/// <summary>
/// Extracts words from a document content read with <paramref name="sourceReader"/> and returns them as an observable sequence
/// </summary>
/// <param name="sourceReader">The reader to read the document content from</param>
/// <returns>Observable sequence with a document words, one by one</returns>
/// <remarks>
/// Reads the content from the <paramref name="sourceReader"/> line by line and extracts words from each line
/// with the specified <see cref="ILineParser"/> instance.
/// </remarks>
public IObservable <string> ExtractWords([NotNull] TextReader sourceReader)
{
if (sourceReader == null)
{
throw new ArgumentNullException("sourceReader");
}
if (!UseAsync)
{
return(Observable.Create <string>(obs =>
{
var scheduler = Scheduler.Default;
var subscription = new SingleAssignmentDisposable();
subscription.Disposable = scheduler.Schedule(() =>
{
try
{
while (!subscription.IsDisposed)
{
var line = sourceReader.ReadLine();
if (line == null)
{
break;
}
foreach (var word in lineParser.ExtractWords(line))
{
obs.OnNext(word);
}
}
obs.OnCompleted();
}
catch (Exception e)
{
obs.OnError(e);
}
});
return subscription;
}));
}
else
{
// async version is much more slower
return(Observable.Create <string>(async(obs, cancel) =>
{
try
{
while (!cancel.IsCancellationRequested)
{
var line = await sourceReader.ReadLineAsync().ConfigureAwait(false);
if (line == null)
{
break;
}
foreach (var word in lineParser.ExtractWords(line))
{
obs.OnNext(word);
}
}
obs.OnCompleted();
}
catch (Exception e)
{
obs.OnError(e);
}
}).SubscribeOn(Scheduler.Default));
}
}
