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                 = head:{Header} body:Body  /ws
  60 Header              = "header" attrs:{ KeyVal */ ws }  /ws
  61 Body                = {Section}*/ws
  62 Section             = SectionHeader Paragraph* /ws
  63 SectionHeader       = "==" SectionHeaderBody "=="
  64 SectionHeaderBody   =  "=" SectionHeaderBody "="
  65                     >      !ws alphanum++ !ws
  66
  67 sp       = " "**
  68 blank    = !sp "\n" !sp "\n" !ws
  69
  70 KeyVal       = key:word "=" val:text /ws
  71 wp           = w++
  72 num          = [0-9]++
  73
  74 Paragraph   = Blockquote:: { "\"\" "    text }
  75             > HR::         { "---" "-"*      }
  76             > P::          { 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                 Itemx
  88              > "[]":: { UL:: uli+/ws }           Itemx
  89              | "[]":: { OL:: oli+/ws }           Itemx
  90              > "[]":: pre                        Itemx
  91              > "[]":: link                       Itemx
  92              > "[]":: structured                 Itemx
  93              > "[]":: styled                     Itemx
  94              > "[]":: (Chars:: alphanum++)       Itemx
  95              > "[]":: "\"" text "\""             Itemx
  96              > "[]":: (Symbol:: sym++)           Itemx
  97              > "[]":: { Block:: text }           Itemx
  98
  99 blockquote   = Blockquote:: "\"\"" (block | text "\"\"")
 100
 101 pre          = Verbatim:: "[verbatim]" { ~[]+ } /ws   // FIXME doesn't work
 102
 103 styled       = Underline::     "__" text "__"
 104              | Footnote::      "((" text "))"
 105              | TT::            "[[" text "]]"
 106              | Citation::       "[" word "]"
 107              | Strikethrough:: "!!" text "!!"
 108              | Superscript::   "^^" (word|block)
 109              | Subscript::     ",," (word|block)
 110              | Smallcap::      "\\sc" block
 111              | Bold::           "++" text "++"
 112              | Keyword::         "!" (word|block)
 113              | Italic::         "**" text "**"
 114
 115 block         = { text }
 116
 117 link          = LinkText:: text:({ text })      "->" href:(url|email)
 118               > LinkChars:: text:alphanum++  !ws "->" href:(url|email)
 119
 120 structured    = command & "\\" !([a-zA-Z0-9]++) block?
 121               > glyph
 122               > email
 123               > url
 124
 125 glyph        = Euro:: "(e)" | "(r)" | "(c)" | "(tm)" | emdash:: "--" | "..."
 126
 127 command      = Today::     "\\today"
 128              | LineBreak:: "\\br"
 129
 130 // URLs //////////////////////////////////////////////////////////////////////////////
 131
 132 // interesting opportunity to show off boolean grammars here: define other
 133 // subtypes of url (ftp, etc) as conjunctions, but the "master pattern"
 134 // only gets parsed once
 135
 136 urlpath      = urlchar*
 137 username     = [a-zA-Z0-9;/?:&=$\-_.+]++
 138 password     = [a-zA-Z0-9;/?:&=$\-_.+]++
 139 urlc         = [a-zA-Z0-9;/?:&=$\-_.+@]
 140 urlv         = urlc | [%]
 141 urlchar      = urlc
 142              | urlescape:: "%" [0-9] [0-9]
 143 url          = Mailto:: "mailto" ":"   email -> ~urlv
 144              > URL::
 145                   method:method
 146                   "://"
 147                   login:url_login?
 148                   host:host
 149                   port:(":" nums)?
 150                   path:("/" urlpath)?
 151                      -> ~urlv
 152 url_login    = Login:: username:username password:(":" password) "@"
 153 method       = [+\-.a-z0-9]+
 154 domain       = (part +/ ".") -> ~"."
 155 part         = [a-zA-Z0-9\-]++
 156 // interesting use of boolean grammars
 157 //            &~ ([\-0-9] ~[]* | ~[]* [\-0-9])
 158
 159 email        = Email:: user:username "@" host:host -> ~[.]
 160 nums         = [0-9]++
 161 host         = IP::  nums "." nums "." nums "." nums
 162              | DNS:: domain
 163
 164
 165
 166 // Tokens ///////////////////////////////////////////////////////////////////
 167
 168 word       = alphanum++
 169            | quoted
 170
 171 quoted     = "\"" ((~[\"\\] | escaped)+) "\""
 172            | "\"\"":: "\"\""
 173 escaped    = lf:: "\\n"
 174            | cr:: "\\r"
 175            | "\\" ~[nr]
 176
 177
 178 // Chars ///////////////////////////////////////////////////////////////
 179
 180 alpha      = [a-zA-Z]
 181 alphanum   = [a-zA-Z0-9]
 182 sym        = ~[a-zA-Z0-9\ \r\n=\">]
 183
 184