public SToken Add(string token)
{
if (string.IsNullOrWhiteSpace(token))
{
return(null);
}
return(Context(db =>
{
SToken sToken = new SToken
{
Systime = DateTime.Now,
Createtime = DateTime.Now,
Invalid = false,
Operator = "hjh",
Token = token,
Tid = TRandom.Instance.GetRandomString(20)
};
db.STokens.InsertOnSubmit(sToken);
db.SubmitChanges();
return sToken;
}));
}
private Task SendValidatingEmailAsync(SToken token)
{
return(Task.Run(() =>
{
token.ValidationCode = CodeGenerator.GetCode(8);
logger.LogInformation("Validation Code for User(id: " + token.UserID + ") is: " + token.ValidationCode);
var mailMsg = new MailMessage();
mailMsg.From = new MailAddress("*****@*****.**", "Do-Not-Reply");
mailMsg.To.Add(new MailAddress(token.Email));
mailMsg.Subject = "线上教学系统：密码找回验证码";
mailMsg.Body = "请勿将验证码泄露给他人！\n验证码：" + token.ValidationCode;
var client = new SmtpClient();
client.Host = "smtp.qq.com";
client.Port = 587;
client.EnableSsl = true;
client.Credentials = new NetworkCredential("*****@*****.**", "dlshuyxgvyodbjbc");
try
{
client.SendAsync(mailMsg, null);
}
catch (Exception e)
{
logger.LogError(e.Message);
throw new Exception("邮件发送失败");
}
}));
}
public async Task <string> AddSTokenAsync(User userInfo, string email)
{
if (userInfo == null)
{
throw new Exception("用户不存在");
}
if (userInfo.Email != email)
{
throw new Exception("邮箱错误");
}
var t = new SToken(userInfo.Id, email);
var tID = CodeGenerator.GetCode(20);
t.ID = tID;
await SendValidatingEmailAsync(t);
tokenMap.Add(tID, t);
return(tID);
}
public EDocument(string documentText)
{
DocumentText = documentText;
RootTocken = new SToken();
SToken newTocken = new SToken();
RootTocken.Configure(0, string.Empty + "Program", SElementType.Undefined, string.Empty + " * ");
newTocken.Configure(0, string.Empty + "..", SElementType.Undefined, string.Empty);
RootTocken.TokenNodeType = SElementType.Undefined;
newTocken.TokenNodeType = SElementType.Undefined;
RootTocken.Link(SToken.NULL, SToken.NULL, null, null);
newTocken.Link(rootTocken, SToken.NULL, null, null);
RootTocken.TokenChilds.Add(newTocken);
parentToken = RootTocken;
currentToken = RootTocken.TokenChilds[0];
InitializePersons();
}
/*
private int p;
private string p_2;
private SElementType sElementType;
private string p_3;
*/
#endregion NULL
#region DEFINE
internal void Link(SToken tokenParent, SToken tokenBefore, SToken tokenAfter, List<SToken> tokenChilds)
{
TokenParent = tokenParent;
TokenBefore = tokenBefore;
if (tokenAfter != null)
{
if (TokenAfter == null)
{
TokenAfter = new List<SToken>();
TokenAfter.Add(tokenAfter);
if (tokenAfter.TokenChilds == null)
{
tokenAfter.TokenChilds = new List<SToken>();
}
}
}
if (tokenChilds == null)
{
tokenChilds = new List<SToken>();
}
TokenChilds = tokenChilds;
}
private void Wrap(SToken token)
{
while (token != null)
{
//if (token.TokenType == SElementType.Symbol)
{
CodeFile newCodeFile = new CodeFile();
ElementNamespace namespaceItem = new ElementNamespace();
ElementClass classItem = new ElementClass();
bool isNeededCodeFileCreation = true;
foreach (CodeFile existentCodeFile in program.CodeFiles)
{
if (existentCodeFile.FileName == ToUp(token.TokenString))
{
isNeededCodeFileCreation = false;
newCodeFile = existentCodeFile;
}
}
if (isNeededCodeFileCreation)
{
newCodeFile.FileName = ToUp(token.TokenString);
program.CodeFiles.Add(newCodeFile);
namespaceItem.Name = "GeneratedApplication";
newCodeFile.Namespaces.Add(namespaceItem);
classItem.Name = ToUp(token.TokenString);
namespaceItem.ElementClasses.Add(classItem);
}
if (token.TokenProperties != null)
{
SToken tokenToRead = token;
if (token.TokenProperties[0].TokenNodeType == SElementType.FoldOpen)
{
foreach (SToken localTokenF in token.TokenProperties[0].TokenChilds)
{
SToken localToken = localTokenF;
while (localToken != null)
{
Wrap(localToken);
ElementProperty newProperty = new ElementProperty();
switch (localToken.TokenMultiplicity)
{
case SElementMultiplicity.Some:
newProperty.Name = ToUp(localToken.TokenString) + "s";
newProperty.PropertyType = "List<" + ToUp(localToken.TokenString) + ">";
break;
case SElementMultiplicity.Explicit:
newProperty.Name = ToUp(localToken.TokenString) + "s";
newProperty.PropertyType = "" + ToUp(localToken.TokenString) + " " + token.TokenMultiplicityLevel + "";
break;
default:
newProperty.Name = ToUp(localToken.TokenString) + "P";
newProperty.PropertyType = ToUp(localToken.TokenString);
break;
}
//newProperty.PropertyType = localToken.TokenString;
newProperty.ElementClassName = ToUp(localToken.TokenString);
classItem.ElementProperties.Add(newProperty);
if (localToken.TokenAfter != null)
{
localToken = localToken.TokenAfter[0];
}
else
{
localToken = null;
}
}
}
}
else
foreach (SToken localToken in token.TokenProperties)
{
Wrap(localToken);
ElementProperty newProperty = new ElementProperty();
switch (localToken.TokenMultiplicity)
{
case SElementMultiplicity.Some:
newProperty.Name = ToUp(localToken.TokenString) + "s";
newProperty.PropertyType = "List<" + ToUp(localToken.TokenString) + ">";
break;
case SElementMultiplicity.Explicit:
newProperty.Name = ToUp(localToken.TokenString) + "s";
newProperty.PropertyType = "" + ToUp(localToken.TokenString) + " " + token.TokenMultiplicityLevel + "";
break;
default:
newProperty.Name = ToUp(localToken.TokenString) + "P";
newProperty.PropertyType = ToUp(localToken.TokenString);
break;
}
//newProperty.PropertyType = localToken.TokenString;
newProperty.ElementClassName = ToUp(localToken.TokenString);
classItem.ElementProperties.Add(newProperty);
}
}
if (token.TokenActivities != null)
{
foreach (SToken localToken in token.TokenActivities)
