private Map<String, List<Integer>> getCommitsBetween(
Map<String, List<VcsCommitInfo>> first, Map<String, List<VcsCommitInfo>> second) {
Map<String, List<Integer>> result = new HashMap<>();
for (String filename : first.keySet()) {
// System.out.println(filename+":");
List<VcsCommitInfo> commitsFromFirst = first.get(filename);
List<VcsCommitInfo> commitsFromSecond = second.get(filename);
List<Integer> intervals = new ArrayList<>();
if (commitsFromSecond == null) result.put(filename, Arrays.asList(-1));
else {
Iterator<VcsCommitInfo> firstIt = commitsFromFirst.iterator();
Iterator<VcsCommitInfo> secondIt = commitsFromSecond.iterator();
while (firstIt.hasNext() && secondIt.hasNext()) {
VcsCommitInfo topCommit = firstIt.next();
// System.out.println("top: "+topCommit);
while (secondIt.hasNext()) {
VcsCommitInfo bugCommit = secondIt.next();
// System.out.println("bugfix: "+bugCommit);
if (bugCommit.getDate().after(topCommit.getDate())) {
intervals.add(
getNumberOfCommitsBetween(topCommit.getCommitId(), bugCommit.getCommitId()));
break;
}
}
}
if (firstIt.hasNext() && !secondIt.hasNext()) intervals.add(-1);
}
result.put(filename, intervals);
}
return result;
}
private void generateTestAndTrainSets(Collection<VcsCommitInfo> allCommits) {
testSet = new ArrayList<>();
trainSet = new ArrayList<>();
for (VcsCommitInfo commit : allCommits) {
if (Math.random() > TEST_SET_FRACTION) {
trainSet.add(commit.getCommitId());
} else {
testSet.add(commit.getCommitId());
}
}
}
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);
}
}
}
}
}
private Map<String, Integer> processIndices(Collection<VcsCommitInfo> commits) {
int index = 0;
Map<String, Integer> result = new HashMap<>();
for (VcsCommitInfo commitInfo : commits) {
result.put(commitInfo.getCommitId(), ++index);
}
return result;
}