public void WriteCodes2()
{
EnsureCodeMapBuild();
//int tot = 0;
//foreach ( var pair in _codeMap ) {
// if ( pair.Key != pair.Value ) {
// Debug.WriteLine( $"{pair.Key:X4}:{pair.Value:X4}" );
// tot++;
// }
//}
var map = new InterleaveMap();
for (int i = 0; i <= ushort.MaxValue; i++)
{
if (_codeMap.TryGetValue((ushort)i, out ushort value))
{
map.Add((char)i, i ^ value);
}
}
using (var w = File.Create(@"F:\Dev\GitHub\ecl.Unicode\src\GenIOCMap\Scanner2.bin"))
using (var b = new BinaryWriter(w)) {
map.Save(b);
WriteComposites(b);
}
}
public void WriteScriptMap()
{
_loader.EnsureDataLoaded();
_loader.EnsureScriptsLoaded();
EnsureCodeMapBuild();
var map = new Dictionary <int, ScriptSystem>();
foreach (var entry in _loader.GetCodePoints(0, ushort.MaxValue))
{
if (entry.Script != 0)
{
ScriptSystem sys = GetScriptSystem(entry.Script);
map.Add(entry.CodeValue, sys);
}
}
#if true1
using (var w = File.CreateText(@"f:\_tests\Del\ScriptMap.txt")) {
for (int i = 0; i <= ushort.MaxValue; i++)
{
if (map.TryGetValue(i, out ScriptSystem value))
{
}
string text = $"{i:X4} ";
if (_loader.TryGetEntry(i, out UnicodeEntry entry))
{
if (IsPrintable(entry.Category))
{
text += entry + " ";
}
if (entry.Script != 0)
{
text += "[" + entry.Script + "] ";
}
text += entry.Name + " ";
}
w.WriteLine(text + value);
}
}
#else
InterleaveMap imap = new InterleaveMap();
for (int i = 0; i <= ushort.MaxValue; i++)
{
if (map.TryGetValue(i, out ScriptSystem value))
{
imap.Add((char)i, (int)value);
}
}
using (var w = File.Create(@"f:\_tests\Del\ScriptMap.bin"))
using (var b = new BinaryWriter(w)){
imap.SaveByte(b);
}
#endif
}
public void BuildToUpperMap(bool text = false)
{
var map = new InterleaveMap();
foreach (KeyValuePair <int, int> pair in _iocMap)
{
if (!_iocMap.TryGetValue(pair.Value, out int key) ||
key != pair.Key)
{
Console.WriteLine("{0:X6} != {1:X6}", key, pair.Key);
}
else
{
UnicodeEntry left = _loader[pair.Key];
//var right = _loader[ pair.Value ];
if (left.Uppercase != 0)
{
if (left.Uppercase != pair.Value)
{
Console.WriteLine("{0:X6} != {1:X6}", key, pair.Key);
}
else
{
int diff = pair.Value ^ pair.Key;
map.Add((char)pair.Key, diff);
}
}
else if (left.LowerCase != pair.Value)
{
Console.WriteLine("{0:X6} != {1:X6}", key, pair.Key);
}
}
}
if (text)
{
using (var w = File.CreateText(@"../../bin/OrdinalIgnoreCaseMap.cs")) {
map.Write(w);
}
}
else
{
using (var w = File.Create(@"../../bin/OrdinalIgnoreCaseMap.bin"))
using (var b = new BinaryWriter(w)) {
map.Save(b);
}
}
}
public void WriteCodes3()
{
EnsureCodeMapBuild();
var map = new InterleaveMap();
for (int i = 0; i <= ushort.MaxValue; i++)
{
if (_codeMap.TryGetValue((ushort)i, out ushort value) && value >= 32)
{
map.AddSet((char)i, i ^ value);
}
else
{
map.Add((char)i, 0);
}
}
using (var w = File.Create(@"F:\Dev\GitHub\ecl.Unicode\src\GenIOCMap\Scanner2.bin"))
using (var b = new BinaryWriter(w)) {
map.SaveMasked(b);
WriteComposites(b);
}
}
public void WriteBreakMap()
{
EnumRange <WordBreak>[] breaks = _loader.LoadWordBreak();
var map = new Dictionary <int, byte>();
foreach (var range in breaks)
