lexers/yaml.lua

   1 -- Copyright 2006-2016 Mitchell mitchell.att.foicica.com. See LICENSE.
   2 -- YAML LPeg lexer.
   3 -- It does not keep track of indentation perfectly.
   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 = 'yaml'}
  10
  11 -- Whitespace.
  12 local indent = #l.starts_line(S(' \t')) *
  13                (token(l.WHITESPACE, ' ') + token('indent_error', '\t'))^1
  14 local ws = token(l.WHITESPACE, S(' \t')^1 + l.newline^1)
  15
  16 -- Comments.
  17 local comment = token(l.COMMENT, '#' * l.nonnewline^0)
  18
  19 -- Strings.
  20 local string = token(l.STRING, l.delimited_range("'") + l.delimited_range('"'))
  21
  22 -- Numbers.
  23 local integer = l.dec_num + l.hex_num + '0' * S('oO') * R('07')^1
  24 local special_num = '.' * word_match({'inf', 'nan'}, nil, true)
  25 local number = token(l.NUMBER, special_num + l.float + integer)
  26
  27 -- Timestamps.
  28 local ts = token('timestamp', l.digit * l.digit * l.digit * l.digit * -- year
  29                               '-' * l.digit * l.digit^-1 * -- month
  30                               '-' * l.digit * l.digit^-1 * -- day
  31                               ((S(' \t')^1 + S('tT'))^-1 * -- separator
  32                                l.digit * l.digit^-1 * -- hour
  33                                ':' * l.digit * l.digit * -- minute
  34                                ':' * l.digit * l.digit * -- second
  35                                ('.' * l.digit^0)^-1 * -- fraction
  36                                ('Z' + -- timezone
  37                                 S(' \t')^0 * S('-+') * l.digit * l.digit^-1 *
  38                                 (':' * l.digit * l.digit)^-1)^-1)^-1)
  39
  40 -- Constants.
  41 local constant = token(l.CONSTANT,
  42                        word_match({'null', 'true', 'false'}, nil, true))
  43
  44 -- Types.
  45 local type = token(l.TYPE, '!!' * word_match({
  46   -- Collection types.
  47   'map', 'omap', 'pairs', 'set', 'seq',
  48   -- Scalar types.
  49   'binary', 'bool', 'float', 'int', 'merge', 'null', 'str', 'timestamp',
  50   'value', 'yaml'
  51 }, nil, true) + '!' * l.delimited_range('<>'))
  52
  53 -- Document boundaries.
  54 local doc_bounds = token('document', l.starts_line(P('---') + '...'))
  55
  56 -- Directives
  57 local directive = token('directive', l.starts_line('%') * l.nonnewline^1)
  58
  59 local word = (l.alpha + '-' * -l.space) * (l.alnum + '-')^0
  60
  61 -- Keys and literals.
  62 local colon = S(' \t')^0 * ':' * (l.space + -1)
  63 local key = token(l.KEYWORD,
  64                   #word * (l.nonnewline - colon)^1 * #colon *
  65                   P(function(input, index)
  66                     local line = input:sub(1, index - 1):match('[^\r\n]+$')
  67                     return not line:find('[%w-]+:') and index
  68                   end))
  69 local value = #word * (l.nonnewline - l.space^0 * S(',]}'))^1
  70 local block = S('|>') * S('+-')^-1 * (l.newline + -1) * function(input, index)
  71   local rest = input:sub(index)
  72   local level = #rest:match('^( *)')
  73   for pos, indent, line in rest:gmatch('() *()([^\r\n]+)') do
  74     if indent - pos < level and line ~= ' ' or level == 0 and pos > 1 then
  75       return index + pos - 1
  76     end
  77   end
  78   return #input + 1
  79 end
  80 local literal = token('literal', value + block)
  81
  82 -- Indicators.
  83 local anchor = token(l.LABEL, '&' * word)
  84 local alias = token(l.VARIABLE, '*' * word)
  85 local tag = token('tag', '!' * word * P('!')^-1)
  86 local reserved = token(l.ERROR, S('@`') * word)
  87 local indicator_chars = token(l.OPERATOR, S('-?:,[]{}!'))
  88
  89 M._rules = {
  90   {'indent', indent},
  91   {'whitespace', ws},
  92   {'comment', comment},
  93   {'doc_bounds', doc_bounds},
  94   {'key', key},
  95   {'literal', literal},
  96   {'timestamp', ts},
  97   {'number', number},
  98   {'constant', constant},
  99   {'type', type},
 100   {'indicator', tag + indicator_chars + alias + anchor + reserved},
 101   {'directive', directive},
 102 }
 103
 104 M._tokenstyles = {
 105   indent_error = 'back:%(color.red)',
 106   document = l.STYLE_CONSTANT,
 107   literal = l.STYLE_DEFAULT,
 108   timestamp = l.STYLE_NUMBER,
 109   tag = l.STYLE_CLASS,
 110   directive = l.STYLE_PREPROCESSOR,
 111 }
 112
 113 M._FOLDBYINDENTATION = true
 114
 115 return M