basics.java

基于数据挖掘的决策树改进算法和贝叶斯改进算法

package shared;
import java.lang.*;
import java.io.*;

/** Every file uses the members of the basics class.  Here we define some constants, an
 * initialization function and termination function. Note that since
 * initialization order of static instances is not defined within different
 * files, it is important that all initialization be done here, or at least
 * that any other initialization will not depend on anything initialized here.
 * This is especially a problem with constructors that call fatal_error and
 * output to err, since err may not be initialized at that stage. Note the
 * dependency between LONG_MAX and MAX_LONG_LEN.
 *
 * @author James Louis	8/16/2001	Ported to Java.
 * @author Eric Eros		5/16/97	Created MLC class
 * @author Ronny Kohavi	7/13/93	Initial revision (.c)
 * @author Ronny Kohavi	8/26/93	Initial revision (.h)
 */

public class Basics {
    //Compilation directives - JAL
    /** TRUE if DBG sections of code should be executed, FALSE otherwise. Default is FALSE**/
    public static boolean DBG = false;
    /** TRUE if DBGSLOW sections of code should be executed, FALSE otherwise. Default is FALSE**/
    public static boolean DBGSLOW = false;
    
    /** The probability interval value used for calculating confidence.
     */    
    public static double CONFIDENCE_INTERVAL_PROBABILITY;
    
    // MLC++ - Machine Learning Library in -*- C++ -*-
    // See Descrip.txt for terms and conditions relating to use and distribution.
    
    // This is an include file.  For a full description of the class and
    // functions, see the file "basicCore.c".
    
    
    /** Default value for opening protections.
     */
    public static int defaultOpenProt;
    /** The prefix character for displaying error strings.
     */
    public static char ERROR_PREFIX;
    /** The character used to display when a line is being wrapped.
     */
    public static char WRAP_INDENT;
    /** Bad exit status for program.
     */
    public static int BAD_STATUS;
    /** Value used for error catching.
     */
    public static String fatal_expected;
    /** TRUE if mineset is being used, FALSE otherwise.
     */
    public static boolean mineset; // for those messages specific to SGI's MineSet.
    //obs   public static char *minesetVersionStr;
    /** The version of mineset being used.
     */
    public static String minesetVersionStr;
    
    //obs   public static char *err_text;
    /** Standard error text.
     */
    public static String err_text;
    //obs   public static MLCOStream err;
    /** Writer for output of error information.
     */
    public static Writer err;
    
    // note that MString.c assumes that Real is typedef'd to a double.
    /** Maximum value for real numbers.
     */
    public static double REAL_MAX;
    /** Minimum number for real numbers.
     */
    public static double REAL_MIN;
    /** Maximum number for stored real numbers.
     */
    public static float STORED_REAL_MAX;
    /** Minimum number for stored real numbers.
     */
    public static float STORED_REAL_MIN;
    
    // For cases where a variable's value may be used as a flag to indicate
    //   it has not yet been set.  These are extreme negative values.
    //   See basicCore.c
    /** Number that designates undefined variance values.
     */
    public static double UNDEFINED_VARIANCE;
    /** Number that represents undefined real values.
     */
    public static double UNDEFINED_REAL;
    /** Number that represents undefined integer values.
     */
    public static int  UNDEFINED_INT;
    
    // Default wrap width for an MStream, this guarantees that it is off initially
    /** Number of characters allowed before a line is word wrapped.
     */    
    public static int DEFAULT_WRAP_WIDTH;
    // Default wrap prefix for an MStream -3 spaces
    //obs extern const char* DEFAULT_WRAP_PREFIX;
    /** Prefix added to word wrapped lines.
     */    
    public static String DEFAULT_WRAP_PREFIX;
    
/*
 
// These are defined here because of a Bug in CFront that declared
//   INT_MIN wrong (without (int)).  What happens is that 2147483648
//   does not fit in int, so it is made unsigned, then unary minus subtracts
//   from max unsigned int to get the same number.
#ifdef CFRONT
#define SHORT_MAX      32767
#define LONG_MAX       2147483647
#define INT_MAX        LONG_MAX
 
// defining LONG_MIN as -2147483648 does not work because the minus is
//   a unary operator and the number causes it to be unsigned.  We can
//   cast to (long), but ObjectCenter gives a warning... This method
//   is adapted from GNU limits.h
#define SHORT_MIN      (-SHORT_MAX-1)
#define LONG_MIN       (-LONG_MAX-1)
#define INT_MIN        LONG_MIN
 
