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 // consider ++bold++ and **italic**?
  23 // \br
  24 // nonbreaking text?
  25 // ellipsis detection (...)
  26 // degree: 15^o
  27 // Arrows: <- -> => <= <->
  28
  29 // textblocks:
  30 //  - attention, caution, danger, error, hint, important, note, tip, warning
  31 // definition
  32 // sidebar
  33 // figure-with-caption
  34 // epigraph (end-of-chapter note)
  35 // compound paragraph (??)
  36 // csv-table?
  37 // table of contents
  38 // header, footer
  39 // #include
  40
  41 // simple macros (#define) (\define)
  42 // today's date
  43
  44 // table representation
  45 //
  46 //  \table
  47 //     a bbb c
  48 //     ddd   e
  49 //
  50 //     [a] ...
  51 //     [b] ...
  52 //     [c] ...
  53
  54 // FIXME: these have to go at the top so they have their dropAll bit set before PreSequence.build...
  55 ws         = w**
  56 w          =  [\r\n\ ]
  57 nw         = ~[\r\n\ ]
  58
  59 //////////////////////////////////////////////////////////////////////////////
  60
  61 s                   = top:: Doc
  62
  63 Doc                 = doc::    {Header} Body                   /ws
  64 Header              = header:: "header" { kv */ ws }           /ws
  65 Body                = body::   Section*/ws
  66 Section             = { section:: SectionHeader Paragraph* /ws }
  67 SectionHeader       = "==" SectionHeaderBody "=="
  68 SectionHeaderBody   =  "=" SectionHeaderBody "="
  69                     >      !ws text !ws
  70
  71 sp       = " "**
  72 blank    = !sp "\n" !sp "\n" !ws
  73
  74 kv           = kv1:: word "=" text /ws
  75 wp           = w++
  76 num          = [0-9]++
  77 Paragraph    = blockquote:: { "\"\"" !ws  text }
  78              > hr::         { "---" "-"*      }
  79              > p::          { text }
  80
  81 onums        = nums !(". "|") ")
  82 any          = ~[]*
  83
  84 uli          = li:: "* "          (!ws text &~ any (oli|uli))
  85 oli          = li:: ("# "|onums) (!ws text &~ any (oli|uli))
  86
  87 //
  88
  89 text         = text:: Item
  90 Itemx        = !ws Item
  91              | ()
  92 Item         = blockquote
  93              > "[]":: { ul:: uli+/ws }          Itemx
  94              | "[]":: { ol:: oli+/ws }          Itemx
  95              > "[]":: pre                       Itemx
  96              > "[]":: structured                Itemx
  97              > "[]":: structuredx               Itemx
  98              > "[]":: styled                    Itemx
  99              > "[]":: qtext                     Itemx
 100              > "[]":: (stringify:: alphanum++)  Itemx
 101              > "[]":: symbol                    Itemx
 102              > "[]":: (stringify:: sym++)       Itemx
 103              > "[]":: Paragraph                 Itemx
 104
 105 blockquote   = blockquote:: "\"\"" text "\"\""
 106              | blockquote:: "\"\"" block
 107
 108 qtext        = quoted:: "\"" text "\""
 109 pre          = verbatim:: "[verbatim]" { ~[]+ } /ws   // FIXME doesn't work
 110
 111 styled       = underline:: "__" text "__"
 112              | footnote:: "((" text "))"
 113              | ( tt:: "[[" text "]]"
 114              | citation::   "[" text "]"
 115                )
 116              | strikethrough:: "!!" text "!!"
 117              | superscript:: "^^" (word|block)
 118              | subscript:: ",," (word|block)
 119              | smallcap:: "\\sc" block
 120              | bold:: "**" text "**"
 121              | keyword:: "!" (word|block)
 122              > it:: "*" text "*"
 123
 124 //
 125
 126 block   = { text }
 127 structured   = link:: { text } "->" (url|email)
 128               //> alphanum++ "->" (url|email) => link
 129 structuredx   = glyph
 130               > email
 131               > url
 132
 133 glyph        = "(r)" | "(c)" | "(tm)" | "--"  // euro symbol?
 134              | today:: "\\today" -> ~[a-z]
 135
 136
 137 // URLs //////////////////////////////////////////////////////////////////////////////
 138
 139 // interesting opportunity to show off boolean grammars here: define other
 140 // subtypes of url (ftp, etc) as conjunctions, but the "master pattern"
 141 // only gets parsed once
 142
 143 urlpath      = urlchar*
 144 username     = stringify:: [a-zA-Z0-9;/?:&=$\-_.+]++
 145 password     = stringify:: [a-zA-Z0-9;/?:&=$\-_.+]++
 146 urlchar      = [a-zA-Z0-9;/?:&=$\-_.+@]
 147              | "%":: "%" [0-9] [0-9]
 148 url          = "mailto" ":"   email
 149              > url::   method "://" url_login? host (":" nums)? ("/" urlpath)?
 150 url_login    = login:: username (":" password) "@"
 151 method       = stringify:: [+\-.a-z0-9]+
 152 domain       = domain:: (part +/ ".") -> ~"."
 153 part         = stringify:: [a-zA-Z0-9\-]++
 154 // interesting use of boolean grammars
 155 //            &~ ([\-0-9] ~[]* | ~[]* [\-0-9])
 156
 157 email        = emailaddr:: username "@" host -> ~[.]
 158 nums         = stringify:: [0-9]++
 159 host         = ip:: nums "." nums "." nums "." nums
 160              | domain
 161
 162
 163
 164 // Tokens ///////////////////////////////////////////////////////////////////
 165
 166 word       = stringify:: alphanum++
 167            | quoted
 168
 169 quoted     = "\"" ((~[\"\\] | escaped)+) "\""
 170            | "":: "\"\""
 171 escaped    = "\n":: "\\n"
 172            | "\r":: "\\r"
 173            | "\\" ~[nr]
 174
 175
 176 // Chars ///////////////////////////////////////////////////////////////
 177
 178 alpha      = [a-zA-Z]
 179 //num        = [0-9]
 180 alphanum   = [a-zA-Z0-9]
 181 sym        = ~[a-zA-Z0-9\ \r\n=\">]
 182
 183