{
Wrap(localToken);
ElementMethod newMethod = new ElementMethod();
newMethod.Name = ToUp(localToken.TokenString);
newMethod.ElementAccessType = ElementMethod.DataAccessType.isPublic;
//newProperty.PropertyType = localToken.TokenString;
newMethod.ElementClassName = ToUp(localToken.TokenString);
classItem.ElementMethods.Add(newMethod);
}
}
List<string> influencesA = new List<string>();
List<string> influencesB = new List<string>();
List<string> influencesC = new List<string>();
List<string> influencesD = new List<string>();
if (token.TokenInfluenceA != null)
{
foreach (SToken localToken in token.TokenInfluenceA)
{
/*{
foreach (SToken tokenA in token.TokenInfluenceA)
{
if (token.TokenString.Equals(tokenA))
{
System.Windows.Forms.MessageBox.Show("No need to repeat yourself");
break;
}
}
}*/
if (!influencesA.Contains(localToken.TokenString))
{
influencesA.Add(localToken.TokenString);
}
else
{
System.Windows.Forms.MessageBox.Show("No need to repeat yourself!");
break;
}
if (influencesB.Contains(localToken.TokenString))
{
System.Windows.Forms.MessageBox.Show("Your statement is inconsistent!");
break;
}
if (influencesD.Contains(localToken.TokenString))
{
System.Windows.Forms.MessageBox.Show("You defined a rebound effect!");
break;
}
//Wrap(localToken);
ElementMethod newMethod = new ElementMethod();
newMethod.Name = ToUp(localToken.TokenString);
newMethod.ElementAccessType = ElementMethod.DataAccessType.isPublic;
newMethod.IsStatic = true;
//newProperty.PropertyType = localToken.TokenString;
newMethod.Name = "Increase" + ToUp(localToken.TokenString);
classItem.ElementMethods.Add(newMethod);
}
}
if (token.TokenInfluenceB != null)
{
foreach (SToken localToken in token.TokenInfluenceB)
{
if (!influencesB.Contains(localToken.TokenString))
{
influencesB.Add(localToken.TokenString);
}
else
{
System.Windows.Forms.MessageBox.Show("No need to repeat yourself!");
break;
}
if (influencesA.Contains(localToken.TokenString))
{
System.Windows.Forms.MessageBox.Show("Your statement is inconsistent!");
break;
}
if (influencesC.Contains(localToken.TokenString))
{
System.Windows.Forms.MessageBox.Show("You defined a rebound effect!");
break;
}
//Wrap(localToken);
ElementMethod newMethod = new ElementMethod();
newMethod.Name = ToUp(localToken.TokenString);
newMethod.ElementAccessType = ElementMethod.DataAccessType.isPublic;
newMethod.IsStatic = true;
//newProperty.PropertyType = localToken.TokenString;
newMethod.Name = "Decrease" + ToUp(localToken.TokenString);
classItem.ElementMethods.Add(newMethod);
}
}
if (token.TokenInfluenceC != null)
{
foreach (SToken localToken in token.TokenInfluenceC)
{
if (!influencesC.Contains(localToken.TokenString))
{
influencesC.Add(localToken.TokenString);
}
else
{
System.Windows.Forms.MessageBox.Show("No need to repeat yourself!");
break;
}
if (influencesD.Contains(localToken.TokenString))
{
System.Windows.Forms.MessageBox.Show("Your statement is inconsistent!");
break;
}
if (influencesB.Contains(localToken.TokenString))
{
System.Windows.Forms.MessageBox.Show("You defined a rebound effect!");
break;
}
Wrap(localToken);
ElementMethod newMethod = new ElementMethod();
newMethod.Name = ToUp(localToken.TokenString);
newMethod.ElementAccessType = ElementMethod.DataAccessType.isPublic;
newMethod.IsStatic = true;
//newProperty.PropertyType = localToken.TokenString;
newMethod.Name = "IncreasedBy" + ToUp(localToken.TokenString);
classItem.ElementMethods.Add(newMethod);
}
}
if (token.TokenInfluenceD != null)
{
foreach (SToken localToken in token.TokenInfluenceD)
{
if (!influencesD.Contains(localToken.TokenString))
{
influencesD.Add(localToken.TokenString);
}
else
{
System.Windows.Forms.MessageBox.Show("No need to repeat yourself!");
break;
}
if (influencesC.Contains(localToken.TokenString))
{
System.Windows.Forms.MessageBox.Show("Your statement is inconsistent!");
break;
}
if (influencesA.Contains(localToken.TokenString))
{
System.Windows.Forms.MessageBox.Show("You defined a rebound effect!");
break;
}
Wrap(localToken);
ElementMethod newMethod = new ElementMethod();
newMethod.Name = ToUp(localToken.TokenString);
newMethod.ElementAccessType = ElementMethod.DataAccessType.isPublic;
newMethod.IsStatic = true;
//newProperty.PropertyType = localToken.TokenString;
newMethod.Name = "DecreasedBy" + ToUp(localToken.TokenString);
classItem.ElementMethods.Add(newMethod);
}
}
}
foreach (SToken tokenlist in token.TokenChilds)
{
Wrap(tokenlist);
}
if (token.TokenAfter != null)
{
token = token.TokenAfter[0];
}
else
{
token = null;
}
}
}
public string ToString(SToken token)
{
int acoladeBracketBlock = 0;
int angleBracketBlock = 0;
int rightBracketBlock = 0;
int roundBracketBlock = 0;
StringBuilder tokenCollectionAsString = new StringBuilder();
while (token != null)
{
string color;
switch (token.TokenType)
{
case SElementType.Asignement:
color = "red";
break;
case SElementType.Comparator:
color = "orange";
break;
case SElementType.Incertitude:
color = "magenta";
break;
case SElementType.Iterator:
color = "orange";
break;
case SElementType.Logic:
color = "red";
break;
case SElementType.Multiplicity:
color = "green";
break;
case SElementType.Operator:
color = "blue";
break;
case SElementType.Selector:
color = "green";
break;
case SElementType.Symbol:
color = "purple";
break;
default:
color = "black";
break;
}
//*** if (((!(token.TokenAfter == null) && !(token.TokenAfter.TokenAfter == null) && (!(token.TokenAfter.TokenAfter.TokenAfter == null && !string.IsNullOrEmpty(token.TokenAfter.TokenString) && ")]>}".Contains(token.TokenAfter.TokenString[0]))))) || token.TokenType == SElementType.Undefined)
