arff: fixed soybean due to capitalization differences

src/arff/arff.cpp

@@ -90,6 +90,7 @@ namespace ARFF {
         for (AttributeType type : attributeList) {
             std::cout << type.attribute << " (" << type.values.size() << "):";
             for (std::string value : type.values) {
+                if (value == "?") { continue; }
                 std::cout << " " << value;
             }
             std::cout << '\n';
@@ -113,10 +114,10 @@ namespace ARFF {
         parser >> token;
         // Signifies beginning of data
         // Might add a boolean later to mark this
-        if (token == "@data") {
+        if (token == "@data" || token == "@DATA") {
             return;
         }
-        if (token == "@relation") {
+        if (token == "@relation" || token == "@RELATION") {
             parser >> token;
             relation = token;
             LogInfo("ARFF/Attribute", "Relation set: " + relation);
@@ -125,8 +126,10 @@ namespace ARFF {
         parser >> token;
         attributeList.emplace_back(token);
         LogInfo("ARFF/Attribute", "Added attribute: " + token);
-        while (parser >> token) {
+        while (std::getline(parser, token, ',')) {
             // Clean token from outside pieces
+            token.erase(std::remove(token.begin(), token.end(), ' '), token.end());
+            token.erase(std::remove(token.begin(), token.end(), '\t'), token.end());
             token.erase(std::remove(token.begin(), token.end(), '{'), token.end());
             token.erase(std::remove(token.begin(), token.end(), '}'), token.end());
             token.erase(std::remove(token.begin(), token.end(), ','), token.end());
@@ -135,6 +138,8 @@ namespace ARFF {
             attributeList.back().AddValue(token);
             LogInfo("ARFF/Attribute", "Added value: " + token);
         }
+        // Additional missing value case
+        attributeList.back().AddValue("?");
     }
 
     // Add data to runtime database
@@ -147,6 +152,7 @@ namespace ARFF {
         LogInfo("ARFF/Data", "Added id: " + std::to_string(database.back().id));
         for (int i = 0; i < attributeList.size(); ++i) {
             std::getline(parser, token, ',');
+            token.erase(std::remove(token.begin(), token.end(), ' '), token.end());
             token.erase(std::remove(token.begin(), token.end(), '\r'), token.end());
             token.erase(std::remove(token.begin(), token.end(), '\n'), token.end());
             database.back().values.at(i) = token;