lexers/php.lua

   1 -- Copyright 2006-2016 Mitchell mitchell.att.foicica.com. See LICENSE.
   2 -- PHP LPeg lexer.
   3
   4 local l = require('lexer')
   5 local token, word_match = l.token, l.word_match
   6 local P, R, S, V = lpeg.P, lpeg.R, lpeg.S, lpeg.V
   7
   8 local M = {_NAME = 'php'}
   9
  10 -- Whitespace.
  11 local ws = token(l.WHITESPACE, l.space^1)
  12
  13 -- Comments.
  14 local line_comment = (P('//') + '#') * l.nonnewline^0
  15 local block_comment = '/*' * (l.any - '*/')^0 * P('*/')^-1
  16 local comment = token(l.COMMENT, block_comment + line_comment)
  17
  18 -- Strings.
  19 local sq_str = l.delimited_range("'")
  20 local dq_str = l.delimited_range('"')
  21 local bt_str = l.delimited_range('`')
  22 local heredoc = '<<<' * P(function(input, index)
  23   local _, e, delimiter = input:find('([%a_][%w_]*)[\n\r\f]+', index)
  24   if delimiter then
  25     local _, e = input:find('[\n\r\f]+'..delimiter, e)
  26     return e and e + 1
  27   end
  28 end)
  29 local string = token(l.STRING, sq_str + dq_str + bt_str + heredoc)
  30 -- TODO: interpolated code.
  31
  32 -- Numbers.
  33 local number = token(l.NUMBER, l.float + l.integer)
  34
  35 -- Keywords.
  36 local keyword = token(l.KEYWORD, word_match{
  37   'and', 'array', 'as', 'bool', 'boolean', 'break', 'case',
  38   'cfunction', 'class', 'const', 'continue', 'declare', 'default',
  39   'die', 'directory', 'do', 'double', 'echo', 'else', 'elseif',
  40   'empty', 'enddeclare', 'endfor', 'endforeach', 'endif',
  41   'endswitch', 'endwhile', 'eval', 'exit', 'extends', 'false',
  42   'float', 'for', 'foreach', 'function', 'global', 'if', 'include',
  43   'include_once', 'int', 'integer', 'isset', 'list', 'new', 'null',
  44   'object', 'old_function', 'or', 'parent', 'print', 'real',
  45   'require', 'require_once', 'resource', 'return', 'static',
  46   'stdclass', 'string', 'switch', 'true', 'unset', 'use', 'var',
  47   'while', 'xor', '__class__', '__file__', '__function__',
  48   '__line__', '__sleep', '__wakeup'
  49 })
  50
  51 -- Variables.
  52 local word = (l.alpha + '_' + R('\127\255')) * (l.alnum + '_' + R('\127\255'))^0
  53 local variable = token(l.VARIABLE, '$' * word)
  54
  55 -- Identifiers.
  56 local identifier = token(l.IDENTIFIER, word)
  57
  58 -- Operators.
  59 local operator = token(l.OPERATOR, S('!@%^*&()-+=|/.,;:<>[]{}') + '?' * -P('>'))
  60
  61 M._rules = {
  62   {'whitespace', ws},
  63   {'keyword', keyword},
  64   {'identifier', identifier},
  65   {'string', string},
  66   {'variable', variable},
  67   {'comment', comment},
  68   {'number', number},
  69   {'operator', operator},
  70 }
  71
  72 -- Embedded in HTML.
  73 local html = l.load('html')
  74
  75 -- Embedded PHP.
  76 local php_start_rule = token('php_tag', '<?' * ('php' * l.space)^-1)
  77 local php_end_rule = token('php_tag', '?>')
  78 l.embed_lexer(html, M, php_start_rule, php_end_rule)
  79
  80 M._tokenstyles = {
  81   php_tag = l.STYLE_EMBEDDED
  82 }
  83
  84 local _foldsymbols = html._foldsymbols
  85 _foldsymbols._patterns[#_foldsymbols._patterns + 1] = '<%?'
  86 _foldsymbols._patterns[#_foldsymbols._patterns + 1] = '%?>'
  87 _foldsymbols._patterns[#_foldsymbols._patterns + 1] = '/%*'
  88 _foldsymbols._patterns[#_foldsymbols._patterns + 1] = '%*/'
  89 _foldsymbols._patterns[#_foldsymbols._patterns + 1] = '//'
  90 _foldsymbols._patterns[#_foldsymbols._patterns + 1] = '#'
  91 _foldsymbols._patterns[#_foldsymbols._patterns + 1] = '[{}()]'
  92 _foldsymbols.php_tag = {['<?'] = 1, ['?>'] = -1}
  93 _foldsymbols[l.COMMENT]['/*'], _foldsymbols[l.COMMENT]['*/'] = 1, -1
  94 _foldsymbols[l.COMMENT]['//'] = l.fold_line_comments('//')
  95 _foldsymbols[l.COMMENT]['#'] = l.fold_line_comments('#')
  96 _foldsymbols[l.OPERATOR] = {['{'] = 1, ['}'] = -1, ['('] = 1, [')'] = -1}
  97 M._foldsymbols = _foldsymbols
  98
  99 return M