Bringing C code up to R library state

MANIFEST.in

@@ -1,5 +1,4 @@
 include README.md
-include requirements.txt
 include setup.py
 include pyproject.toml
 include cubist/src/*

cubist/src/construct.c

@@ -194,7 +194,8 @@ RRuleSet ConstructRuleSet(int ModelNo)
 /*       ----------------  */
 {
   RRuleSet RS;
-  char Msg[20];
+  size_t size = 20;
+  char Msg[size];
   CaseNo i;
   RuleNo r;
   float TempMTSize;
@@ -227,9 +228,9 @@ RRuleSet ConstructRuleSet(int ModelNo)
   ForEach(r, 1, RS->SNRules) { RS->SRule[r]->MNo = ModelNo; }
 
   if (MEMBERS > 1) {
-    sprintf(Msg, "Model %d:", ModelNo + 1);
+    snprintf(Msg, size, "Model %d:", ModelNo + 1);
   } else {
-    sprintf(Msg, "Model:");
+    snprintf(Msg, size, "Model:");
   }
 
   PrintRules(RS, Msg);
@@ -370,7 +371,8 @@ void AttributeUsage(void)
   int m;
   RuleNo r;
   Attribute Att, BestAtt;
-  char U1[5], U2[5];
+  size_t size = 5;
+  char U1[size], U2[size];
 
   /*  Initialise counts  */
 
@@ -406,8 +408,8 @@ void AttributeUsage(void)
     if (!BestAtt)
       break;
 
-    sprintf(U1, "%3.0f%%", rint((100.0 * SumCond[BestAtt]) / SumCases));
-    sprintf(U2, "%3.0f%%", rint((100.0 * SumModel[BestAtt]) / SumCases));
+    snprintf(U1, size, "%3.0f%%", rint((100.0 * SumCond[BestAtt]) / SumCases));
+    snprintf(U2, size, "%3.0f%%", rint((100.0 * SumModel[BestAtt]) / SumCases));
 
     fprintf(Of, "\t  %4s   %4s    %s\n",
             (SumCond[BestAtt] >= 0.01 * SumCases ? U1 : " "),

cubist/src/contin.c

@@ -3,7 +3,7 @@
 /*  Copyright 2010 Rulequest Research Pty Ltd.                           */
 /*                                                                       */
 /*  This file is part of Cubist GPL Edition, a single-threaded version   */
-/*  of Cubist release 2.07.                                           */
+/*  of Cubist release 2.07.                                              */
 /*                                                                       */
 /*  Cubist GPL Edition is free software: you can redistribute it and/or  */
 /*  modify it under the terms of the GNU General Public License as       */
@@ -285,4 +285,4 @@ ContValue GreatestValueBelow(ContValue Th)
   }
 
   return SRec[Low].V;
-}
+}

cubist/src/defns.h

@@ -445,11 +445,6 @@ typedef struct _elt_rec {
 /*									 */
 /*************************************************************************/
 
-/* for R c functions */
-
-void GetRNGstate(void);
-void PutRNGstate(void);
-
 /* cubist.c */
 
 int Xmain(int, char *[]);
@@ -574,7 +569,7 @@ void SmoothModels(Tree T, Tree Parent, CaseNo Fp, CaseNo Lp);
 double ErrVariance(double *Model, CaseNo Fp, CaseNo Lp, double *Err);
 double Smooth(double PVal, int PCases, double LVal, int LCases);
 void FindErrors(Tree T, CaseNo Fp, CaseNo Lp);
-void CValToStr(ContValue CV, Attribute Att, String DS);
+void CValToStr(ContValue CV, Attribute Att, String DS, size_t DS_size);
 Boolean FindWeakestSubtree(Tree T);
 float TreeValue(Tree T, DataRec Case);
 float MedianResid(CaseNo Fp, CaseNo Lp, double Want);
@@ -619,9 +614,9 @@ int Denominator(ContValue Val);
 int FracBase(Attribute Att);
 int GetInt(String S, int N);
 int DateToDay(String DS);
-void DayToDate(int DI, String Date);
+void DayToDate(int DI, String Date, size_t DT_size);
 int TimeToSecs(String TS);
-void SecsToTime(int Secs, String Time);
+void SecsToTime(int Secs, String Time, size_t DT_size);
 void SetTSBase(int y);
 int TStampToMins(String TS);
 void Cleanup(void);

cubist/src/discr.c

@@ -216,4 +216,4 @@ void DiscreteTest(Tree Node, Attribute Att, Set *Subset)
       CopyBits(Bytes, Subset[S], Node->Subset[S]);
     }
   }
-}
+}

cubist/src/extern.h

@@ -102,4 +102,4 @@ extern Boolean LOCK;
 extern CaseCount MINITEMS;
 extern int MAXRULES;
 
