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 w          = " " | "\n" | "\r"
  52 ws         =  "()":: w**
  53 //           |  "()":: w** "#" (~[\n])* "\n" ws
  54 nw         = ~[\r\n\ ]
  55
  56 //////////////////////////////////////////////////////////////////////////////
  57
  58 s                   = Doc
  59
  60 Doc                 = head:Header ws! body:Body
  61 Header              = { "\\header" { KeyVal */ ws } /ws }
  62 Body                = { Section } */ws
  63 Section             = SectionHeader ws! Paragraph*
  64 SectionHeader       = "==" SectionHeaderBody "=="
  65 SectionHeaderBody   =  "=" SectionHeaderBody "="
  66                     >      ws! alphanum++ ws!
  67
  68 sp       = " "**
  69 blank    = sp! "\n" sp! "\n" ws!
  70
  71 KeyVal       = key:bareword "=" val:text /ws
  72 wp           = w++
  73 num          = [0-9]++
  74
  75 Paragraph   = { Blockquote:: "\"\" "    text }
  76             > HR::         { "---" "-"*      }
  77             > P::          { text            }
  78
  79 onums        = nums (". "|") ")!
  80 any          = ~[]*
  81
  82 uli          =  "* "         (ws! text &~ any (oli|uli)!)
  83 oli          = onums!        (ws! text &~ any (oli|uli)!)
  84
  85 text         = Item
  86
  87 Item*/ws     =
  88                blockquote
  89 //             > ^"#" ws! { ~[]* }
  90              > { UL:: uli+/ws }
  91              | { OL:: oli+/ws }
  92              > Verbatim
  93              > InlineGrammar
  94              > link
  95              > structured
  96              > styled
  97              > (Chars:: alphanum++)
  98              > "\"" text "\""
  99              > (Symbol:: sym++)
 100              > { Block:: text }
 101
 102 word = Chars:: bareword
 103
 104 blockquote = "adsfafewag"
 105 //blockquote   = Blockquote:: "\"\"" (block | text "\"\"")
 106
 107 Verbatim      = "[verbatim]" ws! { (~[])++ }
 108
 109 #import meta.g as meta
 110 InlineGrammar = "\grammar"   ws! { meta.Grammar }
 111
 112 styled       = Underline::     "__" text "__"
 113              | Footnote::      "((" text "))"
 114              | TT::            "[[" text "]]"
 115              | Citation::       "[" word "]"
 116              | Strikethrough:: "!!" text "!!"
 117              | Superscript::   "^^" (word|block)
 118              | Subscript::     ",," (word|block)
 119              | Smallcap::      "\\sc" block
 120              | Bold::           "++" text "++"
 121              | Keyword::         "!" (word|block)
 122              | Italic::         "**" text "**"
 123
 124 block         = { text }
 125
 126 link          = text:({ text }|word)  "->" href:(url|email)
 127
 128 structured    = command & "\\" ([a-zA-Z0-9]++)! block?
 129               > glyph
 130               > email
 131               > url
 132
 133 glyph        = euro::     "(e)"
 134              | r::        "(r)"
 135              | c::        "(c)"
 136              | tm::       "(tm)"
 137              | emdash::   "--"
 138              | ellipses:: "..."
 139              | cent::     "\\cent"
 140
 141 command      = "\\" [a-z]++
 142
 143 // URLs //////////////////////////////////////////////////////////////////////////////
 144
 145 // interesting opportunity to show off boolean grammars here: define other
 146 // subtypes of url (ftp, etc) as conjunctions, but the "master pattern"
 147 // only gets parsed once
 148
 149 urlpath      = urlchar* -> ~urlv           // this ~urlv should be handled by url! bug!
 150 username     = [a-zA-Z0-9;/?:&=$\-_.+]++
 151 password     = [a-zA-Z0-9;/?:&=$\-_.+]++
 152 urlc         = [a-zA-Z0-9;/?:&=$\-_.+@]
 153 urlv         = urlc | [%]
 154 urlchar      = urlc
 155              | urlescape:: "%" [0-9] [0-9]
 156 url          = Mailto:: "mailto" ":"   email -> ~urlv
 157              > URL::
 158                   method:method
 159                   "://"
 160                   login:url_login?
 161                   host:host
 162                   port:(":" nums)?
 163                   path:("/" urlpath)?
 164                      -> ~urlv
 165 url_login    = Login:: username:username password:(":" password) "@"
 166 method       = [+\-.a-z0-9]+
 167 domain       = (part +/ ".") -> ~"."
 168 part         = [a-zA-Z0-9\-]++
 169 // interesting use of boolean grammars
 170 //            &~ ([\-0-9] ~[]* | ~[]* [\-0-9])
 171
 172 email        = user:username "@" host:host -> ~[.]
 173 nums         = [0-9]++
 174 host         = IP::  nums "." nums "." nums "." nums
 175              | DNS:: domain
 176
 177
 178
 179 // Tokens ///////////////////////////////////////////////////////////////////
 180
 181 bareword   = alphanum++
 182            | quoted
 183
 184 quoted     = "\"" ((~[\"\\] | escaped)+) "\""
 185            | "\"\"":: "\"\""
 186 escaped    = lf:: "\\n"
 187            | cr:: "\\r"
 188            | "\\" ~[nr]
 189
 190
 191 // Chars ///////////////////////////////////////////////////////////////
 192
 193 alpha      = [a-zA-Z]
 194 alphanum   = [a-zA-Z0-9]
 195 sym        = ~[a-zA-Z0-9\ \r\n=\">]
 196 //sym        = [,()]
 197
 198