private async Task <bool> PushIndexChangesAsync()
{
// The "old" data in this case is the download count data that was last indexed by this job (or
// initialized by Db2AzureSearch).
_logger.LogInformation("Fetching old download count data from blob storage.");
var oldResult = await _downloadDataClient.ReadLatestIndexedAsync(
AccessConditionWrapper.GenerateEmptyCondition(),
_stringCache);
// The "new" data in this case is from the statistics pipeline.
_logger.LogInformation("Fetching new download count data from blob storage.");
var newData = await _auxiliaryFileClient.LoadDownloadDataAsync();
_logger.LogInformation("Removing invalid IDs and versions from the old data.");
CleanDownloadData(oldResult.Data);
_logger.LogInformation("Removing invalid IDs and versions from the new data.");
CleanDownloadData(newData);
// Fetch the download overrides from the auxiliary file. Note that the overriden downloads are kept
// separate from downloads data as the original data will be persisted to auxiliary data, whereas the
// overriden data will be persisted to Azure Search.
_logger.LogInformation("Overriding download count data.");
var downloadOverrides = await _auxiliaryFileClient.LoadDownloadOverridesAsync();
var overridenDownloads = newData.ApplyDownloadOverrides(downloadOverrides, _logger);
_logger.LogInformation("Detecting download count changes.");
var changes = _downloadSetComparer.Compare(oldResult.Data, overridenDownloads);
var idBag = new ConcurrentBag <string>(changes.Keys);
_logger.LogInformation("{Count} package IDs have download count changes.", idBag.Count);
if (!changes.Any())
{
return(false);
}
_logger.LogInformation(
"Starting {Count} workers pushing download count changes to Azure Search.",
_options.Value.MaxConcurrentBatches);
await ParallelAsync.Repeat(
() => WorkAsync(idBag, changes),
_options.Value.MaxConcurrentBatches);
_logger.LogInformation("All of the download count changes have been pushed to Azure Search.");
_logger.LogInformation("Uploading the new download count data to blob storage.");
await _downloadDataClient.ReplaceLatestIndexedAsync(newData, oldResult.Metadata.GetIfMatchCondition());
return(true);
}
private async Task <bool> PushIndexChangesAsync()
{
// The "old" data in this case is the download count data that was last indexed by this job (or
// initialized by Db2AzureSearch).
_logger.LogInformation("Fetching old download count data from blob storage.");
var oldResult = await _downloadDataClient.ReadLatestIndexedAsync(
AccessConditionWrapper.GenerateEmptyCondition(),
_stringCache);
// The "new" data in this case is from the statistics pipeline.
_logger.LogInformation("Fetching new download count data from blob storage.");
var newData = await _auxiliaryFileClient.LoadDownloadDataAsync();
_logger.LogInformation("Removing invalid IDs and versions from the old downloads data.");
CleanDownloadData(oldResult.Data);
_logger.LogInformation("Removing invalid IDs and versions from the new downloads data.");
CleanDownloadData(newData);
_logger.LogInformation("Detecting download count changes.");
var changes = _downloadSetComparer.Compare(oldResult.Data, newData);
_logger.LogInformation("{Count} package IDs have download count changes.", changes.Count);
// The "old" data is the popularity transfers data that was last indexed by this job (or
// initialized by Db2AzureSearch).
_logger.LogInformation("Fetching old popularity transfer data from blob storage.");
var oldTransfers = await _popularityTransferDataClient.ReadLatestIndexedAsync(
AccessConditionWrapper.GenerateEmptyCondition(),
_stringCache);
// The "new" data is the latest popularity transfers data from the database.
_logger.LogInformation("Fetching new popularity transfer data from database.");
var newTransfers = await GetPopularityTransfersAsync();
_logger.LogInformation("Applying download transfers to download changes.");
ApplyDownloadTransfers(
newData,
oldTransfers.Data,
newTransfers,
changes);
var idBag = new ConcurrentBag <string>(changes.Keys);
_logger.LogInformation("{Count} package IDs need to be updated.", idBag.Count);
if (!changes.Any())
{
return(false);
}
_logger.LogInformation(
"Starting {Count} workers pushing download count changes to Azure Search.",
_options.Value.MaxConcurrentBatches);
await ParallelAsync.Repeat(
() => WorkAndRetryAsync(idBag, changes),
_options.Value.MaxConcurrentBatches);
_logger.LogInformation("All of the download count changes have been pushed to Azure Search.");
_logger.LogInformation("Uploading the new download count data to blob storage.");
await _downloadDataClient.ReplaceLatestIndexedAsync(newData, oldResult.Metadata.GetIfMatchCondition());
_logger.LogInformation("Uploading the new popularity transfer data to blob storage.");
await _popularityTransferDataClient.ReplaceLatestIndexedAsync(
newTransfers,
oldTransfers.Metadata.GetIfMatchCondition());
return(true);
}
