写経 p87

いろいろ間違えた。

import StringIO



class Token:
  def __init__(self, raw):
    self.raw = raw

  def __eq__(self, x):
    assert isinstance(x, Token)
    return self.raw == x.raw


class Num(Token):
  def __init__(self, raw):
    Token.__init__(self, raw)
    self.value = int(raw)

class Identifier(Token):
  pass

class Div(Token):
  pass

class Mod(Token):
  pass

class Done(Token):
  pass



buf = None
lineno = None
lookahead = None
symtable = dict()


def lexan():
  while True:
    t = buf.read(1)
    if t.isspace():
      pass #skip white spaces
    elif t == '\n':
      lineno = lineno + 1
    elif t.isdigit():
      tokenval = None
      tokenval = int(t)
      t = buf.read(1)
      while t.isdigit():
        tokenval = tokenval * 10 + int(t)
        t = buf.read(1)
      buf.seek(-1, 1)
      return Num(tokenval)
    elif t.isalpha():
      lexbuf = ''
      while t.isalpha() or t.isdigit():
        lexbuf += t
        t = buf.read(1)
        # no size error...
      # lexbuf += EOS python string do not need explicit \0
      if t != '':
        buf.seek(-1, 1)

      # it is better idea to tweek constructor of Identity as a flyweight pattern?
      try:
        i = symtable[lexbuf]
      except KeyError:
        i = Identifier(lexbuf)
        symtable[lexbuf] = i
      return i

    elif t == '': #EOF
      return Done(t)

    else:
      # tokenval = None
      return Token(t) #, tokenval)


def parse():
  global lookahead
  lookahead = lexan()
  while not isinstance(lookahead, Done):
    expr()
    match(';')

def expr():
  global lookahead
  t = None
  term()
  while True:
    if lookahead.raw in ('+', '-'):
      t = lookahead
      match(lookahead)
      term()
      emit(t)
      continue
    else:
      return

def term():
  global lookahead
  factor()
  while True:
    if lookahead.raw in ('*', '/') or isinstance(lookahead, (Div, Mod)):
      t = lookahead
      match(lookahead)
      factor()
      emit(t)
      continue
    else:
      return


def factor():
  global lookahead
  if lookahead.raw == '(':
    match('(')
    expr()
    match(')')
  elif isinstance(lookahead, Num):
    emit(lookahead)
    match(lookahead)
  elif isinstance(lookahead, Identifier):
    emit(lookahead)
    match(lookahead) #Identifier)
  else:
    error("syntax error")
def match(t):
  global lookahead
  if isinstance(t, Token):
    if lookahead == t:
      lookahead = lexan()
    else:
      error("syntax error")
  elif isinstance(t, str):
    if lookahead.raw == t:
      lookahead = lexan()
    else:
      error('lookahead is %s, match arg is %s'%(lookahead.raw,  t))
  else:
    error("syntax error")

def error(msg):
  raise msg + ' at %d'%(lineno)


def emit(t):
  if False:
    pass
  elif isinstance(t, Div):
    print 'Div'
  elif isinstance(t, Mod):
    print 'Mod'
  elif isinstance(t, Num):
    print '%d'%(t.value)
  elif isinstance(t, Identifier):
    print t.raw
  else:
    print "token %s"%(t.raw)



def init():
  global symtable
  symtable['div'] = Div('div')
  symtable['mod'] = Mod('mod')


buf = StringIO.StringIO('2+3 * 5;\n 12 div 7 mod 2;')
lineno = 1

init()
parse()