lexers/dmd.lua

   1 -- Copyright 2006-2016 Mitchell mitchell.att.foicica.com. See LICENSE.
   2 -- D LPeg lexer.
   3 -- Heavily modified by Brian Schott (@Hackerpilot on Github).
   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 = 'dmd'}
  10
  11 -- Whitespace.
  12 local ws = token(l.WHITESPACE, l.space^1)
  13
  14 -- Comments.
  15 local line_comment = '//' * l.nonnewline_esc^0
  16 local block_comment = '/*' * (l.any - '*/')^0 * P('*/')^-1
  17 local nested_comment = l.nested_pair('/+', '+/')
  18 local comment = token(l.COMMENT, line_comment + block_comment + nested_comment)
  19
  20 -- Strings.
  21 local sq_str = l.delimited_range("'", true) * S('cwd')^-1
  22 local dq_str = l.delimited_range('"') * S('cwd')^-1
  23 local lit_str = 'r' * l.delimited_range('"', false, true) * S('cwd')^-1
  24 local bt_str = l.delimited_range('`', false, true) * S('cwd')^-1
  25 local hex_str = 'x' * l.delimited_range('"') * S('cwd')^-1
  26 local other_hex_str = '\\x' * (l.xdigit * l.xdigit)^1
  27 local del_str = l.nested_pair('q"[', ']"') * S('cwd')^-1 +
  28                 l.nested_pair('q"(', ')"') * S('cwd')^-1 +
  29                 l.nested_pair('q"{', '}"') * S('cwd')^-1 +
  30                 l.nested_pair('q"<', '>"') * S('cwd')^-1 +
  31                 P('q') * l.nested_pair('{', '}') * S('cwd')^-1
  32 local string = token(l.STRING, del_str + sq_str + dq_str + lit_str + bt_str +
  33                                hex_str + other_hex_str)
  34
  35 -- Numbers.
  36 local dec = l.digit^1 * ('_' * l.digit^1)^0
  37 local hex_num = l.hex_num * ('_' * l.xdigit^1)^0
  38 local bin_num = '0' * S('bB') * S('01_')^1
  39 local oct_num = '0' * S('01234567_')^1
  40 local integer = S('+-')^-1 * (hex_num + oct_num + bin_num + dec)
  41 local number = token(l.NUMBER, (l.float + integer) * S('uUlLdDfFi')^-1)
  42
  43 -- Keywords.
  44 local keyword = token(l.KEYWORD, word_match{
  45   'abstract', 'align', 'asm', 'assert', 'auto', 'body', 'break', 'case', 'cast',
  46   'catch', 'const', 'continue', 'debug', 'default', 'delete',
  47   'deprecated', 'do', 'else', 'extern', 'export', 'false', 'final', 'finally',
  48   'for', 'foreach', 'foreach_reverse', 'goto', 'if', 'import', 'immutable',
  49   'in', 'inout', 'invariant', 'is', 'lazy', 'macro', 'mixin', 'new', 'nothrow',
  50   'null', 'out', 'override', 'pragma', 'private', 'protected', 'public', 'pure',
  51   'ref', 'return', 'scope', 'shared', 'static', 'super', 'switch',
  52   'synchronized', 'this', 'throw','true', 'try', 'typeid', 'typeof', 'unittest',
  53   'version', 'virtual', 'volatile', 'while', 'with', '__gshared', '__thread',
  54   '__traits', '__vector', '__parameters'
  55 })
  56
  57 -- Types.
  58 local type = token(l.TYPE, word_match{
  59   'alias', 'bool', 'byte', 'cdouble', 'cent', 'cfloat', 'char', 'class',
  60   'creal', 'dchar', 'delegate', 'double', 'enum', 'float', 'function',
  61   'idouble', 'ifloat', 'int', 'interface', 'ireal', 'long', 'module', 'package',
  62   'ptrdiff_t', 'real', 'short', 'size_t', 'struct', 'template', 'typedef',
  63   'ubyte', 'ucent', 'uint', 'ulong', 'union', 'ushort', 'void', 'wchar',
  64   'string', 'wstring', 'dstring', 'hash_t', 'equals_t'
  65 })
  66
  67 -- Constants.
  68 local constant = token(l.CONSTANT, word_match{
  69   '__FILE__', '__LINE__', '__DATE__', '__EOF__', '__TIME__', '__TIMESTAMP__',
  70   '__VENDOR__', '__VERSION__', '__FUNCTION__', '__PRETTY_FUNCTION__',
  71   '__MODULE__',
  72 })
  73
  74 local class_sequence = token(l.TYPE, P('class') + P('struct')) * ws^1 *
  75                                      token(l.CLASS, l.word)
  76
  77 -- Identifiers.
  78 local identifier = token(l.IDENTIFIER, l.word)
  79
  80 -- Operators.
  81 local operator = token(l.OPERATOR, S('?=!<>+-*$/%&|^~.,;()[]{}'))
  82
  83 -- Properties.
  84 local properties = (type + identifier + operator) * token(l.OPERATOR, '.') *
  85   token(l.VARIABLE, word_match{
  86     'alignof', 'dig', 'dup', 'epsilon', 'idup', 'im', 'init', 'infinity',
  87     'keys', 'length', 'mangleof', 'mant_dig', 'max', 'max_10_exp', 'max_exp',
  88     'min', 'min_normal', 'min_10_exp', 'min_exp', 'nan', 'offsetof', 'ptr',
  89     're', 'rehash', 'reverse', 'sizeof', 'sort', 'stringof', 'tupleof',
  90     'values'
  91   })
  92
  93 -- Preprocs.
