fooscanner.cpp

用于词法分析的词法分析器

/*
  $Id: FooScanner.cpp,v 1.5 1997/02/02 01:31:01 matt Exp $

  FOO scanner class.
  
  (c) Feb 96 Matt Phillips
  */

#include <ctype.h>
#include <util/checks.h>

#include "FooScanner.h"

void FooScanner::getIdent ()
{
  while (isalnum (input.peek ()) || input.peek () == '_')
  {
    input.get ();
  }

  symbol = SymIdent;
}

void FooScanner::handleBackslash ()
{
  input.skip ();
  char chr = input.peek ();

  // handle \xx format
  if (isxdigit (chr))
  {
    int value = 0;
    
    // read up to 2 hex digits
    for (int count = 2;
	 count > 0 && !input.eof () && isxdigit (input.peek ());
	 count--, input.skip ())
    {
      value *= 16;
      if (isdigit (input.peek ()))
	value += input.peek () - '0';
      else
	value += tolower (input.peek ()) - 'a' + 10;
    }
    chr = (char)value;
  } else
    input.skip ();

  input.pushBuffer (chr);
}

void FooScanner::getString ()
{
  input.clearBuffer ();

  while (input.peek () != '"' && !input.eof ())
  {
    if (input.peek () == '\\')
      handleBackslash ();
    else
      input.get ();
  }

  input.skip ();		// lose terminating "

  if (input.eof ())
    warning ("unterminated string");

  symbol = SymString;
}

void FooScanner::getNext ()
{
  input.clearBuffer ();

  // skip whitespace/comments
  for (;;)
  {
    if (isspace (input.peek ()))
    {
      input.skip ();
    } else if (input.peek () == '-')
    {
      while (!input.inState (input.getState (),
			     InputStream::stEndLine))
      {
	input.skip ();
      }
      input.skip ();
    } else
      break;
  }

  if (!input.eof ())
  {
    symLine = input.getLine ();
    symCol = input.getColumn ();

    if (isalnum (input.peek ()) || input.peek () == '_')
    {
      getIdent ();
    } else
    {
      int chr = input.peek ();
      input.get ();
	 
      switch (chr)
      {
      case '"':
	getString ();
	break;
      case '(':
	symbol = SymLBracket;
	break;
      case ')':
	symbol = SymRBracket;
	break;
      case '[':
	symbol = SymLSBracket;
	break;
      case ']':
	symbol = SymRSBracket;
	break;
      default:
	input.get ();
	error ("illegal symbol");
      }
    }
  } else
  {
    symbol = SymEOF;
  }
}