/// <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(MessageSensitivity.UserData, "Reusing cache '{0}'", _cacheFile);
}
else
{
ch.Info(MessageSensitivity.UserData, "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);
}
}
/// <summary>
/// Constructor.
/// </summary>
/// <param name="inputSchema">existing schema</param>
/// <param name="names">new columns</param>
/// <param name="types">corresponding types</param>
public TypeReplacementSchema(DataViewSchema inputSchema, string[] names, DataViewType[] types)
{
_schemaInput = new ExtendedSchema(inputSchema);
if (names == null || names.Length == 0)
{
throw Contracts.ExceptEmpty("The extended schema must contain new names.");
}
if (types == null || types.Length != names.Length)
{
throw Contracts.Except("names and types must have the same length.");
}
_types = new Dictionary <int, DataViewType>();
_mappedColumns = new Dictionary <int, int>();
Contracts.Assert(types.Length == names.Length);
int index;
for (int i = 0; i < names.Length; ++i)
{
index = SchemaHelper.GetColumnIndex(inputSchema, names[i]);
_types[index] = types[i];
_mappedColumns[inputSchema.Count + i] = index;
}
}
public static Delegate GetGetterChoice <T1, T2, T3>(DataViewRowCursor cur, DataViewSchema.Column col)
{
Delegate res = null;
try
{
res = cur.GetGetter <T1>(col);
if (res != null)
{
return(res);
}
}
catch (Exception)
{
}
try
{
res = cur.GetGetter <T2>(col);
if (res != null)
{
return(res);
}
}
catch (Exception)
{
}
try
{
res = cur.GetGetter <T3>(col);
if (res != null)
{
return(res);
}
}
catch (Exception)
{
}
if (res == null)
{
throw Contracts.ExceptNotImpl($"Unable to get a getter for column {col} of type {typeof(T1)} or {typeof(T2)} or {typeof(T3)} from schema\n{SchemaHelper.ToString(cur.Schema)}.");
}
return(res);
}
/// <summary>
/// Returns the getters for all columns.
/// </summary>
public static Delegate[] GetAllGetters(DataViewRowCursor cur)
{
var sch = cur.Schema;
var res = new List <Delegate>();
for (int i = 0; i < sch.Count; ++i)
{
if (sch[i].IsHidden)
{
continue;
}
var getter = GetColumnGetter(cur, SchemaHelper._dc(i, cur), sch);
if (getter == null)
{
throw Contracts.Except($"Unable to get getter for column {i} from schema\n{SchemaHelper.ToString(sch)}.");
}
res.Add(getter);
}
return(res.ToArray());
}
/// <summary>
/// Parses a schema and creates an ExtendedSchema for it.
/// </summary>
public static ExtendedSchema Parse(string schema, IChannel ch = null)
{
var cols = schema.Split(new char[] { ' ', ';' }, StringSplitOptions.RemoveEmptyEntries);
var tlcols = new List <TextLoader.Column>();
for (int i = 0; i < cols.Length; ++i)
{
if (cols[i].StartsWith("col="))
{
cols[i] = cols[i].Substring(4);
}
if (cols[i].Contains(":Vec<"))
{
var dot = cols[i].Split(':');
if (dot.Length != 3)
{
throw ch != null?ch.Except("Expects 3 parts in '{0}'", cols[i])
: Contracts.Except("Expects 3 parts in '{0}'", cols[i]);
}
if (!dot[1].StartsWith("Vec<"))
{
throw ch != null?ch.Except("Unable to parse '{0}'", cols[i])
: Contracts.Except("Unable to parse '{0}'", cols[i]);
}
if (!dot[1].EndsWith(">"))
{
throw ch != null?ch.Except("Unable to parse '{0}'", cols[i])
: Contracts.Except("Unable to parse '{0}'", cols[i]);
}
var temp = dot[1].Substring(4);
temp = temp.Substring(0, temp.Length - 1);
var splc = temp.Split(',');
if (splc.Length != 2)
{
throw ch != null?ch.Except("Unable to parse '{0}'", cols[i])
: Contracts.Except("Unable to parse '{0}'", cols[i]);
}
int last, nb;
try
{
last = int.Parse(dot[2]);
nb = int.Parse(splc[1]);
}
catch
{
throw ch != null?ch.Except("Unable to parse '{0}'", cols[i])
: Contracts.Except("Unable to parse '{0}'", cols[i]);
}
dot[1] = splc[0];
dot[2] = string.Format("{0}-{1}", last, last + nb - 1);
cols[i] = string.Format("{0}:{1}:{2}", dot[0], dot[1], dot[2]);
}
else if (cols[i].Contains(":Key<"))
{
var dot = cols[i].Split(':');
if (dot.Length == 4)
{
// GroupId:Key<U4, Min:0>:8
dot = new[] { dot[0], dot[1] + ":" + dot[2], dot[3] };
if (dot[1].Contains(",Min:0>"))
{
dot[1] = dot[1].Replace(",Min:0>", ",0-*>");
}
else
{
throw ch != null?ch.Except("Expects 3 parts in '{0}'", cols[i])