if (token.TokenAfter != null)
{
tokenCollectionAsString.Append("<span style='color:" + color + "'>" + System.Security.SecurityElement.Escape(pluralize(token)) + "</span> ");
//tokenCollectionAsString.Append("<span style='color:" + color + "'>" + System.Security.SecurityElement.Escape(token.TokenString) + "</span> ");
}
else
{
tokenCollectionAsString.Append("<span style='color:" + color + "'>" + System.Security.SecurityElement.Escape(pluralize(token)) + "</span> ");
//tokenCollectionAsString.Append("<span style='color:" + color + "'>" + System.Security.SecurityElement.Escape(token.TokenString) + "</span> ");
}
if (token.TokenChilds != null)
foreach (SToken codeTokenlist in token.TokenChilds)
{
tokenCollectionAsString.Append(ToString(codeTokenlist));
/*if (token.TokenType == SElementType.AcoladeBracketBlock && tokenCollectionAsString.ToString().TrimEnd().EndsWith(";"))
{
tokenCollectionAsString.Append("</br>");
}*/
}
if (token.TokenProperties != null)
foreach (SToken codeTokenlist in token.TokenProperties)
{
tokenCollectionAsString.Append(ToString(codeTokenlist));
/*if (token.TokenType == SElementType.AcoladeBracketBlock && tokenCollectionAsString.ToString().TrimEnd().EndsWith(";"))
{
tokenCollectionAsString.Append("</br>");
}*/
}
if (token.TokenActivities != null)
foreach (SToken codeTokenlist in token.TokenActivities)
{
tokenCollectionAsString.Append(ToString(codeTokenlist));
/*if (token.TokenType == SElementType.AcoladeBracketBlock && tokenCollectionAsString.ToString().TrimEnd().EndsWith(";"))
{
tokenCollectionAsString.Append("</br>");
}*/
}
token = GetNext(token);
}
return tokenCollectionAsString.ToString();
}
private List<SToken> ExplodeStoken(SToken sToken, SToken token, string form)
{
List<SToken> tokens = new List<SToken>();
//foreach (string form in forms)
{
string text = sToken.TokenString;
string textBefore = string.Empty;
string textAfter = string.Empty;
if (text.Contains(form))
{
textBefore = " "+text.Substring(0, text.IndexOf(form)).Trim()+" ";
textAfter = " "+text.Substring(text.IndexOf(form) + form.Length).Trim()+" ";
}
// while (text.Contains(form))
{
SToken tokenInside = new SToken();
SToken tokenBefore = new SToken();
SToken tokenAfter = new SToken();
tokenInside.Configure(token.TokenId, token.TokenString, token.TokenType, token.TokenCode);
tokenInside.Link(sToken.TokenParent, tokenBefore, tokenAfter, sToken.TokenChilds);
tokenInside.TokenMultiplicity = sToken.TokenMultiplicity;
tokenInside.TokenMultiplicityLevel = sToken.TokenMultiplicityLevel;
tokenBefore.Configure(sToken.TokenId, textBefore, sToken.TokenType, sToken.TokenCode);
tokenBefore.Link(sToken.TokenParent, sToken.TokenBefore, tokenInside, token.TokenChilds);
tokenAfter.Configure(sToken.TokenId, textAfter, sToken.TokenType, sToken.TokenCode);
//*** tokenAfter.Link(sToken.TokenParent, tokenInside, sToken.TokenAfter, token.TokenChilds);
//*** tokenInside.TokenAfter = tokenAfter;
if (string.IsNullOrEmpty(tokenBefore.TokenString.Trim()))
{
tokenInside.TokenBefore = tokenBefore.TokenBefore;
}
if (string.IsNullOrEmpty(tokenAfter.TokenString.Trim()))
{
//*** tokenInside.TokenAfter = tokenAfter.TokenAfter;
}
if (sToken.TokenMultiplicity != SElementMultiplicity.Undefined)
{
tokenInside.TokenMultiplicity = sToken.TokenMultiplicity;
tokenBefore.TokenMultiplicity = sToken.TokenMultiplicity;
tokenInside.TokenMultiplicity = sToken.TokenMultiplicity;
}
else
{
tokenInside.TokenMultiplicity = token.TokenMultiplicity;
tokenBefore.TokenMultiplicity = token.TokenMultiplicity;
tokenInside.TokenMultiplicity = token.TokenMultiplicity;
}
tokenAfter.TokenBefore = tokenBefore;
if (sToken.TokenBefore != null)
{
//*** sToken.TokenBefore.TokenAfter = tokenBefore;
}
//*** if (sToken.TokenAfter != null)
{
//*** sToken.TokenAfter.TokenBefore = tokenAfter;
}
if (!string.IsNullOrEmpty(tokenBefore.TokenString.Trim()))
{
tokens.Add(tokenBefore);
}
tokens.Add(tokenInside);
if (!string.IsNullOrEmpty(tokenAfter.TokenString.Trim()))
{
tokens.Add(tokenAfter);
}
}
}
return tokens;
}
public void AdoptMultiplicity(int multiplicityId, string multiplicatorText, List<string> forms, string code)
{
SToken token = new SToken();
//token.Configure();
//token.Link();
//multiplicityId, SElementType.Multiplicity, multiplicatorText, SToken.NULL, SToken.NULL, code);
StringBuilder tokenCollectionAsString = new StringBuilder();
SToken localCurrentToken = currentToken;
while (localCurrentToken != null)
{
for (int i = 0; i < forms.Count; i++)
{
if (localCurrentToken.TokenType == SElementType.Undefined)
{
if (localCurrentToken.TokenString.Contains(forms[i]))
{
List<SToken> explodedToken = ExplodeStoken(localCurrentToken, token, forms[i]);
tokenCollection.InsertRange(tokenCollection.IndexOf(localCurrentToken), explodedToken);
tokenCollection.Remove(localCurrentToken);
i = 0;
localCurrentToken = explodedToken[explodedToken.Count - 1];
}
}
}
//*** localCurrentToken = localCurrentToken.TokenAfter;
}
}
void AppendToken(string c)
{
SElementType[] conditionType = new SElementType[] { SElementType.CollectionAnd, SElementType.CollectionOr };
SToken newToken = SetNext(currentToken);
newToken.Configure(0, "..", SElementType.Symbol, "");
newToken.TokenNodeType = SElementType.Undefined;
if (!sticked && nextIsAPropertyFoTheCurrent)
{
if (currentToken.TokenProperties == null)
{
currentToken.TokenProperties = new List<SToken>();
}
newToken.Link(currentToken.TokenParent, null, null, null);
currentToken.TokenProperties.Add(newToken);
newToken.TokenDescribed = currentToken;
currentToken = newToken;
nextIsAPropertyFoTheCurrent = false;
return;
}
if (!sticked && conditionType.Contains(currentToken.TokenParent.TokenType))
