public MorphAutomat(Lemmatizer lemmatizer,InternalMorphLanguage language, char annotChar)
: base(lemmatizer,language, annotChar) {
}
public FormInfo(Lemmatizer parent, AutomAnnotationInner a, string inputWordForm, bool found) {
_innerAnnot = a;
_parent = parent;
_found = found;
_inputWordBase = inputWordForm;
var m = FlexiaModel[a.ItemNo];
var flexLength = m.FlexiaStr.Length;
// It can be so( if CLemmatizer::PredictByDataBase was used) that
// the flexion is not suffix of m_InputWordBase, but only part of it.
// If so, then we cannot generate paradigm, since the current form cannot be
// divided into two parts: the base and a known flexion.
if (_found
|| ((_inputWordBase.Length >= flexLength)
&& (_inputWordBase.Substring(_inputWordBase.Length - flexLength) == m.FlexiaStr))) {
_flexiaWasCut = true;
var rempos = _inputWordBase.Length - m.FlexiaStr.Length;
if (rempos < _inputWordBase.Length) {
_inputWordBase = _inputWordBase.Remove(rempos);
}
} else {
_flexiaWasCut = false;
}
var lemmPrefix = _parent.Prefixes[_innerAnnot.PrefixNo];
if (_found
|| ((_inputWordBase.Substring(0, lemmPrefix.Length) == lemmPrefix)
&& (_inputWordBase.Substring(lemmPrefix.Length, m.PrefixStr.Length) == m.PrefixStr))) {
_inputWordBase = _inputWordBase.Remove(0, lemmPrefix.Length + m.PrefixStr.Length);
_prefixesWereCut = true;
} else {
_prefixesWereCut = false;
}
}
public void Load(Lemmatizer lemmatizer, string prefix, FileManager manager)
{
using (var file = manager.GetFile(lemmatizer.Registry, prefix + Constants.HomoweightBinPath, lemmatizer.CodePage)) {
Tools.LoadList(file, _homoWeights);
}
using (var file = manager.GetFile(lemmatizer.Registry, prefix + Constants.WordweightBinPath, lemmatizer.CodePage)) {
Tools.LoadList(file, _wordWeights);
}
}
public ABCEncoder(Lemmatizer lemmatizer,InternalMorphLanguage language, char annotChar) {
_lemmatizer = lemmatizer;
_language = language;
_annotChar=annotChar;
_alphabetSize = InitAlphabet(language, _code2Alphabet, _alphabet2Code, _annotChar);
_alphabetSizeWithoutAnnotator = InitAlphabet(language,_code2AlphabetWithoutAnnotator,_alphabet2CodeWithoutAnnotator,(char)257/* non-exeting symbol */);
if (_alphabetSizeWithoutAnnotator + 1 != _alphabetSize) {
throw new MorphException("_alphabetSizeWithoutAnnotator + 1 != _alphabetSize");
}
}
public ABCEncoder(Lemmatizer lemmatizer, InternalMorphLanguage language, char annotChar)
{
_lemmatizer = lemmatizer;
_language = language;
_annotChar = annotChar;
_alphabetSize = InitAlphabet(language, _code2Alphabet, _alphabet2Code, _annotChar);
_alphabetSizeWithoutAnnotator = InitAlphabet(language, _code2AlphabetWithoutAnnotator, _alphabet2CodeWithoutAnnotator, (char)257 /* non-exeting symbol */);
if (_alphabetSizeWithoutAnnotator + 1 != _alphabetSize)
{
throw new MorphException("_alphabetSizeWithoutAnnotator + 1 != _alphabetSize");
}
}
public FormInfo(Lemmatizer parent, AutomAnnotationInner a, string inputWordForm, bool found)
{
_innerAnnot = a;
_parent = parent;
_found = found;
_inputWordBase = inputWordForm;
var m = FlexiaModel[a.ItemNo];
var flexLength = m.FlexiaStr.Length;
_tools = new Tools();
// It can be so( if CLemmatizer::PredictByDataBase was used) that
// the flexion is not suffix of m_InputWordBase, but only part of it.
// If so, then we cannot generate paradigm, since the current form cannot be
// divided into two parts: the base and a known flexion.
if (_found ||
((_inputWordBase.Length >= flexLength) &&
(_inputWordBase.Substring(_inputWordBase.Length - flexLength) == m.FlexiaStr)))
{
_flexiaWasCut = true;
var rempos = _inputWordBase.Length - m.FlexiaStr.Length;
if (rempos < _inputWordBase.Length)
{
_inputWordBase = _inputWordBase.Remove(rempos);
}
}
else
{
_flexiaWasCut = false;
}
var lemmPrefix = _parent.Prefixes[_innerAnnot.PrefixNo];
if (_found ||
((_inputWordBase.Substring(0, lemmPrefix.Length) == lemmPrefix) &&
(_inputWordBase.Substring(lemmPrefix.Length, m.PrefixStr.Length) == m.PrefixStr)))
{
_inputWordBase = _inputWordBase.Remove(0, lemmPrefix.Length + m.PrefixStr.Length);
_prefixesWereCut = true;
}
else
{
_prefixesWereCut = false;
}
}
public void AttachLemmatizer(Lemmatizer parent) {
_parent = parent;
}
public PredictBase(Lemmatizer lemmatizer, InternalMorphLanguage lang)
{
_suffixAutomat = new MorphAutomat(lemmatizer, lang, Constants.MorphAnnotChar);
}
public MorphAutomat(Lemmatizer lemmatizer, InternalMorphLanguage language, char annotChar)
: base(lemmatizer, language, annotChar)
{
_tools = new Tools();
}
public PredictBase(Lemmatizer lemmatizer,InternalMorphLanguage lang) {
_suffixAutomat = new MorphAutomat(lemmatizer,lang, Constants.MorphAnnotChar);
}
public void AttachLemmatizer(Lemmatizer parent)
{
_parent = parent;
}