private void FillList <T>(ICollection <T> list, QueryDefinition qd, EntityDefinition ed) where T : IEntity
{
var sql = GetSqlHelper.GetSql(Settings.Current.Building.SourceEngine.Database,
qd.GetSql(Settings.Current.Building.Vendor, Settings.Current.Building.SourceSchema), Settings.Current.Building.SourceSchema);
if (string.IsNullOrEmpty(sql))
{
return;
}
var keys = new Dictionary <string, bool>();
using (var connection = new OdbcConnection(Settings.Current.Building.SourceConnectionString))
{
connection.Open();
using (var c = new OdbcCommand(sql, connection))
{
c.CommandTimeout = 30000;
using (var reader = c.ExecuteReader())
{
while (reader.Read())
{
Concept conceptDef = null;
if (ed.Concepts != null && ed.Concepts.Any())
{
conceptDef = ed.Concepts[0];
}
var concept = (T)ed.GetConcepts(conceptDef, reader, null).ToList()[0];
var key = concept.GetKey();
if (key == null)
{
continue;
}
if (keys.ContainsKey(key))
{
continue;
}
keys.Add(key, false);
list.Add(concept);
if (CurrentState != BuilderState.Running)
{
break;
}
}
}
}
}
}
public IEnumerable <List <KeyValuePair <string, string> > > GetPersonKeys(long batches, int batchSize)
{
var batch = new List <KeyValuePair <string, string> >(batchSize);
var query = GetSqlHelper.GetSql(Settings.Current.Building.SourceEngine.Database, Settings.Current.Building.BatchScript, Settings.Current.Building.SourceSchemaName);
foreach (var reader in dbSource.GetPersonKeys(query, batches, batchSize))
{
if (batch.Count == batchSize)
{
yield return(batch);
batch.Clear();
}
var id = reader[0].ToString().Trim();
var source = reader[1].ToString().Trim(); // ? replace to reader[1].ToString(), without Trim()
batch.Add(new KeyValuePair <string, string>(id, source));
}
yield return(batch);
}
public IEnumerable <List <KeyValuePair <string, string> > > GetPersonKeys(long batches, int batchSize)
{
var batch = new List <KeyValuePair <string, string> >(batchSize);
var query = GetSqlHelper.GetSql(_settings.SourceEngine.Database, _settings.BatchScript, _settings.ConversionSettings.SourceSchema);
foreach (var reader in _dbSource.GetPersonKeys(query, batches, batchSize))
{
if (batch.Count == batchSize)
{
yield return(batch);
batch.Clear();
}
var id = reader[0].ToString().Trim();
var source = reader[1].ToString().Trim();
batch.Add(new KeyValuePair <string, string>(id, source));
}
yield return(batch);
}
public KeyValuePair <string, Exception> Load(IDatabaseEngine sourceEngine, string sourceSchemaName, List <QueryDefinition> sourceQueryDefinitions, OdbcConnection sourceConnection, string vendor)
{
var fileName = string.Empty;
var query = string.Empty;
var connectionString = string.Empty;
try
{
var timer = new Stopwatch();
timer.Start();
foreach (var qd in sourceQueryDefinitions)
{
if (qd.Providers != null)
{
continue;
}
if (qd.Locations != null)
{
continue;
}
if (qd.CareSites != null)
{
continue;
}
fileName = qd.FileName;
var sql = GetSqlHelper.GetSql(sourceEngine.Database,
qd.GetSql(vendor, sourceSchemaName),
sourceSchemaName);
if (string.IsNullOrEmpty(sql))
{
continue;
}
var q = string.Format(sql, ChunkId);
using (var cdm = sourceEngine.GetCommand(q, sourceConnection))
{
cdm.CommandTimeout = 30000;
using (var reader =
sourceEngine.ReadChunkData(sourceConnection, cdm, qd, ChunkId,
Prefix))
{
