hooks.c

数据挖掘中的决策树生成的经典的see5软件的源代码。

		Dv = Which(Name, AttValName[Att], 1, MaxAttVal[Att]);
		if ( ! Dv )
		{
		    if ( StatBit(Att, DISCRETE) )
		    {
			if ( Train )
			{
			    /*  Add value to list  */

			    if ( MaxAttVal[Att] >= (long) AttValName[Att][0] )
			    {
				XError(TOOMANYVALS, AttName[Att],
					 (char *) AttValName[Att][0] - 1);
				Dv = MaxAttVal[Att];
			    }
			    else
			    {
				Dv = ++MaxAttVal[Att];
				AttValName[Att][Dv]   = strdup(Name);
				AttValName[Att][Dv+1] = "<other>"; /* no free */
			    }
			}
			else
			{
			    /*  Set value to "<other>"  */

			    Dv = MaxAttVal[Att] + 1;
			}
		    }
		    else
		    {
			XError(BADATTVAL, AttName[Att], Name);
			Dv = UNKNOWN;
		    }
		}
		DVal(DVec, Att) = Dv;
	    }
	    else
	    {
		/*  Continuous value  */

		if ( TStampVal(Att) )
		{
		    CVal(DVec, Att) = Cv = TStampToMins(Name);
		    if ( Cv >= 1E9 )	/* long time in future */
		    {
			XError(BADTSTMP, AttName[Att], Name);
			DVal(DVec, Att) = UNKNOWN;
		    }
		}
		else
		if ( DateVal(Att) )
		{
		    CVal(DVec, Att) = Cv = DateToDay(Name);
		    if ( Cv < 1 )
		    {
			XError(BADDATE, AttName[Att], Name);
			DVal(DVec, Att) = UNKNOWN;
		    }
		}
		else
		if ( TimeVal(Att) )
		{
		    CVal(DVec, Att) = Cv = TimeToSecs(Name);
		    if ( Cv < 0 )
		    {
			XError(BADTIME, AttName[Att], Name);
			DVal(DVec, Att) = UNKNOWN;
		    }
		}
		else
		{
		    CVal(DVec, Att) = strtod(Name, &EndName);
		    if ( EndName == Name || *EndName != '\0' )
		    {
			XError(BADATTVAL, AttName[Att], Name);
			DVal(DVec, Att) = UNKNOWN;
		    }
		}

		CheckValue(DVec, Att);
	    }
	}

	if ( ClassAtt )
	{
	    if ( Discrete(ClassAtt) )
	    {
		Class(DVec) = XDVal(DVec, ClassAtt);
	    }
	    else
	    if ( Unknown(DVec, ClassAtt) || NotApplic(DVec, ClassAtt) )
	    {
		Class(DVec) = 0;
	    }
	    else
	    {
		/*  Find appropriate segment using class thresholds  */

		Cv = CVal(DVec, ClassAtt);

		for ( Dv = 1 ; Dv < MaxClass && Cv > ClassThresh[Dv] ; Dv++ )
		    ;

		Class(DVec) = Dv;
	    }
	}
	else
	{
	    if ( ! ReadName(Df, Name, 1000, '\00') )
	    {
		XError(HITEOF, Fn, "");
		FreeLastCase(DVec);
		return Nil;
	    }

	    Class(DVec) = Dv = Which(Name, ClassName, 1, MaxClass);
	}

	return DVec;
    }
    else
    {
	return Nil;
    }
}



/*************************************************************************/
/*									 */
/*	Store a label or ignored value in IValStore			 */
/*									 */
/*************************************************************************/


int StoreIVal(String S)
/*  ---------  */
{
    int		StartIx, Length;

    if ( (Length=strlen(S) + 1) + IValsOffset > IValsSize )
    {
	if ( IgnoredVals )
	{
	    Realloc(IgnoredVals, IValsSize += 32768, char);
	}
	else
	{
	    IValsSize   = 32768;
	    IValsOffset = 0;
	    IgnoredVals = Alloc(IValsSize, char);
	}
    }

