private void Copy(DocBin docBin)
{
_docs = docBin._docs;
// I'd rather copy Python object no matter the serialization mode
// If set to DotNet, the variable will be initialized to null
// disregarding its current value which might be a default object
_pyDocBin = docBin._pyDocBin;
if (Serialization.Selected == Serialization.Mode.SpacyAndDotNet)
{
using (Py.GIL())
{
dynamic spacy = Py.Import("spacy");
dynamic pyVocab = spacy.vocab.Vocab.__call__();
dynamic pyDocs = _pyDocBin.get_docs(pyVocab);
dynamic builtins = Py.Import("builtins");
dynamic listDocs = builtins.list(pyDocs);
var pyCount = new PyInt(builtins.len(listDocs));
var count = pyCount.ToInt32();
for (var i = 0; i < count; i++)
{
dynamic pyDoc = listDocs[i];
_docs[i].PyDoc = pyDoc;
_docs[i].Vocab.PyVocab = pyDoc.vocab;
}
}
}
}
public void FromBytes(byte[] bytes)
{
if (Serialization.Selected == Serialization.Mode.Spacy)
{
using (Py.GIL())
{
var pyObj = ToPython.GetBytes(bytes);
_pyDocBin.from_bytes(pyObj);
}
}
else
{
var stream = new MemoryStream(bytes);
var settings = new XmlReaderSettings();
settings.IgnoreComments = true;
settings.IgnoreWhitespace = true;
var reader = XmlReader.Create(stream, settings);
var docBin = new DocBin();
docBin.ReadXml(reader);
Copy(docBin);
}
}
public void FromDisk(string pathFile)
{
if (Serialization.Selected == Serialization.Mode.Spacy)
{
using (Py.GIL())
{
var pyPath = new PyString(pathFile);
_pyDocBin.from_disk(pyPath);
}
}
else
{
using var stream = new FileStream(pathFile, FileMode.Open, FileAccess.Read);
var settings = new XmlReaderSettings();
settings.IgnoreComments = true;
settings.IgnoreWhitespace = true;
var reader = XmlReader.Create(stream, settings);
var docBin = new DocBin();
docBin.ReadXml(reader);
Copy(docBin);
}
}
