public SortedDictionary <string, long> Compare(
DownloadData oldData,
DownloadData newData)
{
if (newData.Count == 0)
{
throw new InvalidOperationException("The new data should not be empty.");
}
var stopwatch = Stopwatch.StartNew();
// We use a very simplistic algorithm here. Find the union of both ID sets and compare each download count.
var uniqueIds = new HashSet <string>(
oldData.Keys.Concat(newData.Keys),
StringComparer.OrdinalIgnoreCase);
_logger.LogInformation(
"There are {OldCount} IDs in the old data, {NewCount} IDs in the new data, and {TotalCount} IDs in total.",
oldData.Count,
newData.Count,
uniqueIds.Count);
var result = new SortedDictionary <string, long>(StringComparer.OrdinalIgnoreCase);
var decreaseCount = 0;
foreach (var id in uniqueIds)
{
// Detect download count decreases and emit a metric. This is not necessarily wrong because there have
// been times that we manually delete spoofed download counts.
DetectDownloadCountDecreases(oldData, newData, id, ref decreaseCount);
var oldCount = oldData.GetDownloadCount(id);
var newCount = newData.GetDownloadCount(id);
if (oldCount != newCount)
{
result.Add(id, newCount);
}
}
_logger.LogInformation("There are {Count} package IDs with download count changes.", result.Count);
_logger.LogInformation("There are {Count} package versions with download count decreases.", decreaseCount);
if (decreaseCount > _options.Value.MaxDownloadCountDecreases)
{
throw new InvalidOperationException("Too many download count decreases are occurring.");
}
stopwatch.Stop();
_telemetryService.TrackDownloadSetComparison(oldData.Count, newData.Count, result.Count, stopwatch.Elapsed);
return(result);
}