    StartIx = IValsOffset;
    strcpy(IgnoredVals + StartIx, S);
    IValsOffset += Length;

    return StartIx;
}



/*************************************************************************/
/*									 */
/*	Check for bad continuous value					 */
/*									 */
/*************************************************************************/


void CheckValue(Description DVec, Attribute Att)
/*   ----------  */
{
    ContValue	Cv;

    Cv = CVal(DVec, Att);
    if ( ! finite(Cv) )
    {
	Error(BADNUMBER, AttName[Att], "");

	CVal(DVec, Att) = UNKNOWN;
    }
}



/*************************************************************************/
/*									 */
/*	Routines to handle implicitly-defined attributes		 */
/*									 */
/*************************************************************************/


char	*Buff;			/* buffer for input characters */
int	BuffSize, BN;		/* size and index of next character */

EltRec	*TStack;		/* expression stack model */
int	TStackSize, TSN;	/* size of stack and index of next entry */

int	DefSize, DN;		/* size of definition and next element */

Boolean PreviousError;		/* to avoid parasytic errors */

AttValue _UNK,			/* quasi-constant for unknown value */
	 _NA;			/* ditto for not applicable */


#define FailSyn(Msg)	 {DefSyntaxError(Msg); return false;}
#define FailSem(Msg)	 {DefSemanticsError(Fi, Msg, OpCode); return false;}

typedef  union  _xstack_elt
         {
            DiscrValue  _discr_val;
            ContValue   _cont_val;
            String      _string_val;
         }
	 XStackElt;

#define	cval		_cont_val
#define	sval		_string_val
#define	dval		_discr_val



/*************************************************************************/
/*									 */
/*	A definition is handled in two stages:				 */
/*	  - The definition is read (up to a line ending with a period)	 */
/*	    replacing multiple whitespace characters with one space	 */
/*	  - The definition is then read (using a recursive descent	 */
/*	    parser), building up a reverse polish expression		 */
/*	Syntax and semantics errors are flagged				 */
/*									 */
/*************************************************************************/


void ImplicitAtt(FILE *Nf)
/*   -----------  */
{
#ifdef CUBIST
    _UNK.cval = UNKNOWN;
#else
    _UNK.dval = UNKNOWN;
#endif
    _NA.dval  = NA;

    /*  Get definition as a string in Buff  */

    ReadDefinition(Nf);

    PreviousError = false;
    BN = 0;

    /*  Allocate initial stack and attribute definition  */

    TStack = Alloc(TStackSize=50, EltRec);
    TSN = 0;

    AttDef[MaxAtt] = Alloc(DefSize = 100, DefElt);
    DN = 0;

    /*  Parse Buff as an expression terminated by a period  */

    Expression();
    if ( ! Find(".") ) DefSyntaxError("'.' ending definition");

    /*  Final check -- defined attribute must not be of type String  */

    if ( ! PreviousError )
    {
	if ( DN == 1 && DefOp(AttDef[MaxAtt][0]) == OP_ATT )
	{
	    Error(SAMEATT, AttName[ (long) DefSVal(AttDef[MaxAtt][0]) ], Nil);
	    PreviousError = true;
	}

	if ( TStack[0].Type == 'B' )
	{
	    /*  Defined attributes should never have a value N/A  */

	    MaxAttVal[MaxAtt] = 3;
	    AttValName[MaxAtt] = AllocZero(4, String);
	    AttValName[MaxAtt][1] = strdup("??");
	    AttValName[MaxAtt][2] = strdup("t");
	    AttValName[MaxAtt][3] = strdup("f");
	}
	else
	{
	    MaxAttVal[MaxAtt] = 0;
	}
    }

    if ( PreviousError )
    {
	DN = 0;
	SpecialStatus[MaxAtt] = EXCLUDE;
    }

    /*  Write a terminating marker  */

    DefOp(AttDef[MaxAtt][DN]) = OP_END;

