public bool TranslateSentence(ref string sentence, bool force = false)
{
bool result = force || CanTranslateSentence(sentence);
if (result)
{
sentence = Sentences[sentence]();
}
else
{
List <string> words = sentence.GetWords().ToList();
result = CanTranslateWords(words);
if (result)
{
List <string> translatedWords = new List <string>(words);
TranslateWords(translatedWords, true);
StringBuilder translatedSentence = new StringBuilder(sentence);
words.ForEach((w, i) => translatedSentence.Replace(w, translatedWords[i]));
sentence = translatedSentence.ToString();
}
}
return(result);
}
public override string ToString()
{
string result = "";
Sentences.ToList().ForEach((x) => result += x.ToString());
return(result);
}
public void CopyTo(ISentence[] array, int arrayIndex)
{
if (array != null && array.Length > 0)
{
Sentences.CopyTo(array, arrayIndex);
}
}
private void CalculateDictionary()
{
if (Dictionary == null)
{
List <Token> allWords = new List <Token>();
Sentences.ForEach(s =>
{
allWords.AddRange(s.Words);
});
Features = allWords.Where(w => w.IsAlpha).Select(x => x.Lemma).Distinct().OrderBy(x => x).ToList();
Dictionary = new List <Tuple <string, int> >();
allWords.Select(x => x.Lemma)
.Distinct()
.OrderBy(x => x)
.ToList()
.ForEach(word =>
{
Dictionary.Add(new Tuple <string, int>(word, allWords.Count(x => x.Lemma == word)));
});
}
}
public void Add(ISentence sentence)
{
if (sentence != null)
{
Sentences.Add(sentence);
}
}
private void fillSentenseArray()
{
lineArray = new Sentences[tokenList[tokenList.Count - 1].Column];
int i = 0;
int currentColumn = 1;
string errorMessage = null;
while (tokenList.Count - 1 >= i)
{
List <Word> lineList = new List <Word>();
for (; i < tokenList.Count && tokenList[i].Column == currentColumn; i++)
{
lineList.Add(new Word(tokenList[i]));
if (tokenList[i].GetTypeWord() == typeArr[12])
{
errorMessage = "error type of one of operands";
}
}
if (errorMessage != null)
{
lineArray[currentColumn - 1] = new Sentences(lineList, currentColumn, errorMessage);
currentColumn++;
errorMessage = null;
continue;
}
lineArray[currentColumn - 1] = new Sentences(lineList, currentColumn);
currentColumn++;
}
for (int c = 0; c < lineArray.Length; c++)
{
lineArray[c].printLine();
}
}
private void AnalyseText()
{
IsBusy = true;
Sentences.Clear();
var text = Text.Replace("\r\n", "\r");
// 去除行首的发言人
{
var reg = new Regex(@"^.*:", RegexOptions.Multiline);
text = reg.Replace(text, "");
}
var fkf = new[] { '.', '\"', '\r', '?', '!' };
var cs = text.Split(fkf);
foreach (var l in cs)
{
var s = l.Trim();
if (string.IsNullOrEmpty(s))
{
continue;
}
Sentences.Add(new SentenceData
{
Sentence = l.Trim()
});
}
InitRange();
RaisePropertyChanged("CanSave");
FindSign();
MakeUtfText();
IsBusy = false;
}
public string ToHtmlVersion()
{
switch (Type)
{
case SectionType.PaperTitle:
return($"<p style =\"font-weight: bold; font-size: 20px\">{Sentences[0].ToStringVersion()}</p>");
case SectionType.SectionTitle:
return($"<p style =\"font-weight: bold; font-size: 16px\">{Sentences[0].ToStringVersion()}</p>");
case SectionType.Text:
return($"<p style =\"font-size: 14px\">{string.Join(" ", Sentences.Select(x => x.ToStringVersion()))}</p>");
case SectionType.ReferencesList:
var sb = new StringBuilder();
foreach (var reference in References)
{
string referedToString, referedToStyle, oldSource;
referedToString = reference.ReferedTo ? "Есть ссылка в статье" : "Нет ссылки в статье";
