public Doc(Vocab vocab)
{
_vocab = vocab;
using (Py.GIL())
{
dynamic spacy = Py.Import("spacy");
dynamic pyVocab = vocab.PyVocab;
PyDoc = spacy.tokens.doc.Doc.__call__(pyVocab);
}
}
private void Copy(Doc doc)
{
// 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
PyDoc = doc.PyDoc;
_text = doc._text;
_vocab = doc._vocab;
_tokens = doc._tokens;
_sentences = doc._sentences;
_nounChunks = doc._nounChunks;
_ents = doc._ents;
}
internal Doc(dynamic doc, string text)
{
PyDoc = doc;
_vocab = null;
_text = text;
}
internal Doc(dynamic doc)
{
PyDoc = doc;
_vocab = null;
}
public void ReadXml(XmlReader reader)
{
var serializationMode = Serialization.Selected;
reader.MoveToContent();
Debug.Assert(reader.Name == $"{Serialization.Prefix}:Doc");
reader.ReadStartElement();
if (serializationMode == Serialization.Mode.SpacyAndDotNet)
{
Debug.Assert(reader.Name == $"{Serialization.Prefix}:PyObj");
var bytesB64 = reader.ReadElementContentAsString();
var bytes = Convert.FromBase64String(bytesB64);
using (Py.GIL())
{
dynamic spacy = Py.Import("spacy");
dynamic pyVocab = spacy.vocab.Vocab.__call__();
PyDoc = spacy.tokens.doc.Doc.__call__(pyVocab);
var pyBytes = ToPython.GetBytes(bytes);
PyDoc.from_bytes(pyBytes);
_vocab = new Vocab(PyDoc.vocab);
}
}
Debug.Assert(Serialization.Selected != Serialization.Mode.Spacy);
Debug.Assert(reader.Name == $"{Serialization.Prefix}:Text");
_text = reader.ReadElementContentAsString();
Debug.Assert(reader.Name == $"{Serialization.Prefix}:Vocab");
_vocab = new Vocab(null);
_vocab.ReadXml(reader);
Debug.Assert(reader.Name == $"{Serialization.Prefix}:Tokens");
_tokens = new List <Token>();
var isEmpty = reader.IsEmptyElement;
reader.ReadStartElement();
if (!isEmpty)
{
while (reader.MoveToContent() != XmlNodeType.EndElement)
{
Debug.Assert(reader.Name == $"{Serialization.Prefix}:Token");
reader.ReadStartElement();
if (reader.NodeType != XmlNodeType.EndElement)
{
var token = new Token();
token.ReadXml(reader);
_tokens.Add(token);
reader.ReadEndElement();
}
}
reader.ReadEndElement();
}
foreach (var token in _tokens)
{
token.RestoreHead(_tokens);
}
Debug.Assert(reader.Name == $"{Serialization.Prefix}:Sentences");
_sentences = new List <Span>();
isEmpty = reader.IsEmptyElement;
reader.ReadStartElement();
if (!isEmpty)
{
while (reader.MoveToContent() != XmlNodeType.EndElement)
{
Debug.Assert(reader.Name == $"{Serialization.Prefix}:Sent");
reader.ReadStartElement();
if (reader.NodeType != XmlNodeType.EndElement)
{
var sent = new Span();
sent.ReadXml(reader);
_sentences.Add(sent);
reader.ReadEndElement();
}
}
reader.ReadEndElement();
}
Debug.Assert(reader.Name == $"{Serialization.Prefix}:NounChunks");
_nounChunks = new List <Span>();
isEmpty = reader.IsEmptyElement;
reader.ReadStartElement();
if (!isEmpty)
{
while (reader.MoveToContent() != XmlNodeType.EndElement)
{
Debug.Assert(reader.Name == $"{Serialization.Prefix}:NounChunk");
reader.ReadStartElement();
if (reader.NodeType != XmlNodeType.EndElement)
{
var nChunk = new Span();
nChunk.ReadXml(reader);
_nounChunks.Add(nChunk);
reader.ReadEndElement();
}
}
reader.ReadEndElement();
}
Debug.Assert(reader.Name == $"{Serialization.Prefix}:Ents");
_ents = new List <Span>();
reader.ReadStartElement();
while (reader.MoveToContent() != XmlNodeType.EndElement)
{
Debug.Assert(reader.Name == $"{Serialization.Prefix}:Ent");
reader.ReadStartElement();
if (reader.NodeType != XmlNodeType.EndElement)
{
var ent = new Span();
ent.ReadXml(reader);
_ents.Add(ent);
reader.ReadEndElement();
}
}
reader.ReadEndElement();
}
public List <Doc> GetDocs(Vocab vocab)
{
return(Interop.GetListFromGenerator(_pyDocBin?.get_docs(vocab.PyVocab), ref _docs));
}