  94 local annotation = token('annotation', '@' * l.word^1)
  95 local preproc = token(l.PREPROCESSOR, '#' * l.nonnewline^0)
  96
  97 -- Traits.
  98 local traits_list = token('traits', word_match{
  99   'allMembers', 'classInstanceSize', 'compiles', 'derivedMembers',
 100   'getAttributes', 'getMember', 'getOverloads', 'getProtection', 'getUnitTests',
 101   'getVirtualFunctions', 'getVirtualIndex', 'getVirtualMethods', 'hasMember',
 102   'identifier', 'isAbstractClass', 'isAbstractFunction', 'isArithmetic',
 103   'isAssociativeArray', 'isFinalClass', 'isFinalFunction', 'isFloating',
 104   'isIntegral', 'isLazy', 'isNested', 'isOut', 'isOverrideFunction', 'isPOD',
 105   'isRef', 'isSame', 'isScalar', 'isStaticArray', 'isStaticFunction',
 106   'isUnsigned', 'isVirtualFunction', 'isVirtualMethod', 'parent'
 107 })
 108
 109 local scopes_list = token('scopes', word_match{'exit', 'success', 'failure'})
 110
 111 -- versions
 112 local versions_list = token('versions', word_match{
 113   'AArch64', 'AIX', 'all', 'Alpha', 'Alpha_HardFloat', 'Alpha_SoftFloat',
 114   'Android', 'ARM', 'ARM_HardFloat', 'ARM_SoftFloat', 'ARM_SoftFP', 'ARM_Thumb',
 115   'assert', 'BigEndian', 'BSD', 'Cygwin', 'D_Coverage', 'D_Ddoc', 'D_HardFloat',
 116   'DigitalMars', 'D_InlineAsm_X86', 'D_InlineAsm_X86_64', 'D_LP64',
 117   'D_NoBoundsChecks', 'D_PIC', 'DragonFlyBSD', 'D_SIMD', 'D_SoftFloat',
 118   'D_Version2', 'D_X32', 'FreeBSD', 'GNU', 'Haiku', 'HPPA', 'HPPA64', 'Hurd',
 119   'IA64', 'LDC', 'linux', 'LittleEndian', 'MIPS32', 'MIPS64', 'MIPS_EABI',
 120   'MIPS_HardFloat', 'MIPS_N32', 'MIPS_N64', 'MIPS_O32', 'MIPS_O64',
 121   'MIPS_SoftFloat', 'NetBSD', 'none', 'OpenBSD', 'OSX', 'Posix', 'PPC', 'PPC64',
 122   'PPC_HardFloat', 'PPC_SoftFloat', 'S390', 'S390X', 'SDC', 'SH', 'SH64',
 123   'SkyOS', 'Solaris', 'SPARC', 'SPARC64', 'SPARC_HardFloat', 'SPARC_SoftFloat',
 124   'SPARC_V8Plus', 'SysV3', 'SysV4', 'unittest', 'Win32', 'Win64', 'Windows',
 125   'X86', 'X86_64'
 126 })
 127
 128 local versions = token(l.KEYWORD, 'version') * l.space^0 *
 129                  token(l.OPERATOR, '(') * l.space^0 * versions_list
 130
 131 local scopes = token(l.KEYWORD, 'scope') * l.space^0 *
 132                token(l.OPERATOR, '(') * l.space^0 * scopes_list
 133
 134 local traits = token(l.KEYWORD, '__traits') * l.space^0 *
 135                token(l.OPERATOR, '(') * l.space^0 * traits_list
 136
 137 local func = token(l.FUNCTION, l.word) *
 138              #(l.space^0 * (P('!') * l.word^-1 * l.space^-1)^-1 * P('('))
 139
 140 M._rules = {
 141   {'whitespace', ws},
 142   {'class', class_sequence},
 143   {'traits', traits},
 144   {'versions', versions},
 145   {'scopes', scopes},
 146   {'keyword', keyword},
 147   {'variable', properties},
 148   {'type', type},
 149   {'function', func},
 150   {'constant', constant},
 151   {'string', string},
 152   {'identifier', identifier},
 153   {'comment', comment},
 154   {'number', number},
 155   {'preproc', preproc},
 156   {'operator', operator},
 157   {'annotation', annotation},
 158 }
 159
 160 M._tokenstyles = {
 161   annotation = l.STYLE_PREPROCESSOR,
 162   traits = l.STYLE_CLASS,
 163   versions = l.STYLE_CONSTANT,
 164   scopes = l.STYLE_CONSTANT
 165 }
 166
 167 M._foldsymbols = {
 168   _patterns = {'[{}]', '/[*+]', '[*+]/', '//'},
 169   [l.OPERATOR] = {['{'] = 1, ['}'] = -1},
 170   [l.COMMENT] = {
 171     ['/*'] = 1, ['*/'] = -1, ['/+'] = 1, ['+/'] = -1,
 172     ['//'] = l.fold_line_comments('//')
 173   }
 174 }
 175
 176 return M