referedToStyle = reference.ReferedTo ? "style=\"color: green;\"" : "style=\"color: red;\"";
oldSource = reference.Year != 0 && reference.Year < 1990 ? "<span style=\"color: red;\">Устаревший источник</span>" : "";
sb.Append($"<span>{reference.Original.Original}</span> <span {referedToStyle}>{referedToString}</span> {oldSource}\n");
}
return($"<p style =\"font-size: 14px\">{sb.ToString()}</p>");
default:
return("Что то не так");
}
}
public Text(IEnumerable <ISentence> sentences) : this()
{
foreach (var sentence in sentences)
{
Sentences.Add(sentence);
}
}
public void AddSentence(int sentence)
{
if (!Sentences.Contains(sentence))
{
Sentences.Add(sentence);
}
}
public TokenizedText(GraphemeString graphemeString, TokenRange tokenRange)
{
GraphemeString = graphemeString;
TokenRange = tokenRange;
for (TokenNode start = tokenRange.Start, node = tokenRange.Start; ;)
{
if (node.Value.Type == TokenType.FullStop || node.Value.Type == TokenType.QuestionMark || node.Value.Type == TokenType.ExclamationPoint)
{
for (node = node.Next; node.Value.Type == TokenType.FullStop; node = node.Next)
{
;
}
Sentences.Add(new TokenizedSentence(new TokenRange(start, node)));
start = node;
continue;
}
if (node.Value.Type == TokenType.Eof)
{
if (start != node)
{
Sentences.Add(new TokenizedSentence(new TokenRange(start, node)));
}
break;
}
node = node.Next;
}
}
protected override async Task <string[]> GetPostAsync()
{
StringBuilder str = new StringBuilder();
if (_firstTurn)
{
str.AppendLine(Sentences.ReversiIntro(GetGuildId(), GetTurnName()));
}
_firstTurn = false;
str.AppendLine("```");
str.AppendLine(" A B C D E F G H");
str.AppendLine(" ┌─┬─┬─┬─┬─┬─┬─┬─┐");
for (int i = 0; i < 8; i++)
{
str.Append((i + 1).ToString() + "│");
for (int y = 0; y < 8; y++)
{
str.Append(_board[i, y]);
str.Append("│");
}
str.AppendLine();
if (i != 7)
{
str.AppendLine(" ├─┼─┼─┼─┼─┼─┼─┼─┤");
}
}
str.AppendLine(" └─┴─┴─┴─┴─┴─┴─┴─┘");
str.AppendLine("```");
return(new[] { str.ToString() });
}
public void Write(Sentences text)
{
hasAlreadyWritten = true;
StopAllCoroutines();
currentText = text;
StartWriting();
}
public Sentences(Sentences other) : this(udpipe_csharpPINVOKE.new_Sentences__SWIG_1(Sentences.getCPtr(other)), true)
{
if (udpipe_csharpPINVOKE.SWIGPendingException.Pending)
{
throw udpipe_csharpPINVOKE.SWIGPendingException.Retrieve();
}
}
public SentencesEnumerator(Sentences collection)
{
collectionRef = collection;
currentIndex = -1;
currentObject = null;
currentSize = collectionRef.Count;
}
//문장 분해
public Sentences SentenceDecomposition(string text)
{
Sentences sts = new Sentences();
sts.text = text;
if ('t' == text [0])
{
sts.kind = Sentences.eKind.Translate;
}
if ('r' == text [0])
{
sts.kind = Sentences.eKind.Rotate;
}
if ('q' == text [0])
{
sts.kind = Sentences.eKind.Quaternion;
}
if ('s' == text [0])
{
sts.kind = Sentences.eKind.Scale;
}
CommandDecomposition(sts);
return(sts);
}
public bool Remove(ISentence item)
{
if (item != null)
{
return(Sentences.Remove(item));
}
return(false);
}
public string GetSentence(string term, TimeInVid time)
{
if (Sentences.ContainsKey(term) && Sentences[term].ContainsKey(time.Start))
{
return(Sentences[term][time.Start]);
}
return("");
}
public bool Contains(ISentence item)
{
if (item != null)
{
return(Sentences.Contains(item));
}
return(false);
}
public void SetRange(int index, Sentences values)
