lex.py

M5,一个功能强大的多处理器系统模拟器.很多针对处理器架构,性能的研究都使用它作为模拟平台

                # If token is processed by a function, call it
                newtok = func(tok)

                # Every function must return a token, if nothing, we just move to next token
                if not newtok:
                    lexpos = self.lexpos        # This is here in case user has updated lexpos.
                    break

                # Verify type of the token.  If not in the token map, raise an error
                if not self.lexoptimize:
                    if not self.lextokens.has_key(newtok.type):
                        raise LexError, ("%s:%d: Rule '%s' returned an unknown token type '%s'" % (
                            func.func_code.co_filename, func.func_code.co_firstlineno,
                            func.__name__, newtok.type),lexdata[lexpos:])

                return newtok
            else:
                # No match, see if in literals
                if lexdata[lexpos] in self.lexliterals:
                    tok = LexToken()
                    tok.value = lexdata[lexpos]
                    tok.lineno = self.lineno
                    tok.lexer = self
                    tok.type = tok.value
                    tok.lexpos = lexpos
                    self.lexpos = lexpos + 1
                    return tok

                # No match. Call t_error() if defined.
                if self.lexerrorf:
                    tok = LexToken()
                    tok.value = self.lexdata[lexpos:]
                    tok.lineno = self.lineno
                    tok.type = "error"
                    tok.lexer = self
                    tok.lexpos = lexpos
                    self.lexpos = lexpos
                    newtok = self.lexerrorf(tok)
                    if lexpos == self.lexpos:
                        # Error method didn't change text position at all. This is an error.
                        raise LexError, ("Scanning error. Illegal character '%s'" % (lexdata[lexpos]), lexdata[lexpos:])
                    lexpos = self.lexpos
                    if not newtok: continue
                    return newtok

                self.lexpos = lexpos
                raise LexError, ("Illegal character '%s' at index %d" % (lexdata[lexpos],lexpos), lexdata[lexpos:])

        self.lexpos = lexpos + 1
        if self.lexdata is None:
             raise RuntimeError, "No input string given with input()"
        return None

# -----------------------------------------------------------------------------
# _validate_file()
#
# This checks to see if there are duplicated t_rulename() functions or strings
# in the parser input file.  This is done using a simple regular expression
# match on each line in the filename.
# -----------------------------------------------------------------------------

def _validate_file(filename):
    import os.path
    base,ext = os.path.splitext(filename)
    if ext != '.py': return 1        # No idea what the file is. Return OK

    try:
        f = open(filename)
        lines = f.readlines()
        f.close()
    except IOError:
        return 1                       # Oh well

    fre = re.compile(r'\s*def\s+(t_[a-zA-Z_0-9]*)\(')
    sre = re.compile(r'\s*(t_[a-zA-Z_0-9]*)\s*=')
    counthash = { }
    linen = 1
    noerror = 1
    for l in lines:
        m = fre.match(l)
        if not m:
            m = sre.match(l)
        if m:
            name = m.group(1)
            prev = counthash.get(name)
            if not prev:
                counthash[name] = linen
            else:
                print >>sys.stderr, "%s:%d: Rule %s redefined. Previously defined on line %d" % (filename,linen,name,prev)
                noerror = 0
        linen += 1
    return noerror

# -----------------------------------------------------------------------------
# _funcs_to_names()
#
# Given a list of regular expression functions, this converts it to a list
# suitable for output to a table file
# -----------------------------------------------------------------------------

def _funcs_to_names(funclist):
    result = []
    for f in funclist:
         if f and f[0]:
             result.append((f[0].__name__,f[1]))
         else:
             result.append(f)
    return result

# -----------------------------------------------------------------------------
# _names_to_funcs()
#
# Given a list of regular expression function names, this converts it back to
# functions.
# -----------------------------------------------------------------------------

def _names_to_funcs(namelist,fdict):
     result = []
     for n in namelist:
          if n and n[0]:
              result.append((fdict[n[0]],n[1]))
          else:
              result.append(n)
     return result

# -----------------------------------------------------------------------------
# _form_master_re()
#
# This function takes a list of all of the regex components and attempts to
# form the master regular expression.  Given limitations in the Python re
# module, it may be necessary to break the master regex into separate expressions.
# -----------------------------------------------------------------------------

def _form_master_re(relist,reflags,ldict,toknames):
    if not relist: return []
    regex = "|".join(relist)
    try:
        lexre = re.compile(regex,re.VERBOSE | reflags)