{
currentToken = currentToken.TokenParent.TokenAfter[currentToken.TokenParent.TokenAfter.Count-1];
}
newToken.Link(currentToken.TokenParent, currentToken, null, null);
//currentToken = newToken;
AdoptAsCurrent(newToken);
}
private SToken AppendToken(string tokenText, string c)
{
//ParseDocument();
if (!sticked)
{
if (currentToken.TokenParent.TokenNodeType == SElementType.FoldOpen || currentToken.TokenParent.TokenType == SElementType.CollectionAnd || currentToken.TokenParent.TokenType == SElementType.CollectionOr)
{
//*** currentToken.TokenBefore.TokenAfter = currentToken.TokenAfter; // link
//*** currentToken.TokenBefore = currentToken.TokenParent.TokenAfter;
//*** currentToken.TokenParent.TokenAfter.TokenAfter = currentToken;
currentToken.TokenParent = currentToken.TokenParent.TokenParent;
}
else
{
SToken newToken = new SToken();
newToken.Configure(0, "", SElementType.Undefined, "");
newToken.Link(currentToken.TokenParent, currentToken, null, null);
// if the multilplicity is not defined get the colection multiplicity
if (currentToken.TokenMultiplicity == SElementMultiplicity.Undefined)
{
newToken.TokenMultiplicity = currentToken.TokenParent.TokenMultiplicity;
}
//currentToken = newToken;
currentToken.TokenString = c;
//*** currentToken.TokenAfter = newToken;
//*** currentToken = currentToken.TokenAfter;
}
}
else
{
currentToken.TokenString = c;
//currentToken = newToken;
}
sticked = false;
return currentToken;
}
private void AddBlock(string tokenText, string c)
{
currentToken.TokenString = tokenText + c;
SToken aToken = new SToken();
aToken.Configure(0, string.Empty, SElementType.Undefined, "");
aToken.Link(SToken.NULL, SToken.NULL, parentToken, new List<SToken>());
parentToken.TokenChilds.Add(aToken);
currentToken = aToken;
}
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;
}
}
public Sentence(XmlNode n_sent)
{
text = n_sent.SelectSingleNode("text").InnerText;
// токены
tokens = new List <SToken>();
int token_index = 0;
foreach (XmlNode n_token in n_sent.SelectNodes("tokens/token"))
{
SToken t = new SToken(token_index, n_token);
tokens.Add(t);
token_index++;
}
// дерево зависимостей
List <int> root_index = new List <int>();
Dictionary <int, int> child2parent = new Dictionary <int, int>();
Dictionary <KeyValuePair <int, int>, string> edge_type = new Dictionary <KeyValuePair <int, int>, string>();
Dictionary <int, List <int> > parent2child = new Dictionary <int, List <int> >();
foreach (XmlNode n_token in n_sent.SelectNodes("syntax_tree/node"))
{
int child_index = int.Parse(n_token["token"].InnerText);
if (n_token.Attributes["is_root"] != null && n_token.Attributes["is_root"].Value == "true")
{
root_index.Add(child_index);
}
else
{
int parent_index = int.Parse(n_token["parent"].InnerText);
child2parent.Add(child_index, parent_index);
edge_type.Add(new KeyValuePair <int, int>(child_index, parent_index), n_token["link_type"].InnerText);
List <int> child_idx;
if (!parent2child.TryGetValue(parent_index, out child_idx))
{
child_idx = new List <int>();
parent2child.Add(parent_index, child_idx);
}
child_idx.Add(child_index);
}
}
nodes = new List <SNode>();
for (int inode = 0; inode < tokens.Count; ++inode)
{
SNode n = new SNode();
n.index = inode;
n.word = tokens[inode];
nodes.Add(n);
}
// проставим родителей и детей в каждом узле
for (int inode = 0; inode < nodes.Count; ++inode)
{
SNode node = nodes[inode];
if (!root_index.Contains(node.index))
{
SNode parent_node = nodes[child2parent[node.index]];
node.parent = parent_node;
parent_node.children.Add(node);
parent_node.edge_types.Add(edge_type[new KeyValuePair <int, int>(node.index, parent_node.index)]);
}
else
{
root = node;
}
}
}
private void RegisterPerson()
{
switch (currentToken.TokenMultiplicity)
{
case SElementMultiplicity.One:
tokenHe = currentToken;
break;
case SElementMultiplicity.Explicit:
tokenHe = currentToken;
break;
}
}
public Sentence(XmlNode n_sent)
{
text = n_sent.SelectSingleNode("text").InnerText;
// токены
tokens = new List<SToken>();
int token_index = 0;
foreach (XmlNode n_token in n_sent.SelectNodes("tokens/token"))
{
SToken t = new SToken(token_index, n_token);
tokens.Add(t);
token_index++;
}
// дерево зависимостей
List<int> root_index = new List<int>();
Dictionary<int, int> child2parent = new Dictionary<int, int>();
Dictionary<KeyValuePair<int, int>, string> edge_type = new Dictionary<KeyValuePair<int, int>, string>();
Dictionary<int, List<int>> parent2child = new Dictionary<int, List<int>>();
foreach (XmlNode n_token in n_sent.SelectNodes("syntax_tree/node"))
{
int child_index = int.Parse(n_token["token"].InnerText);
if (n_token.Attributes["is_root"] != null && n_token.Attributes["is_root"].Value == "true")
root_index.Add(child_index);
else
{
int parent_index = int.Parse(n_token["parent"].InnerText);
child2parent.Add(child_index, parent_index);
edge_type.Add(new KeyValuePair<int, int>(child_index, parent_index), n_token["link_type"].InnerText);
List<int> child_idx;
if (!parent2child.TryGetValue(parent_index, out child_idx))
{
child_idx = new List<int>();
parent2child.Add(parent_index, child_idx);
}
child_idx.Add(child_index);
}
}
nodes = new List<SNode>();
for (int inode = 0; inode < tokens.Count; ++inode)
{
SNode n = new SNode();
n.index = inode;
n.word = tokens[inode];
nodes.Add(n);
}
// проставим родителей и детей в каждом узле
for (int inode = 0; inode < nodes.Count; ++inode)
{
SNode node = nodes[inode];
if (!root_index.Contains(node.index))
{
SNode parent_node = nodes[child2parent[node.index]];
node.parent = parent_node;
parent_node.children.Add(node);
