public void MethodGetCharInfoShouldNeverFail()
{
for (int i = 0; i <= 0x10FFFF; i++)
{
UnicodeInfo.GetCharInfo(i);
}
}
private static List <IndexEntry> BuildUnicodeIndex()
{
var sw = Stopwatch.StartNew();
var result = new List <IndexEntry>();
var blocks = UnicodeInfo.GetBlocks();
foreach (var block in blocks)
{
foreach (var codepoint in block.CodePointRange)
{
if (char.IsSurrogate((char)codepoint))
{
continue;
}
var charInfo = UnicodeInfo.GetCharInfo(codepoint);
var displayText = charInfo.Name;
if (displayText != null)
{
result.Add(new(charInfo, displayText, displayText.ToUpperInvariant()));
}
}
}
Console.WriteLine("Index built in " + sw.Elapsed);
return(result);
}
public static Models.CodePoint[] GetCodePoints(int offset, int limit)
{
if (offset < 0 || offset > 0x10FFFF)
{
throw new ArgumentOutOfRangeException(nameof(offset));
}
if (limit < 1)
{
throw new ArgumentOutOfRangeException(nameof(limit));
}
var codePoints = new List <Models.CodePoint>();
int max = Math.Min(offset + limit - 1, 0x10FFFF);
for (int i = offset; i <= max; i++)
{
var charInfo = UnicodeInfo.GetCharInfo(i);
if (charInfo.Category != UnicodeCategory.OtherNotAssigned)
{
codePoints.Add(GetCodePoint(charInfo));
}
}
return(codePoints.ToArray());
}
private string GetResult(int value)
{
var sb = new StringBuilder();
char ch = (char)value;
bool isSurrogate = char.IsSurrogate(ch);
string text = null;
if (!isSurrogate)
{
text = char.ConvertFromUtf32(value);
sb.AppendLine(DivClass(Escape(text), "charSample"));
}
if (descriptions.TryGetValue(value, out string description))
{
sb.AppendLine(Div(description));
}
var info = UnicodeInfo.GetCharInfo(value);
sb.AppendLine(TableStart("smallTable"));
sb.AppendLine(Tr(Td(Gray("Code point:")), Td($"{value} (U+{value.ToHex()})")));
sb.AppendLine(Tr(Td(Gray("Category:")), Td(CharUnicodeInfo.GetUnicodeCategory(ch).ToString())));
sb.AppendLine(Tr(Td(Gray("Block:")), Td(info.Block)));
sb.AppendLine(Tr(Td(Gray("Escape:")), Td(DivClass(GetEscapeString(value), "fixed"))));
if (text != null)
{
sb.AppendLine(Tr(Td(Gray("UTF-8:")), Td(GetUtf8(text))));
}
sb.AppendLine("</table>");
return(sb.ToString());
}
public void Show(FontFamily font, char glyph, string code)
{
lblGlyph.FontFamily = font;
lblGlyph.Content = glyph.ToString();
this.Title = $"Details - U+{code}";
var info = UnicodeInfo.GetCharInfo(glyph);
lblName.Content = info.Name.Replace("WITH", "\nWITH");
lblName.ToolTip = info.Name;
tbxCode.Text = $"U+{code} (&#x{code};) [Alt+{Convert.ToInt32(code, 16),0:D4}]";
if (!string.IsNullOrEmpty(info.OldName))
{
PrintCharInfo("Old Name", info.OldName);
}
PrintCharInfo("Category", info.Category.ToString());
PrintCharInfo("Block", info.Block);
PrintCharInfo("Canoncial Combining Class", info.CanonicalCombiningClass.ToString());
PrintCharInfo("Bidirectional Class", info.BidirectionalClass.ToString());
PrintCharInfo("Contributory Properties", info.ContributoryProperties.ToString());
PrintCharInfo("Core Properties", info.CoreProperties.ToString());
this.Show();
}
private static void PrintCodePointInfo(int codePoint)
{
var charInfo = UnicodeInfo.GetCharInfo(codePoint);
Console.WriteLine(UnicodeInfo.GetDisplayText(charInfo));
Console.WriteLine("U+" + codePoint.ToString("X4"));
Console.WriteLine(charInfo.Name ?? charInfo.OldName);
Console.WriteLine(charInfo.Category);
}
public override string ToDescriptionString()
{
var info = UnicodeInfo.GetCharInfo(codePoint);
return(base.ToLongString() + "\n" +
"Kun: " + info.JapaneseKunReading + "\n" +
"On: " + info.JapaneseOnReading + "\n" +