    Free(Buff);
    Free(TStack);
}



/*************************************************************************/
/*									 */
/*	Read the text of a definition.  Skip comments, collapse		 */
/*	multiple whitespace characters.					 */
/*									 */
/*************************************************************************/


void ReadDefinition(FILE *f)
/*   --------------  */
{
    Boolean	LastWasPeriod=false;
    char	c;

    Buff = Alloc(BuffSize=50, char);
    BN = 0;

    while ( true )
    {
	c = InChar(f);

	if ( c == '|' ) SkipComment;

	if ( c == EOF || c == '\n' && LastWasPeriod )
	{
	    /*  The definition is complete.  Add a period if it's
		not there already and terminate the string  */

	    if ( ! LastWasPeriod ) Append('.');
	    Append(0);

	    return;
	}

	if ( Space(c) )
	{
	    Append(' ');
	}
	else
	if ( c == '\\' )
	{
	    /*  Escaped character -- bypass any special meaning  */

	    Append(InChar(f));
	}
	else
	{
	    LastWasPeriod = ( c == '.' );
	    Append(c);
	}
    }
}



/*************************************************************************/
/*									 */
/*	Append a character to Buff, resizing it if necessary		 */
/*									 */
/*************************************************************************/


void Append(char c)
/*   ------  */
{
    if ( c == ' ' && (! BN || Buff[BN-1] == ' ' ) ) return;

    if ( BN >= BuffSize )
    {
	Realloc(Buff, BuffSize += 50, char);
    }

    Buff[BN++] = c;
}



/*************************************************************************/
/*									 */
/*	Recursive descent parser with syntax error checking.		 */
/*	The reverse polish is built up by calls to Dump() and DumpOp(),	 */
/*	which also check for semantic validity.				 */
/*									 */
/*	For possible error messages, each routine also keeps track of	 */
/*	the beginning of the construct that it recognises (in Fi).	 */
/*									 */
/*************************************************************************/


Boolean Expression()
/*      ----------  */
{
    int		Fi=BN;

    if ( Buff[BN] == ' ' ) BN++;

    if ( ! Conjunct() ) FailSyn("expression");

    while ( Find("or") )
    {
	BN += 2;

	if ( ! Conjunct() ) FailSyn("expression");

	DumpOp(OP_OR, Fi);
    }

    return true;
}



Boolean Conjunct()
/*      --------  */
{
    int		Fi=BN;

    if ( ! SExpression() ) FailSyn("expression");

    while ( Find("and") )
    {
	BN += 3;

	if ( ! SExpression() ) FailSyn("expression");

	DumpOp(OP_AND, Fi);
    }

    return true;
}



String RelOps[] = {">=", "<=", "!=", "<>", ">", "<", "=", (String) 0};

Boolean SExpression()
/*      -----------  */
{
    int		o, Fi=BN;

    if ( ! AExpression() ) FailSyn("expression");

    if ( (o = FindOne(RelOps)) >= 0 )
    {
	BN += strlen(RelOps[o]);

	if ( ! AExpression() ) FailSyn("expression");

	DumpOp(( o == 0 ? OP_GE :
		 o == 1 ? OP_LE :
		 o == 4 ? OP_GT :
		 o == 5 ? OP_LT :
		 o == 2 || o == 3 ?
			( TStack[TSN-1].Type == 'S' ? OP_SNE : OP_NE ) :
			( TStack[TSN-1].Type == 'S' ? OP_SEQ : OP_EQ ) ), Fi);
    }

    return true;
}



String AddOps[] = {"+", "-", (String) 0};

Boolean AExpression()
/*      -----------  */
{
    int		o, Fi=BN;

    if ( Buff[BN] == ' ' ) BN++;

    if ( (o = FindOne(AddOps)) >= 0 )
    {
	BN += 1;
    }

    if ( ! Term() ) FailSyn("expression");

    if ( o == 1 ) DumpOp(OP_UMINUS, Fi);

    while ( (o = FindOne(AddOps)) >= 0 )
    {
	BN += 1;

	if ( ! Term() ) FailSyn("arithmetic expression");

