public char joined; // Resulting character.
#region IComparable Members
// Note we're sorting these to get the longest ones first.
public int CompareTo(object obj)
{
if (obj is LigatureEntry)
{
LigatureEntry le = (LigatureEntry)obj;
return(-chars.Length.CompareTo(le.chars.Length));
}
Debug.Assert(false, "Object is not a LigatureEntry");
return(0);
}
public static void Init()
{
//
// Mirroring data.
//
string[] mirrorData = Storage4.ReadAllLines(@"Content\Xml\Unicode\BidiMirroring.txt", StorageSource.TitleSpace);
if (mirrorData != null)
{
for (int i = 0; i < mirrorData.Length; i++)
{
if (!String.IsNullOrEmpty(mirrorData[i]) && !mirrorData[i].StartsWith("#"))
{
string[] fields = mirrorData[i].Split(new char[] { ';', '#' });
// Note, this will throw when it tries to parse the lines in the header.
// Feel free to ignore those lines.
try
{
int code = int.Parse(fields[0], NumberStyles.HexNumber);
int mirror = int.Parse(fields[1], NumberStyles.HexNumber);
mirrorDict.Add((char)code, (char)mirror);
}
catch
{
}
}
} // end of loop over each line.
}
string[] unicodeData = Storage4.ReadAllLines(@"Content\Xml\Unicode\UnicodeData.txt", StorageSource.TitleSpace);
if (unicodeData != null)
{
for (int i = 0; i < unicodeData.Length; i++)
{
string[] fields = unicodeData[i].Split(';');
int code = 0;
// If we throw trying to parse the code, just skip this glyph.
try
{
code = int.Parse(fields[0], NumberStyles.HexNumber);
}
catch
{
continue;
}
UnicodeCharData data = new UnicodeCharData();
data.c = (char)code;
data.name = fields[1];
data.type = fields[4];
data.decompositionMapping = fields[5];
data.mirrored = fields[9] == "Y";
// Ignore Arabic contextual forms for ligatures. This was causing too much breakage.
if (data.name.Contains("LIGATURE") && data.name.Contains("ARABIC"))
{
continue;
}
if (data.decompositionMapping.StartsWith("<isolated>"))
{
// Parse may throw. If it does, just ignore this line.
try
{
data.baseForm = (char)int.Parse(fields[5].Substring(fields[5].Length - 4), NumberStyles.HexNumber);
DecompMappingData dm = null;
decompMappingDict.TryGetValue(data.baseForm, out dm);
if (dm == null)
{
dm = new DecompMappingData();
decompMappingDict.Add(data.baseForm, dm);
dm.c = data.baseForm;
}
dm.isolated = data.c;
}
catch { }
}
else if (data.decompositionMapping.StartsWith("<initial>"))
{
// Parse may throw. If it does, just ignore this line.
try
{
data.baseForm = (char)int.Parse(fields[5].Substring(fields[5].Length - 4), NumberStyles.HexNumber);
DecompMappingData dm = null;
decompMappingDict.TryGetValue(data.baseForm, out dm);
if (dm == null)
{
dm = new DecompMappingData();
decompMappingDict.Add(data.baseForm, dm);
dm.c = data.baseForm;
}
dm.initial = data.c;
}
catch { }
}
else if (data.decompositionMapping.StartsWith("<medial>"))
{
// Parse may throw. If it does, just ignore this line.
try
{
data.baseForm = (char)int.Parse(fields[5].Substring(fields[5].Length - 4), NumberStyles.HexNumber);
DecompMappingData dm = null;
decompMappingDict.TryGetValue(data.baseForm, out dm);
if (dm == null)
{
dm = new DecompMappingData();
decompMappingDict.Add(data.baseForm, dm);
dm.c = data.baseForm;
}
dm.medial = data.c;
}
catch { }
}
else if (data.decompositionMapping.StartsWith("<final>"))
{
// Parse may throw. If it does, just ignore this line.
try
{
data.baseForm = (char)int.Parse(fields[5].Substring(fields[5].Length - 4), NumberStyles.HexNumber);
DecompMappingData dm = null;
decompMappingDict.TryGetValue(data.baseForm, out dm);
if (dm == null)
{
dm = new DecompMappingData();
decompMappingDict.Add(data.baseForm, dm);
dm.c = data.baseForm;
}
dm.final = data.c;
}
catch { }
}
// Get ligature data from Decompostion Mapping field. Note that we're filtering
// on only getting Arabic ligatures and ignoring other languages. This is because
// it's too easy to find ligatures with no matching glyphs.
if (data.name.Contains("ARABIC LIGATURE") && !string.IsNullOrEmpty(data.decompositionMapping))
{
// HACK Finding too many missing characters so only look for the one required ligature.
if (data.c != 0xfefb)
{
continue;
}
// Parse out the chars that combine to form this ligature.
int index = data.decompositionMapping.LastIndexOf('>');
string[] args = data.decompositionMapping.Substring(index + 1).Trim().Split(' ');
// Parse may throw. If it does, just ignore this ligature.
try
{
char c0 = (char)int.Parse(args[0], NumberStyles.HexNumber);
char[] chars = new char[args.Length - 1];
for (int c = 0; c < args.Length - 1; c++)
{
chars[c] = (char)int.Parse(args[c + 1], NumberStyles.HexNumber);
}
// Create or add to the ligature entry. Note that the entry
// uses the first character of the pair for the key.
LigatureData ld = null;
ligatureDict.TryGetValue(c0, out ld);
// Create new dictionary entry if needed.
if (ld == null)
{
ld = new LigatureData();
ld.c = c0;
ld.ligatures = new List <LigatureEntry>();
ligatureDict.Add(c0, ld);
}
// Add new entry.
LigatureEntry le = new LigatureEntry();
le.chars = chars;
le.joined = data.c;
ld.ligatures.Add(le);
}
catch { }
}
if (data.mirrored)
{
mirrorDict.TryGetValue(data.c, out data.mirroredChar);
}
if (code >= 0 && code <= 0xFFFF) //UTF-16 BMP code points only
{
bool inRange = data.name.EndsWith(", First>");
if (inRange) // Add all characters within a specified range.
{
data.name.Replace(", First", String.Empty); // Remove range indicator from name .
fields = unicodeData[++i].Split(';');
// Parse may throw. If it does, just ignore this range?
// Not totally sure if this is the right thing to do...
try
{
int endCharCode = int.Parse(fields[0], NumberStyles.HexNumber);
if (!fields[1].EndsWith(", Last>"))
{
throw new Exception("Expected end-of-range indicator.");
}
for (int codeInRange = code; codeInRange <= endCharCode; codeInRange++)
{
charDict.Add((char)codeInRange, data);
}
}
catch { }
}
else
{
charDict.Add((char)code, data);
}
}
} // end of loop over each line.
}
// For ligatures, we need to be greedy and try to apply longer conversions
// before shorter ones. So, sort all the ligatures in the dictionary by
// the number of characters putting the longest first.
foreach (LigatureData ld in ligatureDict.Values)
{
ld.ligatures.Sort();
}
} // end of Init()
