Esempio n. 1
0
        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
        }
Esempio n. 2
0
        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
        }