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 ws! body:Body
  60 Header              = { "\\header" { KeyVal */ ws } /ws }
  61 Body                = { Section } */ws
  62 Section             = SectionHeader ws! Paragraph*
  63 SectionHeader       = "==" SectionHeaderBody "=="
  64 SectionHeaderBody   =  "=" SectionHeaderBody "="
  65                     >      ws! alphanum++ ws!
  66
  67 sp       = " "**
  68 blank    = sp! "\n" sp! "\n" ws!
  69
  70 KeyVal       = key:bareword "=" 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
  86 Item*/ws     =
  87                blockquote
  88              > { UL:: uli+/ws }
  89              | { OL:: oli+/ws }
  90              > Verbatim
  91              > link
  92              > structured
  93              > styled
  94              > (Chars:: alphanum++)
  95              > "\"" text "\""
  96              > (Symbol:: sym++)
  97              > { Block:: text }
  98
  99 word = Chars:: bareword
 100
 101 blockquote = "adsfafewag"
 102 //blockquote   = Blockquote:: "\"\"" (block | text "\"\"")
 103
 104 Verbatim     = "[verbatim]" ws! { (~[])++ }
 105
 106 styled       = Underline::     "__" text "__"
 107              | Footnote::      "((" text "))"
 108              | TT::            "[[" text "]]"
 109              | Citation::       "[" word "]"
 110              | Strikethrough:: "!!" text "!!"
 111              | Superscript::   "^^" (word|block)
 112              | Subscript::     ",," (word|block)
 113              | Smallcap::      "\\sc" block
 114              | Bold::           "++" text "++"
 115              | Keyword::         "!" (word|block)
 116              | Italic::         "**" text "**"
 117
 118 block         = { text }
 119
 120 link          = text:({ text }|word)  "->" href:(url|email)
 121
 122 structured    = command & "\\" ([a-zA-Z0-9]++)! block?
 123               > glyph
 124               > email
 125               > url
 126
 127 glyph        = euro::     "(e)"
 128              | r::        "(r)"
 129              | c::        "(c)"
 130              | tm::       "(tm)"
 131              | emdash::   "--"
 132              | ellipses:: "..."
 133              | cent::     "\\cent"
 134
 135 command      = "\\" [a-z]++
 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* -> ~urlv           // this ~urlv should be handled by url! bug!
 144 username     = [a-zA-Z0-9;/?:&=$\-_.+]++
 145 password     = [a-zA-Z0-9;/?:&=$\-_.+]++
 146 urlc         = [a-zA-Z0-9;/?:&=$\-_.+@]
 147 urlv         = urlc | [%]
 148 urlchar      = urlc
 149              | urlescape:: "%" [0-9] [0-9]
 150 url          = Mailto:: "mailto" ":"   email -> ~urlv
 151              > URL::
 152                   method:method
 153                   "://"
 154                   login:url_login?
 155                   host:host
 156                   port:(":" nums)?
 157                   path:("/" urlpath)?
 158                      -> ~urlv
 159 url_login    = Login:: username:username password:(":" password) "@"
 160 method       = [+\-.a-z0-9]+
 161 domain       = (part +/ ".") -> ~"."
 162 part         = [a-zA-Z0-9\-]++
 163 // interesting use of boolean grammars
 164 //            &~ ([\-0-9] ~[]* | ~[]* [\-0-9])
 165
 166 email        = user:username "@" host:host -> ~[.]
 167 nums         = [0-9]++
 168 host         = IP::  nums "." nums "." nums "." nums
 169              | DNS:: domain
 170
 171
 172
 173 // Tokens ///////////////////////////////////////////////////////////////////
 174
 175 bareword   = alphanum++
 176            | quoted
 177
 178 quoted     = "\"" ((~[\"\\] | escaped)+) "\""
 179            | "\"\"":: "\"\""
 180 escaped    = lf:: "\\n"
 181            | cr:: "\\r"
 182            | "\\" ~[nr]
 183
 184
 185 // Chars ///////////////////////////////////////////////////////////////
 186
 187 alpha      = [a-zA-Z]
 188 alphanum   = [a-zA-Z0-9]
 189 sym        = ~[a-zA-Z0-9\ \r\n=\">]
 190 //sym        = [,()]
 191
 192