lexers/lisp.lua

   1 -- Copyright 2006-2016 Mitchell mitchell.att.foicica.com. See LICENSE.
   2 -- Lisp 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 = 'lisp'}
   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 local word = l.alpha * (l.alnum + '_' + '-')^0
  19
  20 -- Strings.
  21 local literal = "'" * word
  22 local dq_str = l.delimited_range('"')
  23 local string = token(l.STRING, literal + dq_str)
  24
  25 -- Numbers.
  26 local number = token(l.NUMBER, P('-')^-1 * l.digit^1 * (S('./') * l.digit^1)^-1)
  27
  28 -- Keywords.
  29 local keyword = token(l.KEYWORD, word_match({
  30   'defclass', 'defconstant', 'defgeneric', 'define-compiler-macro',
  31   'define-condition', 'define-method-combination', 'define-modify-macro',
  32   'define-setf-expander', 'define-symbol-macro', 'defmacro', 'defmethod',
  33   'defpackage', 'defparameter', 'defsetf', 'defstruct', 'deftype', 'defun',
  34   'defvar',
  35   'abort', 'assert', 'block', 'break', 'case', 'catch', 'ccase', 'cerror',
  36   'cond', 'ctypecase', 'declaim', 'declare', 'do', 'do*', 'do-all-symbols',
  37   'do-external-symbols', 'do-symbols', 'dolist', 'dotimes', 'ecase', 'error',
  38   'etypecase', 'eval-when', 'flet', 'handler-bind', 'handler-case', 'if',
  39   'ignore-errors', 'in-package', 'labels', 'lambda', 'let', 'let*', 'locally',
  40   'loop', 'macrolet', 'multiple-value-bind', 'proclaim', 'prog', 'prog*',
  41   'prog1', 'prog2', 'progn', 'progv', 'provide', 'require', 'restart-bind',
  42   'restart-case', 'restart-name', 'return', 'return-from', 'signal',
  43   'symbol-macrolet', 'tagbody', 'the', 'throw', 'typecase', 'unless',
  44   'unwind-protect', 'when', 'with-accessors', 'with-compilation-unit',
  45   'with-condition-restarts', 'with-hash-table-iterator',
  46   'with-input-from-string', 'with-open-file', 'with-open-stream',
  47   'with-output-to-string', 'with-package-iterator', 'with-simple-restart',
  48   'with-slots', 'with-standard-io-syntax',
  49   't', 'nil'
  50 }, '-'))
  51
  52 -- Identifiers.
  53 local identifier = token(l.IDENTIFIER, word)
  54
  55 -- Operators.
  56 local operator = token(l.OPERATOR, S('<>=*/+-`@%()'))
  57
  58 -- Entities.
  59 local entity = token('entity', '&' * word)
  60
  61 M._rules = {
  62   {'whitespace', ws},
  63   {'keyword', keyword},
  64   {'identifier', identifier},
  65   {'string', string},
  66   {'comment', comment},
  67   {'number', number},
  68   {'operator', operator},
  69   {'entity', entity},
  70 }
  71
  72 M._tokenstyles = {
  73   entity = l.STYLE_VARIABLE
  74 }
  75
  76 M._foldsymbols = {
  77   _patterns = {'[%(%)%[%]{}]', '#|', '|#', ';'},
  78   [l.OPERATOR] = {
  79     ['('] = 1, [')'] = -1, ['['] = 1, [']'] = -1, ['{'] = 1, ['}'] = -1
  80   },
  81   [l.COMMENT] = {['#|'] = 1, ['|#'] = -1, [';'] = l.fold_line_comments(';')}
  82 }
  83
  84 return M