[26] | 1 | /*************************************************************************/ |
---|
| 2 | /* */ |
---|
| 3 | /* Get case descriptions from data file */ |
---|
| 4 | /* -------------------------------------- */ |
---|
| 5 | /* */ |
---|
| 6 | /*************************************************************************/ |
---|
| 7 | |
---|
| 8 | |
---|
| 9 | #include "defns.i" |
---|
| 10 | #include "types.i" |
---|
| 11 | #include "extern.i" |
---|
| 12 | |
---|
| 13 | #define Inc 2048 |
---|
| 14 | |
---|
| 15 | |
---|
| 16 | |
---|
| 17 | /*************************************************************************/ |
---|
| 18 | /* */ |
---|
| 19 | /* Read raw case descriptions from file with given extension. */ |
---|
| 20 | /* */ |
---|
| 21 | /* On completion, cases are stored in array Item in the form */ |
---|
| 22 | /* of Descriptions (i.e. arrays of attribute values), and */ |
---|
| 23 | /* MaxItem is set to the number of data items. */ |
---|
| 24 | /* */ |
---|
| 25 | /*************************************************************************/ |
---|
| 26 | |
---|
| 27 | GetData(Extension) |
---|
| 28 | /* -------- */ |
---|
| 29 | String Extension; |
---|
| 30 | { |
---|
| 31 | FILE *Df, *fopen(); |
---|
| 32 | char Fn[100]; |
---|
| 33 | ItemNo i=0, j, ItemSpace=0; |
---|
| 34 | Description GetDescription(); |
---|
| 35 | |
---|
| 36 | /* Open data file */ |
---|
| 37 | |
---|
| 38 | strcpy(Fn, FileName); |
---|
| 39 | strcat(Fn, Extension); |
---|
| 40 | if ( ! ( Df = fopen(Fn, "r") ) ) Error(0, Fn, ""); |
---|
| 41 | |
---|
| 42 | do |
---|
| 43 | { |
---|
| 44 | MaxItem = i; |
---|
| 45 | |
---|
| 46 | /* Make sure there is room for another item */ |
---|
| 47 | |
---|
| 48 | if ( i >= ItemSpace ) |
---|
| 49 | { |
---|
| 50 | if ( ItemSpace ) |
---|
| 51 | { |
---|
| 52 | ItemSpace += Inc; |
---|
| 53 | Item = (Description *) |
---|
| 54 | realloc(Item, ItemSpace*sizeof(Description)); |
---|
| 55 | } |
---|
| 56 | else |
---|
| 57 | { |
---|
| 58 | Item = (Description *) |
---|
| 59 | malloc((ItemSpace=Inc)*sizeof(Description)); |
---|
| 60 | } |
---|
| 61 | } |
---|
| 62 | |
---|
| 63 | Item[i] = GetDescription(Df); |
---|
| 64 | |
---|
| 65 | } while ( Item[i] != Nil && ++i ); |
---|
| 66 | |
---|
| 67 | fclose(Df); |
---|
| 68 | MaxItem = i - 1; |
---|
| 69 | } |
---|
| 70 | |
---|
| 71 | |
---|
| 72 | |
---|
| 73 | /*************************************************************************/ |
---|
| 74 | /* */ |
---|
| 75 | /* Read a raw case description from file Df. */ |
---|
| 76 | /* */ |
---|
| 77 | /* For each attribute, read the attribute value from the file. */ |
---|
| 78 | /* If it is a discrete valued attribute, find the associated no. */ |
---|
| 79 | /* of this attribute value (if the value is unknown this is 0). */ |
---|
| 80 | /* */ |
---|
| 81 | /* Returns the Description of the case (i.e. the array of */ |