	DumpOp((char)(OP_PLUS + o), Fi);
    }

    return true;
}



String MultOps[] = {"*", "/", "%", (String) 0};

Boolean Term()
/*      ----  */
{
    int		o, Fi=BN;

    if ( ! Factor() ) FailSyn("expression");

    while ( (o = FindOne(MultOps)) >= 0 )
    {
	BN += 1;

	if ( ! Factor() ) FailSyn("arithmetic expression");

	DumpOp((char)(OP_MULT + o), Fi);
    }

    return true;
}



Boolean Factor()
/*      ----  */
{
    int		Fi=BN;

    if ( ! Primary() ) FailSyn("value");

    while ( Find("^") )
    {
	BN += 1;

	if ( ! Primary() ) FailSyn("exponent");

	DumpOp(OP_POW, Fi);
    }

    return true;
}



Boolean Primary()
/*      -------  */
{
    if ( Atom() )
    {
	return true;
    }
    else
    if ( Find("(") )
    {
	BN++;
	if ( ! Expression() ) FailSyn("expression in parentheses");
	if ( ! Find(")") ) FailSyn("')'");
	BN++;
	return true;
    }
    else
    {
	FailSyn("attribute, value, or '('");
    }
}



String Funcs[] = {"sin", "cos", "tan", "log", "exp", "int", (String) 0};

Boolean Atom()
/*      ----  */
{
    char	*EndPtr, *Str, Date[11], Time[9];
    int		o, FirstBN, Fi=BN;
    ContValue	F;
    Attribute	Att;

    if ( Buff[BN] == ' ' ) BN++;

    if ( Buff[BN] == '"' )
    {
	FirstBN = ++BN;
	while ( Buff[BN] != '"' )
	{
	    if ( ! Buff[BN] ) FailSyn("closing '\"'");
	    BN++;
	}

	/*  Make a copy of the string without double quotes  */

	Buff[BN] = '\00';
	Str = strdup(Buff + FirstBN);

	Buff[BN++] = '"';
	Dump(OP_STR, 0, Str, Fi);
    }
    else
    if ( (Att = FindAttName()) )
    {
	BN += strlen(AttName[Att]);

	Dump(OP_ATT, 0, (String) (long) Att, Fi);
    }
    else
    if ( isdigit(Buff[BN]) )
    {
	/*  Check for date or time first  */

	if ( ( Buff[BN+4] == '/' && Buff[BN+7] == '/' ||
	       Buff[BN+4] == '-' && Buff[BN+7] == '-' )&&
	     isdigit(Buff[BN+1]) && isdigit(Buff[BN+2]) &&
		isdigit(Buff[BN+3]) &&
	     isdigit(Buff[BN+5]) && isdigit(Buff[BN+6]) &&
	     isdigit(Buff[BN+8]) && isdigit(Buff[BN+9]) )
	{
	    memcpy(Date, Buff+BN, 10);
	    Date[10] = '\00';
	    if ( (F = DateToDay(Date)) == 0 )
	    {
		Error(BADDEF1, Date, "date");
	    }

	    BN += 10;
	}
	else
	if ( Buff[BN+2] == ':' && Buff[BN+5] == ':' &&
	     isdigit(Buff[BN+1]) &&
	     isdigit(Buff[BN+3]) && isdigit(Buff[BN+4]) &&
	     isdigit(Buff[BN+6]) && isdigit(Buff[BN+7]) )
	{
	    memcpy(Time, Buff+BN, 8);
	    Time[8] = '\00';
	    if ( (F = TimeToSecs(Time)) == 0 )
	    {
		Error(BADDEF1, Time, "time");
	    }

	    BN += 8;
	}
	else
	{
	    F = strtod(Buff+BN, &EndPtr);

	    /*  Check for period after integer  */

	    if ( EndPtr > Buff+BN+1 && *(EndPtr-1) == '.' )
	    {
		EndPtr--;
	    }

	    BN = EndPtr - Buff;