lexers/verilog.lua

   1 -- Copyright 2006-2016 Mitchell mitchell.att.foicica.com. See LICENSE.
   2 -- Verilog 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 = 'verilog'}
   9
  10 -- Whitespace.
  11 local ws = token(l.WHITESPACE, l.space^1)
  12
  13 -- Comments.
  14 local line_comment = '//' * l.nonnewline^0
  15 local block_comment = '/*' * (l.any - '*/')^0 * P('*/')^-1
  16 local comment = token(l.COMMENT, line_comment + block_comment)
  17
  18 -- Strings.
  19 local string = token(l.STRING, l.delimited_range('"'))
  20
  21 -- Numbers.
  22 local bin_suffix = S('bB') * S('01_xXzZ')^1
  23 local oct_suffix = S('oO') * S('01234567_xXzZ')^1
  24 local dec_suffix = S('dD') * S('0123456789_xXzZ')^1
  25 local hex_suffix = S('hH') * S('0123456789abcdefABCDEF_xXzZ')^1
  26 local number = token(l.NUMBER, (l.digit + '_')^1 + "'" *
  27                                (bin_suffix + oct_suffix + dec_suffix +
  28                                 hex_suffix))
  29
  30 -- Keywords.
  31 local keyword = token(l.KEYWORD, word_match({
  32   'always', 'assign', 'begin', 'case', 'casex', 'casez', 'default', 'deassign',
  33   'disable', 'else', 'end', 'endcase', 'endfunction', 'endgenerate',
  34   'endmodule', 'endprimitive', 'endspecify', 'endtable', 'endtask', 'for',
  35   'force', 'forever', 'fork', 'function', 'generate', 'if', 'initial', 'join',
  36   'macromodule', 'module', 'negedge', 'posedge', 'primitive', 'repeat',
  37   'release', 'specify', 'table', 'task', 'wait', 'while',
  38   -- Compiler directives.
  39   '`include', '`define', '`undef', '`ifdef', '`ifndef', '`else', '`endif',
  40   '`timescale', '`resetall', '`signed', '`unsigned', '`celldefine',
  41   '`endcelldefine', '`default_nettype', '`unconnected_drive',
  42   '`nounconnected_drive', '`protect', '`endprotect', '`protected',
  43   '`endprotected', '`remove_gatename', '`noremove_gatename', '`remove_netname',
  44   '`noremove_netname', '`expand_vectornets', '`noexpand_vectornets',
  45   '`autoexpand_vectornets',
  46   -- Signal strengths.
  47   'strong0', 'strong1', 'pull0', 'pull1', 'weak0', 'weak1', 'highz0', 'highz1',
  48   'small', 'medium', 'large'
  49 }, '`01'))
  50
  51 -- Function.
  52 local func = token(l.FUNCTION, word_match({
  53   '$stop', '$finish', '$time', '$stime', '$realtime', '$settrace',
  54   '$cleartrace', '$showscopes', '$showvars', '$monitoron', '$monitoroff',
  55   '$random', '$printtimescale', '$timeformat', '$display',
  56   -- Built-in primitives.
  57   'and', 'nand', 'or', 'nor', 'xor', 'xnor', 'buf', 'bufif0', 'bufif1', 'not',
  58   'notif0', 'notif1', 'nmos', 'pmos', 'cmos', 'rnmos', 'rpmos', 'rcmos', 'tran',
  59   'tranif0', 'tranif1', 'rtran', 'rtranif0', 'rtranif1', 'pullup', 'pulldown'
  60 }, '$01'))
  61
  62 -- Types.
  63 local type = token(l.TYPE, word_match({
  64   'integer', 'reg', 'time', 'realtime', 'defparam', 'parameter', 'event',
  65   'wire', 'wand', 'wor', 'tri', 'triand', 'trior', 'tri0', 'tri1', 'trireg',
  66   'vectored', 'scalared', 'input', 'output', 'inout',
  67   'supply0', 'supply1'
  68 }, '01'))
  69
  70 -- Identifiers.
  71 local identifier = token(l.IDENTIFIER, l.word)
  72
  73 -- Operators.
  74 local operator = token(l.OPERATOR, S('=~+-/*<>%&|^~,:;()[]{}'))
  75
  76 M._rules = {
  77   {'whitespace', ws},
  78   {'number', number},
  79   {'keyword', keyword},
  80   {'function', func},
  81   {'type', type},
  82   {'identifier', identifier},
  83   {'string', string},
  84   {'comment', comment},
  85   {'operator', operator},
  86 }
  87
  88 M._foldsymbols = {
  89   _patterns = {'[a-z]+', '[%(%){}]', '/%*', '%*/', '//'},
  90   [l.KEYWORD] = {
  91     case = 1, casex = 1, casez = 1, endcase = -1, ['function'] = 1,
  92     endfunction = -1, fork = 1, join = -1, table = 1, endtable = -1, task = 1,
  93     endtask = -1, generate = 1, endgenerate = -1, specify = 1, endspecify = -1,
  94     primitive = 1, endprimitive = -1, ['module'] = 1, endmodule = -1, begin = 1,
  95     ['end'] = -1
  96   },
  97   [l.OPERATOR] = {['('] = 1, [')'] = -1, ['{'] = 1, ['}'] = -1},
  98   [l.COMMENT] = {['/*'] = 1, ['*/'] = -1, ['//'] = l.fold_line_comments('//')}
  99 }
 100
 101 return M