parent_node.edge_types.Add(edge_type[new KeyValuePair<int, int>(node.index, parent_node.index)]);
}
else
{
root = node;
}
}
}
void AppendTokenDown()
{
SToken newToken = new SToken();
newToken.Configure(0, "append", SElementType.Undefined, "");
newToken.Link(currentToken.TokenParent, currentToken, null, null);
//*** currentToken.TokenAfter = newToken;
//*** currentToken = currentToken.TokenAfter;
}
public bool AdoptSymbol(int symbolId, string symbolText, List<string> forms, string code, SElementMultiplicity multiplicity)
{
bool simbolFound = false;
SToken token = new SToken();
if (currentToken.TokenBefore != null)
{
token = currentToken;
}
token.Configure(symbolId, symbolText, SElementType.Symbol, "[" + symbolId + ":0]");
token.Link(currentToken.TokenParent, currentToken, null, null);
if (currentToken.TokenBefore != null)
{
token.TokenBefore = currentToken.TokenBefore;
}
if (token.TokenMultiplicity == SElementMultiplicity.Undefined)
token.TokenMultiplicity = multiplicity;
StringBuilder tokenCollectionAsString = new StringBuilder();
SToken localCurrentToken = currentToken;
while (localCurrentToken != null)
{
for (int i = 0; i < forms.Count; i++)
{
if (localCurrentToken.TokenType == SElementType.Undefined)
{
if (localCurrentToken.TokenString.Contains(forms[i]))
{
List<SToken> explodedToken = ExplodeStoken(localCurrentToken, token, forms[i]);
// Try child
int ii = localCurrentToken.TokenParent.TokenChilds.IndexOf(localCurrentToken);
if (ii > -1)
{
localCurrentToken.TokenParent.TokenChilds.Insert(localCurrentToken.TokenParent.TokenChilds.IndexOf(localCurrentToken), explodedToken[0]);
localCurrentToken.TokenParent.TokenChilds.Remove(localCurrentToken);
i = 0;
//if (explodedToken[explodedToken.Count - 1]!=null)
localCurrentToken = explodedToken[explodedToken.Count - 1];
simbolFound = true;
}
// Try properties
if (currentToken.TokenDescribed != null)
{
ii = localCurrentToken.TokenDescribed.TokenProperties.IndexOf(localCurrentToken);
if (ii > -1)
{
localCurrentToken.TokenDescribed.TokenProperties.Insert(localCurrentToken.TokenDescribed.TokenProperties.IndexOf(localCurrentToken), explodedToken[0]);
localCurrentToken.TokenDescribed.TokenProperties.Remove(localCurrentToken);
i = 0;
//if (explodedToken[explodedToken.Count - 1]!=null)
localCurrentToken = explodedToken[explodedToken.Count - 1];
simbolFound = true;
}
}
}
}
}
localCurrentToken = GetNext(localCurrentToken);
}
return simbolFound;
}
void FoldToken(string c)
{
bool foldNeeded = true;
SElementType[] conditionType = new SElementType[] { SElementType.CollectionAnd, SElementType.CollectionOr };
if (currentToken.TokenBefore != null) // if there is nothing before fold
{
if (conditionType.Contains(currentToken.TokenBefore.TokenType)) // if there is another terminated fold before then fold
{
//foldNeeded = false;
}
if (currentToken.TokenBefore.TokenType == SElementType.Collection && currentToken.TokenString == "and") // Fold needed ?
{
foldNeeded = false;
}
if (currentToken.TokenParent.TokenNodeType == SElementType.FoldOpen)
{
foldNeeded = false;
}
}
if (foldNeeded)
{
SToken firstToFold = currentToken;
if (currentToken.TokenBefore != null)
{
if (currentToken.TokenBefore.TokenBefore != null && currentToken.TokenBefore.TokenNodeType == SElementType.FoldClose && currentToken.TokenBefore.TokenBefore.TokenNodeType != SElementType.CollectionAnd)
{
while (firstToFold.TokenBefore != null && firstToFold.TokenNodeType != SElementType.FoldOpen)
{
firstToFold = firstToFold.TokenBefore;
}
}
else
{
firstToFold = currentToken.TokenBefore;
}
}
// Insert Folding Markers
SToken bracketOpenToken = new SToken();
SToken bracketCloseToken = new SToken();
bracketOpenToken.Configure(0, "(", SElementType.Collection, "");
bracketCloseToken.Configure(0, ")", SElementType.Collection, "");
bracketOpenToken.TokenNodeType = SElementType.FoldOpen;
bracketCloseToken.TokenNodeType = SElementType.FoldClose;
if (currentToken.TokenBefore != null)
{
bracketOpenToken.TokenMultiplicity = currentToken.TokenBefore.TokenMultiplicity;
}
bracketOpenToken.Link(firstToFold.TokenParent, firstToFold.TokenBefore, bracketCloseToken, new List<SToken>() { firstToFold });
bracketCloseToken.Link(firstToFold.TokenParent, bracketOpenToken, null, null);
firstToFold.Link(bracketOpenToken, null, null, firstToFold.TokenChilds);
if (bracketOpenToken.TokenBefore != null)
{
bracketOpenToken.TokenBefore.TokenAfter[0] = bracketOpenToken;
}
SToken localToken = firstToFold;
while (localToken != null && localToken.TokenAfter != null)
{
localToken.TokenParent = bracketOpenToken;
localToken = localToken.TokenAfter[0];
}
// If bracketOpenToken is the first in clilds replace the lider
if (bracketOpenToken.TokenBefore == null)
{
bracketOpenToken.TokenParent.TokenChilds.Remove(firstToFold);
bracketOpenToken.TokenParent.TokenChilds.Add(bracketOpenToken);
}
//currentToken = bracketCloseToken;
}
switch (c)
{
case "and":
currentToken.TokenParent.TokenType = SElementType.CollectionAnd;
currentToken.TokenParent.TokenString = "( && ";
break;
case "or":
currentToken.TokenParent.TokenType = SElementType.CollectionOr;
currentToken.TokenParent.TokenString = "( || ";
break;
}
if (currentToken.TokenParent.TokenNodeType != SElementType.FoldOpen)
{
//currentToken = currentToken.TokenParent;
}
else
{
//AppendToken();
}
}
public void AdoptLogic(int logicId, string logicText, List<string> forms, string code)
{
SToken token = new SToken();
token.Configure(logicId, code, SElementType.Logic, logicText);
token.Link(SToken.NULL, SToken.NULL, null, null);
StringBuilder tokenCollectionAsString = new StringBuilder();
SToken localCurrentToken = currentToken;