while (reader.Read())
{
PopulateData(qd, reader);
}
}
}
}
timer.Stop();
}
catch (Exception e)
{
var info = new StringBuilder();
info.AppendLine("SourceEngine=" + sourceEngine);
info.AppendLine("SourceConnectionString=" + connectionString);
info.AppendLine("File name=" + fileName);
info.AppendLine("Query:");
info.AppendLine(query);
return(new KeyValuePair <string, Exception>(info.ToString(), e));
}
return(new KeyValuePair <string, Exception>(null, null));
}
public string GetSql(Database sourceDatabase, Vendors vendor, string schemaName)
{
return(GetSqlHelper.GetSql(sourceDatabase, GetSql(vendor, schemaName), schemaName));
}
private IEnumerable <string> GetParts()
{
var folder = $"{Settings.Settings.Current.Building.Vendor}/{Settings.Settings.Current.Building.Id}/raw";
var fileName = string.Empty;
foreach (var qd in Settings.Settings.Current.Building.SourceQueryDefinitions)
{
//var sql = qd.GetSql(Settings.Current.Building.SourceEngine.Database,
// Settings.Current.Building.Vendor, Settings.Current.Building.SourceSchemaName);
var sql = GetSqlHelper.GetSql(Settings.Settings.Current.Building.SourceEngine.Database,
qd.GetSql(Settings.Settings.Current.Building.Vendor, Settings.Settings.Current.Building.SourceSchemaName), Settings.Settings.Current.Building.SourceSchemaName);
if (qd.Persons == null)
{
continue;
}
if (string.IsNullOrEmpty(sql))
{
continue;
}
fileName = qd.FileName;
break;
}
if (string.IsNullOrEmpty(fileName))
{
fileName = Settings.Settings.Current.Building.SourceQueryDefinitions[0].FileName;
}
using (var client = new AmazonS3Client(Settings.Settings.Current.S3AwsAccessKeyId, Settings.Settings.Current.S3AwsSecretAccessKey, Amazon.RegionEndpoint.USEast1))
{
var request = new ListObjectsV2Request
{
BucketName = Settings.Settings.Current.Bucket,
Prefix = $"{folder}/{_chunkId}/{fileName}/{fileName}"
};
ListObjectsV2Response response;
do
{
var responseTask = client.ListObjectsV2Async(request);
responseTask.Wait();
response = responseTask.Result;
var partIndexes = new HashSet <string>();
var fn = fileName;
foreach (var entry in response.S3Objects)
{
var end = entry.Key.LastIndexOf(fn, StringComparison.InvariantCultureIgnoreCase) + fn.Length;
var key = entry.Key.Replace(entry.Key.Substring(0, end), "");
key = key.Substring(0, key.IndexOf('_'));
partIndexes.Add(key);
}
foreach (var partIndex in partIndexes)
{
yield return(partIndex);
}
request.ContinuationToken = response.NextContinuationToken;
} while (response.IsTruncated);
}
}
public void Process()
{
try
{
var dbChunk = new DbChunk(Settings.Settings.Current.Building.BuilderConnectionString);
var timer = new Stopwatch();
timer.Start();
var folder = $"{Settings.Settings.Current.Building.Vendor}/{Settings.Settings.Current.Building.Id}/raw";
Parallel.ForEach(Settings.Settings.Current.Building.SourceQueryDefinitions, qd =>
{
if (qd.Providers != null)
{
return;
}
if (qd.Locations != null)
{
return;
}
if (qd.CareSites != null)
{
return;
}
//var sql = qd.GetSql(Settings.Current.Building.SourceEngine.Database,
// Settings.Current.Building.Vendor, Settings.Current.Building.SourceSchemaName);
var sql = GetSqlHelper.GetSql(Settings.Settings.Current.Building.SourceEngine.Database,
qd.GetSql(Settings.Settings.Current.Building.Vendor, Settings.Settings.Current.Building.SourceSchemaName), Settings.Settings.Current.Building.SourceSchemaName);
