Class Tokenizer

tools/test.py:1106–1171

A simple string tokenizer that chops expressions into variables, parens and operators

class Tokenizer(object):
  """A simple string tokenizer that chops expressions into variables,
  parens and operators"""

  def __init__(self, expr):
    self.index = 0
    self.expr = expr
    self.length = len(expr)
    self.tokens = None

  def Current(self, length = 1):
    if not self.HasMore(length): return ""
    return self.expr[self.index:self.index+length]

  def HasMore(self, length = 1):
    return self.index < self.length + (length - 1)

  def Advance(self, count = 1):
    self.index = self.index + count

  def AddToken(self, token):
    self.tokens.append(token)

  def SkipSpaces(self):
    while self.HasMore() and self.Current().isspace():
      self.Advance()

  def Tokenize(self):
    self.tokens = [ ]
    while self.HasMore():
      self.SkipSpaces()
      if not self.HasMore():
        return None
      if self.Current() == '(':
        self.AddToken('(')
        self.Advance()
      elif self.Current() == ')':
        self.AddToken(')')
        self.Advance()
      elif self.Current() == '$':
        self.AddToken('$')
        self.Advance()
      elif self.Current() == ',':
        self.AddToken(',')
        self.Advance()
      elif IsAlpha(self.Current()):
        buf = ""
        while self.HasMore() and IsAlpha(self.Current()):
          buf += self.Current()
          self.Advance()
        self.AddToken(buf)
      elif self.Current(2) == '&&':
        self.AddToken('&&')
        self.Advance(2)
      elif self.Current(2) == '||':
        self.AddToken('||')
        self.Advance(2)
      elif self.Current(2) == '==':