regexpscanner.cpp

用于词法分析的词法分析器

/*
  $Id: RegexpScanner.cpp,v 1.5 1997/02/23 13:35:19 matt Exp $

  Regular expression scanner class.
  
  (c) Apr 1996 Matthew Phillips.
  */

#include "RegexpScanner.h"

int RegexpScanner::handleVariable ()
{
  input.clearBuffer ();

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

    input.get ();
  }

  input.skip ();		// lose terminating }

  return SymVariable;
}

char RegexpScanner::readDecNum ()
{
  int value = 0;
 
  // read up to 3 decimal digits
  for (int count = 3;
       count > 0 && !input.eof () && isdigit (input.peek ());
       count--, input.skip ())
  {
    value *= 10;
    value += input.peek () - '0';
  }

  return (char)value;
}

char RegexpScanner::readHexNum ()
{
  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;
  }

  return (char)value;
}

int RegexpScanner::handleBackslash ()
{
  char chr = input.peek ();

  switch (chr)
  {
  case 'n':
    chr = '\n';
    input.skip ();
    break;
  case 't':
    chr = '\t';
    input.skip ();
    break;
  case 'v':
    chr = '\v';
    input.skip ();
    break;
  case 'a':
    chr = '\a';
    input.skip ();
    break;
  case 'f':
    chr = '\f';
    input.skip ();
    break;
  case 'b':
    chr = '\b';
    input.skip ();
    break;
  case 'r':
    chr = '\r';
    input.skip ();
    break;
  case 'x':
  case 'X':
    input.skip ();
    chr = readHexNum ();
    break;
  default:
    if (isdigit (chr))
    {
      chr = readDecNum ();
    } else
      input.skip ();
  }

  input.setBuffer (chr);

  return SymChar;
}

void RegexpScanner::getNext ()
{
  symLine = input.getLine ();
  symCol = input.getColumn ();

  if (!input.eof ())
  {
    input.clearBuffer ();

    int chr = input.peek ();
    input.get ();
    
    switch (chr)
    {
    case '.':
      symbol = SymDot;
      break;
    case '(':
      symbol = SymLBracket;
      break;
    case ')':
      symbol = SymRBracket;
      break;
    case '[':
      symbol = SymLSBracket;
      break;
    case ']':
      symbol = SymRSBracket;
      break;
    case '<':
      symbol = handleVariable ();
      break;
    case '~':
      symbol = SymTilde;
      break;
    case '?':
      symbol = SymQuestion;
      break;
    case '^':
      symbol = SymCaret;
      break;
    case '*':
      symbol = SymStar;
      break;
    case '+':
      symbol = SymPlus;
      break;
    case '-':
      symbol = SymMinus;
      break;
    case '|':
      symbol = SymBar;
      break;
    case '\\':
      symbol = handleBackslash ();
      break;
    default:
      symbol = SymChar;
    }
  } else
  {
    symbol = SymEOF;
  }
}