if (string.IsNullOrEmpty(sql))
{
return;
}
qd.FieldHeaders = new Dictionary <string, int>(StringComparer.OrdinalIgnoreCase);
var metadataKey = $"{folder}/metadata/{qd.FileName + ".txt"}";
using (var client = new AmazonS3Client(Settings.Settings.Current.S3AwsAccessKeyId, Settings.Settings.Current.S3AwsSecretAccessKey, Amazon.RegionEndpoint.USEast1))
using (var stream = new MemoryStream())
using (var sr = new StreamReader(stream))
{
var request = new GetObjectRequest {
BucketName = Settings.Settings.Current.Bucket, Key = metadataKey
};
var getObject = client.GetObjectAsync(request);
getObject.Wait();
using (var response = getObject.Result.ResponseStream)
{
response.CopyTo(stream);
}
stream.Position = 0;
var index = 0;
foreach (var fieldName in sr.ReadLine().Split(new[] { ',' }, StringSplitOptions.RemoveEmptyEntries))
{
try
{
qd.FieldHeaders.Add(fieldName, index);
index++;
}
catch (Exception)
{
throw new Exception("[RestoreMetadataFromS3] fieldName duplication: " + fieldName + " - " + qd.FileName);
}
}
}
});
Parallel.ForEach(GetParts(), new ParallelOptions {
MaxDegreeOfParallelism = 2
}, p =>
{
Logger.Write(_chunkId, LogMessageTypes.Info, "load part=" + p);
var part = new DatabaseChunkPart(_chunkId, _createPersonBuilder, p, 0);
LoadPart(part, p);
part.Build();
SavePart(part, p);
});
Logger.Write(_chunkId, LogMessageTypes.Info,
$"Loaded - {timer.ElapsedMilliseconds} ms | {(GC.GetTotalMemory(false) / 1024f) / 1024f} Mb");
dbChunk.ChunkComplete(_chunkId, Settings.Settings.Current.Building.Id.Value);
}
catch (Exception e)
{
Logger.WriteError(_chunkId, e);
throw;
}
}
public void Load()
{
var timer = new Stopwatch();
timer.Start();
//Parallel.ForEach(Settings.Current.Building.SourceQueryDefinitions, new ParallelOptions { MaxDegreeOfParallelism = 1 }, qd =>
//Parallel.ForEach(Settings.Current.Building.SourceQueryDefinitions, qd =>
foreach (var qd in Settings.Settings.Current.Building.SourceQueryDefinitions)
{
if (qd.Providers != null)
{
continue;
}
if (qd.Locations != null)
{
continue;
}
if (qd.CareSites != null)
{
continue;
}
//var sql = qd.GetSql(Settings.Current.Building.SourceEngine.Database,
//Settings.Current.Building.Vendor, Settings.Current.Building.SourceSchemaName);
var sql = GetSqlHelper.GetSql(Settings.Settings.Current.Building.SourceEngine.Database,
qd.GetSql(Settings.Settings.Current.Building.Vendor, Settings.Settings.Current.Building.SourceSchemaName), Settings.Settings.Current.Building.SourceSchemaName);
if (string.IsNullOrEmpty(sql))
{
continue;
}
if (Settings.Settings.Current.Building.SourceEngine.Database != Database.Redshift)
{
var q = string.Format(sql, ChunkId);
using (var conn = Settings.Settings.Current.Building.SourceEngine.GetConnection(Settings.Settings.Current.Building.SourceConnectionString))
using (var cdm = Settings.Settings.Current.Building.SourceEngine.GetCommand(q, conn))
using (var reader = Settings.Settings.Current.Building.SourceEngine.ReadChunkData(conn, cdm, qd, ChunkId, Prefix))
{
while (reader.Read())
{
PopulateData(qd, reader);
}
}
}
else
{
using (var reader = Settings.Settings.Current.Building.SourceEngine.ReadChunkData(null, null, qd, ChunkId, Prefix))
{
while (reader.Read())
{
PopulateData(qd, reader);