while (localCurrentToken != null)
{
for (int i = 0; i < forms.Count; i++)
{
if (localCurrentToken.TokenType == SElementType.Undefined)
{
if (localCurrentToken.TokenString.Contains(forms[i]))
{
List<SToken> explodedToken = ExplodeStoken(localCurrentToken, token, forms[i]);
tokenCollection.AddRange(explodedToken);
tokenCollection.Remove(localCurrentToken);
i = 0;
localCurrentToken = explodedToken[explodedToken.Count - 1];
}
}
}
//*** localCurrentToken = localCurrentToken.TokenAfter;
}
}
private SToken GetNext(SToken token)
{
if (token.TokenAfter == null)
{
return null;
}
else
{
return token.TokenAfter[0];
}
}
/// <summary>
/// Create a collection of tokens from the input string
/// </summary>
/// <returns></returns>
public List<SToken> TokenizeInput(string tokenizableString)
{
List<SToken> tokenizedInput = new List<SToken>();
SToken newToken = new SToken();
newToken.Configure(0, string.Empty, SElementType.Undefined, tokenizableString);
newToken.Link(SToken.NULL, SToken.NULL, null, null);
tokenizedInput.Add(newToken);
return tokenizedInput;
}
private SToken SetNext(SToken token)
{
SToken newToken = new SToken();
if (token.TokenAfter == null)
{
token.TokenAfter = new List<SToken>();
token.TokenAfter.Add(newToken);
}
else
{
if (newEvaluation)
{
token.TokenAfter.Add(newToken);
}
}
newEvaluation = false;
return newToken;
}
string pluralize(SToken token)
{
switch (token.TokenMultiplicity)
{
case SElementMultiplicity.All:
return "*" + pluralize(token.TokenString);
break;
case SElementMultiplicity.One:
return "1" + token.TokenString;
break;
case SElementMultiplicity.Explicit:
switch (token.TokenMultiplicityLevel)
{
case 0:
return "No " + token.TokenString + "s";
case 1:
return "One " + token.TokenString + "s";
default:
return token.TokenMultiplicityLevel+"X " + token.TokenString + "s";
break;
}
break;
case SElementMultiplicity.Some:
return "" + token.TokenString + "s[]";
break;
}
return token.TokenString;
}
private bool SubstituteSymbols(SToken localCurrentToken)
{
try
{
bool simbolFound = false;
//SToken localCurrentToken = currentToken;
//while (localCurrentToken.TokenBefore != null && localCurrentToken.TokenBefore.TokenType == SElementType.Undefined)
{
}
while (localCurrentToken != null)
{
int offset = 0;
if (localCurrentToken.TokenType == SElementType.Undefined)
while (!NothingMoreToreplace(localCurrentToken, offset))
{
string candidate = GetFirstCandidateToReplace(localCurrentToken, offset);
candidate = candidate.Trim();
offset += candidate.Length;
while (!string.IsNullOrEmpty(candidate))
{
bool found = false;
string partialCandidate = candidate;
while (!found)
{
SObjectCollection objectCollection = new SObjectCollection();
objectCollection.AddByName(singularize(partialCandidate), 0); // logics.getSymbsBySubstring(partialCandidate, 0, 100);
if (objectCollection.SObject.Count < 1)
{
if (partialCandidate.LastIndexOf(' ') > -1)
{
partialCandidate = partialCandidate.Substring(0, partialCandidate.LastIndexOf(' '));
}
else
{
candidate = candidate.Substring(partialCandidate.Length).Trim();//
/*for (int i = 0; i < candidate.Length; i++)
{
if (char.IsWhiteSpace(candidate[i]))
{
ReplaceAt++;
}
else
{
candidate = candidate.Substring(i);
return;
}
}*/
//string substitute = "[0:0]";
/***>> if (Symbol.CreateUnknown && (char.IsWhiteSpace(SubsritutedQuery[SubsritutedQuery.Length - 1]) || (SubsritutedQuery[SubsritutedQuery.Length - 1] == '!')))
{
Symbol symbol = new Symbol(partialCandidate, true);
substitute = "[" + symbol.SymbolID[0] + ":" + symbol.SymbolID[1] + "]";
} <<***/
//SubsritutedQuery = SubsritutedQuery.Substring(0, ReplaceAt) + substitute + SubsritutedQuery.Substring(ReplaceAt + partialCandidate.Length);
//ReplaceAt = ReplaceAt + substitute.Length + 1; // because i have a spave after substitude
found = true;
}
}
else
{
bool needReplaceAtPluse = false;
string substitute = "";
foreach (SymbolID symbol in objectCollection.SObject)
{
substitute = symbol.ToString();
simbolFound = AdoptSymbol(symbol.Location.A, symbol.Names[0], new List<string>() { partialCandidate }, "[" + symbol.Location.A + ":0]", pluralize(symbol.Names[0]).Equals(partialCandidate) ? SElementMultiplicity.All : SElementMultiplicity.One);
currentToken = currentToken.TokenParent.TokenChilds[0];
//*** while (currentToken.TokenAfter != null)
{
//*** currentToken = currentToken.TokenAfter;
}
//currentToken.TokenParent.TokenChilds
//currentToken.TokenType = localCurrentToken.TokenType;
//currentToken.TokenNodeType = localCurrentToken.TokenNodeType;
//currentToken.TokenMultiplicity = localCurrentToken.TokenMultiplicity;
//currentToken = currentToken.TokenParent.TokenChilds[0];
//*** if (currentToken.TokenAfter != null)
{
//*** currentToken.TokenAfter.TokenBefore = currentToken;
}
//offset = 0;
}
if (partialCandidate.Length < candidate.Length)
{
candidate = candidate.Substring(partialCandidate.Length + 1); // this is not possible if we are at the end of candidate
needReplaceAtPluse = true;
}
else
{
candidate = string.Empty;
}
if (needReplaceAtPluse)
{
//ReplaceAt++;
}
found = true;
}
}
}
}
localCurrentToken = GetNext(localCurrentToken);
}
if (simbolFound && false)
{
SToken newToken = new SToken();
newToken.Configure(0, "", SElementType.Undefined, "");
newToken.Link(currentToken.TokenParent, currentToken, null, null);
//*** currentToken.TokenAfter = newToken;
currentToken = newToken;
}
return simbolFound;
}
catch (Exception ss)
{
}
return false;
}
private List<SToken> ExplodeStoken(SToken sToken, SToken token, string form)
{
List<SToken> tokens = new List<SToken>();
//foreach (string form in forms)
{
