private SortInDataFrameTransform(IHost host, ModelLoadContext ctx, IDataView input) : base(host, input)
{
Host.CheckValue(input, "input");
Host.CheckValue(ctx, "ctx");
_sortColumn = ctx.Reader.ReadString();
Host.AssertValue(_sortColumn);
int index;
var schema = input.Schema;
if (!schema.TryGetColumnIndex(_sortColumn, out index))
{
Contracts.Check(false, "sortColumn not found in input schema.");
}
var type = schema.GetColumnType(index);
Host.Check(!type.IsVector(), "sortColumn cannot be a vector.");
_reverse = ctx.Reader.ReadBoolean();
_numThreads = ctx.Reader.ReadInt32();
if (_numThreads < 0)
{
_numThreads = null;
}
_transform = CreateTemplatedTransform();
}
public ExtendedCacheTransform(IHostEnvironment env, Arguments args, IDataView input)
: base(env, RegistrationName, input)
{
Host.CheckValue(args, "args");
Host.CheckUserArg(args.inDataFrame || !string.IsNullOrEmpty(args.cacheFile), "cacheFile cannot be empty if inDataFrame is false.");
Host.CheckUserArg(!args.async || args.inDataFrame, "inDataFrame must be true if async is true.");
Host.CheckUserArg(!args.numTheads.HasValue || args.numTheads > 0, "numThread must be > 0 if specified.");
var saverSettings = args.saverSettings as ICommandLineComponentFactory;
Host.CheckValue(saverSettings, nameof(saverSettings));
_saverSettings = string.Format("{0}{{{1}}}", saverSettings.Name, saverSettings.GetSettingsString());
_saverSettings = _saverSettings.Replace("{}", "");
if (!_saverSettings.ToLower().StartsWith("binary"))
{
throw env.ExceptNotSupp("Only binary format is supported.");
}
_inDataFrame = args.inDataFrame;
_cacheFile = args.cacheFile;
_reuse = args.reuse;
_async = args.async;
_numThreads = args.numTheads;
var saver = ComponentCreation.CreateSaver(Host, _saverSettings);
if (saver == null)
{
throw Host.Except("Cannot parse '{0}'", _saverSettings);
}
_pipedTransform = CreatePipeline(env, input);
}
public SortInDataFrameState(IHostEnvironment host, IDataView input, int sortColumn, bool reverse, int?numThreads)
{
_host = host.Register("SortInDataFrameState");
_host.CheckValue(input, "input");
_source = input;
_reverse = reverse;
_lock = new object();
_autoView = null;
_canShuffle = sortColumn < 0;
_numThreads = numThreads;
_sortColumn = sortColumn;
}
/// <summary>
/// Creation of the pipeline knowing parameters _inDataFrame, _cacheFile, _reuse.
/// </summary>
protected IDataTransform CreatePipeline(IHostEnvironment env, IDataView input)
{
if (_inDataFrame)
{
if (_async)
{
var view = new CacheDataView(env, input, null);
var tr = new PassThroughTransform(env, new PassThroughTransform.Arguments(), view);
return(tr);
}
else
{
var args = new SortInDataFrameTransform.Arguments()
{
numThreads = _numThreads, sortColumn = null
};
var tr = new SortInDataFrameTransform(env, args, input);
return(tr);
}
}
else
{
string nt = _numThreads > 0 ? string.Format("{{t={0}}}", _numThreads) : string.Empty;
using (var ch = Host.Start("Caching data..."))
{
if (_reuse && File.Exists(_cacheFile))
{
ch.Info("Reusing cache '{0}'", _cacheFile);
}
else
{
ch.Info("Building cache '{0}'", _cacheFile);
var saver = ComponentCreation.CreateSaver(env, _saverSettings);
using (var fs0 = Host.CreateOutputFile(_cacheFile))
DataSaverUtils.SaveDataView(ch, saver, input, fs0, true);
}
}
var loader = ComponentCreation.CreateLoader(env, string.Format("binary{{{0}}}", nt),
new MultiFileSource(_cacheFile));
SchemaHelper.CheckSchema(Host, input.Schema, loader.Schema);
var copy = ComponentCreation.CreateTransform(env, "skip{s=0}", loader);
return(copy);
}
}
public SortInDataFrameTransform(IHostEnvironment env, Arguments args, IDataView input)
: base(env, RegistrationName, input)
{
Host.CheckValue(args, "args");
Host.CheckUserArg(!args.numThreads.HasValue || args.numThreads.Value > 0, "numThreads cannot be negative.");
if (!string.IsNullOrEmpty(args.sortColumn))
{
int index;
var schema = input.Schema;
if (!schema.TryGetColumnIndex(args.sortColumn, out index))
{
Contracts.Check(false, "sortColumn not found in input schema.");
}
var type = schema.GetColumnType(index);
Host.Check(!type.IsVector(), "sortColumn cannot be a vector.");
}
_reverse = args.reverse;
_sortColumn = args.sortColumn;
_numThreads = args.numThreads;
_transform = CreateTemplatedTransform();
}
private ExtendedCacheTransform(IHost host, ModelLoadContext ctx, IDataView input) :
base(host, input)
{
Host.CheckValue(input, "input");
Host.CheckValue(ctx, "ctx");
_inDataFrame = ctx.Reader.ReadBoolean();
_async = ctx.Reader.ReadBoolean();
_numThreads = ctx.Reader.ReadInt32();
host.Check(_numThreads > -2, "_numThreads");
if (_numThreads < 0)
{
_numThreads = null;
}
_saverSettings = ctx.Reader.ReadString();
if (_inDataFrame)
{
_cacheFile = null;
_reuse = false;
}
else
{
_cacheFile = ctx.Reader.ReadString();
_reuse = ctx.Reader.ReadBoolean();
host.CheckValue(_cacheFile, "_cacheFile");
}
var saver = ComponentCreation.CreateSaver(Host, _saverSettings);
if (saver == null)
{
throw Host.Except("Cannot parse '{0}'", _saverSettings);
}
_pipedTransform = CreatePipeline(host, input);
}
public static SortInDataFrameTransform Create(IHostEnvironment env, ModelLoadContext ctx, IDataView input)
{
Contracts.CheckValue(env, "env");
var h = env.Register(RegistrationName);
h.CheckValue(ctx, "ctx");
h.CheckValue(input, "input");
ctx.CheckAtModel(GetVersionInfo());
return(h.Apply("Loading Model", ch => new SortInDataFrameTransform(h, ctx, input)));
}