[InlineData(SoraniReh, SoraniOpenYeh, SoraniWaw, SoraniYeh)] // رێوی => ڕێوی
public void Should_Change_Initial_r_To_R(params char[] chars)
{
var text = new string(chars);
var normalized = Sorani.ToStandardSorani(text);
Assert.Equal(SoraniHeavyReh, normalized[0]);
}
private void AssertSimpleSoraniNormalizer(string text, char find, char replace)
{
var normalized = Sorani.ToStandardSorani(text);
Assert.True(normalized.Contains(replace));
Assert.False(normalized.Contains(find));
}
[InlineData(SoraniSeen, SoraniAe, SoraniReh, SoraniDal, SoraniAlef, SoraniReh)] // سەردار
public void Should_Not_Change_Medial_r_To_R(params char[] chars)
{
var text = new string(chars);
var normalized = Sorani.ToStandardSorani(text);
Assert.Equal(text, normalized);
}
[Fact] // ازاد => ئازاد
public void Should_Insert_Hamza_To_The_Begenning_Of_The_Words_That_Start_With_Aelf()
{
var text = $"{SoraniAlef}{SoraniZeh}{SoraniAlef}{SoraniDal}"; // ازاد
var normalized = Sorani.ToStandardSorani(text);
Assert.Equal(SoraniHamza, normalized[0]);
Assert.Equal(SoraniAlef, normalized[1]);
}
[InlineData(SoraniBeh, SoraniHamza, SoraniGaf, SoraniWaw, SoraniMeem, SoraniAlef, SoraniNoon)] // بئگومان => بێگومان
public void Should_Replace_Hamza_Followed_By_Consonants_With_OpenYeh(params char[] chars)
{
var text = new string(chars);
var normalized = Sorani.ToStandardSorani(text);
Assert.True(normalized.Contains(SoraniOpenYeh));
Assert.False(normalized.Contains(SoraniHamza));
}
[InlineData(SoraniLam, SoraniAlef, SoraniReh)] // لار
public void Should_Not_Replace_Lam_Not_Followed_By_Fatha(params char[] chars)
{
var text = new string(chars);
var normalized = Sorani.ToStandardSorani(text);
Assert.True(normalized.Contains(SoraniLam));
Assert.False(normalized.Contains(SoraniHeavyLam));
}
[Fact] // ئئستا => ئێستا
public void Should_Replace_Second_Hamza_But_Not_First_Hamza()
{
var text = $"{SoraniHamza}{SoraniHamza}{SoraniSeen}{SoraniTeh}{SoraniAlef}"; // ئێستا
var normalized = Sorani.ToStandardSorani(text);
Assert.Equal(SoraniHamza, normalized[0]);
Assert.Equal(SoraniOpenYeh, normalized[1]);
}
[InlineData(SoraniKaf, SoraniReh, ArabicKasra, SoraniYeh, SoraniNoon)] // کرِین => کڕین
public void Should_Replace_Reh_Followed_By_Kasra_With_HeavyReh(params char[] chars)
{
var text = new string(chars);
var normalized = Sorani.ToStandardSorani(text);
Assert.True(normalized.Contains(SoraniHeavyReh));
Assert.False(normalized.Contains(ArabicKasra));
Assert.False(normalized.Contains(SoraniReh));
}
public void Normalize_A_Sentence()
{
var text = "ثيَش هةوليَر ضووم بؤ سليَماني. لةويَ ضووم بؤ بازارِ. زؤر شتي جوانم كرِي.";
var expected = "پێش هەولێر چووم بۆ سلێمانی. لەوێ چووم بۆ بازاڕ. زۆر شتی جوانم کڕی.";
var normalized = Sorani.ToStandardSorani(text);
Assert.Equal(expected, normalized);
}
[InlineData(SoraniHeavyReh, ArabicWawWithHamzaAbove, SoraniJeh)] // ڕؤژ => ڕۆژ
public void Should_Replace_Waw_Followed_By_Fatha_To_Oe(params char[] chars)
{
var text = new string(chars);
var normalized = Sorani.ToStandardSorani(text);
Assert.True(normalized.Contains(SoraniOe));
Assert.False(normalized.Contains(ArabicFatha));
Assert.False(normalized.Contains(SoraniWaw));
Assert.False(normalized.Contains(ArabicWawWithHamzaAbove));
}
[InlineData(SoraniHeavyReh, ArabicYeh, ArabicFatha, SoraniZeh)] // ڕيَز => ڕێز
public void Should_Replace_Yeh_Followed_By_Kasra_To_OpenYeh(params char[] chars)
{
var text = new string(chars);
var normalized = Sorani.ToStandardSorani(text);
Assert.True(normalized.Contains(SoraniOpenYeh));
Assert.False(normalized.Contains(ArabicFatha));
Assert.False(normalized.Contains(ArabicYeh));
Assert.False(normalized.Contains(ArabicAlefMaksura));
}
[InlineData(LatinNine, SoraniNine)] // 9 => ٩
public void Should_Replace_Non_Standard_Char(char find, char replace)
{
var text = find.ToString();
var normalized = Sorani.ToStandardSorani(text);
Assert.True(normalized == replace.ToString());
text = $"{SoraniAlef}{find}";
AssertSimpleSoraniNormalizer(text, find, replace);
text = $"{find}{SoraniAlef}";
AssertSimpleSoraniNormalizer(text, find, replace);
text = $"{SoraniBeh}{find}";
AssertSimpleSoraniNormalizer(text, find, replace);
text = $"{find}{SoraniBeh}";
AssertSimpleSoraniNormalizer(text, find, replace);
}
public void Normalize_A_Very_Long_Text(bool includeSpace)
{
var timer = new Stopwatch();
var random = new Random();
var space = includeSpace ? new char[] { Space } : new char[] { };
var ignoredChars = new char[] { ArabicSad, ArabicKasra, ArabicFatha, ZeroWidthNonJoiner };
var troublesomeChars = NonStandardSoraniAlphabet.Except(ignoredChars);
var charSet = troublesomeChars.Union(SoraniAlphabet)
.Union(space)
.ToArray();
var text = new string(Enumerable.Repeat(charSet, 10000)
.Select(set => set[random.Next(charSet.Length)]).ToArray());
timer.Start();
var normalized = Sorani.ToStandardSorani(text);
timer.Stop();
Assert.True(timer.ElapsedMilliseconds < 1000);
Assert.False(normalized.Any(c => troublesomeChars.Contains(c)));
}
public void Passing_Null_Should_Throw_Exception()
{
Assert.Throws <ArgumentNullException>(() =>
Sorani.ToStandardSorani(null));
}
public void Should_Return_Empty_String_When_Given_Empty_String()
{
var normalized = Sorani.ToStandardSorani(string.Empty);
Assert.Equal(string.Empty, normalized);
}