info.Definition + "\n");
}
public void RadicalUnicodeCategories()
{
foreach (var radical in lookup.AllRadicals)
{
Console.Write(radical.ToString());
Console.Write(": ");
var info = UnicodeInfo.GetCharInfo(radical.Utf32);
Console.WriteLine(info.Name);
}
}
public void MethodGetCharInfoShouldHaveCoherentResults()
{
for (int i = 0; i <= 0x10FFFF; i++)
{
var charInfo = UnicodeInfo.GetCharInfo(i);
Assert.Equal(charInfo.Name, UnicodeInfo.GetName(i));
Assert.Equal(charInfo.Category, UnicodeInfo.GetCategory(i));
Assert.Equal(UnicodeInfo.GetDisplayText(charInfo), UnicodeInfo.GetDisplayText(i));
}
}
public void RadicalStrokeCountShouldHaveExpectedResults()
{
var char5E7A = UnicodeInfo.GetCharInfo(0x5E7A);
Assert.NotEmpty(char5E7A.UnicodeRadicalStrokeCounts);
Assert.False(char5E7A.UnicodeRadicalStrokeCounts[0].IsSimplified);
Assert.Equal(52, char5E7A.UnicodeRadicalStrokeCounts[0].Radical);
Assert.Equal(0, char5E7A.UnicodeRadicalStrokeCounts[0].StrokeCount);
var char2A6D6 = UnicodeInfo.GetCharInfo(0x2A6D6);
Assert.NotEmpty(char2A6D6.UnicodeRadicalStrokeCounts);
Assert.False(char2A6D6.UnicodeRadicalStrokeCounts[0].IsSimplified);
Assert.Equal(214, char2A6D6.UnicodeRadicalStrokeCounts[0].Radical);
Assert.Equal(20, char2A6D6.UnicodeRadicalStrokeCounts[0].StrokeCount);
}
public static CodePoint FromInt(int codePoint)
{
var info = UnicodeInfo.GetCharInfo(codePoint);
switch (info.Block)
{
case "Hiragana":
return(new Hiragana(codePoint));
case "Katakana":
return(new Katakana(codePoint));
case "CJK Unified Ideographs":
return(new Kanji(codePoint));
default:
return(new CodePoint(codePoint));
}
}
public void CharacterInfoShouldHaveExpectedResults(int codePoint, UnicodeCategory expectedCategory, UnicodeNumericType expectedNumericType, string expectedNumericValue, string expectedName, string expectedBlock)
{
var info = UnicodeInfo.GetCharInfo(codePoint);
Assert.Equal(codePoint, info.CodePoint);
Assert.Equal(expectedCategory, info.Category);
Assert.Equal(expectedNumericType, info.NumericType);
if (expectedNumericValue != null)
{
Assert.Equal(UnicodeRationalNumber.Parse(expectedNumericValue), info.NumericValue);
}
else
{
Assert.Null(info.NumericValue);
}
Assert.Equal(expectedName, info.Name);
Assert.Equal(expectedBlock, UnicodeInfo.GetBlockName(codePoint));
Assert.Equal(expectedBlock, info.Block);
}
private void Initialize()
{
var builder = new IndexBuilder();
for (int i = 0; i <= 0x10FFFF; i++)
{
var charInfo = UnicodeInfo.GetCharInfo(i);
string name = charInfo.Name;
if (!string.IsNullOrEmpty(name))
{
foreach (string word in name.Split(i == 0x1180 ? SimplifiedWordSeparators : WordSeparators))
{
builder.AddWord(word, i);
}
}
if (!string.IsNullOrEmpty(charInfo.OldName))
{
foreach (string word in RemoveForbiddenCharacters(charInfo.OldName).Split(WordSeparators))
{
builder.AddWord(word, i);
}
}
if (charInfo.NameAliases.Count > 0)
{
foreach (var nameAlias in charInfo.NameAliases)
{
foreach (string word in nameAlias.Name.Split(WordSeparators))
{
builder.AddWord(word, i);
}
}
}
}
_wordIndex = builder.Build();
GC.Collect();
}
private void BuildUnicodeList()
{
var blocks = UnicodeInfo.GetBlocks();
foreach (var block in blocks)
{
foreach (var codepoint in block.CodePointRange)
{
if (char.IsSurrogate((char)codepoint))
{
continue;
}
var charInfo = UnicodeInfo.GetCharInfo(codepoint);
var displayText = charInfo.Name;
if (displayText != null)
{
descriptions[codepoint] = displayText;
}
}
}
}
private void blockSelection_SelectedIndexChanged(object sender, EventArgs e)
{