// MAX_LONG_LEN is used to define array dimensions so it cannot be
//    extern const int.
// LONG_MAX is [phone] which is 10 digits +1 for sign.
// Note that this is an upper bound and MAX_LONG can be lower.
#define MAX_LONG_LEN   11
// Similar to LONG_MAX, 5 digits +1 for sign.
#define MAX_SHORT_LEN 6
#define	UCHAR_MAX     255  // max value of an "unsigned char"
 
// Just use standard "limits.h" if we're not using CFRONT.
//   We still need to set some of the symbols
#else
#include <limits.h>
#define SHORT_MAX      32767
#define SHORT_MIN      (-SHORT_MAX-1)
#define MAX_LONG_LEN   11
#define MAX_SHORT_LEN 6
#endif
 
 
// In the following, DBL_DIG and FLT_DIG are defined in limits.h
// This is needed to set the precision of the stream when doing Real I/O
//   This value is trunc((DSIGNIF - 1) / log2(10)),
//   or trunc((FSIGNIF - 1) / log2(10)), where DSIGNIF is the
//   number of significant bits for a double (defined in values.h),
//   FSIGNIF is likewise defined for a float, and 1 is subtracted for
//   the sign bit.
   public static int REAL_MANTISSA_LEN = DBL_DIG; // 15
   public static int STORED_REAL_MANTISSA_LEN = FLT_DIG; //  6
   public static int MAX_ECVT_DIGITS = DBL_DIG+2; // 17
   public static int MAX_ECVT_WIDTH = 84;  // from man page for ecvt_r
 */
    
    // Note:  Should the size of Real and/or StoredReal change, the above
    //   numbers must change correspondingly.  Also, REAL_EPSILON and
    //   STORED_REAL_EPSILON (in basicCore.c) must change.
    
    // Because Reals can be displayed with full digits,
    // MAX_REAL_LEN is set to a large number
    // MAX_REAL_DIGITS specifies the maximum number of digits which may
    //   appear before the decimal point in a Real.
    /** Maximum number of digits for real numbers.
     */    
    public static int MAX_REAL_DIGITS = 800;
    /** Maximum length for real number display.
     */    
    public static int MAX_REAL_LEN = 2048;
    /** Maximum length for stored real number display.
     */    
    public static int MAX_STORED_REAL_LEN = 2048;
    /** MAX_WIDTH is the maximum width allowed for real output.
     */    
    public static int MAX_WIDTH = 2048;
    
    
    
    
    
/*
#ifndef _basic_h
#define _basic_h 1
 
#include <machine.h>
char is_compiled_fast(); // returns TRUE if library compiled in fast mode
 
#include <stdlib.h>
#include <math.h>
 
// There's sometimes a need to give a reference to something that
//   is invalid.  the SGI compiler warns about NULL ref, so this
//   avoids the warning and is just as bad, i.e., an access will
//   cause program to abort.
#define NULL_REF 1
 
// make sure nobody has defined Bool via #define.  Xlib.h does this.
#ifdef Bool
#error Bool is already defined.  Please undefine before including basics.h
#endif
 
// knock out any alternate TRUE or FALSE if defined
#ifdef TRUE
#undef TRUE
#endif
#ifdef FALSE
#undef FALSE
#endif
 
#ifdef _BOOL
  typedef bool Bool;  // After George Boole
# define TRUE true
# define FALSE false
#else
  typedef char Bool;
# define TRUE Bool(!0)
# define FALSE Bool(0)
#endif
 
#ifndef FAST
#define DBG(stmts) if (GlobalOptions::debugLevel >= 1) {stmts;} else
#else
#define DBG(stmts)
#endif
 
#ifdef TEMPL_MAIN
   #define TEMPL_GENERATOR(name) int main()
#else
   #define TEMPL_GENERATOR(name) int name()
#endif
 
 
// DBGSLOW() should only be used for especially expensive code.
#ifndef FAST
#define DBGSLOW(stmts) if (GlobalOptions::debugLevel >= 2) {stmts;} else
#else
#define DBGSLOW(stmts)
#endif
 
// DBG_DECLARE() is intended for use in class or function declarations,
//   where "if" statements are not allowed.
// Note that when using this macro, the semicolon must be INSIDE,
//   since empty declarations are not allowed (r9.2)
#ifndef FAST
#define DBG_DECLARE(stmts) stmts
#else
#define DBG_DECLARE(stmts)
#endif
 
 
 
typedef double Real;
typedef float StoredReal; // for arrays etc.
 