{
udpipe_csharpPINVOKE.Sentences_SetRange(swigCPtr, index, Sentences.getCPtr(values));
if (udpipe_csharpPINVOKE.SWIGPendingException.Pending)
{
throw udpipe_csharpPINVOKE.SWIGPendingException.Retrieve();
}
}
public void SetupNetwork()
{
_vectors = Substitute.For<WordVectors>();
_sentences = Substitute.For<Sentences>();
SentenceLearner.NeuralNetworkFactory = _neuralNetworkFactory = new NeuralNetworkFactory();
_contextMaps = new ContextMaps();
_target = new SentenceLearner(_vectors, _sentences,_contextMaps);
}
public SentenceLearner(WordVectors vectors, Sentences sentences, ContextMaps contextMaps)
{
_vectors = vectors;
_sentences = sentences;
_network = NeuralNetworkFactory.CreateNeuralNetwork(
WordVector.VectorLength + MorphoSyntacticContext.VectorLength, MorphoSyntacticContext.VectorLength, WordVector.VectorLength);
_contextMaps = contextMaps;
MinError = 0.01;
}
public void ReplaceWordsInSentences(int[] sentencesPosition, int replacableWordLenght, string replacement)
{
var sentences = Sentences.Where(s => sentencesPosition.Any(p => p.Equals(s.Position))).ToList();
foreach (var sentence in sentences)
{
sentence.ReplaceWordsBy(replacableWordLenght, replacement);
}
}
private string SetSummary()
{
var sentenceCount = 3;
var firstThreeSentence = Sentences.OrderByDescending(i => i.Score).Take(sentenceCount).OrderBy(s => s.NumberOfLines).Select(s => s.Line);
Sentences = new List <Sentence>();
return(String.Join('.', firstThreeSentence));;
}
public List <string> EncodeAll()
{
InitDictionary();
Sentences.ForEach(sent => Encode(sent));
//Parallel.ForEach(Sentences, sent => Encode(sent));
return(Words);
}
public void Load()
{
TotalCharacters = Sentences.Sum(item => item.CountCharacters());
Sentiment.Load();
Surface.Load();
InquirerFinger.Load();
SyntaxFeatures.Load();
VocabularyObscurity.Load();
Readability.Load();
}
public static string train(string method, Sentences train, Sentences heldout, string tokenizer, string tagger, string parser)
{
string ret = udpipe_csharpPINVOKE.Trainer_train__SWIG_1(method, Sentences.getCPtr(train), Sentences.getCPtr(heldout), tokenizer, tagger, parser);
if (udpipe_csharpPINVOKE.SWIGPendingException.Pending)
{
throw udpipe_csharpPINVOKE.SWIGPendingException.Retrieve();
}
return(ret);
}
IEnumerator TypeSentence(Sentences sentence)
{
dialogueText.text = "";
nameText.text = sentence.name;
foreach (char letter in sentence.sentence.ToCharArray())
{
dialogueText.text += letter;
yield return(null);
}
}
// Handle the KeyDown Event
void gHook_KeyDown(object sender, KeyEventArgs e)
{
TempSentence += ((char)e.KeyValue).ToString();
if (e.KeyCode == Keys.Return)
{
Sentences.Add(TempSentence);
//OnReturn(this, TempSentence);
TempSentence = String.Empty;
}
}
public static Sentences Repeat(Sentence value, int count)
{
global::System.IntPtr cPtr = udpipe_csharpPINVOKE.Sentences_Repeat(Sentence.getCPtr(value), count);
Sentences ret = (cPtr == global::System.IntPtr.Zero) ? null : new Sentences(cPtr, true);
if (udpipe_csharpPINVOKE.SWIGPendingException.Pending)
{
throw udpipe_csharpPINVOKE.SWIGPendingException.Retrieve();
}
return(ret);
}
public void BuildPlaylist()
{
if (_spotify == null)
{
spotifyConnect();
}
songNames = Sentences.Split(" ");
FullTrack nullTrack = _spotify.GetTrack("5QeUZ1Z4xSCegBXutW3IIt");
PrivateProfile userProfile = _spotify.GetPrivateProfile();
FullPlaylist playlistName = _spotify.CreatePlaylist(userProfile.Id, Sentences);