charView.Items.Clear();
UnicodeBlock newB = UnicodeInfo.GetBlocks()[blockSelection.SelectedIndex];
int i = newB.CodePointRange.FirstCodePoint;
int range_upper = newB.CodePointRange.LastCodePoint;
UnicodeCharInfo uc = new UnicodeCharInfo();
int itemind = 0;
string name = "RESERVED CODEPOINT";
while (i <= range_upper)
{
uc = UnicodeInfo.GetCharInfo(i);
charView.Items.Add(UnicodeInfo.GetDisplayText(uc));
if (uc.Name != null)
{
name = uc.Name;
}
charView.Items[itemind].ToolTipText = name + "\r\nCode point: U+" + BitConverter.ToString(BitConverter.GetBytes(uc.CodePoint).Reverse().ToArray()).Replace("-", "");
i++;
itemind++;
}
}
static void Main(string[] args)
{
//https://docs.microsoft.com/ja-jp/dotnet/api/system.char?view=netcore-3.1#Relationship
//https://github.com/GoldenCrystal/NetUnicodeInfo
if (args.Length != 2)
{
Environment.Exit(0);
}
int s = int.Parse(args[0]);
int e = int.Parse(args[1]);
if (s > e)
{
Environment.Exit(0);
}
int c = e - s + 1;
if (s < 0)
{
Environment.Exit(0);
}
if (e < 0 || e > 1114112)
{
Environment.Exit(0);
}
if (c < 0 || c > 1114112)
{
Environment.Exit(0);
}
List <int> lowerCodePointList = Enumerable.Range(s, c).Where(i => i < 55296).ToList();
List <int> upperCodePointList = Enumerable.Range(s, c).Where(i => i > 57343).ToList();
List <int> codePointList = new List <int> ();
if (0 != lowerCodePointList.Count)
{
codePointList.AddRange(lowerCodePointList);
}
if (0 != upperCodePointList.Count)
{
codePointList.AddRange(upperCodePointList);
}
if (0 == codePointList.Count)
{
Environment.Exit(0);
}
int grp = 0;
List <Dictionary <string, string> > summaryList = new List <Dictionary <string, string> > ();
foreach (var codePoint in codePointList)
{
grp++;
string item = Char.ConvertFromUtf32(codePoint);
List <char> charList = item.ToCharArray().ToList();
var charInfo = UnicodeInfo.GetCharInfo(codePoint);
int grpseq = 0;
foreach (Char ch in charList)
{
Dictionary <string, string> detailDict = new Dictionary <string, string> ();
grpseq++;
detailDict.Add(GRP, grp.ToString());
detailDict.Add(GRPSEQ, grpseq.ToString());
detailDict.Add(CODEPOINT, codePoint.ToString());
detailDict.Add(GRAPH, item);
detailDict.Add(NAME, charInfo.Name);
detailDict.Add(NUMERIC_VALUE, charInfo.NumericValue.ToString() == "" ? DEFAULT_NONE_STRING_VALUE : charInfo.NumericValue.ToString());
detailDict.Add(BIDIRECTIONAL_CLASS, charInfo.BidirectionalClass.ToString() == "" ? DEFAULT_NONE_STRING_VALUE : charInfo.BidirectionalClass.ToString());
detailDict.Add(BIDIRECTIONAL_MIRRORED, charInfo.BidirectionalMirrored.ToString() == "" ? DEFAULT_NONE_STRING_VALUE : charInfo.BidirectionalMirrored.ToString());
detailDict.Add(BLOCK, charInfo.Block.ToString() == "" ? DEFAULT_NONE_STRING_VALUE : charInfo.Block.ToString());
detailDict.Add(CANONICAL_COMBINING_CLASS, charInfo.CanonicalCombiningClass.ToString() == "" ? DEFAULT_NONE_STRING_VALUE : charInfo.CanonicalCombiningClass.ToString());
detailDict.Add(CANTONESE_READING, charInfo.CantoneseReading == null ? DEFAULT_NONE_STRING_VALUE : charInfo.CantoneseReading);
detailDict.Add(CATEGORY, charInfo.Category.ToString() == "" ? DEFAULT_NONE_STRING_VALUE : charInfo.Category.ToString());
detailDict.Add(DECOMPOSITION_MAPPING, charInfo.DecompositionMapping == null ? DEFAULT_NONE_STRING_VALUE : charInfo.DecompositionMapping);
detailDict.Add(DECOMPOSITION_TYPE, charInfo.DecompositionType.ToString() == "" ? DEFAULT_NONE_STRING_VALUE : charInfo.DecompositionType.ToString());
detailDict.Add(DEFINITION, charInfo.Definition == null ? DEFAULT_NONE_STRING_VALUE : charInfo.Definition);