        # Build the index to function map for the matching engine
        lexindexfunc = [ None ] * (max(lexre.groupindex.values())+1)
        for f,i in lexre.groupindex.items():
            handle = ldict.get(f,None)
            if type(handle) in (types.FunctionType, types.MethodType):
                lexindexfunc[i] = (handle,toknames[handle.__name__])
            elif handle is not None:
                # If rule was specified as a string, we build an anonymous
                # callback function to carry out the action
                if f.find("ignore_") > 0:
                    lexindexfunc[i] = (None,None)
                else:
                    lexindexfunc[i] = (None, toknames[f])

        return [(lexre,lexindexfunc)],[regex]
    except Exception,e:
        m = int(len(relist)/2)
        if m == 0: m = 1
        llist, lre = _form_master_re(relist[:m],reflags,ldict,toknames)
        rlist, rre = _form_master_re(relist[m:],reflags,ldict,toknames)
        return llist+rlist, lre+rre

# -----------------------------------------------------------------------------
# def _statetoken(s,names)
#
# Given a declaration name s of the form "t_" and a dictionary whose keys are
# state names, this function returns a tuple (states,tokenname) where states
# is a tuple of state names and tokenname is the name of the token.  For example,
# calling this with s = "t_foo_bar_SPAM" might return (('foo','bar'),'SPAM')
# -----------------------------------------------------------------------------

def _statetoken(s,names):
    nonstate = 1
    parts = s.split("_")
    for i in range(1,len(parts)):
         if not names.has_key(parts[i]) and parts[i] != 'ANY': break
    if i > 1:
       states = tuple(parts[1:i])
    else:
       states = ('INITIAL',)

    if 'ANY' in states:
       states = tuple(names.keys())

    tokenname = "_".join(parts[i:])
    return (states,tokenname)

# -----------------------------------------------------------------------------
# lex(module)
#
# Build all of the regular expression rules from definitions in the supplied module
# -----------------------------------------------------------------------------
def lex(module=None,object=None,debug=0,optimize=0,lextab="lextab",reflags=0,nowarn=0):
    global lexer
    ldict = None
    stateinfo  = { 'INITIAL' : 'inclusive'}
    error = 0
    files = { }
    lexobj = Lexer()
    lexobj.lexdebug = debug
    lexobj.lexoptimize = optimize
    global token,input

    if nowarn: warn = 0
    else: warn = 1

    if object: module = object

    if module:
        # User supplied a module object.
        if isinstance(module, types.ModuleType):
            ldict = module.__dict__
        elif isinstance(module, _INSTANCETYPE):
            _items = [(k,getattr(module,k)) for k in dir(module)]
            ldict = { }
            for (i,v) in _items:
                ldict[i] = v
        else:
            raise ValueError,"Expected a module or instance"
        lexobj.lexmodule = module

    else:
        # No module given.  We might be able to get information from the caller.
        try:
            raise RuntimeError
        except RuntimeError:
            e,b,t = sys.exc_info()
            f = t.tb_frame
            f = f.f_back           # Walk out to our calling function
            ldict = f.f_globals    # Grab its globals dictionary

    if optimize and lextab:
        try:
            lexobj.readtab(lextab,ldict)
            token = lexobj.token
            input = lexobj.input
            lexer = lexobj
            return lexobj

        except ImportError:
            pass

    # Get the tokens, states, and literals variables (if any)
    if (module and isinstance(module,_INSTANCETYPE)):
        tokens   = getattr(module,"tokens",None)
        states   = getattr(module,"states",None)
        literals = getattr(module,"literals","")
    else:
        tokens   = ldict.get("tokens",None)
        states   = ldict.get("states",None)
        literals = ldict.get("literals","")

    if not tokens:
        raise SyntaxError,"lex: module does not define 'tokens'"
    if not (isinstance(tokens,types.ListType) or isinstance(tokens,types.TupleType)):
        raise SyntaxError,"lex: tokens must be a list or tuple."

    # Build a dictionary of valid token names
    lexobj.lextokens = { }
    if not optimize:
        for n in tokens:
            if not _is_identifier.match(n):
                print >>sys.stderr, "lex: Bad token name '%s'" % n
                error = 1
            if warn and lexobj.lextokens.has_key(n):
                print >>sys.stderr, "lex: Warning. Token '%s' multiply defined." % n
            lexobj.lextokens[n] = None
    else:
        for n in tokens: lexobj.lextokens[n] = None

    if debug:
        print "lex: tokens = '%s'" % lexobj.lextokens.keys()

    try:
         for c in literals:
               if not (isinstance(c,types.StringType) or isinstance(c,types.UnicodeType)) or len(c) > 1:
                    print >>sys.stderr, "lex: Invalid literal %s. Must be a single character" % repr(c)
                    error = 1
                    continue

    except TypeError:
         print >>sys.stderr, "lex: Invalid literals specification. literals must be a sequence of characters."
         error = 1

    lexobj.lexliterals = literals

    # Build statemap
    if states:
         if not (isinstance(states,types.TupleType) or isinstance(states,types.ListType)):
              print >>sys.stderr, "lex: states must be defined as a tuple or list."