 125               > url
 126
 127 glyph        = Euro:: "(e)" | "(r)" | "(c)" | "(tm)" | "--" | "..."
 128
 129 command      = Today::     "\\today"
 130              | LineBreak:: "\\br"
 131
 132 // URLs //////////////////////////////////////////////////////////////////////////////
 133
 134 // interesting opportunity to show off boolean grammars here: define other
 135 // subtypes of url (ftp, etc) as conjunctions, but the "master pattern"
 136 // only gets parsed once
 137
 138 urlpath      = urlchar*
 139 username     = [a-zA-Z0-9;/?:&=$\-_.+]++
 140 password     = [a-zA-Z0-9;/?:&=$\-_.+]++
 141 urlc         = [a-zA-Z0-9;/?:&=$\-_.+@]
 142 urlv         = urlc | [%]
 143 urlchar      = urlc
 144              | urlescape:: "%" [0-9] [0-9]
 145 url          = "mailto" ":"   email -> ~urlv
 146              > URL:: method:method "://" login:url_login? host:host port:(":" nums)? path:("/" urlpath)? -> ~urlv
 147 url_login    = Login:: username:username password:(":" password) "@"
 148 method       = [+\-.a-z0-9]+
 149 domain       = (part +/ ".") -> ~"."
 150 part         = [a-zA-Z0-9\-]++
 151 // interesting use of boolean grammars
 152 //            &~ ([\-0-9] ~[]* | ~[]* [\-0-9])
 153
 154 email        = Email:: user:username "@" host:host -> ~[.]
 155 nums         = [0-9]++
 156 host         = IP::  nums "." nums "." nums "." nums
 157              | DNS:: domain
 158
 159
 160
 161 // Tokens ///////////////////////////////////////////////////////////////////
 162
 163 word       = alphanum++
 164            | quoted
 165
 166 quoted     = "\"" ((~[\"\\] | escaped)+) "\""
 167            | "":: "\"\""
 168 escaped    = lf:: "\\n"
 169            | cr:: "\\r"
 170            | "\\" ~[nr]
 171
 172
 173 // Chars ///////////////////////////////////////////////////////////////
 174
 175 alpha      = [a-zA-Z]
 176 //num        = [0-9]
 177 alphanum   = [a-zA-Z0-9]
 178 sym        = ~[a-zA-Z0-9\ \r\n=\">]
 179
 180