// To prevent accidental copy construction, most copy constructors
// take a second "dummy" argument.  Initially this dummy argument was
// forced to be an integer and that integer was checked to be 0,
// The new standard is that it would be much simpler to have the dummy
// argument be an actual enumerated type.
// Example usage:
//	Foo.h: Foo(const Foo& foo, CtorDummy);
//	Client.h: Foo foo2(foo1, ctorDummy);
enum CtorDummy {ctorDummy=0};
 
// Also, there is an enumerated type for dummy arguments.
enum ArgDummy {argDummy=0};
 
// Declarations for the parallel execution using pthread library.  Currently
// only entropy discretizors are parallelized.  Do not define this constant if
// you do not want to use pthreads
 
// #define PTHREADS
 
#ifdef PTHREADS
#include <malloc.h>
#include <unistd.h>
#include <pthread.h>
#include <sys/shm.h>
#include <sys/types.h>
#include <sys/resource.h>
#include <sys/prctl.h>
#define PTHR_DECL(stmts)	stmts
#else
#define PTHR_DECL(stmts)
#endif // PTHREADS
 
#include <time.h>
#include <MString.h>
#include <GlobalOptions.h>
 
// Due to archaic 8+3 PC naming conventions, strstream.h
// is sometimes called strstrea.h
#ifdef PC_MSVC
#include <strstrea.h>
#else
#include <strstream.h>
#endif
 
#ifdef __GNUC__
typedef timespec timespec_t;
#endif
 
/* G++ and other smart and proper compilers define null as a variable, to
 * escape type sensitive situations. However, this breaks operator overloading
 * for comparason operators that require NULL to be 0, (traditional C++)
 *
#undef NULL
#define NULL 0
 
 
#include <error.h>
 
// MLCInit class to force consistent initialization order
class MLCInit {
   NO_DEFAULT_OPS(MLCInit);
public:
   // Public data
   static int count;
   // Methods
   MLCInit() { if(count++ == 0) startup(); }
   ~MLCInit() { if(--count == 0) shutdown(); }
 
   void startup();
   void shutdown();
};
 
// static instance of MLCInit:  used for force initialization order.
// note that this instance is DEFINED here in the header file; this
// forces its definition to occur BEFORE all other definitions in
// each file which includes basics.h, and thus before all other
// static initialization.
static MLCInit mlcInit;
 
// Enumerated type to declare constructors which indicate that a static
// instance should be initialized externally
enum UseExternalCtor { useExternalCtor = 1 };
 
// This macro causes a fatal_error if the given condition is FALSE.
// The behavior is similar to the C/C++ assert (see include/assert.h).
// The macro is executed as a DBG statement.
 
// These allow sharing the same message string across files to
//   save space (also help with gp_overflow(5) errors.
extern const char *ASSERT_FAILURE_MSG;
extern const char *ASSERT_FILE_MSG;
extern const char *ASSERT_LINE_MSG;
 
// ASSERTs for debugging purposes should be in DBG().
#define ASSERT(stmt) \
      ((stmt)?((void)0): \
       (void)(err << ASSERT_FAILURE_MSG << # stmt <<  \
        ASSERT_FILE_MSG << __FILE__ << \
        ASSERT_LINE_MSG << __LINE__ << fatal_error))
 
// Old version below caused compiler core dump on SGI and ConstCat
//      (void)((stmt) ||                                                     \
//      ((err << "MLC++ internal error: assertion failed: " # stmt   \
//	", file " << __FILE__ << ", line " << __LINE__ << fatal_error), 0))
 
 
#define ABORT_IF_REACHED \
   err << "MLC++ internal error: unexpected condition in file " \
       << __FILE__ << ", line " << __LINE__ << fatal_error
 
 
#include <MLCStream.h>
extern MLCIStream Mcin;
extern MLCOStream Mcout;
extern MLCOStream Mcerr;
 
// DBG*(code) needs a trailing semicolon outside, i.e. DBG(x=3);
// The last statement inside does not need a semicolon.  The semicolon
//   outside allows proper indentation in Emacs C++ mode.
//   and is also required inside IFs to generate an empty statement.
 
// The "else" is so DBG will work even if it is inside an if statement.
//   (the else in the definition assures proper matching of the else
//    in the code, and also makes use of the semicolon...):
//
//       if (foo)
//          DBG(bar);
//       else
//          kuku();
 
// This is the size of the buffer used for reading and writing files.