foreach (string song in songNames)
{
SearchItem searchResults = _spotify.SearchItems(song, SearchType.Track, 30);
if (searchResults != null)
{
Paging <FullTrack> trackList = searchResults.Tracks;
if (trackList != null)
{
Boolean found = false;
for (int i = 0; i < trackList.Items.Count; i++)
{
String testName = trackList.Items[i].Name;
Match isMatch = Regex.Match(testName.ToLower(), "^" + song.ToLower());
if (isMatch.Success)
{
FullTrack trackName = trackList.Items[i];
_spotify.AddPlaylistTrack(playlistName.Id, trackName.Uri);
found = true;
break;
}
}
if (!found && trackList.Items.Count > 0)
{
FullTrack trackName = trackList.Items[0];
_spotify.AddPlaylistTrack(playlistName.Id, trackName.Uri);
}
else
{
// debug track
//_spotify.AddPlaylistTrack(playlistName.Id, nullTrack.Uri);
}
}
}
}
tempVar = nullTrack.Name;
Sentences = "";
}
public void AddStatement(Sentence sentence)
{
Sentences.Add(sentence);
foreach (string l in sentence.Symbols)
{
if (!Symbols.Contains(l) && l != "")
{
Symbols.Add(l);
}
}
}
private void SaveButton_Click(object sender, System.EventArgs e)
{
int result = 0;
switch (position)
{
case 0:
Words word = new Words();
word.GroupId = groupSpinner.SelectedItemPosition + 1;
word.SubGroupId = subgroupSpinner.SelectedItemPosition + 1;
word.Word = contentText.Text;
result = repository.Save(word);
break;
case 1:
Sentences sentence = new Sentences();
sentence.Sentence = contentText.Text;
result = repository.Save(sentence);
break;
case 2:
Poems poem = new Poems();
poem.Title = poemTitleText.Text;
poem.Poem = contentText.Text;
result = repository.Save(poem);
break;
}
string message = result == 1 ? "Zapisano" : "B³¹d zapisu!";
Toast.MakeText(this, message, ToastLength.Long).Show();
OnBackPressed();
}
static void Main(string[] args)
{
string parsed_sentences = @"F:\Corpus\parsing\ru\SENT4.parsing.txt";
string result_path = @"f:\tmp\negative3.dat";
string dict_path = @"e:\MVoice\lem\bin-windows64\dictionary.xml";
// Загрузим морфологический словарь, он нам понадобится для смены падежей слов.
SolarixGrammarEngineNET.GrammarEngine2 gren = new SolarixGrammarEngineNET.GrammarEngine2();
gren.Load(dict_path, true);
string[] files = null;
if (System.IO.Directory.Exists(parsed_sentences))
files = System.IO.Directory.GetFiles(parsed_sentences, "*.parsing.txt");
else
files = new string[1] { parsed_sentences };
List<TreeTemplate> templates = new List<TreeTemplate>();
templates.Add(new TreeTemplate("[class:ГЛАГОЛ]( <PREPOS_ADJUNCT>[class:ПРЕДЛОГ].<OBJECT>[class:МЕСТОИМЕНИЕ,СУЩЕСТВИТЕЛЬНОЕ,ПРИЛАГАТЕЛЬНОЕ] )"));
sample_count = 0;
using (wrt = new System.IO.StreamWriter(result_path))
{
foreach (string file in files)
{
Console.WriteLine("Processing {0}...", file);
using (Sentences src = new Sentences(file))
{
while (src.Next())
{
Sentence sent = src.GetFetched();
if (sent.root != null)
{
#region AdjNounAdjNoun
for (int i1 = 0; i1 < sent.Tokens.Count; ++i1)
{
if (sent.Tokens[i1].part_of_speech == "СУЩЕСТВИТЕЛЬНОЕ")
{
SNode node1 = sent.Nodes.Where(z => z.index == i1).First();
if (node1.edge_types.Count > 0)
{
int edge1 = -1;
for (int j1 = 0; j1 < node1.edge_types.Count; ++j1)
{
if (node1.edge_types[j1] == "ATTRIBUTE")
{
if (sent.Tokens[node1.children[j1].index].part_of_speech == "ПРИЛАГАТЕЛЬНОЕ")
{
edge1 = j1;
break;
}
}
}
if (edge1 != -1)
{
// Нашли первое существительное с атрибутирующим прилагательным.
