lexers/lua.lua

   1 -- Copyright 2006-2016 Mitchell mitchell.att.foicica.com. See LICENSE.
   2 -- Lua LPeg lexer.
   3 -- Original written by Peter Odding, 2007/04/04.
   4
   5 local l = require('lexer')
   6 local token, word_match = l.token, l.word_match
   7 local P, R, S = lpeg.P, lpeg.R, lpeg.S
   8
   9 local M = {_NAME = 'lua'}
  10
  11 -- Whitespace.
  12 local ws = token(l.WHITESPACE, l.space^1)
  13
  14 local longstring = lpeg.Cmt('[' * lpeg.C(P('=')^0) * '[',
  15                             function(input, index, eq)
  16                               local _, e = input:find(']'..eq..']', index, true)
  17                               return (e or #input) + 1
  18                             end)
  19
  20 -- Comments.
  21 local line_comment = '--' * l.nonnewline^0
  22 local block_comment = '--' * longstring
  23 local comment = token(l.COMMENT, block_comment + line_comment)
  24
  25 -- Strings.
  26 local sq_str = l.delimited_range("'")
  27 local dq_str = l.delimited_range('"')
  28 local string = token(l.STRING, sq_str + dq_str) +
  29                token('longstring', longstring)
  30
  31 -- Numbers.
  32 local lua_integer = P('-')^-1 * (l.hex_num + l.dec_num)
  33 local number = token(l.NUMBER, l.float + lua_integer)
  34
  35 -- Keywords.
  36 local keyword = token(l.KEYWORD, word_match{
  37   'and', 'break', 'do', 'else', 'elseif', 'end', 'false', 'for', 'function',
  38   'goto', 'if', 'in', 'local', 'nil', 'not', 'or', 'repeat', 'return', 'then',
  39   'true', 'until', 'while'
  40 })
  41
  42 -- Functions.
  43 local func = token(l.FUNCTION, word_match{
  44   'assert', 'collectgarbage', 'dofile', 'error', 'getmetatable', 'ipairs',
  45   'load', 'loadfile', 'next', 'pairs', 'pcall', 'print', 'rawequal', 'rawget',
  46   'rawset', 'require', 'select', 'setmetatable', 'tonumber', 'tostring', 'type',
  47   'xpcall',
  48   -- Added in 5.2.
  49   'rawlen'
  50 })
  51
  52 -- Deprecated functions.
  53 local deprecated_func = token('deprecated_function', word_match{
  54   -- Deprecated in 5.2.
  55   'getfenv', 'loadstring', 'module', 'setfenv', 'unpack'
  56 })
  57
  58 -- Constants.
  59 local constant = token(l.CONSTANT, word_match{
  60   '_G', '_VERSION',
  61   -- Added in 5.2.
  62   '_ENV'
  63 })
  64
  65 -- Libraries.
  66 local library = token('library', word_match({
  67   -- Coroutine.
  68   'coroutine', 'coroutine.create', 'coroutine.resume', 'coroutine.running',
  69   'coroutine.status', 'coroutine.wrap', 'coroutine.yield',
  70   -- Coroutine added in 5.3.
  71   'coroutine.isyieldable',
  72   -- Module.
  73   'package', 'package.cpath', 'package.loaded', 'package.loadlib',
  74   'package.path', 'package.preload',
  75   -- Module added in 5.2.
  76   'package.config', 'package.searchers', 'package.searchpath',
  77   -- UTF-8 added in 5.3.
  78   'utf8', 'utf8.char', 'utf8.charpattern', 'utf8.codepoint', 'utf8.codes',
  79   'utf8.len', 'utf8.offset',
  80   -- String.
  81   'string', 'string.byte', 'string.char', 'string.dump', 'string.find',
  82   'string.format', 'string.gmatch', 'string.gsub', 'string.len', 'string.lower',
  83   'string.match', 'string.rep', 'string.reverse', 'string.sub', 'string.upper',
  84   -- String added in 5.3.
  85   'string.pack', 'string.packsize', 'string.unpack',
  86   -- Table.
  87   'table', 'table.concat', 'table.insert', 'table.remove', 'table.sort',
  88   -- Table added in 5.2.
  89   'table.pack', 'table.unpack',
  90   -- Table added in 5.3.
  91   'table.move',
  92   -- Math.
  93   'math', 'math.abs', 'math.acos', 'math.asin', 'math.atan', 'math.ceil',
  94   'math.cos', 'math.deg', 'math.exp', 'math.floor', 'math.fmod', 'math.huge',
  95   'math.log', 'math.max', 'math.min', 'math.modf', 'math.pi', 'math.rad',
  96   'math.random', 'math.randomseed', 'math.sin', 'math.sqrt', 'math.tan',
  97   -- Math added in 5.3.
