/// <summary>
/// Преобразовать наречие в прилагательное (это пока только для русского языка)
/// </summary>
/// <param name="adverb">наречие</param>
/// <param name="bi">род число падеж</param>
/// <return>прилагательное</return>
public static string ConvertAdverbToAdjective(string adverb, MorphBaseInfo bi)
{
if (adverb == null || (adverb.Length < 4))
{
return(null);
}
char last = adverb[adverb.Length - 1];
if (last != 'О' && last != 'Е')
{
return(adverb);
}
string var1 = adverb.Substring(0, adverb.Length - 1) + "ИЙ";
string var2 = adverb.Substring(0, adverb.Length - 1) + "ЫЙ";
MorphWordForm bi1 = GetWordBaseInfo(var1, null, false, false);
MorphWordForm bi2 = GetWordBaseInfo(var2, null, false, false);
string var = var1;
if (!bi1.Class.IsAdjective && bi2.Class.IsAdjective)
{
var = var2;
}
if (bi == null)
{
return(var);
}
return(m_Morph.GetWordform(var, MorphClass.Adjective, bi.Gender, bi.Case, bi.Number, MorphLang.Unknown, null) ?? var);
}
/// <summary>
/// Получить вариант написания словоформы
/// </summary>
/// <param name="word">слово</param>
/// <param name="morphInfo">морфологическая информация</param>
/// <return>вариант написания</return>
public static string GetWordform(string word, MorphBaseInfo morphInfo)
{
if (morphInfo == null || string.IsNullOrEmpty(word))
{
return(word);
}
MorphClass cla = morphInfo.Class;
if (cla.IsUndefined)
{
MorphWordForm mi0 = GetWordBaseInfo(word, null, false, false);
if (mi0 != null)
{
cla = mi0.Class;
}
}
string word1 = word;
foreach (char ch in word)
{
if (char.IsLower(ch))
{
word1 = word.ToUpper();
break;
}
}
MorphWordForm wf = morphInfo as MorphWordForm;
string res = m_Morph.GetWordform(word1, cla, morphInfo.Gender, morphInfo.Case, morphInfo.Number, morphInfo.Language, wf);
if (string.IsNullOrEmpty(res))
{
return(word);
}
return(res);
}
public void CopyFromWordForm(MorphWordForm src)
{
base.CopyFrom(src);
UndefCoef = src.UndefCoef;
NormalCase = src.NormalCase;
NormalFull = src.NormalFull;
Misc = src.Misc;
}
/// <summary>
/// Получить для словоформы род\число\падеж
/// </summary>
/// <param name="word">словоформа</param>
/// <param name="lang">возможный язык</param>
/// <param name="isCaseNominative">исходное слово в именительном падеже (иначе считается падеж любым)</param>
/// <param name="inDictOnly">при true не строить гипотезы для несловарных слов</param>
/// <return>базовая морфологическая информация</return>
public static MorphWordForm GetWordBaseInfo(string word, MorphLang lang = null, bool isCaseNominative = false, bool inDictOnly = false)
{
List <MorphToken> mt = m_Morph.Run(word, false, lang, false, null);
MorphWordForm bi = new MorphWordForm();
MorphClass cla = new MorphClass();
if (mt != null && mt.Count > 0)
{
for (int k = 0; k < 2; k++)
{
bool ok = false;
foreach (MorphWordForm wf in mt[0].WordForms)
{
if (k == 0)
{
if (!wf.IsInDictionary)
{
continue;
}
}
else if (wf.IsInDictionary)
{
continue;
}
if (isCaseNominative)
{
if (!wf.Case.IsNominative && !wf.Case.IsUndefined)
{
continue;
}
}
cla.Value |= wf.Class.Value;
bi.Gender |= wf.Gender;
bi.Case |= wf.Case;
bi.Number |= wf.Number;
if (wf.Misc != null && bi.Misc == null)
{
bi.Misc = wf.Misc;
}
ok = true;
}
if (ok || inDictOnly)
{
break;
}
}
}
bi.Class = cla;
return(bi);
}
/// <summary>
/// Лемма (вариант морфологической нормализации)
/// </summary>
public string GetLemma()
{
if (m_Lemma != null)
{
return(m_Lemma);
}
string res = null;
if (WordForms != null && WordForms.Count > 0)
{
if (WordForms.Count == 1)
{
