vis

a vi-like editor based on Plan 9's structural regular expressions
wsf.lua

(3123B)
      1 -- Copyright 2006-2025 Mitchell. See LICENSE.
      2 -- WSF LPeg lexer (based on XML).
      3 -- Contributed by Jeff Stone.
      4 
      5 local lexer = lexer
      6 local P, S = lpeg.P, lpeg.S
      7 
      8 local lex = lexer.new(...)
      9 
     10 -- Comments.
     11 lex:add_rule('comment', lex:tag(lexer.COMMENT, lexer.range('<!--', '-->')))
     12 
     13 -- Elements.
     14 local identifier = (lexer.alpha + S('_-')) * (lexer.alnum + S('_-'))^0
     15 local tag = lex:tag(lexer.TAG, '<' * P('/')^-1 * identifier)
     16 lex:add_rule('tag', tag)
     17 
     18 -- Closing tags.
     19 local tag_close = lex:tag(lexer.TAG, P('/')^-1 * '>')
     20 lex:add_rule('tag_close', tag_close)
     21 
     22 -- Equals.
     23 -- TODO: performance is terrible on large files.
     24 local in_tag = P(function(input, index)
     25 	local before = input:sub(1, index - 1)
     26 	local s, e = before:find('<[^>]-$'), before:find('>[^<]-$')
     27 	if s and e then return s > e end
     28 	if s then return true end
     29 	return input:find('^[^<]->', index) ~= nil
     30 end)
     31 
     32 local equals = lex:tag(lexer.OPERATOR, '=') -- * in_tag
     33 -- lex:add_rule('equals', equals)
     34 
     35 -- Attributes.
     36 local ws = lex:get_rule('whitespace')
     37 local attribute_eq = lex:tag(lexer.ATTRIBUTE, identifier) * ws^-1 * equals
     38 lex:add_rule('attribute', attribute_eq)
     39 
     40 -- Strings.
     41 local sq_str = lexer.range("'", false, false)
     42 local dq_str = lexer.range('"', false, false)
     43 local string = lex:tag(lexer.STRING, lexer.after_set('=', sq_str + dq_str))
     44 lex:add_rule('string', string)
     45 
     46 -- Numbers.
     47 local number = lex:tag(lexer.NUMBER, lexer.dec_num * P('%')^-1)
     48 lex:add_rule('number', lexer.after_set('=', number)) -- * in_tag)
     49 
     50 -- Entities.
     51 local predefined = lex:tag(lexer.CONSTANT_BUILTIN .. '.entity',
     52 	'&' * lexer.word_match('lt gt amp apos quot') * ';')
     53 local general = lex:tag(lexer.CONSTANT .. '.entity', '&' * identifier * ';')
     54 lex:add_rule('entity', predefined + general)
     55 
     56 -- Fold points.
     57 local function disambiguate_lt(text, pos, line, s) return not line:find('^</', s) and 1 or -1 end
     58 lex:add_fold_point(lexer.TAG, '<', disambiguate_lt)
     59 lex:add_fold_point(lexer.TAG, '/>', -1)
     60 lex:add_fold_point(lexer.COMMENT, '<!--', '-->')
     61 
     62 -- Finally, add JavaScript and VBScript as embedded languages
     63 
     64 -- Tags that start embedded languages.
     65 local embed_start_tag = tag * (ws * attribute_eq * ws^-1 * string)^0 * ws^-1 * tag_close
     66 local embed_end_tag = tag * tag_close
     67 
     68 -- Embedded JavaScript.
     69 local js = lexer.load('javascript')
     70 local js_start_rule = #(P('<script') * (P(function(input, index)
     71 	if input:find('^%s+language%s*=%s*(["\'])[jJ][ava]*[sS]cript%1', index) then return true end
     72 end) + '>')) * embed_start_tag -- <script language="javascript">
     73 local js_end_rule = #P('</script>') * embed_end_tag -- </script>
     74 lex:embed(js, js_start_rule, js_end_rule)
     75 
     76 -- Embedded VBScript.
     77 local vbs = lexer.load('vb', 'vbscript')
     78 local vbs_start_rule = #(P('<script') * (P(function(input, index)
     79 	if input:find('^%s+language%s*=%s*(["\'])[vV][bB][sS]cript%1', index) then return true end
     80 end) + '>')) * embed_start_tag -- <script language="vbscript">
     81 local vbs_end_rule = #P('</script>') * embed_end_tag -- </script>
     82 lex:embed(vbs, vbs_start_rule, vbs_end_rule)
     83 
     84 lexer.property['scintillua.comment'] = '<!--|-->'
     85 lexer.property['scintillua.angle.braces'] = '1'
     86 
     87 return lex