---|
| 82 | /* attribute values). */ |
---|
| 83 | /* */ |
---|
| 84 | /*************************************************************************/ |
---|
| 85 | |
---|
| 86 | |
---|
| 87 | Description GetDescription(Df) |
---|
| 88 | /* --------------- */ |
---|
| 89 | FILE *Df; |
---|
| 90 | { |
---|
| 91 | Attribute Att; |
---|
| 92 | char name[500], *endname, *CopyString(); |
---|
| 93 | Boolean ReadName(); |
---|
| 94 | int Dv; |
---|
| 95 | float Cv; |
---|
| 96 | Description Dvec; |
---|
| 97 | double strtod(); |
---|
| 98 | |
---|
| 99 | if ( ReadName(Df, name) ) |
---|
| 100 | { |
---|
| 101 | Dvec = (Description) calloc(MaxAtt+2, sizeof(AttValue)); |
---|
| 102 | |
---|
| 103 | ForEach(Att, 0, MaxAtt) |
---|
| 104 | { |
---|
| 105 | if ( SpecialStatus[Att] == IGNORE ) |
---|
| 106 | { |
---|
| 107 | /* Skip this value */ |
---|
| 108 | |
---|
| 109 | DVal(Dvec, Att) = 0; |
---|
| 110 | } |
---|
| 111 | else |
---|
| 112 | if ( MaxAttVal[Att] || SpecialStatus[Att] == DISCRETE ) |
---|
| 113 | { |
---|
| 114 | /* Discrete value */ |
---|
| 115 | |
---|
| 116 | if ( ! ( strcmp(name, "?") ) ) |
---|
| 117 | { |
---|
| 118 | Dv = 0; |
---|
| 119 | } |
---|
| 120 | else |
---|
| 121 | { |
---|
| 122 | Dv = Which(name, AttValName[Att], 1, MaxAttVal[Att]); |
---|
| 123 | if ( ! Dv ) |
---|
| 124 | { |
---|
| 125 | if ( SpecialStatus[Att] == DISCRETE ) |
---|
| 126 | { |
---|
| 127 | /* Add value to list */ |
---|
| 128 | |
---|
| 129 | Dv = ++MaxAttVal[Att]; |
---|
| 130 | if ( Dv > (int) AttValName[Att][0] ) |
---|
| 131 | { |
---|
| 132 | printf("\nToo many values for %s (max %d)\n", |
---|
| 133 | AttName[Att], (int) AttValName[Att][0]); |
---|
| 134 | exit(1); |
---|
| 135 | } |
---|
| 136 | |
---|
| 137 | AttValName[Att][Dv] = CopyString(name); |
---|
| 138 | } |
---|
| 139 | else |
---|
| 140 | { |
---|
| 141 | Error(4, AttName[Att], name); |
---|
| 142 | } |
---|
| 143 | } |
---|
| 144 | } |
---|
| 145 | DVal(Dvec, Att) = Dv; |
---|
| 146 | } |
---|
| 147 | else |
---|
| 148 | { |
---|
| 149 | /* Continuous value */ |
---|
| 150 | |
---|
| 151 | if ( ! ( strcmp(name, "?") ) ) |
---|
| 152 | { |
---|
| 153 | Cv = Unknown; |
---|
| 154 | } |
---|
| 155 | else |
---|
| 156 | { |
---|
| 157 | Cv = strtod(name, &endname); |
---|
| 158 | if ( endname == name || *endname != '\0' ) |
---|
| 159 | Error(4, AttName[Att], name); |
---|
| 160 | } |
---|
| 161 | CVal(Dvec, Att) = Cv; |
---|
| 162 | } |
---|
| 163 | |
---|
| 164 | ReadName(Df, name); |
---|
| 165 | } |
---|
| 166 | |
---|
| 167 | if ( (Dv = Which(name, ClassName, 0, MaxClass)) < 0 ) |
---|
| 168 | { |
---|
| 169 | Error(5, "", name); |
---|
| 170 | Dv = 0; |
---|
| 171 | } |
---|
| 172 | Class(Dvec) = Dv; |
---|
| 173 | |
---|
| 174 | return Dvec; |
---|
| 175 | } |
---|
| 176 | else |
---|
| 177 | { |
---|
| 178 | return Nil; |
---|
| 179 | } |
---|
| 180 | } |
---|