elexscanner.cpp

用于词法分析的词法分析器

/*
  $Id: ElexScanner.cpp,v 1.9 1997/05/18 10:58:17 matt Exp $

  Elex script scanner class.

  (c) 1996 Matt Phillips.
  */

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

#include "ElexScanner.h"

struct KeyWord
{
  char *name;
  int symbol;
};

static KeyWord keyWords [] = 
{
  {"scanner", ElexScanner::SymScanner},
  {"symbols", ElexScanner::SymSymbols},
  {"defines", ElexScanner::SymDefines},
  {"begin", ElexScanner::SymBegin},
  {"end", ElexScanner::SymEnd},
  {"on", ElexScanner::SymOn},
  {0, 0}			// sentinel
};

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

  while (input.peek () != '"')
  {
    if (input.eof ())
    {
      error ("unterminated string");
      break;
    }

    if (input.peek () == '\\') input.get ();

    input.get ();
  }

  input.skip ();		// lose terminating "
}

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

  // is it a keyword?  
  symbol = SymIdent;

  string str (getText ());

  for (KeyWord *keyword = keyWords; keyword->name; keyword++)
  {
    if (keyword->name == str)
    {
      symbol = keyword->symbol;
      break;
    }
  }
}

void ElexScanner::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;
  };

  symLine = input.getLine ();
  symCol = input.getColumn ();

  if (!input.eof ())
  {
    if (isalpha (input.peek ()))
    {
      getIdent ();
    } else
    {
      int chr = input.peek ();
      input.get ();
	 
      switch (chr)
      {
      case ':':
	symbol = SymColon;
	break;
      case '=':
	symbol = SymEquals;
	break;
      case ',':
	symbol = SymComma;
	break;
      case '<':
	symbol = SymLessThan;
	break;
      case '"':
	symbol = SymString;
	getString ();
	break;
      default:
	input.get ();
	error ("illegal symbol");
      }
    }
  } else
  {
    symbol = SymEOF;
  }
}