string text = sToken.TokenString;
string textBefore = string.Empty;
string textAfter = string.Empty;
if (text.Contains(form))
{
textBefore = " "+text.Substring(0, text.IndexOf(form)).Trim()+" ";
textAfter = " "+text.Substring(text.IndexOf(form) + form.Length).Trim()+" ";
}
// while (text.Contains(form))
{
SToken tokenInside = new SToken();
tokenInside.Configure(token.TokenId, token.TokenString, token.TokenType, token.TokenCode);
tokenInside.Link(SToken.NULL, SToken.NULL, null, null);
SToken tokenBefore = new SToken();
tokenBefore.Configure(sToken.TokenId, textBefore, sToken.TokenType, sToken.TokenCode);
tokenBefore.Link(sToken.TokenBefore, tokenInside, null, null);
SToken tokenAfter = new SToken();
tokenAfter.Configure(sToken.TokenId, textAfter, sToken.TokenType, sToken.TokenCode);
//*** tokenAfter.Link(tokenInside, sToken.TokenAfter, null, null);
tokenInside.TokenBefore = tokenBefore;
if (sToken.TokenBefore != null)
{
//*** sToken.TokenBefore.TokenAfter = tokenBefore;
}
//*** if (sToken.TokenAfter != null)
{
//*** sToken.TokenAfter.TokenBefore = tokenAfter;
}
//*** tokenInside.TokenAfter = tokenAfter;
tokens.Add(tokenBefore);
tokens.Add(tokenInside);
tokens.Add(tokenAfter);
}
}
return tokens;
}
private SToken AdoptAsCurrent(SToken token)
{
if (lastToken != currentToken && lastTokenCollection != currentToken && isPreoperand)
{
if (currentToken.TokenMultiplicity == SElementMultiplicity.One)
{
lastToken = currentToken;
}
else
{
lastTokenCollection = currentToken;
}
currentToken = token;
}
else
{
currentToken = token;
}
return currentToken;
}
private void AddPropertyToClass(SToken token, SToken localToken, SElementMultiplicity sElementMultiplicity)
{
throw new NotImplementedException();
}
private string GetFirstCandidateToReplace(SToken currentToken, int offset)
{
bool found = false;
int start = -1, end = -1;
for (int i = offset; i < currentToken.TokenString.Length; i++)
{
if (char.IsLetter(currentToken.TokenString[i]))
{
if (start < 0)
{
start = i;
end = i;
found = true;
}
else
{
end = i;
}
}
else
{
if (found)
{
// if (!char.IsWhiteSpace(SubsritutedQuery[i]))
{
ReplaceAt = start;
return currentToken.TokenString.Substring(start, end + 1 - start);
}
}
}
}
if (found)
{
ReplaceAt = start;
return currentToken.TokenString.Substring(start, end + 1 - start);
}
else
return string.Empty;
}
private void InitializePersons()
{
tokenI = new SToken();
tokenYou = new SToken();
tokenShe = new SToken();
tokenHe = new SToken();
tokenWe = new List<SToken>();
tokenYous = new List<SToken>();
tokenThey = new List<SToken>();
tokenI.TokenString = "Florin";
tokenYou.TokenString = "Irina";
tokenShe.TokenString = "Irina";
tokenI.TokenMultiplicity = SElementMultiplicity.One;
tokenYou.TokenMultiplicity = SElementMultiplicity.One;
tokenShe.TokenMultiplicity = SElementMultiplicity.One;
tokenHe.TokenMultiplicity = SElementMultiplicity.One;
tokenI.TokenType = SElementType.Symbol;
tokenYou.TokenType = SElementType.Symbol;
tokenShe.TokenType = SElementType.Symbol;
tokenHe.TokenType = SElementType.Symbol;
tokenWe.Add(tokenI);
tokenWe.Add(tokenYou);
}
private bool NothingMoreToreplace(SToken currentToken, int offset)
{
for (int i = offset; i < currentToken.TokenString.Length; i++)
{
if (char.IsLetter(currentToken.TokenString[i]))
{ return false; }
}
return true;
}
public void ParseAsNatural()
{
StringBuilder currentPhrase = new StringBuilder();
bool handled = false;
foreach (string c in documentText.Split(new char[] { ' ' }, StringSplitOptions.RemoveEmptyEntries))
{
switch (c)
{
case ".":
//currentToken.TokenString = currentPhrase.ToString();
handled = true;
AddBlock(currentPhrase.ToString(), c);
currentPhrase = new StringBuilder();
break;
case "he":
case "she":
case "it":
handled = true;
currentToken = LastToken;
break;
case "increase": // Set influenc
case "decrease":
case "increasedBy": // Set influence
case "decreasedBy":
handled = true;
SetInfluence(currentToken, c);
currentPhrase = new StringBuilder();
break;
case "they":
handled = true;
newEvaluation = true;
CurrentToken = lastTokenCollection;
AppendToken(c);
//SetNext();
break;
case ",":
if (FoldNeeded)
{
CreateNewCollection();
}
else
{
AppendToCollection();
}
break;
case "and":
case "or":
EscapeFromCollection(c);
//currentToken.TokenString = currentPhrase.ToString();
handled = true;
//Append();
//AddToCollection(currentPhrase.ToString(), c);
currentPhrase = new StringBuilder();
//AppendT(c);
break;
case "can": // Set a function
handled = true;
if (currentToken.TokenDescribed == null)
currentToken.TokenDescribed = currentToken.TokenBefore;
if (currentToken.TokenDescribed.TokenActivities == null)
{
currentToken.TokenDescribed.TokenActivities = new List<SToken>();
}
currentToken.TokenDescribed.TokenActivities.Add(currentToken);
currentToken.Link(currentToken.TokenParent, null, null, null);
currentPhrase = new StringBuilder();
isPreoperand = false;
break;
case "have": // Set a property
handled = true;
currentToken.TokenDescribed = currentToken.TokenBefore;
if (currentToken.TokenDescribed.TokenProperties == null)
{
currentToken.TokenDescribed.TokenProperties = new List<SToken>();
}
currentToken.TokenDescribed.TokenProperties.Add(currentToken);
currentToken.TokenDescribed.TokenAfter = null;
currentToken.Link(currentToken.TokenParent, null, null, null);
currentPhrase = new StringBuilder();
nextIsAPropertyFoTheCurrent = true;
isPreoperand = false;
break;
case "a":
case "an":
case "some":
case "many":
case "all":
case "one":
case "no":
case "1":
case "0":
handled = true;
SetMultiplicityFlag(c);
currentPhrase = new StringBuilder();
break;
default:
// is it a multiplication factor?