  98   'math.maxinteger', 'math.mininteger', 'math.tointeger', 'math.type',
  99   'math.ult',
 100   -- IO.
 101   'io', 'io.close', 'io.flush', 'io.input', 'io.lines', 'io.open', 'io.output',
 102   'io.popen', 'io.read', 'io.stderr', 'io.stdin', 'io.stdout', 'io.tmpfile',
 103   'io.type', 'io.write',
 104   -- OS.
 105   'os', 'os.clock', 'os.date', 'os.difftime', 'os.execute', 'os.exit',
 106   'os.getenv', 'os.remove', 'os.rename', 'os.setlocale', 'os.time',
 107   'os.tmpname',
 108   -- Debug.
 109   'debug', 'debug.debug', 'debug.gethook', 'debug.getinfo', 'debug.getlocal',
 110   'debug.getmetatable', 'debug.getregistry', 'debug.getupvalue',
 111   'debug.sethook', 'debug.setlocal', 'debug.setmetatable', 'debug.setupvalue',
 112   'debug.traceback',
 113   -- Debug added in 5.2.
 114   'debug.getuservalue', 'debug.setuservalue', 'debug.upvalueid',
 115   'debug.upvaluejoin',
 116 }, '.'))
 117
 118 -- Deprecated libraries.
 119 local deprecated_library = token('deprecated_library', word_match({
 120   -- Module deprecated in 5.2.
 121   'package.loaders', 'package.seeall',
 122   -- Table deprecated in 5.2.
 123   'table.maxn',
 124   -- Math deprecated in 5.2.
 125   'math.log10',
 126   -- Math deprecated in 5.3.
 127   'math.atan2', 'math.cosh', 'math.frexp', 'math.ldexp', 'math.pow',
 128   'math.sinh', 'math.tanh',
 129   -- Bit32 deprecated in 5.3.
 130   'bit32', 'bit32.arshift', 'bit32.band', 'bit32.bnot', 'bit32.bor',
 131   'bit32.btest', 'bit32.extract', 'bit32.lrotate', 'bit32.lshift',
 132   'bit32.replace', 'bit32.rrotate', 'bit32.rshift', 'bit32.xor',
 133   -- Debug deprecated in 5.2.
 134   'debug.getfenv', 'debug.setfenv'
 135 }, '.'))
 136
 137 -- Identifiers.
 138 local identifier = token(l.IDENTIFIER, l.word)
 139
 140 -- Labels.
 141 local label = token(l.LABEL, '::' * l.word * '::')
 142
 143 -- Operators.
 144 local operator = token(l.OPERATOR, S('+-*/%^#=<>&|~;:,.{}[]()'))
 145
 146 M._rules = {
 147   {'whitespace', ws},
 148   {'keyword', keyword},
 149   {'function', func + deprecated_func},
 150   {'constant', constant},
 151   {'library', library + deprecated_library},
 152   {'identifier', identifier},
 153   {'string', string},
 154   {'comment', comment},
 155   {'number', number},
 156   {'label', label},
 157   {'operator', operator},
 158 }
 159
 160 M._tokenstyles = {
 161   longstring = l.STYLE_STRING,
 162   deprecated_function = l.STYLE_FUNCTION..',italics',
 163   library = l.STYLE_TYPE,
 164   deprecated_library = l.STYLE_TYPE..',italics'
 165 }
 166
 167 local function fold_longcomment(text, pos, line, s, match)
 168   if match == '[' then
 169     if line:find('^%[=*%[', s) then return 1 end
 170   elseif match == ']' then
 171     if line:find('^%]=*%]', s) then return -1 end
 172   end
 173   return 0
 174 end
 175
 176 M._foldsymbols = {
 177   _patterns = {'%l+', '[%({%)}]', '[%[%]]', '%-%-'},
 178   [l.KEYWORD] = {
 179     ['if'] = 1, ['do'] = 1, ['function'] = 1, ['end'] = -1, ['repeat'] = 1,
 180     ['until'] = -1
 181   },
 182   [l.COMMENT] = {
 183     ['['] = fold_longcomment, [']'] = fold_longcomment,
 184     ['--'] = l.fold_line_comments('--')
 185   },
 186   longstring = {['['] = 1, [']'] = -1},
 187   [l.OPERATOR] = {['('] = 1, ['{'] = 1, [')'] = -1, ['}'] = -1}
 188 }
 189
 190 return M