res = WordForms[0].NormalFull ?? WordForms[0].NormalCase;
}
if (res == null && !CharInfo.IsAllLower)
{
foreach (MorphWordForm m in WordForms)
{
if (m.Class.IsProperSurname)
{
string s = m.NormalFull ?? m.NormalCase ?? "";
if (LanguageHelper.EndsWithEx(s, "ОВ", "ЕВ", null, null))
{
res = s;
break;
}
}
else if (m.Class.IsProperName && m.IsInDictionary)
{
return(m.NormalCase);
}
}
}
if (res == null)
{
MorphWordForm best = null;
foreach (MorphWordForm m in WordForms)
{
if (best == null)
{
best = m;
}
else if (this.CompareForms(best, m) > 0)
{
best = m;
}
}
res = best.NormalFull ?? best.NormalCase;
}
}
if (res != null)
{
if (LanguageHelper.EndsWithEx(res, "АНЫЙ", "ЕНЫЙ", null, null))
{
res = res.Substring(0, res.Length - 3) + "ННЫЙ";
}
else if (LanguageHelper.EndsWith(res, "ЙСЯ"))
{
res = res.Substring(0, res.Length - 2);
}
else if (LanguageHelper.EndsWith(res, "АНИЙ") && res == Term)
{
foreach (MorphWordForm wf in WordForms)
{
if (wf.IsInDictionary)
{
return(res);
}
}
return(res.Substring(0, res.Length - 1) + "Е");
}
return(res);
}
return(Term ?? "?");
}
int CompareForms(MorphWordForm x, MorphWordForm y)
{
string vx = x.NormalFull ?? x.NormalCase;
string vy = y.NormalFull ?? y.NormalCase;
if (vx == vy)
{
return(0);
}
if (string.IsNullOrEmpty(vx))
{
return(1);
}
if (string.IsNullOrEmpty(vy))
{
return(-1);
}
char lastx = vx[vx.Length - 1];
char lasty = vy[vy.Length - 1];
if (x.Class.IsProperSurname && !CharInfo.IsAllLower)
{
if (LanguageHelper.EndsWithEx(vx, "ОВ", "ЕВ", "ИН", null))
{
if (!y.Class.IsProperSurname)
{
return(-1);
}
}
}
if (y.Class.IsProperSurname && !CharInfo.IsAllLower)
{
if (LanguageHelper.EndsWithEx(vy, "ОВ", "ЕВ", "ИН", null))
{
if (!x.Class.IsProperSurname)
{
return(1);
}
if (vx.Length > vy.Length)
{
return(-1);
}
if (vx.Length < vy.Length)
{
return(1);
}
return(0);
}
}
if (x.Class == y.Class)
{
if (x.Class.IsAdjective)
{
if (lastx == 'Й' && lasty != 'Й')
{
return(-1);
}
if (lastx != 'Й' && lasty == 'Й')
{
return(1);
}
if (!LanguageHelper.EndsWith(vx, "ОЙ") && LanguageHelper.EndsWith(vy, "ОЙ"))
{
return(-1);
}
if (LanguageHelper.EndsWith(vx, "ОЙ") && !LanguageHelper.EndsWith(vy, "ОЙ"))
{
return(1);
}
}
if (x.Class.IsNoun)
{
if (x.Number == MorphNumber.Singular && y.Number == MorphNumber.Plural && vx.Length <= (vy.Length + 1))
{
return(-1);
}
if (x.Number == MorphNumber.Plural && y.Number == MorphNumber.Singular && vx.Length >= (vy.Length - 1))
{
return(1);
}
}
if (vx.Length < vy.Length)
{
return(-1);
}
if (vx.Length > vy.Length)
{
return(1);
}
return(0);
}
if (x.Class.IsAdverb)
{
return(1);
}
if (x.Class.IsNoun && x.IsInDictionary)
{
if (y.Class.IsAdjective && y.IsInDictionary)
{
if (!y.Misc.Attrs.Contains("к.ф."))
{
return(1);
}
}
return(-1);
}
if (x.Class.IsAdjective)
{
if (!x.IsInDictionary && y.Class.IsNoun && y.IsInDictionary)
{
return(1);
}
return(-1);
}
if (x.Class.IsVerb)
{
if (y.Class.IsNoun || y.Class.IsAdjective || y.Class.IsPreposition)
{
return(1);
}
return(-1);
}
if (y.Class.IsAdverb)
{
return(-1);
}
if (y.Class.IsNoun && y.IsInDictionary)
{
return(1);
}
if (y.Class.IsAdjective)
{
if (((x.Class.IsNoun || x.Class.IsProperSecname)) && x.IsInDictionary)
{
return(-1);
}
if (x.Class.IsNoun && !y.IsInDictionary)
{
if (vx.Length < vy.Length)
{
return(-1);
}
}
return(1);
}
if (y.Class.IsVerb)
{
if (x.Class.IsNoun || x.Class.IsAdjective || x.Class.IsPreposition)
{
return(-1);
}
if (x.Class.IsProper)
{
return(-1);
}
return(1);
}
if (vx.Length < vy.Length)
{
return(-1);
}
if (vx.Length > vy.Length)
{
return(1);
}
return(0);
}