require 'dentaku/token' require 'dentaku/token_matcher' require 'dentaku/token_scanner' module Dentaku class Tokenizer SCANNERS = [ TokenScanner.new(:whitespace, '\s+'), TokenScanner.new(:numeric, '(\d+(\.\d+)?|\.\d+)', lambda{|raw| raw =~ /\./ ? raw.to_f : raw.to_i }), TokenScanner.new(:string, '"[^"]*"', lambda{|raw| raw.gsub(/^"|"$/, '') }), TokenScanner.new(:string, "'[^']*'", lambda{|raw| raw.gsub(/^'|'$/, '') }), TokenScanner.new(:operator, '\+|-|\*|\/', lambda do |raw| case raw when '+' then :add when '-' then :subtract when '*' then :multiply when '/' then :divide end end), TokenScanner.new(:grouping, '\(|\)|,', lambda do |raw| case raw when '(' then :open when ')' then :close when ',' then :comma end end), TokenScanner.new(:comparator, '<=|>=|!=|<>|<|>|=', lambda do |raw| case raw when '<=' then :le when '>=' then :ge when '!=' then :ne when '<>' then :ne when '<' then :lt when '>' then :gt when '=' then :eq end end), TokenScanner.new(:combinator, '(and|or)\b', lambda {|raw| raw.strip.to_sym }), TokenScanner.new(:function, '(if|round)\b', lambda {|raw| raw.strip.to_sym }), TokenScanner.new(:identifier, '[A-Za-z_]+', lambda {|raw| raw.to_sym }) ] LPAREN = TokenMatcher.new(:grouping, :open) RPAREN = TokenMatcher.new(:grouping, :close) def tokenize(string) nesting = 0 tokens = [] input = string.dup.downcase until input.empty? raise "parse error at: '#{ input }'" unless SCANNERS.any? do |scanner| if token = scanner.scan(input) raise "unexpected zero-width match (:#{ token.category }) at '#{ input }'" if token.length == 0 nesting += 1 if LPAREN == token nesting -= 1 if RPAREN == token raise "too many closing parentheses" if nesting < 0 tokens << token unless token.is?(:whitespace) input.slice!(0, token.length) true else false end end end raise "too many opening parentheses" if nesting > 0 tokens end end end