source: trunk/CrossPare/decent/epsilon/query/addLabels.eol @ 76

import "../libraries/arff/common.eol";
import "../libraries/decent/common.eol";
import "../libraries/decent/logging.eol";

"Running addlabels".log(1);
var start = Native("java.lang.System").currentTimeMillis();


var s = ",";
var targetAttributes = new Map;
//TODO: export options
targetAttributes.put("Artifact.Target.BugFix.AverageWeight",0.1);

/*
targetAttributes.put("Artifact.Target.Refactoring.AverageWeight",0.1);
targetAttributes.put("Artifact.Target.Fix.AverageWeight",0.1);
targetAttributes.put("Artifact.Target.IssueCount.AverageWeight",0.1);
targetAttributes.put("Artifact.Target.IssueReference.AverageWeight",0.1);
targetAttributes.put("Artifact.Target.UsersPerIssue.AverageWeight",0.1);
targetAttributes.put("Artifact.Target.CommentsPerIssue.AverageWeight",0.1);

targetAttributes.put("Artifact.Target.BugFix.Shared.AverageWeight",0.1);
targetAttributes.put("Artifact.Target.Refactoring.Shared.AverageWeight",0.1);
targetAttributes.put("Artifact.Target.Fix.Shared.AverageWeight",0.1);
targetAttributes.put("Artifact.Target.IssueReference.Shared.AverageWeight",0.1);

targetAttributes.put("Artifact.Target.BugFix.Churn.AverageWeight",0.1);
targetAttributes.put("Artifact.Target.Refactoring.Churn.AverageWeight",0.1);
targetAttributes.put("Artifact.Target.Fix.Churn.AverageWeight",0.1);
targetAttributes.put("Artifact.Target.IssueReference.Churn.AverageWeight",0.1);

targetAttributes.put("Artifact.Target.BugFix.Size.AverageWeight",0.1);
targetAttributes.put("Artifact.Target.Refactoring.Size.AverageWeight",0.1);
targetAttributes.put("Artifact.Target.Fix.Size.AverageWeight",0.1);
targetAttributes.put("Artifact.Target.IssueReference.Size.AverageWeight",0.1);
*/

//does not work with binary resources
//var modelFile = new Native("java.io.File") (ARFFx.getModelFile());
var modelFile = new Native("java.io.File") (ARFFx.getModelImpl().getURI().toString().replaceAll("^file:",""));

//var CONFIDENCE = "CONFIDENCE".getARFFAttribute();
//var LABEL = "LABEL".getARFFAttribute();

//TODO: move to common
//TODO: remove once established at earlier steps
var nestedAnonymousClassFilter = "\"[\\w]+\\$[\\d]+.*\"";

var threshold = 100;

//ARTIFACTS
for (arffx in ARFFx!Model.allInstances().select(x|x.data.size() > 0)) {
        arffx.checkTargetAttributes(targetAttributes);
        arffx.checkForCompleteness();

        //these will be recalculated for the bags..
        arffx.setConfidenceThresholds(targetAttributes);
        arffx.assignClassAndConfidence(targetAttributes);

}


var end = Native("java.lang.System").currentTimeMillis();
var duration = end - start;
("Duration: "+duration.toMinutes().round(5)).log(1);


operation ARFFx!Model checkTargetAttributes(targetAttributes : Map) {
        var notFound = targetAttributes.keySet().select(x|not self.attributes.exists(a|a.name = x));
        for (a in notFound) {
                targetAttributes.remove(a);
        }
}

//slow? 
operation ARFFx!Model assignClassAndConfidence(targetAttributes : Map) {
        for (baseAttribute in targetAttributes.keySet()) {
                for (i in self.data) {
                        i.assignClassAndConfidence(baseAttribute, targetAttributes.get(baseAttribute));
                }
        }
}

operation ARFFx!Model setConfidenceThresholds(targetAttributes : Map) {
        //self.name.println();  
        //TODO: store as meta data
        //TODO: store mean divisor as meta-data 
        for (a in targetAttributes.keySet) {
                //("  "+a +" -> "+ targetAttributes.get(a)).println();
                var v = self.data.collect(x|x.getValue(a.getARFFAttribute(self)).asDouble());
                //v = v.normalizeMinMax(0.asDouble(), 1.asDouble());
                //("  "+v.getMin()+" : "+v.getMax()+" : "+v.getMean()+" : "+v.getVariance()+" : "+v.getStandardDeviation()).println();
                var t = (v.getMean()/2).round(4);
                //("  "+a +" -> "+ t).println();
                targetAttributes.put(a,t);
                //("  Non-zero:\t\t"+v.select(x|x <> 0).size()).println();
                //("  Above threshold:\t"+v.select(x|x > t).size()).println();
        }
        //"  updated".println();
        //("    "+targetAttributes).println();
}

operation Collection setConfidenceThresholds(targetAttributes : Map) {
        //TODO: store as meta data
        //TODO: store mean divisor as meta-data 
        for (a in targetAttributes.keySet) {
                var v = new Sequence();
                for (arffx in self) {
                        v.addAll(arffx.data.collect(x|x.getValue(a.getARFFAttribute(arffx)).asDouble()));
                }
                //v = v.normalizeMinMax(0.asDouble(), 1.asDouble());
                //("  "+v.getMin()+" : "+v.getMax()+" : "+v.getMean()+" : "+v.getVariance()+" : "+v.getStandardDeviation()).println();
                var t = (v.getMean()/2).round(4);
                targetAttributes.put(a,t);
                //("  Non-zero:\t\t"+v.select(x|x <> 0).size()).println();
                //("  Above threshold:\t"+v.select(x|x > t).size()).println();
        }
}


operation ARFFx!Model checkForCompleteness() {
        var line = 1;
        var NameAttribute = "Artifact.Name".getARFFAttribute(self);             
        for (i in self.data.select(x|not x.getValue(NameAttribute).matches(nestedAnonymousClassFilter))) {
        //for (i in self.data) {
                line = line+1;
                if (i.values.size() <> self.attributes.size()) {
                        (self.name+" : Line "+line+" : Value and attribute counts do not match : " + i.values.size() +" vs "+ self.attributes.size()).log(1);
                        (i.getValues(s).substring(1)).log(1);
                        i.printMissingAttributes();
                }
        }
}

operation ARFFx!Instance printMissingAttributes() {
        for (a in self.eContainer.attributes) {
                if (not self.values.exists(v|v.ofAttribute = a)) {
                        ("  Missing attribute: "+a.name).log(1);
                }
        }
}


operation ARFFx!Instance assignClassAndConfidence(baseAttribute : String, threshold : Real) : OrderedSet {
        var confidenceAttribute = "CONFIDENCE."+baseAttribute;
        var labelAttribute = "LABEL."+baseAttribute;
        
        //TODO: add attributes to filter
        var base = self.getValue(baseAttribute.getARFFAttribute(self.eContainer()));
        var label = "false";
        var confidence = "high";
        if (base.asDouble() > threshold) {
                label = "true";
        }
        //TODO: also export as parameters
        if (base.asDouble() < 1.01*threshold and base.asDouble() > 0.09*threshold) {
                confidence = "low";
        }
        self.updateValue(confidence, confidenceAttribute);
        self.updateValue(label, labelAttribute);
}

operation ARFFx!Instance getValues(s : String) : String {
        var line = "";
        for (v in self.values) {
                line = line  + s + v.content;
                //line = line  + s + v.ofAttribute.name+"="+v.content;
        }
        return line.replace("NaN","0.0").substring(0); //Substring? why?
}