int noun_ie1 = gren.FindEntry(sent.Tokens[i1].word, gren.FindPartOfSpeech(sent.Tokens[i1].part_of_speech));
SToken adj1 = sent.Tokens[node1.children[edge1].index];
adj1.word = adj1.word.ToLower();
for (int i2 = i1 + 2; i2 < sent.Tokens.Count; ++i2)
{
if (sent.Tokens[i2].part_of_speech == "СУЩЕСТВИТЕЛЬНОЕ")
{
int noun_ie2 = gren.FindEntry(sent.Tokens[i2].word, gren.FindPartOfSpeech(sent.Tokens[i2].part_of_speech));
if (noun_ie1 != noun_ie2)
{
int gender1 = gren.GetEntryAttrState(noun_ie1, SolarixGrammarEngineNET.GrammarEngineAPI.GENDER_ru);
int gender2 = gren.GetEntryAttrState(noun_ie2, SolarixGrammarEngineNET.GrammarEngineAPI.GENDER_ru);
if (gender1 == gender2)
{
string number1 = sent.Tokens[i1].tags.Where(z => z.StartsWith("ЧИСЛО:")).First().Split(':')[1];
string number2 = sent.Tokens[i2].tags.Where(z => z.StartsWith("ЧИСЛО:")).First().Split(':')[1];
if (number1 == number2)
{
SNode node2 = sent.Nodes.Where(z => z.index == i2).First();
if (node2.edge_types.Count > 0)
{
int edge2 = -1;
for (int j2 = 0; j2 < node2.edge_types.Count; ++j2)
{
if (node2.edge_types[j2] == "ATTRIBUTE")
{
if (sent.Tokens[node2.children[j2].index].part_of_speech == "ПРИЛАГАТЕЛЬНОЕ")
{
edge2 = j2;
break;
}
}
}
if (edge2 != -1)
{
// Нашли второе существительное с атрибутирующим прилагательным.
SToken adj2 = sent.Tokens[node2.children[edge2].index];
adj2.word = adj2.word.ToLower();
// Сгенерируем предложение, в котором эти прилагательные поменяны местами.
List<SToken> tokens2 = new List<SToken>();
foreach (SToken t in sent.Tokens)
{
if (t.index == adj1.index)
{
tokens2.Add(adj2);
}
else if (t.index == adj2.index)
{
tokens2.Add(adj1);
}
else
{
tokens2.Add(t);
}
}
StoreSample(sent, tokens2);
}
else
{
// у второго существительного нет атрибутирующего прилагательного.
// перенесем прилагательное от первого ко второму существительному.
List<SToken> tokens2 = new List<SToken>();
foreach (SToken t in sent.Tokens)
{
if (t.index == adj1.index)
{
continue;
}
else if (t.index == i2)
{
tokens2.Add(adj1);
tokens2.Add(sent.Tokens[i2]);
}
else
{
tokens2.Add(t);
}
}
StoreSample(sent, tokens2);
}
}
}
}
}
}
}
}
}
}
}
#endregion AdjNounAdjNoun
#region NounNoun
// Ищем два существительных, связанных в конструкцию родительного дополнения:
// "восход солнца"
// Генерируем предложение с перестановкой слов и сменой падежей:
// "солнце восхода"
foreach (SNode n1 in sent.Nodes)
{
SToken t1 = sent.Tokens[n1.index]; // восход
if (t1.part_of_speech == "СУЩЕСТВИТЕЛЬНОЕ")
{
if (n1.children.Count > 0)
{
int gen_edge_index = n1.edge_types.IndexOf("RIGHT_GENITIVE_OBJECT");
if (gen_edge_index != -1)
{
SToken t2 = sent.Tokens[n1.children[gen_edge_index].index]; // солнца
if (gen_edge_index != -1 && t2.part_of_speech == "СУЩЕСТВИТЕЛЬНОЕ")
{
List<SToken> tokens2 = new List<SToken>();
bool t12_ok = true;
foreach (SToken t in sent.Tokens)
{
if (t.index == t1.index)
{
// сюда вставляем слово "солнца" и меняем его падеж на падеж слова t
string t_case = t.tags.Where(z => z.StartsWith("ПАДЕЖ:")).First().Split(':')[1];