}
}
}
}
timer.Stop();
Logger.Write(ChunkId, LogMessageTypes.Info,
string.Format(Prefix + ") loaded - {0} ms | {1} Mb", timer.ElapsedMilliseconds,
(GC.GetTotalMemory(false) / 1024f) / 1024f));
}
private void MoveChunkDataToS3()
{
var chunkIds = _dbChunk.GetNotMovedToS3Chunks(Settings.Settings.Current.Building.Id.Value).ToArray();
if (chunkIds.Length == 0)
{
return;
}
var baseFolder =
$"{Settings.Settings.Current.Bucket}/{Settings.Settings.Current.Building.Vendor}/{Settings.Settings.Current.Building.Id}/raw";
Console.WriteLine("S3 raw folder - " + baseFolder);
Parallel.ForEach(Settings.Settings.Current.Building.SourceQueryDefinitions, queryDefinition =>
{
if (queryDefinition.Providers != null)
{
return;
}
if (queryDefinition.Locations != null)
{
return;
}
if (queryDefinition.CareSites != null)
{
return;
}
var sql = GetSqlHelper.GetSql(Settings.Settings.Current.Building.SourceEngine.Database,
queryDefinition.GetSql(Settings.Settings.Current.Building.Vendor,
Settings.Settings.Current.Building.SourceSchemaName),
Settings.Settings.Current.Building.SourceSchemaName);
if (string.IsNullOrEmpty(sql))
{
return;
}
sql = string.Format(sql, chunkIds[0]);
if (queryDefinition.FieldHeaders == null)
{
StoreMetadataToS3(queryDefinition, sql);
}
});
Parallel.ForEach(chunkIds, new ParallelOptions {
MaxDegreeOfParallelism = 2
}, cId =>
{
var chunkId = cId;
Parallel.ForEach(Settings.Settings.Current.Building.SourceQueryDefinitions,
new ParallelOptions {
MaxDegreeOfParallelism = 5
}, queryDefinition =>
{
try
{
if (queryDefinition.Providers != null)
{
return;
}
if (queryDefinition.Locations != null)
{
return;
}
if (queryDefinition.CareSites != null)
{
return;
}
var sql = GetSqlHelper.GetSql(Settings.Settings.Current.Building.SourceEngine.Database,
queryDefinition.GetSql(Settings.Settings.Current.Building.Vendor,
Settings.Settings.Current.Building.SourceSchemaName),
Settings.Settings.Current.Building.SourceSchemaName);
if (string.IsNullOrEmpty(sql))
{
return;
}
sql = string.Format(sql, chunkId);
//if (queryDefinition.FieldHeaders == null)
//{
// StoreMetadataToS3(queryDefinition, sql);
//}
var personIdField = queryDefinition.GetPersonIdFieldName();
var tmpTableName = "#" + queryDefinition.FileName + "_" + chunkId;
var folder = $"{baseFolder}/{chunkId}/{queryDefinition.FileName}";
var fileName = $@"{folder}/{queryDefinition.FileName}";
var unloadQuery = string.Format(@"create table {0} sortkey ({1}) distkey ({1}) as {2}; " +
@"UNLOAD ('select * from {0} order by {1}') to 's3://{3}' " +
@"DELIMITER AS '\t' " +
@"credentials 'aws_access_key_id={4};aws_secret_access_key={5}' " +
@"GZIP ALLOWOVERWRITE PARALLEL ON",
tmpTableName, //0
personIdField, //1
sql, //2
fileName, //3
Settings.Settings.Current.S3AwsAccessKeyId, //4
Settings.Settings.Current.S3AwsSecretAccessKey); //5
using (var connection =
SqlConnectionHelper.OpenOdbcConnection(Settings.Settings.Current.Building
.SourceConnectionString))
using (var c = new OdbcCommand(unloadQuery, connection))
{
c.CommandTimeout = 999999999;
c.ExecuteNonQuery();
}
}
catch (Exception e)
{
Logger.WriteError(chunkId, e);
throw;
}
});
_dbChunk.ChunkCreated(chunkId, Settings.Settings.Current.Building.Id.Value);
Console.WriteLine("Raw data for chunkId=" + chunkId + " is available on S3");
});
}