lexers/markdown.lua

   1 -- Copyright 2006-2015 Mitchell mitchell.att.foicica.com. See LICENSE.
   2 -- Markdown LPeg lexer.
   3
   4 local l = require('lexer')
   5 local token, word_match = l.token, l.word_match
   6 local P, R, S = lpeg.P, lpeg.R, lpeg.S
   7
   8 local M = {_NAME = 'markdown'}
   9
  10 -- Whitespace.
  11 local ws = token(l.WHITESPACE, l.space^1)
  12
  13 -- Block elements.
  14 local header = token('h6', l.starts_line('######') * l.nonnewline^0) +
  15                token('h5', l.starts_line('#####') * l.nonnewline^0) +
  16                token('h4', l.starts_line('####') * l.nonnewline^0) +
  17                token('h3', l.starts_line('###') * l.nonnewline^0) +
  18                token('h2', l.starts_line('##') * l.nonnewline^0) +
  19                token('h1', l.starts_line('#') * l.nonnewline^0)
  20
  21 local blockquote = token(l.STRING,
  22                          lpeg.Cmt(l.starts_line(S(' \t')^0 * '>'),
  23                                   function(input, index)
  24                                     local _, e = input:find('\n[ \t]*\r?\n',
  25                                                             index)
  26                                     return (e or #input) + 1
  27                                   end))
  28
  29 local blockcode = token('code', l.starts_line(P(' ')^4 + P('\t')) * -P('<') *
  30                                 l.nonnewline^0)
  31
  32 local hr = token('hr', lpeg.Cmt(l.starts_line(S(' \t')^0 * lpeg.C(S('*-_'))),
  33                                 function(input, index, c)
  34                                   local line = input:match('[^\n]*', index)
  35                                   line = line:gsub('[ \t]', '')
  36                                   if line:find('[^'..c..']') or #line < 2 then
  37                                     return nil
  38                                   end
  39                                   return (input:find('\n', index) or #input) + 1
  40                                 end))
  41
  42 -- Span elements.
  43 local dq_str = token(l.STRING, l.delimited_range('"', false, true))
  44 local sq_str = token(l.STRING, l.delimited_range("'", false, true))
  45 local paren_str = token(l.STRING, l.delimited_range('()'))
  46 local link = token('link', P('!')^-1 * l.delimited_range('[]') *
  47                            (P('(') * (l.any - S(') \t'))^0 *
  48                             (S(' \t')^1 *
  49                              l.delimited_range('"', false, true))^-1 * ')' +
  50                             S(' \t')^0 * l.delimited_range('[]')) +
  51                            P('http://') * (l.any - l.space)^1)
  52 local link_label = token('link_label', l.delimited_range('[]') * ':') * ws *
  53                    token('link_url', (l.any - l.space)^1) *
  54                    (ws * (dq_str + sq_str + paren_str))^-1
  55
  56 local strong = token('strong', (P('**') * (l.any - '**')^0 * P('**')^-1) +
  57                                (P('__') * (l.any - '__')^0 * P('__')^-1))
  58 local em = token('em',
  59                  l.delimited_range('*', true) + l.delimited_range('_', true))
  60 local code = token('code', (P('``') * (l.any - '``')^0 * P('``')^-1) +
  61                            l.delimited_range('`', true, true))
  62
  63 local escape = token(l.DEFAULT, P('\\') * 1)
  64
  65 local list = token('list',
  66                    l.starts_line(S(' \t')^0 * (S('*+-') + R('09')^1 * '.')) *
  67                    S(' \t'))
  68
  69 M._rules = {
  70   {'header', header},
  71   {'blockquote', blockquote},
  72   {'blockcode', blockcode},
  73   {'hr', hr},
  74   {'list', list},
  75   {'whitespace', ws},
  76   {'link_label', link_label},
  77   {'escape', escape},
  78   {'link', link},
  79   {'strong', strong},
  80   {'em', em},
  81   {'code', code},
  82 }
  83
  84 local font_size = 10
  85 local hstyle = 'fore:$(color.red)'
  86 M._tokenstyles = {
  87   h6 = hstyle,
  88   h5 = hstyle..',size:'..(font_size + 1),
  89   h4 = hstyle..',size:'..(font_size + 2),
  90   h3 = hstyle..',size:'..(font_size + 3),
  91   h2 = hstyle..',size:'..(font_size + 4),
  92   h1 = hstyle..',size:'..(font_size + 5),
  93   code = l.STYLE_EMBEDDED..',eolfilled',
  94   hr = 'back:$(color.black),eolfilled',
  95   link = 'underlined',
  96   link_url = 'underlined',
  97   link_label = l.STYLE_LABEL,
  98   strong = 'bold',
  99   em = 'italics',
 100   list = l.STYLE_CONSTANT,
 101 }
 102
 103 -- Embedded HTML.
 104 local html = l.load('html')
 105 local start_rule = token('tag', l.starts_line(S(' \t')^0 * '<'))
 106 local end_rule = token(l.DEFAULT, P('\n')) -- TODO: l.WHITESPACE causes errors
 107 l.embed_lexer(M, html, start_rule, end_rule)
 108
 109 return M