{
//switch ( range.Value ) {
//case WordBreak.MidLetter:
// continue;
//}
for (int i = range.Begin; i <= range.End; i++)
{
if (!map.ContainsKey(i))
{
map[i] = (byte)range.Value;
}
else
{
Debug.WriteLine("DF");
}
}
}
map.Add(160, (byte)MoreWord.Connector); // Non Breaking space
map.Add('\t', (byte)WordBreak.WSegSpace); // Tab
foreach (var entry in _loader.GetCodePoints(0, ushort.MaxValue))
{
if (!map.ContainsKey(entry.CodeValue))
{
switch (entry.Category)
{
case UnicodeCharacterType.OtherPrivateUse:
map.Add(entry.CodeValue, (byte)MoreWord.Private);
break;
case UnicodeCharacterType.OtherSurrogate:
map.Add(entry.CodeValue, (byte)MoreWord.Surrogate);
break;
case UnicodeCharacterType.OtherControl:
map.Add(entry.CodeValue, (byte)MoreWord.Control);
continue;
case UnicodeCharacterType.LetterUppercase:
case UnicodeCharacterType.LetterLowercase:
case UnicodeCharacterType.LetterTitlecase:
map.Add(entry.CodeValue, 52);
break;
case UnicodeCharacterType.LetterOther:
map.Add(entry.CodeValue, (byte)MoreWord.Ideograph);
break;
case UnicodeCharacterType.NumberOther:
case UnicodeCharacterType.LetterModifier:
map.Add(entry.CodeValue, (byte)MoreWord.Symbol);
break;
case UnicodeCharacterType.NumberDecimalDigit:
case UnicodeCharacterType.MarkEnclosing:
map.Add(entry.CodeValue, (byte)MoreWord.Symbol);
break;
case UnicodeCharacterType.NumberLetter:
case UnicodeCharacterType.SymbolMath:
case UnicodeCharacterType.SymbolCurrency:
case UnicodeCharacterType.SymbolModifier:
case UnicodeCharacterType.SymbolOther:
map.Add(entry.CodeValue, (byte)MoreWord.Symbol);
break;
case UnicodeCharacterType.PunctuationConnector:
map.Add(entry.CodeValue, (byte)MoreWord.Punctuation);
break;
case UnicodeCharacterType.PunctuationDash:
case UnicodeCharacterType.PunctuationOpen:
case UnicodeCharacterType.PunctuationClose:
case UnicodeCharacterType.PunctuationInitialQuote:
case UnicodeCharacterType.PunctuationFinalQuote:
case UnicodeCharacterType.PunctuationOther:
map.Add(entry.CodeValue, (byte)MoreWord.Punctuation);
break;
case UnicodeCharacterType.OtherFormat:
case UnicodeCharacterType.SeparatorSpace:
map.Add(entry.CodeValue, (byte)MoreWord.WhiteSpace);
break;
case UnicodeCharacterType.MarkSpacingCombining:
case UnicodeCharacterType.SeparatorLine:
case UnicodeCharacterType.SeparatorParagraph:
map.Add(entry.CodeValue, (byte)MoreWord.LineSeparator);
break;
default:
map.Add(entry.CodeValue, 60);
break;
}
}
}
#if true1
using (var w = File.CreateText(@"f:\_tests\Del\WordBreak.txt")) {
for (int i = 0; i <= ushort.MaxValue; i++)
{
string text = $"{i:X4} ";
if (_loader.TryGetEntry(i, out UnicodeEntry entry))
{
if (IsPrintable(entry.Category))
{
text += entry + " ";
}
text += entry.Name + " ";
}
map.TryGetValue(i, out var value);
w.WriteLine(text + value);
}
}
#else
InterleaveMap imap = new InterleaveMap();
for (int i = 0; i <= ushort.MaxValue; i++)
{
if (map.TryGetValue(i, out byte value))
{
imap.Add((char)i, value);
}
}
using (var w = File.Create(@"f:\_tests\Del\WordBreak2.bin"))
using (var b = new BinaryWriter(w)){
imap.SaveByte(b);
}
#endif
}