public static DupeCollection FindDupes(MetricDict metrics)
{
Dictionary <AMetric, DupeFileCollection> filesByMetrics = new Dictionary <AMetric, DupeFileCollection>();
foreach (var kv in metrics)
{
if (kv.Value == null)
{
continue;
}
DupeFileCollection dfc;
if (!filesByMetrics.TryGetValue(kv.Value, out dfc))
{
dfc = new DupeFileCollection(kv.Value);
filesByMetrics[kv.Value] = dfc;
}
dfc.Add(kv.Key);
}
DupeCollection dc = new DupeCollection();
foreach (var kv in filesByMetrics)
{
if (kv.Value.Count > 1)
{
dc.Add(kv.Value);
}
}
return(dc);
}
public DuplicateFinderResult FindDuplicates()
{
OnProgress?.Invoke(new BasicProgress(0, "Starting up..."));
SearchOption filesSearchOption = topDirOnly ? SearchOption.TopDirectoryOnly : SearchOption.AllDirectories;
int fileCount = 0;
//get the total count of files asynchronously so we can start collecting metrics using the enumerator
var fileCountTask = new Task(() =>
{
var getFiles = FileFinder.GetFiles(dir, AppSettings.Instance.extensionsToProcess, filesSearchOption);
fileCount = getFiles.Length;
});
fileCountTask.Start();
Dictionary <string, MetricDict> metricsPerGenerator = new Dictionary <string, MetricDict>();
foreach (var gen in generators) //should i make this a parallel foreach too? most of it is IO heavy, so only gain would be for saving and loading metrics
{
MetricDict fileMetrics = null;
if (deleteCache)
{
MetricCache.DeleteCache(dir, gen.ID);
}
else
{
fileMetrics = MetricCache.LoadMetrics(dir, gen.ID);
}
if (fileMetrics == null)
{
var metricGen = new FolderMetricGenerator(gen, dir, AppSettings.Instance.extensionsToProcess, filesSearchOption);
fileMetrics = metricGen.GenerateMetrics(this);
MetricCache.SaveMetrics(fileMetrics, dir, gen.ID);
}
metricsPerGenerator[gen.ID] = fileMetrics;
}
DuplicateFinderResult dfr = new DuplicateFinderResult();
foreach (var gen in generators)
{
DupeCollection dc = PotentialDuplicateFinder.FindDupes(metricsPerGenerator[gen.ID]);
dfr.dupesByGenerator[gen.ID] = dc;
}
OnProgress?.Invoke(new BasicProgress(1, "Done"));
return(dfr);
}
