/// <summary>
/// Replace Shadda + Another Tashkeel with combined form
/// </summary>
public static void FixShaddaCombinations(FastStringBuilder input)
{
/*
* Fix of https://github.com/mnarimani/RTLTMPro/issues/13
*/
int j = 0; // write index
int i = 0; // read index
while (i < input.Length)
{
int curChar = input.Get(i);
int nextChar = i < input.Length - 1 ? input.Get(i + 1) : (char)0;
if ((TashkeelCharacters)curChar == TashkeelCharacters.Shadda &&
ShaddaCombinationMap.ContainsKey((char)nextChar))
{
input.Set(j, ShaddaCombinationMap[(char)nextChar]);
j++;
i += 2;
}
else
{
input.Set(j, curChar);
j++;
i++;
}
}
input.Length = j;
}
/// <summary>
/// Removes tashkeel. Converts general RTL letters to isolated form. Also fixes Farsi and Arabic ی letter.
/// </summary>
/// <param name="text">Input to prepare</param>
/// <param name="farsi"></param>
/// <returns>Prepared input in char array</returns>
public static void FixYah(FastStringBuilder text, bool farsi)
{
for (int i = 0; i < text.Length; i++)
{
if (farsi && text.Get(i) == (int)ArabicGeneralLetters.Ya)
{
text.Set(i, (char)ArabicGeneralLetters.PersianYa);
}
else if (farsi == false && text.Get(i) == (int)ArabicGeneralLetters.PersianYa)
{
text.Set(i, (char)ArabicGeneralLetters.Ya);
}
}
}
/// <summary>
/// Handles the special Lam-Alef connection in the text.
/// </summary>
/// <param name="input"></param>
/// <param name="output"></param>
/// <param name="i">Index of Lam letter</param>
/// <returns><see langword="true" /> if special connection has been made.</returns>
private static bool HandleSpecialLam(FastStringBuilder input, FastStringBuilder output, int i)
{
bool isFixed;
switch (input.Get(i + 1))
{
case (char)ArabicGeneralLetters.AlefMaksoor:
output.Set(i, (char)0xFEF7);
isFixed = true;
break;
case (char)ArabicGeneralLetters.Alef:
output.Set(i, (char)0xFEF9);
isFixed = true;
break;
case (char)ArabicGeneralLetters.AlefHamza:
output.Set(i, (char)0xFEF5);
isFixed = true;
break;
case (char)ArabicGeneralLetters.AlefMad:
output.Set(i, (char)0xFEF3);
isFixed = true;
break;
default:
isFixed = false;
break;
}
if (isFixed)
{
output.Set(i + 1, (char)0xFFFF);
}
return(isFixed);
}
/// <summary>
/// Removes tashkeel from text.
/// </summary>
public static void RemoveTashkeel(FastStringBuilder input)
{
TashkeelLocations.Clear();
int j = 0; // write index
for (int i = 0; i < input.Length; i++)
{
int curChar = input.Get(i);
if (Char32Utils.IsUnicode16Char(curChar) && TashkeelCharactersSet.Contains((char)curChar))
{
TashkeelLocations.Add(new TashkeelLocation((TashkeelCharacters)curChar, i));
}
else
{
input.Set(j, curChar);
j++;
}
}
input.Length = j;
}
/// <summary>
/// Converts English numbers that are outside tags to Persian or Arabic numbers.
/// </summary>
/// <param name="text"></param>
/// <param name="farsi"></param>
/// <returns>Text with converted numbers</returns>
public static void FixNumbersOutsideOfTags(FastStringBuilder text, bool farsi)
{
var englishDigits = new HashSet <char>(EnglishToFarsiNumberMap.Keys);
for (int i = 0; i < text.Length; i++)
{
var iChar = text.Get(i);
// skip valid tags
if (iChar == '<')
{
bool sawValidTag = false;
for (int j = i + 1; j < text.Length; j++)
{
int jChar = text.Get(j);
if ((j == i + 1 && jChar == ' ') || jChar == '<')
{
break;
}
else if (jChar == '>')
{
i = j;
sawValidTag = true;
break;
}
}
if (sawValidTag)
{
continue;
}
}
if (englishDigits.Contains((char)iChar))
{
text.Set(i, farsi ? EnglishToFarsiNumberMap[(char)iChar] : EnglishToHinduNumberMap[(char)iChar]);
}
}
}
/// <summary>
/// Fixes the shape of letters based on their position.
/// </summary>
/// <param name="input"></param>
/// <param name="output"></param>
/// <param name="preserveNumbers"></param>
/// <param name="farsi"></param>
/// <returns></returns>
public static void Fix(FastStringBuilder input, FastStringBuilder output, bool preserveNumbers, bool farsi, bool fixTextTags)
{
FixYah(input, farsi);
output.SetValue(input);
for (int i = 0; i < input.Length; i++)
{
bool skipNext = false;
int iChar = input.Get(i);
// For special Lam Letter connections.
if (iChar == (int)ArabicGeneralLetters.Lam)
{
if (i < input.Length - 1)
{
skipNext = HandleSpecialLam(input, output, i);
if (skipNext)
{
iChar = output.Get(i);
}
}
}
// We don't want to fix tatweel or zwnj character
if (iChar == (int)ArabicGeneralLetters.ArabicTatweel ||
iChar == (int)ArabicGeneralLetters.ZeroWidthNoJoiner)
{
continue;
}
if (iChar < 0xFFFF && TextUtils.IsGlyphFixedArabicCharacter((char)iChar))
{
char converted = GlyphTable.Convert((char)iChar);
if (IsMiddleLetter(input, i))
{
output.Set(i, (char)(converted + 3));
}
else if (IsFinishingLetter(input, i))
{
output.Set(i, (char)(converted + 1));
}
else if (IsLeadingLetter(input, i))
{
output.Set(i, (char)(converted + 2));
}
}
// If this letter as Lam and special Lam-Alef connection was made, We want to skip the Alef
// (Lam-Alef occupies 1 space)
if (skipNext)
{
i++;
}
}
if (!preserveNumbers)
{
if (fixTextTags)
{
FixNumbersOutsideOfTags(output, farsi);
}
else
{
FixNumbers(output, farsi);
}
}
}