lexers/wsf.lua

   1 -- Copyright 2006-2015 Mitchell mitchell.att.foicica.com. See LICENSE.
   2 -- WSF LPeg lexer (based on XML).
   3 -- Contributed by Jeff Stone.
   4
   5 local l = require('lexer')
   6 local token, word_match = l.token, l.word_match
   7 local P, R, S, V = lpeg.P, lpeg.R, lpeg.S, lpeg.V
   8
   9 local M = {_NAME = 'wsf'}
  10
  11 -- Whitespace.
  12 local ws = token(l.WHITESPACE, l.space^1)
  13
  14 -- Comments.
  15 local comment = token(l.COMMENT, '<!--' * (l.any - '-->')^0 * P('-->')^-1)
  16
  17 -- Strings.
  18 local sq_str = l.delimited_range("'", false, true)
  19 local dq_str = l.delimited_range('"', false, true)
  20 local string = #S('\'"') * l.last_char_includes('=') *
  21                token(l.STRING, sq_str + dq_str)
  22
  23 local in_tag = P(function(input, index)
  24   local before = input:sub(1, index - 1)
  25   local s, e = before:find('<[^>]-$'), before:find('>[^<]-$')
  26   if s and e then return s > e and index or nil end
  27   if s then return index end
  28   return input:find('^[^<]->', index) and index or nil
  29 end)
  30
  31 -- Numbers.
  32 local number = #l.digit * l.last_char_includes('=') *
  33                token(l.NUMBER, l.digit^1 * P('%')^-1) * in_tag
  34
  35 local alpha = R('az', 'AZ', '\127\255')
  36 local word_char = l.alnum + S('_-:.??')
  37 local identifier = (l.alpha + S('_-:.??')) * word_char^0
  38
  39 -- Elements.
  40 local element = token('element', '<' * P('/')^-1 * identifier)
  41
  42 -- Attributes.
  43 local attribute = token('attribute', identifier) * #(l.space^0 * '=')
  44
  45 -- Closing tags.
  46 local tag_close = token('element', P('/')^-1 * '>')
  47
  48 -- Equals.
  49 local equals = token(l.OPERATOR, '=') * in_tag
  50
  51 -- Entities.
  52 local entity = token('entity', '&' * word_match{
  53   'lt', 'gt', 'amp', 'apos', 'quot'
  54 } * ';')
  55
  56 M._rules = {
  57   {'whitespace', ws},
  58   {'comment', comment},
  59   {'element', element},
  60   {'tag_close', tag_close},
  61   {'attribute', attribute},
  62   {'equals', equals},
  63   {'string', string},
  64   {'number', number},
  65   {'entity', entity}
  66 }
  67
  68 M._tokenstyles = {
  69   element = l.STYLE_KEYWORD,
  70   attribute = l.STYLE_TYPE,
  71   entity = l.STYLE_OPERATOR
  72 }
  73
  74 M._foldsymbols = {
  75   _patterns = {'</?', '/>', '<!%-%-', '%-%->'},
  76   element = {['<'] = 1, ['/>'] = -1, ['</'] = -1},
  77   [l.COMMENT] = {['<!--'] = 1, ['-->'] = -1},
  78 }
  79
  80 -- Finally, add JavaScript and VBScript as embedded languages
  81
  82 -- Tags that start embedded languages.
  83 M.embed_start_tag = element *
  84                     (ws^1 * attribute * ws^0 * equals * ws^0 * string)^0 *
  85                     ws^0 * tag_close
  86 M.embed_end_tag = element * tag_close
  87
  88 -- Embedded JavaScript.
  89 local js = l.load('javascript')
  90 local js_start_rule = #(P('<script') * (P(function(input, index)
  91   if input:find('^%s+language%s*=%s*(["\'])[jJ][ava]*[sS]cript%1', index) then
  92     return index
  93   end
  94 end) + '>')) * M.embed_start_tag -- <script language="javascript">
  95 local js_end_rule = #('</script' * ws^0 * '>') * M.embed_end_tag -- </script>
  96 l.embed_lexer(M, js, js_start_rule, js_end_rule)
  97
  98 -- Embedded VBScript.
  99 local vbs = l.load('vbscript')
 100 local vbs_start_rule = #(P('<script') * (P(function(input, index)
 101   if input:find('^%s+language%s*=%s*(["\'])[vV][bB][sS]cript%1', index) then
 102     return index
 103   end
 104 end) + '>')) * M.embed_start_tag -- <script language="vbscript">
 105 local vbs_end_rule = #('</script' * ws^0 * '>') * M.embed_end_tag -- </script>
 106 l.embed_lexer(M, vbs, vbs_start_rule, vbs_end_rule)
 107
 108 return M