Language/OS - Multiplatform Resource Library/LANGUAGE OS.iso/ast_comp/cpp-kit.lha/c++kit/Scan.C

C/C++ Source or Header | 1993-04-11 | 5KB | 260 lines

/* 9th April, 1993 Mayan Moudgill
 * Basically is given a file. If asked it will try and scan one of the
 * following tokens from the file.
 *    strings:    "([^"\\\n]|(\.))*"
 *    identifier: [A-Za-z_0-9]+
 *    integer:    [-+]?[0-9]+
 *    character:  .
 * These returns a Token.
 * It can also accept a mark, which returns a Mark, and a reject which
 * when given a Mark, rolls back the state to the Mark, and starts matching
 * from that point onwards. Also, for convenience it can rollback the
 * last token. Also, it can match a character and a character-string.
 * (these are equivalent to trying to scanning a character/string,
 * and rolling back if the character/string does not exactly match the
 * argument).
 *
 * The other facility that the function provides is that it keeps track
 * of the number of lines read in.
 *
 * I can think of at least of two ways of implementing the tokenizing
 * and rollback mechanisms: reading files, and mmap'ing them. This
 * implementation uses mmap.
 */

#include <sys/types.h>
#include <sys/mman.h>
#include <sys/fcntl.h>
#include <sys/stat.h>
#include <ctype.h>

#include "Scan.H"

Scan::Scan()
:  _close(1)
{
}

Scan::Scan(char * name)
{
   strncpy(_name, name, 255);
   _name[255] = 0;
   _fd = open(name, O_RDONLY);
   if( _fd < 0 ) {
      cerr << "could not open " << name << " cause: " << errno << endl;
      _close = 1;
      return;
   }

struct stat buf;
   fstat( _fd, &buf);
   _size =  buf.st_size;
   _file = mmap(0, (int)_size, PROT_READ, MAP_SHARED, _fd, 0);
   if( (int)_file == -1 ) {
      cerr << "could not mmap " << name << " cause: " << errno << endl;
      _close = 1;
   }
   else {
      _close = 0;
      _end = _file + _size;
   }
   if( _size == 0 ) {
      _eof = 1;
   }
   else {
      _eof = 0;
   }
   _oat = _at = _file;
   _onl = _nl = 0;
}

Scan::~Scan()
{
   if( !_close ) {
      munmap( _file, (int)_size);
      close( _fd);
      _close = 1;
   }
}

int     Scan::_space()
{

   if( _start() ) {
      return 1;
   }
   while( isspace(_get())) {
   };
   _backup();
   _wnl = _tnl;
   _wat = _tat;
   return _eof;
}

int     Scan::number(Token& result)
{
int     in;

   if( _space() ) {
      result = Token(0,0);
      return 0;
   }

   in = _get();
   if( in == '-' || in == '+' ) {
      in = _get();
   }
   if( !isdigit(in) ) {
      return 0;
   }
   do {
      in = _get();
   } while( isdigit(in));
   _backup();
   result = Token(_wat, _tat);
   _commit();
   return 1;
}

int     Scan::identifier(Token& result)
{
int     in;

   if( _space() ) {
      result = Token(0,0);
      return 0;
   }

   in = _get();

   if( in != '_' && !isalpha(in) ) {
      result = Token(0, 0);
      return 0;
   }

   do {
      in = _get();
   } while ( in == '_' || isdigit(in) || isalpha(in) );
   _backup();

   result = Token(_wat, _tat);
   _commit();
   return 1;
}

int     Scan::string(Token& result )
{
int     in;

   if( _space() ) {
      result = Token(0,0);
      return 0;
   }

   in = _get();
   if( in != '"' ) {
      result = Token(0,0);
      return 0;
   }
   do {
      in = _get();
      if( in == '\\' ) {
     _get();
     in = _get();
      }
   } while( in != '"' );

   result = Token(_wat, _tat);
   _commit();
   return 1;
}

int    Scan::token(Token& result)
{
int    in;
   if( _space() ) {
      result = Token(0,0);
      return 0;
   }

   in = _get();
   _backup();
   if( in ==  '"') {
      if( string(result) ) {
     result = Token( result.string()+1, result.length() - 2);
     return 1;
      }
      else {
     return 0;
      }
   }
   else {
      return identifier(result);
   }
}

int     Scan::character(Token& result)
{
   if( _start() ) {
      result = Token(0, 0);
      return 0;
   }
   _get();
   result = Token(_wat, 1);
   _commit();
   return 1;
}

int     Scan::match(char c, Token& result)
{
int   in;

   if( _start() ) {
      result = Token(0,0);
      return 0;
   }
   in = _get();
   if( in == c ) {
      _commit();
      result = Token(_wat, 1);
      return 1;
   }
   else {
      result = Token(0,0);
      return 0;
   }
}

int     Scan::match(char * s, Token& result)
{
int   in;

   if( _space() ) {
      result = Token(0,0);
      return 0;
   }

   while( *s != 0  ) {
      in = _get();
      if( in != *s ) {
     break;
      }
      s++;
   }

   if( *s == 0 ) {
      _commit();
      result = Token(_wat, _tat);
      return 1;
   }
   else {
      result = Token(0,0);
      return 0;
   }
}

int      Scan::eof()
{
   return _close || _at == _end;
}