public void ParseHonkingBigFile()
{
Stopwatch sw = new Stopwatch();
using (var stream = File.OpenRead(@"c:\temp\test1.csv"))
{
sw.Restart();
long count = 0;
var settings = this.GetDelimitedParserSettings(
new[] { "field1", "field2" },
',');
var parser = DelimitedParser.Create(settings, stream);
//foreach (var record in parser.Parse(stream))
//{
// count += record.LongCount();
//}
//var a = parser.Parse().Last();
sw.Stop();
var elapsed = sw.Elapsed;
var s = $"elapsed={elapsed}, count={count}, throughput={(stream.Length / (1024 * 1024)) / elapsed.TotalSeconds} MB/s";
Console.WriteLine(s);
}
}
private DelimitedParser CreateDelimitedParser(
string input,
IEnumerable <string> delimitedColumnStrings,
char delimiter)
{
var stream = input.ToStream();
var settings = this.GetDelimitedParserSettings(delimitedColumnStrings, delimiter);
DelimitedParser parser = DelimitedParser.Create(settings, stream);
return(parser);
}
public void SqlBulkCopy_DelimitedParser()
{
string path = Path.Combine(
Directory.GetCurrentDirectory(),
Path.GetRandomFileName());
path = Path.ChangeExtension(path, "mdf");
// Use LocalDB as the temporary database for the test. Substitute with a
// connection string more appropriate to your environment.
string connectionString = $@"Server=(localdb)\MSSQLLocalDB;AttachDbFilename={path};Trusted_Connection=True";
// Build up some fake Cosmos output.
StringBuilder sb = new StringBuilder();
sb.AppendLine("1,http://support.microsoft.com/kb/100,100,0.01,0.25,0.5,0.75,2.1,5.01");
sb.AppendLine("1,http://support.microsoft.com/kb/101,500,0.01,0.25,0.5,0.75,2.1,5.01");
sb.AppendLine("1,http://support.microsoft.com/kb/200,700,0.01,0.25,0.5,0.75,2.1,5.01");
sb.AppendLine("1,http://support.microsoft.com/kb/321,902,0.01,0.25,0.5,0.75,2.1,5.01");
sb.AppendLine("1,http://support.microsoft.com/kb/732,199,0.01,0.25,0.5,0.75,2.1,5.01");
sb.AppendLine("1,http://support.microsoft.com/kb/376,112,0.01,0.25,0.5,0.75,2.1,5.01");
sb.AppendLine("1,http://support.microsoft.com/kb/546,414,0.01,0.25,0.5,0.75,2.1,5.01");
// Read in the fake Cosmos output. Instead of using this stream, why not use
// -> new CosmosStream("http://cosmos05.osdinfra.net:88/cosmos/0365exp.adhoc/my/stream.csv");
var factory = new DelimitedColumnFactory();
var delimitedHeader = new DelimitedHeader(
factory.Create(new[] { "ID:long", "Uri", "Count:long", "Min:double", "P25:double", "P50:double", "P75:double", "P99:double", "Max:double" }));
var settings = new DelimitedParserSettings();
settings.DelimitedHeader = delimitedHeader;
settings.Delimiter = ',';
var parser = DelimitedParser.Create(settings, sb.ToStream());
// Ensure the database is created before bulk loading the data.
using (var context = new Context(connectionString))
{
Assert.False(context.WebMetrics.Any());
}
// Bulk load the data.
using (var transactionScope = new TransactionScope(
TransactionScopeOption.Required,
new TransactionOptions {
IsolationLevel = IsolationLevel.ReadCommitted
}))
using (SqlBulkCopy sqlBulkCopy = new SqlBulkCopy(connectionString))
{
sqlBulkCopy.DestinationTableName = "WebMetrics";
sqlBulkCopy.BatchSize = 10000;
sqlBulkCopy.WriteToServer(parser);
transactionScope.Complete();
}
// Read data back out of the DB...party on it.
using (var context = new Context(connectionString))
{
Assert.Equal(7, context.WebMetrics.Count());
// There are seven unique IDs.
var set = new HashSet <long>(context.WebMetrics.Select(x => x.ID));
Assert.Equal(7, set.Count);
Assert.True(context.WebMetrics.All(x => x.Min == 0.01));
Assert.True(context.WebMetrics.All(x => x.Max == 5.01));
}
}