public IEnumerable<Volume> GetVolumes()
{
using (var stream = Open()) {
var parser = new VolumeParser { Log = Log };
Date created;
using (var reader = parser.Open(stream, out created))
return parser.GetVolumes(reader).ToArray();
}
}
public void Convert(CatalogParser parser, XmlReader reader, Date created,
Stream bookStream, Stream volumeStream)
{
// There are almost 42,000 books and 520,000 files in the project Gutenberg. We set
// up 59 * 10,000 steps here to have the progress bar stretching quite good and still
// having a little space to grow for the project content.
var progress = Log.Action(59, "Converting catalog...");
int bookCount = 0, skippedCount = 0, volumeCount = 0, totalCount = 0;
var bookParser = new BookParser { Log = Log };
var volumeParser = new VolumeParser { Log = Log };
using (var bookWriter = bookParser.Create(bookStream, created))
using (var volumeWriter = volumeParser.Create(volumeStream, created)) {
foreach (var item in parser.GetItems(reader)) {
var book = item as Book;
if (book != null) {
// Book 9140 and a couple of others are invalid.
if (string.IsNullOrEmpty(book.Title)) {
Log.Verbose("Skipping book {0} with no title.", book.Number);
++skippedCount;
} else {
bookParser.Write(bookWriter, book);
}
++bookCount;
} else {
volumeParser.Write(volumeWriter, (Volume) item);
++volumeCount;
}
// One step after every 10,000 items run nicely on my laptop; less loaded the
// CPU with constant reporting and more made the console wait for too long.
if (++totalCount % 10000 == 0)
progress.Continue("{0} items processed...", totalCount);
}
bookParser.Finalize(bookWriter);
volumeParser.Finalize(volumeWriter);
}
Log.Verbose("{0} books and {1} volumes were processed, {2} books skipped.",
bookCount, volumeCount, skippedCount);
progress.Finish();
}
