private void DownloadAzureData()
{
string containerPrefix = null;
string tablePrefix = null;
string clusterId = DetermineClusterId();
if (!Config.FileType.Equals(FileTypesEnum.any))
{
containerPrefix = FileTypes.MapFileTypeUriPrefix(Config.FileType);
tablePrefix = containerPrefix + clusterId.Replace("-", "");
}
if (!string.IsNullOrEmpty(clusterId))
{
// 's-' in prefix may not always be correct
containerPrefix += "s-" + clusterId;
}
if (Config.FileType == FileTypesEnum.table)
{
TableManager tableMgr = new TableManager()
{
IngestCallback = (exportedFile) => { QueueForIngest(exportedFile); }
};
if (tableMgr.Connect())
{
tableMgr.DownloadTables(tablePrefix);
}
}
else
{
BlobManager blobMgr = new BlobManager()
{
IngestCallback = (sourceFileUri) => { QueueForIngest(sourceFileUri); },
ReturnSourceFileLink = Config.IsKustoConfigured() & Config.KustoUseBlobAsSource
};
if (blobMgr.Connect())
{
blobMgr.DownloadContainers(containerPrefix);
}
}
}
private void QueueBlobSegmentDownload(BlobResultSegment blobResultSegment)
{
int parentId = Thread.CurrentThread.ManagedThreadId;
Log.Debug($"enter. current id:{parentId}. results count: {blobResultSegment.Results.Count()}");
long segmentMinDateTicks = Interlocked.Read(ref DiscoveredMinDateTicks);
long segmentMaxDateTicks = Interlocked.Read(ref DiscoveredMaxDateTicks);
foreach (IListBlobItem blob in blobResultSegment.Results)
{
ICloudBlob blobRef = null;
Log.Debug($"parent id:{parentId} current Id:{Thread.CurrentThread.ManagedThreadId}");
if (blob is CloudBlobDirectory)
{
if (!string.IsNullOrEmpty(Config.NodeFilter) && !Regex.IsMatch(blob.Uri.ToString(), Config.NodeFilter, RegexOptions.IgnoreCase))
{
Log.Debug($"blob:{blob.Uri} does not match nodeFilter pattern:{Config.NodeFilter}, skipping...");
continue;
}
DownloadBlobsFromDirectory(blob as CloudBlobDirectory);
Log.Debug("blob is directory.");
continue;
}
Interlocked.Increment(ref TotalFilesEnumerated);
if (Regex.IsMatch(blob.Uri.ToString(), FileFilterPattern, RegexOptions.IgnoreCase))
{
long ticks = Convert.ToInt64(Regex.Match(blob.Uri.ToString(), FileFilterPattern, RegexOptions.IgnoreCase).Groups[1].Value);
if (ticks <Config.StartTimeUtc.Ticks | ticks> Config.EndTimeUtc.Ticks)
{
Interlocked.Increment(ref TotalFilesSkipped);
Log.Debug($"exclude:bloburi file ticks {new DateTime(ticks).ToString("o")} outside of time range:{blob.Uri}");
SetMinMaxDate(ref segmentMinDateTicks, ref segmentMaxDateTicks, ticks);
continue;
}
}
else
{
Log.Debug($"regex not matched: {blob.Uri.ToString()} pattern: {FileFilterPattern}");
}
try
{
Log.Debug($"file Blob: {blob.Uri}");
blobRef = blob.Container.ServiceClient.GetBlobReferenceFromServerAsync(blob.Uri).Result;
}
catch (StorageException se)
{
Interlocked.Increment(ref TotalErrors);
Log.Exception($"getting ref for {blob.Uri}, skipping. {se.Message}");
continue;
}
if (blobRef.Properties.LastModified.HasValue)
{
DateTimeOffset lastModified = blobRef.Properties.LastModified.Value;
SetMinMaxDate(ref segmentMinDateTicks, ref segmentMaxDateTicks, lastModified.Ticks);
if (!string.IsNullOrEmpty(Config.UriFilter) && !Regex.IsMatch(blob.Uri.ToString(), Config.UriFilter, RegexOptions.IgnoreCase))
{
Interlocked.Increment(ref TotalFilesSkipped);
Log.Debug($"blob:{blob.Uri} does not match uriFilter pattern:{Config.UriFilter}, skipping...");
continue;
}
if (Config.FileType != FileTypesEnum.any &&
!FileTypes.MapFileTypeUri(blob.Uri.AbsolutePath).Equals(Config.FileType))
{
Interlocked.Increment(ref TotalFilesSkipped);
Log.Debug($"skipping uri with incorrect file type: {FileTypes.MapFileTypeUri(blob.Uri.AbsolutePath)}");
continue;
}
if (lastModified >= Config.StartTimeUtc && lastModified <= Config.EndTimeUtc)
{
Interlocked.Increment(ref TotalFilesMatched);
if (Config.List)
{
Log.Info($"listing file with timestamp: {lastModified}\r\n file: {blob.Uri.AbsolutePath}");
continue;
}
if (ReturnSourceFileLink)
{
IngestCallback?.Invoke(new FileObject(blob.Uri.AbsolutePath, Config.SasEndpointInfo.BlobEndpoint)
{
Length = blobRef.Properties.Length,
LastModified = lastModified
});
continue;