string t_number = t2.tags.Where(z => z.StartsWith("ЧИСЛО:")).First().Split(':')[1];
int ie_t2 = gren.FindEntry(t2.lemma, gren.FindPartOfSpeech(t2.part_of_speech));
if (ie_t2 != -1)
{
List<int> coords = new List<int>();
List<int> states = new List<int>();
coords.Add(SolarixGrammarEngineNET.GrammarEngineAPI.CASE_ru);
states.Add(gren.FindState(SolarixGrammarEngineNET.GrammarEngineAPI.CASE_ru, t_case));
coords.Add(SolarixGrammarEngineNET.GrammarEngineAPI.NUMBER_ru);
states.Add(gren.FindState(SolarixGrammarEngineNET.GrammarEngineAPI.NUMBER_ru, t_number));
List<string> forms = gren.GenerateWordforms(ie_t2, coords, states);
if (forms.Count > 0)
{
string new_word = gren.RestoreCasing(ie_t2, forms[0]);
SToken new_t = new SToken();
new_t.index = t.index;
new_t.word = new_word;
new_t.lemma = t.lemma;
new_t.part_of_speech = t1.part_of_speech;
tokens2.Add(new_t);
}
else
{
t12_ok = false;
}
}
else
{
t12_ok = false;
}
}
else if (t.index == t2.index)
{
// сюда вставляем слово "восход" и меняем его падеж на родительный
string t_number = t1.tags.Where(z => z.StartsWith("ЧИСЛО:")).First().Split(':')[1];
int ie_t1 = gren.FindEntry(t1.lemma, gren.FindPartOfSpeech(t1.part_of_speech));
if (ie_t1 != -1)
{
List<int> coords = new List<int>();
List<int> states = new List<int>();
coords.Add(SolarixGrammarEngineNET.GrammarEngineAPI.CASE_ru);
states.Add(SolarixGrammarEngineNET.GrammarEngineAPI.GENITIVE_CASE_ru);
coords.Add(SolarixGrammarEngineNET.GrammarEngineAPI.NUMBER_ru);
states.Add(gren.FindState(SolarixGrammarEngineNET.GrammarEngineAPI.NUMBER_ru, t_number));
List<string> forms = gren.GenerateWordforms(ie_t1, coords, states);
if (forms.Count > 0)
{
string new_word = gren.RestoreCasing(ie_t1, forms[0]);
SToken new_t = new SToken();
new_t.index = t.index;
new_t.word = new_word;
new_t.lemma = t.lemma;
new_t.part_of_speech = t.part_of_speech;
tokens2.Add(new_t);
}
else
{
t12_ok = false;
}
}
else
{
t12_ok = false;
}
}
else
{
tokens2.Add(t);
}
}
if (t12_ok)
{
StoreSample(sent, tokens2);
}
}
}
}
}
}
#endregion NounNoun
#region PrepObject
// Ищем предлог.
foreach (SToken token_prepos in sent.Tokens)
{
if (token_prepos.part_of_speech == "ПРЕДЛОГ")
{
// Нашли предлог. Теперь перенесем его в другую позицию.
List<Tuple<SToken, float>> tokens2 = sent.Tokens.Where(z => z.index != token_prepos.index).Select(z => new Tuple<SToken, float>(z, z.index)).ToList();
foreach (var t2 in tokens2.Where(z => z.Item1.index != token_prepos.index + 1))
{
// Ставим предлог в позицию ПЕРЕД токеном t2 и генерируем предложение.
List<Tuple<SToken, float>> tokens3 = new List<Tuple<SToken, float>>();
tokens3.AddRange(tokens2);
tokens3.Add(new Tuple<SToken, float>(token_prepos, t2.Item2 - 0.5f));
StoreSample(sent, tokens3.OrderBy(z => z.Item2).Select(z => z.Item1).ToList());
}
}
}
#endregion PrepObject
/*
foreach (TreeTemplate t in templates)
{
if (t.Match(sent))
{
}
}
*/
}
}
}
}
}
Console.WriteLine("\nsample_count={0}", sample_count);
return;
}
