public void process(String projectId, Date from, Date to) {
reset();
Collection<VcsCommitInfo> allCommits = getCommitsForProject(projectId, from, to);
generateTestAndTrainSets(allCommits);
System.out.println("Processing commits...");
int total = allCommits.size();
int counter = 0;
long startTime = System.currentTimeMillis();
for (VcsCommitInfo commitInfo : allCommits) {
counter++;
if (counter % 100 == 0) {
System.out.println(
counter
+ " / "
+ total
+ " | last 100 processed in "
+ (System.currentTimeMillis() - startTime)
+ " ms");
startTime = System.currentTimeMillis();
}
for (String filename : commitInfo.getAffectedFiles()) {
if (!commitAffections.containsKey(filename)) {
commitAffections.put(filename, new ArrayList<VcsCommitInfo>());
}
commitAffections.get(filename).add(commitInfo);
}
if (VcsCommitInfoUtils.isRelatedToBugIssue(commitInfo)) {
for (String filename : commitInfo.getAffectedFiles()) {
if (!bugFixCommitAffections.containsKey(filename)) {
bugFixCommitAffections.put(filename, new ArrayList<VcsCommitInfo>());
}
bugFixCommitAffections.get(filename).add(commitInfo);
}
}
for (PredictionModel model : myModels) {
if (trainSet.contains(commitInfo.getCommitId())) {
// train the model
model.update(commitInfo, true);
}
if (testSet.contains(commitInfo.getCommitId())) {
// update the model without training and get prediction data
Map<String, List<VcsCommitInfo>> topAppearanceMapForCurrentModel =
appearancesInTop.get(model);
for (String fileInTop : model.getPredictionData().keySet()) {
if (!topAppearanceMapForCurrentModel.containsKey(fileInTop)) {
topAppearanceMapForCurrentModel.put(fileInTop, new ArrayList<VcsCommitInfo>());
}
topAppearanceMapForCurrentModel.get(fileInTop).add(commitInfo);
}
}
}
}
}
public void dumpData(String projectId, Date from, Date to, int part) {
String postfix = "_" + part;
dump(projectId, "time_allFiles" + postfix + ".dat", getAllAffectionToBugFixIntervals());
dump(projectId, "commits_allFiles" + postfix + ".dat", getAllAffectionToBugFixCommitNumbers());
dump(projectId, "bugfixDensity" + postfix + ".dat", getBugFixTimes(projectId, from, to));
int n = 0;
for (PredictionModel model : myModels) {
dump(
projectId,
model.getParameters().getAbbreviation() + postfix + ".dat",
getAllTopToBugFixIntervals(model));
dump(
projectId,
"commits_" + model.getParameters().getAbbreviation() + postfix + ".dat",
getAllTopToBugFixCommitNumbers(model));
}
}