private static Stopwatch ParseDisks(Action<Disk> addToBatch)
{
int i = 0;
var parser = new Parser();
var buffer = new byte[1024 * 1024];// more than big enough for all files
var sp = Stopwatch.StartNew();
using (var bz2 = new BZip2InputStream(File.Open(@"D:\Data\freedb-complete-20120101.tar.bz2", FileMode.Open)))
using (var tar = new TarInputStream(bz2))
{
TarEntry entry;
while ((entry = tar.GetNextEntry()) != null)
{
if (entry.Size == 0 || entry.Name == "README" || entry.Name == "COPYING")
continue;
var readSoFar = 0;
while (true)
{
var read = tar.Read(buffer, readSoFar, ((int)entry.Size) - readSoFar);
if (read == 0)
break;
readSoFar += read;
}
// we do it in this fashion to have the stream reader detect the BOM / unicode / other stuff
// so we can read the values properly
var fileText = new StreamReader(new MemoryStream(buffer, 0, readSoFar)).ReadToEnd();
try
{
var disk = parser.Parse(fileText);
addToBatch(disk);
if (i++ % 1000 == 0)
Console.Write("\r{0} {1:#,#} {2} ", entry.Name, i, sp.Elapsed);
if (i % 50000 == 0)
return sp;
}
catch (Exception e)
{
Console.WriteLine();
Console.WriteLine(entry.Name);
Console.WriteLine(e);
return sp;
}
}
}
return sp;
}
private void ParseEntries()
{
var parser = new Parser();
while (true)
{
var entry = _entries.Take();
if (entry == null)
break;
var disk = parser.Parse(entry);
_disks.Add(disk);
Interlocked.Increment(ref parsed);
}
_disks.Add(null);
}
private static void ParseDisks(BulkInsertOperation insert)
{
int i = 0;
var parser = new Parser();
var buffer = new byte[1024*1024];// more than big enough for all files
using (var bz2 = new BZip2InputStream(File.Open(@"D:\Scratch\freedb-complete-20150101.tar.bz2", FileMode.Open)))
using (var tar = new TarInputStream(bz2))
{
TarEntry entry;
while((entry=tar.GetNextEntry()) != null)
{
if(entry.Size == 0 || entry.Name == "README" || entry.Name == "COPYING")
continue;
var readSoFar = 0;
while(true)
{
var read = tar.Read(buffer, readSoFar, ((int) entry.Size) - readSoFar);
if (read == 0)
break;
readSoFar += read;
}
// we do it in this fashion to have the stream reader detect the BOM / unicode / other stuff
// so we can read the values properly
var fileText = new StreamReader(new MemoryStream(buffer,0, readSoFar)).ReadToEnd();
try
{
var disk = parser.Parse(fileText);
insert.Store(disk);
}
catch (Exception e)
{
Console.WriteLine();
Console.WriteLine(entry.Name);
Console.WriteLine(e);
}
}
}
}