-extern float EXTRAP;
+extern float EXTRAP;

cubist/src/formrules.c

@@ -641,4 +641,4 @@ void FreeFormRuleData(void)
   Succ = Nil;
   Free(CPredVal);
   CPredVal = Nil;
-}
+}

cubist/src/formtree.c

@@ -574,4 +574,4 @@ void FindModelAtts(double *Model)
       GEnv.ModelAtt[++GEnv.NModelAtt] = Att;
     }
   }
-}
+}

cubist/src/getdata.c

@@ -36,8 +36,6 @@
 
 #include "redefine.h"
 #include "transform.h"
-// #include "nmath.h"
-//#include <Rmath.h>
 
 float runif( float min, float max )
 {
@@ -50,9 +48,7 @@ float runif( float min, float max )
 #define AltRandom my_rand()
 static double my_rand(void) {
   double dval;
-  // GetRNGstate();
   dval = runif(0, 1);
-  // PutRNGstate();
   return dval;
 }
 
@@ -83,7 +79,8 @@ void GetData(FILE *Df, Boolean Train, Boolean AllowUnknownTarget)
   Boolean AnyUnknown = binfalse, FirstIgnore = bintrue, *AttMsg;
   Attribute Att;
   ContValue Val, Range;
-  char CVS[20];
+  size_t size = 20;
+  char CVS[size];
   CaseCount *Freq; /* discrete value frequencies */
   CaseNo Count, WantTrain, LeftTrain, WantTest, LeftTest;
   Boolean SelectTrain;
@@ -267,7 +264,7 @@ void GetData(FILE *Df, Boolean Train, Boolean AllowUnknownTarget)
         if (Discrete(Att)) {
           fprintf(Of, "`%s'\n", AttValName[Att][Modal[Att]]);
         } else {
-          CValToStr(AttMean[Att], Att, CVS);
+          CValToStr(AttMean[Att], Att, CVS, size);
           fprintf(Of, "%s\n", CVS);
         }
       }
@@ -600,4 +597,4 @@ void FindLimits(Attribute Att, ContValue *Min, ContValue *Max)
     if (Val < *Min)
       *Min = Val;
   }
-}
+}

cubist/src/getnames.c

@@ -542,4 +542,4 @@ int InChar(FILE *f)
   }
 
   return (int)*LBp++;
-}
+}

cubist/src/global.c

@@ -156,4 +156,4 @@ Boolean LOCK = binfalse; /* bintrue if sample locked */
 CaseCount MINITEMS; /* min rule coverage */
 int MAXRULES = 100; /* max number of rules */
 
-float EXTRAP = 0.1; /* allowed extrapolation from models */
+float EXTRAP = 0.1; /* allowed extrapolation from models */

cubist/src/hash.c

@@ -307,4 +307,4 @@ char *ht_getstr(void *ht, const char *key, char *defval, char *errval) {
   }
 
   return (char *)entry->value;
-}
+}

cubist/src/hash.h

@@ -39,4 +39,4 @@ extern void *ht_getvoid(void *ht, const char *key, void *defval, void *errval);
 // extern int ht_getint(void *ht, const char *key, int defval, int errval);
 extern char *ht_getstr(void *ht, const char *key, char *defval, char *errval);
 
-#endif
+#endif

cubist/src/implicitatt.c

@@ -597,85 +597,87 @@ void DefSyntaxError(String Msg)
 void DefSemanticsError(int Fi, String Msg, int OpCode)
 /*   -----------------  */
 {
-  char Exp[1000], XMsg[1008], Op[1000];
+  size_t size = 1000;
+  size_t size_x = 1008;
+  char Exp[size], XMsg[size_x], Op[size];
 
   if (!PreviousError) {
     /*  Abbreviate the input if necessary  */
 
     if (BN - Fi > 23) {
-      sprintf(Exp, "%.10s...%.10s", Buff + Fi, Buff + BN - 10);
+      snprintf(Exp, size, "%.10s...%.10s", Buff + Fi, Buff + BN - 10);
     } else {
-      sprintf(Exp, "%.*s", BN - Fi, Buff + Fi);
+      snprintf(Exp, size, "%.*s", BN - Fi, Buff + Fi);
     }
 