: Contracts.Except("Expects 3 parts in '{0}'", cols[i]);
}
}
if (dot.Length != 3)
{
throw ch != null?ch.Except("Expects 3 parts in '{0}'", cols[i])
: Contracts.Except("Expects 3 parts in '{0}'", cols[i]);
}
if (!dot[1].StartsWith("Key<"))
{
throw ch != null?ch.Except("Unable to parse '{0}'", cols[i])
: Contracts.Except("Unable to parse '{0}'", cols[i]);
}
if (!dot[1].EndsWith(">"))
{
throw ch != null?ch.Except("Unable to parse '{0}'", cols[i])
: Contracts.Except("Unable to parse '{0}'", cols[i]);
}
var temp = dot[1].Substring(4);
temp = temp.Substring(0, temp.Length - 1);
var splc = temp.Split(',');
if (splc.Length != 2)
{
throw ch != null?ch.Except("Unable to parse '{0}'", cols[i])
: Contracts.Except("Unable to parse '{0}'", cols[i]);
}
dot[1] = string.Format("{0}[{1}]", splc[0], splc[1]);
cols[i] = string.Format("{0}:{1}:{2}", dot[0], dot[1], dot[2]);
}
var t = TextLoader.Column.Parse(cols[i]);
if (t == null)
{
t = TextLoader.Column.Parse(string.Format("{0}:{1}", cols[i], i));
}
if (t == null)
{
throw ch != null?ch.Except("Unable to parse '{0}' or '{1}'", cols[i], string.Format("{0}:{1}", cols[i], i))
: Contracts.Except("Unable to parse '{0}' or '{1}'", cols[i], string.Format("{0}:{1}", cols[i], i));
}
tlcols.Add(t);
}
return(new ExtendedSchema(null, tlcols.Select(c => c.Name).ToArray(),
tlcols.Select(c => SchemaHelper.Convert(c, ch)).ToArray()));
}
void Init(DataKind destKind)
{
_kind = SchemaHelper.GetColumnType <TLabel>().RawKind();
_destKind = destKind;
mapperBL = null;
mapperU1 = null;
mapperU2 = null;
mapperU4 = null;
mapperI4 = null;
mapperR4 = null;
mapperFromBL = null;
mapperFromU1 = null;
mapperFromU2 = null;
mapperFromU4 = null;
mapperFromI4 = null;
mapperFromR4 = null;
switch (destKind)
{
case DataKind.Boolean:
mapperBL = SchemaHelper.GetConverter <TLabel, bool>(out identity);
mapperFromBL = SchemaHelper.GetConverter <bool, TLabel>(out identity);
break;
case DataKind.SByte:
mapperU1 = SchemaHelper.GetConverter <TLabel, byte>(out identity);
mapperFromU1 = SchemaHelper.GetConverter <byte, TLabel>(out identity);
break;
case DataKind.UInt16:
mapperU2 = SchemaHelper.GetConverter <TLabel, ushort>(out identity);
mapperFromU2 = SchemaHelper.GetConverter <ushort, TLabel>(out identity);
break;
case DataKind.UInt32:
mapperU4 = SchemaHelper.GetConverter <TLabel, uint>(out identity);
mapperFromU4 = SchemaHelper.GetConverter <uint, TLabel>(out identity);
break;
case DataKind.Int32:
var temp = SchemaHelper.GetConverter <TLabel, float>(out identity);
mapperI4 = (in TLabel src, ref int dst) =>
{
float v = 0f;
temp(in src, ref v);
dst = (int)v;
};
var temp2 = SchemaHelper.GetConverter <float, TLabel>(out identity);
mapperFromI4 = (in int src, ref TLabel dst) =>
{
float v = (float)src;
temp2(in v, ref dst);
};
break;
case DataKind.Single:
mapperR4 = SchemaHelper.GetConverter <TLabel, float>(out identity);
mapperFromR4 = SchemaHelper.GetConverter <float, TLabel>(out identity);
break;
default:
throw Contracts.ExceptNotSupp("Unsupported kinds {0} --> {1}", _kind, _destKind);
}
}
public TypedConverters()
{
Init(SchemaHelper.GetColumnType <TLabel>().RawKind());
}
/// <summary>
/// Constructor.
/// </summary>
/// <param name="inputSchema">existing schema</param>
/// <param name="names">new columns</param>
/// <param name="types">corresponding types</param>
public TypeReplacementSchema(ISchema inputSchema, string[] names, DataViewType[] types)
{
_schemaInput = inputSchema;
if (names == null || names.Length == 0)
{
throw Contracts.ExceptEmpty("The extended schema must contain new names.");
}
if (types == null || types.Length != names.Length)
{
throw Contracts.Except("names and types must have the same length.");
}
_types = new Dictionary <int, DataViewType>();
_mappedColumns = new Dictionary <int, int>();
Contracts.Assert(types.Length == names.Length);
int index;
for (int i = 0; i < names.Length; ++i)
{
if (!inputSchema.TryGetColumnIndex(names[i], out index))
{
throw Contracts.Except("Unable to find column '{0}' in '{1}'", names[i], SchemaHelper.ToString(inputSchema));
}
_types[index] = types[i];
_mappedColumns[inputSchema.ColumnCount + i] = index;
}
}