int multiplicityLevel = 0;
int.TryParse(c, out multiplicityLevel);
if (multiplicityLevel != 0)
{
handled = true;
SetMultiplicityLevel(multiplicityLevel);
currentPhrase = new StringBuilder();
break;
}
handled = true;
string actualWord = c;
if (c.EndsWith("`s"))
{
nextIsAPropertyFoTheCurrent = true;
actualWord = c.Substring(0, c.Length - 2);
}
CurrentToken.TokenString = actualWord;
SToken existentToken = KnownTockens.FirstOrDefault(p => p.TokenString == actualWord);
if (existentToken != null)
{
CurrentToken = existentToken;
}
else
{
KnownTockens.Add(CurrentToken);
}
if (!SubstituteSymbols(currentToken))
{
if (CurrentToken.TokenMultiplicity == SElementMultiplicity.Undefined)
CurrentToken.TokenMultiplicity = currentToken.TokenParent.TokenMultiplicity;
}
//AppendToken(c);
if (CurrentToken.TokenDescribed != null)
{
if (CurrentToken.TokenDescribed.TokenInfluenceC != null)
{
foreach (SToken token in CurrentToken.TokenDescribed.TokenInfluenceC)
{
if (CurrentToken.TokenDescribed.TokenInfluenceC.Count > 0)
{
if (token.TokenInfluenceA == null)
{
token.TokenInfluenceA = new List<SToken>();
}
}
token.TokenInfluenceA.Add(CurrentToken.TokenDescribed);
}
CurrentToken.TokenDescribed.TokenInfluenceC = null;
}
if (CurrentToken.TokenDescribed.TokenInfluenceD != null)
{
}
}
AppendT(c);
break;
}
if (!handled)
{
currentPhrase.Append(" " + c);
}
else
{
handled = false;
}
//RefreshPersons();
}
}
private SToken FoldBelowWithPrewious(string c, SElementType[] conditionType)
{
// Fold Needed ?
bool foldNeeded = true;
if (currentToken.TokenBefore != null) // if there is nothing before fold
{
if (conditionType.Contains(currentToken.TokenBefore.TokenType)) // if there is another terminated fold before then fold
{
//foldNeeded = false;
}
if (currentToken.TokenBefore.TokenType == SElementType.Collection && c == "and") // Fold needed ?
{
foldNeeded = false;
}
if (true)
{
}
}
// Fold Location ?
SToken firstToFold = currentToken;
if (foldNeeded)
{
if (currentToken.TokenNodeType == SElementType.FoldClose && currentToken.TokenBefore.TokenNodeType != SElementType.CollectionAnd)
{
while (firstToFold.TokenBefore != null && firstToFold.TokenNodeType != SElementType.FoldOpen)
{
firstToFold = firstToFold.TokenBefore;
}
}
// Insert Folding Markers
SToken bracketOpenToken = new SToken();
bracketOpenToken.Configure(0, "(", SElementType.Collection, "");
bracketOpenToken.TokenNodeType = SElementType.FoldOpen;
bracketOpenToken.TokenMultiplicity = currentToken.TokenMultiplicity;
SToken bracketCloseToken = new SToken();
bracketCloseToken.Configure(0, ")", SElementType.Collection, "");
bracketCloseToken.TokenNodeType = SElementType.FoldClose;
bracketOpenToken.Link(firstToFold.TokenParent, firstToFold.TokenBefore, bracketCloseToken, new List<SToken>() { firstToFold });
if (bracketOpenToken.TokenBefore != null)
{
//*** bracketOpenToken.TokenBefore.TokenAfter = bracketOpenToken;
}
SToken localToken = firstToFold;
while (localToken != null)
{
localToken.TokenParent = bracketOpenToken;
//*** localToken = localToken.TokenAfter;
}
bracketCloseToken.Link(firstToFold.TokenParent, bracketOpenToken, null, null);
// If bracketOpenToken is the first in clilds replace the lider
if (bracketOpenToken.TokenBefore == null)
{
bracketOpenToken.TokenParent.TokenChilds.Remove(firstToFold);
bracketOpenToken.TokenParent.TokenChilds.Add(bracketOpenToken);
}
// Fold element
//*** firstToFold.Link(bracketOpenToken, null, firstToFold.TokenAfter, firstToFold.TokenChilds);
}
else
{
SToken localToken = null;
if (currentToken.TokenParent.TokenChilds.Count > 0)
{
localToken = currentToken.TokenParent.TokenChilds[0];
//*** while (localToken.TokenAfter != null)
{
//*** localToken = localToken.TokenAfter;
}
}
currentToken = localToken;
// localToken.TokenAfter = new SToken(0, "", SElementType.Undefined, "");
// currentToken.TokenBefore.TokenAfter = currentToken.TokenAfter;
// currentToken.TokenBefore = localToken;
// currentToken.TokenParent = localToken.TokenParent;
// currentToken.TokenBefore.TokenAfter = currentToken;
}
// Right
//sticked = true; // keep next token into array
return currentToken;
}
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;
}
public void AdoptOperator(int operatorId, string operatorText, List<string> forms, string code)
{
SToken token = new SToken();
token.Configure(operatorId, string.Empty, SElementType.Operator, operatorText);
token.Link(SToken.NULL, SToken.NULL, null, null);
StringBuilder tokenCollectionAsString = new StringBuilder();
SToken localCurrentToken = currentToken;
while (localCurrentToken != null)
{
for (int i = 0; i < forms.Count; i++)
{
if (localCurrentToken.TokenType == SElementType.Undefined)
{
if (localCurrentToken.TokenString.Contains(forms[i]))
{
List<SToken> explodedToken = ExplodeStoken(localCurrentToken, token, forms[i]);
currentToken.TokenParent.TokenChilds.InsertRange(currentToken.TokenParent.TokenChilds.IndexOf(localCurrentToken), explodedToken);
currentToken.TokenParent.TokenChilds.Remove(localCurrentToken);
i = 0;
localCurrentToken = explodedToken[explodedToken.Count - 1];
}
}
}
//*** localCurrentToken = localCurrentToken.TokenAfter;
}
}