     switch (OpCode) {
     case OP_AND:
-      sprintf(Op, "%s", "and");
+      snprintf(Op, size, "%s", "and");
       break;
     case OP_OR:
-      sprintf(Op, "%s", "or");
+      snprintf(Op, size, "%s", "or");
       break;
     case OP_SEQ:
     case OP_EQ:
-      sprintf(Op, "%s", "=");
+      snprintf(Op, size, "%s", "=");
       break;
     case OP_SNE:
     case OP_NE:
-      sprintf(Op, "%s", "<>");
+      snprintf(Op, size, "%s", "<>");
       break;
     case OP_GT:
-      sprintf(Op, "%s", ">");
+      snprintf(Op, size, "%s", ">");
       break;
     case OP_GE:
-      sprintf(Op, "%s", ">=");
+      snprintf(Op, size, "%s", ">=");
       break;
     case OP_LT:
-      sprintf(Op, "%s", "<");
+      snprintf(Op, size, "%s", "<");
       break;
     case OP_LE:
-      sprintf(Op, "%s", "<=");
+      snprintf(Op, size, "%s", "<=");
       break;
     case OP_PLUS:
-      sprintf(Op, "%s", "+");
+      snprintf(Op, size, "%s", "+");
       break;
     case OP_MINUS:
-      sprintf(Op, "%s", "-");
+      snprintf(Op, size, "%s", "-");
       break;
     case OP_UMINUS:
-      sprintf(Op, "%s", "unary -");
+      snprintf(Op, size, "%s", "unary -");
       break;
     case OP_MULT:
-      sprintf(Op, "%s", "*");
+      snprintf(Op, size, "%s", "*");
       break;
     case OP_DIV:
-      sprintf(Op, "%s", "/");
+      snprintf(Op, size, "%s", "/");
       break;
     case OP_MOD:
-      sprintf(Op, "%s", "%");
+      snprintf(Op, size, "%s", "%");
       break;
     case OP_POW:
-      sprintf(Op, "%s", "^");
+      snprintf(Op, size, "%s", "^");
       break;
     case OP_SIN:
-      sprintf(Op, "%s", "sin");
+      snprintf(Op, size, "%s", "sin");
       break;
     case OP_COS:
-      sprintf(Op, "%s", "cos");
+      snprintf(Op, size, "%s", "cos");
       break;
     case OP_TAN:
-      sprintf(Op, "%s", "tan");
+      snprintf(Op, size, "%s", "tan");
       break;
     case OP_LOG:
-      sprintf(Op, "%s", "log");
+      snprintf(Op, size, "%s", "log");
       break;
     case OP_EXP:
-      sprintf(Op, "%s", "exp");
+      snprintf(Op, size, "%s", "exp");
       break;
     case OP_INT:
-      sprintf(Op, "%s", "int");
+      snprintf(Op, size, "%s", "int");
     }
 
-    sprintf(XMsg, "%s with '%s'", Msg, Op);
+    snprintf(XMsg, size_x, "%s with '%s'", Msg, Op);
     Error(BADDEF2, Exp, XMsg);
     PreviousError = bintrue;
   }
@@ -1035,4 +1037,4 @@ AttValue EvaluateDef(Definition D, DataRec Case)
       return ReturnVal;
     }
   }
-}
+}

cubist/src/instance.c

@@ -769,4 +769,4 @@ void FreeInstances(void)
     FreeIndex(KDTree);
     KDTree = Nil;
   }
-}
+}

cubist/src/modelfiles.c

@@ -181,7 +181,6 @@ void WriteFilePrefix(String Extension)
 void ReadFilePrefix(String Extension)
 /*   --------------  */
 {
-
   if (!(Mf = GetFile(Extension, "r")))
     Error(NOFILE, Fn, "");
 
@@ -758,4 +757,4 @@ Set MakeSubset(Attribute Att)
   }
 
   return S;
-}
+}

cubist/src/predict.c

@@ -150,4 +150,4 @@ void FindPredictedValues(RRuleSet *Cttee, CaseNo Fp, CaseNo Lp)
     PredVal(Case[i]) = (USEINSTANCES ? NNEstimate(Cttee, Case[i])
                                      : PredictValue(Cttee, Case[i]));
   }
-}
+}

cubist/src/prunetree.c

@@ -568,4 +568,4 @@ float MedianResid(CaseNo Fp, CaseNo Lp, double Want)
   }
 
   return GEnv.Resid[Fp];
-}
+}

cubist/src/rcubist.c

@@ -63,7 +63,7 @@ int cubistmain(void)
   if (!(F = GetFile(".data", "r")))
     Error(NOFILE, "", "");
   GetData(F, bintrue, binfalse);
-  fprintf(Of, TX_ReadData(MaxCase + 1, MaxAtt));
+  fprintf(Of, TX_ReadData(MaxCase + 1, MaxAtt, FileStem));
 
   if (XVAL && (F = GetFile(".test", "r"))) {
     SaveMaxCase = MaxCase;
@@ -129,4 +129,4 @@ int cubistmain(void)
   fprintf(Of, T_Time, ExecTime() - StartTime);
 
   return 0;
-}
+}