internal static INgramExtractorFactory Create(IHostEnvironment env, NgramHashExtractorArguments extractorArgs,
TermLoaderArguments termLoaderArgs)
{
Contracts.CheckValue(env, nameof(env));
var h = env.Register(LoaderSignature);
h.CheckValue(extractorArgs, nameof(extractorArgs));
h.CheckParam(termLoaderArgs == null, nameof(termLoaderArgs), "Argument cannot be used with NgramHashExtractor, use NgramExtractor instead");
return(new NgramHashExtractorFactory(extractorArgs));
}
public static INgramExtractorFactory Create(IHostEnvironment env, NgramHashExtractorArguments extractorArgs,
TermLoaderArguments termLoaderArgs)
{
Contracts.CheckValue(env, nameof(env));
var h = env.Register(LoaderSignature);
h.CheckValue(extractorArgs, nameof(extractorArgs));
h.CheckValueOrNull(termLoaderArgs);
return(new NgramHashExtractorFactory(extractorArgs, termLoaderArgs));
}
public static IDataTransform Create(NgramHashExtractorArguments extractorArgs, IHostEnvironment env, IDataView input,
ExtractorColumn[] cols, TermLoaderArguments termLoaderArgs = null)
{
Contracts.CheckValue(env, nameof(env));
var h = env.Register(LoaderSignature);
h.CheckValue(extractorArgs, nameof(extractorArgs));
h.CheckValue(input, nameof(input));
h.CheckUserArg(extractorArgs.SkipLength < extractorArgs.NgramLength, nameof(extractorArgs.SkipLength), "Should be less than " + nameof(extractorArgs.NgramLength));
h.CheckUserArg(Utils.Size(cols) > 0, nameof(Arguments.Column), "Must be specified");
h.AssertValueOrNull(termLoaderArgs);
var extractorCols = new Column[cols.Length];
for (int i = 0; i < cols.Length; i++)
{
extractorCols[i] =
new Column
{
Name = cols[i].Name,
Source = cols[i].Source,
FriendlyNames = cols[i].FriendlyNames
};
}
var args = new Arguments
{
Column = extractorCols,
NgramLength = extractorArgs.NgramLength,
SkipLength = extractorArgs.SkipLength,
HashBits = extractorArgs.HashBits,
InvertHash = extractorArgs.InvertHash,
Ordered = extractorArgs.Ordered,
Seed = extractorArgs.Seed,
AllLengths = extractorArgs.AllLengths
};
return(Create(h, args, input, termLoaderArgs));
}
internal static ITransformer Create(NgramHashExtractorArguments extractorArgs, IHostEnvironment env, IDataView input,
ExtractorColumn[] cols)
{
Contracts.CheckValue(env, nameof(env));
var h = env.Register(LoaderSignature);
h.CheckValue(extractorArgs, nameof(extractorArgs));
h.CheckValue(input, nameof(input));
h.CheckUserArg(extractorArgs.SkipLength < extractorArgs.NgramLength, nameof(extractorArgs.SkipLength), "Should be less than " + nameof(extractorArgs.NgramLength));
h.CheckUserArg(Utils.Size(cols) > 0, nameof(Options.Columns), "Must be specified");
var extractorCols = new Column[cols.Length];
for (int i = 0; i < cols.Length; i++)
{
extractorCols[i] =
new Column
{
Name = cols[i].Name,
Source = cols[i].Source,
FriendlyNames = cols[i].FriendlyNames
};
}
var options = new Options
{
Columns = extractorCols,
NgramLength = extractorArgs.NgramLength,
SkipLength = extractorArgs.SkipLength,
NumberOfBits = extractorArgs.NumberOfBits,
MaximumNumberOfInverts = extractorArgs.MaximumNumberOfInverts,
Ordered = extractorArgs.Ordered,
Seed = extractorArgs.Seed,
UseAllLengths = extractorArgs.UseAllLengths
};
return(Create(h, options, input));
}
