async Task Incremental(LoggedConnection db, string table, StageTableCfg t, DateTime latestTs)
{
await DbStore.Optimise(Cfg.Optimise, t.Dir, latestTs.FileSafeTimestamp(), db.Log); // optimise files newer than the last load
var((_, rows, size), dur) = await CopyInto(db, table, t).WithDuration();
db.Log.Information("StageUpdate - {Table} incremental load of {Rows} rows ({Size}) took {Duration}",
table, rows, size.Humanize("#.#"), dur.HumanizeShort());
}
async Task FullLoad(LoggedConnection db, string table, StageTableCfg t)
{
if (t.IsNativeStore)
{
await DbStore.Optimise(Cfg.Optimise, t.Dir, null, db.Log); // optimise all files when performing a full load
}
await db.Execute("truncate table", $"truncate table {table}"); // no transaction, stage tables aren't reported on so don't need to be available
var((_, rows, size), dur) = await CopyInto(db, table, t).WithDuration();
db.Log.Information("StageUpdate - {Table} full load of {Rows} rows ({Size}) took {Duration}",
table, rows, size.Humanize("#.#"), dur.HumanizeShort());
}
AzureBlobFileStore Store(StageTableCfg t) => Stores.Store(t.StoreType);
async Task <(string[] files, long rows, ByteSize size)> CopyInto(ILoggedConnection <IDbConnection> db, string table, StageTableCfg t)
{
var startTime = await db.ExecuteScalar <string>("current time", "select current_timestamp()::string");
var(stage, path) = t.StoreType switch {
DataStoreType.Db => (Cfg.Stage, StorageCfg.DbPath),
DataStoreType.Private => (Cfg.Private, null),
_ => throw new InvalidOperationException($"No warehouse stage for store type {t.StoreType}")
};
var sql = $"copy into {table} from @{new[] {stage, path}.Concat(t.Dir.Tokens).NotNull().Join(" / ")}/ file_format=(type=json)";
await db.Execute("copy into", sql);
// sf should return this info form copy_into (its in their UI, but not in .net or jdbc drivers)
// the int that is return is the # of rows form the first file loaded. So we go get this ourselves
var copyResults = await db.Query <(string fileName, long rows, long size)>("copy results",
"select file_name, row_count, file_size " +
$"from table(information_schema.copy_history(table_name=>'{table}', start_time=>'{startTime}'::timestamp_ltz))");
var res = (copyResults.Select(r => r.fileName).ToArray(), copyResults.Sum(r => r.rows), copyResults.Sum(r => r.size).Bytes());
return(res);
}
}
async Task <(string[] files, long rows, ByteSize size)> CopyInto(LoggedConnection db, string table, StageTableCfg t)
{
var startTime = await db.ExecuteScalar <string>("current time", "select current_timestamp()::string");
var sql = $"copy into {table} from @{Cfg.Stage}/{StorageCfg.DbPath}/{t.Dir}/ file_format=(type=json)";
await db.Execute("copy into", sql);
// sf should return this info form copy_into (its in their UI, but not in .net or jdbc drivers)
// the int that is return is the # of rows form the first file loaded. So we go get this ourselves
var copyResults = await db.Query <(string fileName, long rows, long size)>("copy results",
"select file_name, row_count, file_size " +
$"from table(information_schema.copy_history(table_name=>'{table}', start_time=>'{startTime}'::timestamp_ltz))");
var res = (copyResults.Select(r => r.fileName).ToArray(), copyResults.Sum(r => r.rows), copyResults.Sum(r => r.size).Bytes());
return(res);
}