}
FileObject fileObject = new FileObject(blob.Uri.AbsolutePath, Config.CacheLocation)
{
Length = blobRef.Properties.Length,
LastModified = lastModified
};
Log.Info($"queueing blob with timestamp: {lastModified}\r\n file: {blob.Uri.AbsolutePath}");
InvokeCallback(blob, fileObject);
}
else
{
Interlocked.Increment(ref TotalFilesSkipped);
Log.Debug($"exclude:bloburi {lastModified.ToString("o")} outside of time range:{blob.Uri}");
SetMinMaxDate(ref segmentMinDateTicks, ref segmentMaxDateTicks, lastModified.Ticks);
continue;
}
}
else
{
Log.Error("unable to read blob modified date", blobRef);
TotalErrors++;
}
}
}
private void QueueBlobSegmentDownload(BlobResultSegment blobResultSegment)
{
int parentId = Thread.CurrentThread.ManagedThreadId;
Log.Debug($"enter. current id:{parentId}. results count: {blobResultSegment.Results.Count()}");
foreach (var blob in blobResultSegment.Results)
{
ICloudBlob blobRef = null;
Log.Debug($"parent id:{parentId} current Id:{Thread.CurrentThread.ManagedThreadId}");
if (blob is CloudBlobDirectory)
{
if (!string.IsNullOrEmpty(Config.NodeFilter) && !Regex.IsMatch(blob.Uri.ToString(), Config.NodeFilter, RegexOptions.IgnoreCase))
{
Log.Warning($"blob:{blob.Uri} does not match nodeFilter pattern:{Config.NodeFilter}, skipping...", ConsoleColor.Yellow);
continue;
}
DownloadBlobsFromDirectory(blob as CloudBlobDirectory);
Log.Debug("blob is directory.");
continue;
}
Interlocked.Increment(ref TotalFilesEnumerated);
if (!string.IsNullOrEmpty(Config.UriFilter) && !Regex.IsMatch(blob.Uri.ToString(), Config.UriFilter, RegexOptions.IgnoreCase))
{
Interlocked.Increment(ref TotalFilesSkipped);
Log.Warning($"blob:{blob.Uri} does not match uriFilter pattern:{Config.UriFilter}, skipping...", ConsoleColor.Yellow);
continue;
}
if (Regex.IsMatch(blob.Uri.ToString(), FileFilterPattern, RegexOptions.IgnoreCase))
{
long ticks = Convert.ToInt64(Regex.Match(blob.Uri.ToString(), FileFilterPattern, RegexOptions.IgnoreCase).Groups[1].Value);
if (ticks <Config.StartTimeUtc.Ticks | ticks> Config.EndTimeUtc.Ticks)
{
Interlocked.Increment(ref TotalFilesSkipped);
Log.Debug($"exclude:bloburi ticks outside of time range:{blob.Uri}");
continue;
}
}
try
{
Log.Debug($"file Blob: {blob.Uri}");
blobRef = blob.Container.ServiceClient.GetBlobReferenceFromServerAsync(blob.Uri).Result;
}
catch (StorageException se)
{
Interlocked.Increment(ref TotalErrors);
Log.Exception($"getting ref for {blob.Uri}, skipping. {se.Message}");
continue;
}
if (blobRef.Properties.LastModified.HasValue)
{
DateTimeOffset lastModified = blobRef.Properties.LastModified.Value;
if (Config.FileType != FileTypesEnum.any && !FileTypes.MapFileTypeUri(blob.Uri.AbsolutePath).Equals(Config.FileType))
{
Interlocked.Increment(ref TotalFilesSkipped);
Log.Debug($"skipping uri with incorrect file type: {FileTypes.MapFileTypeUri(blob.Uri.AbsolutePath)}");
continue;
}
if (lastModified >= Config.StartTimeUtc && lastModified <= Config.EndTimeUtc)
{
Interlocked.Increment(ref TotalFilesMatched);
if (Config.List)
{
Log.Info($"listing file with timestamp: {lastModified}\r\n file: {blob.Uri.AbsolutePath}");
continue;
}
if (ReturnSourceFileLink)
{
IngestCallback?.Invoke(new FileObject(blob.Uri.AbsolutePath, Config.SasEndpointInfo.BlobEndpoint)
{
Length = blobRef.Properties.Length
});
continue;
}
FileObject fileObject = new FileObject(blob.Uri.AbsolutePath, Config.CacheLocation);
Log.Info($"queueing blob with timestamp: {lastModified}\r\n file: {blob.Uri.AbsolutePath}");
if (!fileObject.Exists)
{
fileObject.DownloadAction = () =>
{
((CloudBlockBlob)blob).DownloadToStreamAsync(fileObject.Stream.Get(), null,
new BlobRequestOptions()
{
RetryPolicy = new IngestRetryPolicy(),
ParallelOperationThreadCount = Config.Threads
}, null).Wait();
};
IngestCallback?.Invoke(fileObject);
Interlocked.Increment(ref TotalFilesDownloaded);
}
else
{
Log.Warning($"destination file exists. skipping download:\r\n file: {fileObject}");
IngestCallback?.Invoke(fileObject);
}
}
}
else
{
Log.Error("unable to read blob modified date", blobRef);
TotalErrors++;
}
}
}