detailDict.Add(HANGUL_READING, charInfo.HangulReading == null ? DEFAULT_NONE_STRING_VALUE : charInfo.HangulReading);
detailDict.Add(JAPANESE_KUN_READING, charInfo.JapaneseKunReading == null ? DEFAULT_NONE_STRING_VALUE : charInfo.JapaneseKunReading);
detailDict.Add(JAPANESE_ON_READING, charInfo.JapaneseOnReading == null ? DEFAULT_NONE_STRING_VALUE : charInfo.JapaneseOnReading);
detailDict.Add(KOREAN_READING, charInfo.KoreanReading == null ? DEFAULT_NONE_STRING_VALUE : charInfo.KoreanReading);
detailDict.Add(MANDARIN_READING, charInfo.MandarinReading == null ? DEFAULT_NONE_STRING_VALUE : charInfo.MandarinReading);
detailDict.Add(ALIAS_NAME, String.Join(ITEM_JOINER, charInfo.NameAliases.Select(unicodeNameAlias => unicodeNameAlias.Name == null ? DEFAULT_NONE_STRING_VALUE : unicodeNameAlias.Name).ToArray()));
detailDict.Add(ALIAS_KIND, String.Join(ITEM_JOINER, charInfo.NameAliases.Select(unicodeNameAlias => unicodeNameAlias.Kind.ToString() == "" ? DEFAULT_NONE_STRING_VALUE : unicodeNameAlias.Kind.ToString()).ToArray()));
detailDict.Add(NUMERIC_TYPE, charInfo.NumericType.ToString() == "" ? DEFAULT_NONE_STRING_VALUE : charInfo.NumericType.ToString());
detailDict.Add(OLD_NAME, charInfo.OldName == null ? DEFAULT_NONE_STRING_VALUE : charInfo.OldName);
detailDict.Add(SIMPLE_LOWER_CASE_MAPPING, charInfo.SimpleLowerCaseMapping == null ? DEFAULT_NONE_STRING_VALUE : charInfo.SimpleLowerCaseMapping);
detailDict.Add(SIMPLE_TITLE_CASE_MAPPING, charInfo.SimpleTitleCaseMapping == null ? DEFAULT_NONE_STRING_VALUE : charInfo.SimpleTitleCaseMapping);
detailDict.Add(SIMPLE_UPPER_CASE_MAPPING, charInfo.SimpleUpperCaseMapping == null ? DEFAULT_NONE_STRING_VALUE : charInfo.SimpleUpperCaseMapping);
detailDict.Add(SIMPLIFIED_VARIANT, charInfo.SimplifiedVariant == null ? DEFAULT_NONE_STRING_VALUE : charInfo.SimplifiedVariant);
detailDict.Add(TRADITIONAL_VARIANT, charInfo.TraditionalVariant == null ? DEFAULT_NONE_STRING_VALUE : charInfo.TraditionalVariant);
detailDict.Add(UNICODE_RADICAL_STROKE_COUNTS, String.Join(ITEM_JOINER, charInfo.UnicodeRadicalStrokeCounts.Select(unicodeRadicalStrokeCount => unicodeRadicalStrokeCount.Radical.ToString() == "" ? DEFAULT_NONE_STRING_VALUE : unicodeRadicalStrokeCount.Radical.ToString()).ToArray()));
detailDict.Add(VIETNAMESE_READING, charInfo.VietnameseReading == null ? DEFAULT_NONE_STRING_VALUE : charInfo.VietnameseReading);
detailDict.Add(IS_ASCII, String.Join(ITEM_JOINER, item.EnumerateRunes().Select(rune => rune.IsAscii.ToString() == "" ? DEFAULT_NONE_STRING_VALUE : rune.IsAscii.ToString()).ToArray()));
detailDict.Add(IS_BMP, String.Join(ITEM_JOINER, item.EnumerateRunes().Select(rune => rune.IsBmp.ToString() == "" ? DEFAULT_NONE_STRING_VALUE : rune.IsBmp.ToString()).ToArray()));
detailDict.Add(PLANE, String.Join(ITEM_JOINER, item.EnumerateRunes().Select(rune => rune.Plane.ToString() == "" ? DEFAULT_NONE_STRING_VALUE : rune.Plane.ToString()).ToArray()));
summaryList.Add(detailDict);
}
}
//header
Console.WriteLine(String.Join(FS, summaryList[0].Keys.ToArray()));
//body
foreach (var itemDict in summaryList)
{
List <string> keyList = itemDict.Keys.ToList();
foreach (var key in keyList)
{
{
Console.Write(itemDict[key]);
Console.Write(FS);
}
}
Console.WriteLine();
}
}
public static Models.CodePoint GetCodePoint(int codePoint) => GetCodePoint(UnicodeInfo.GetCharInfo(codePoint));
public static ICollection <CharInfoWrapper> GetData(string search)
{
if (search.Length == 0)
{
return(Array.Empty <CharInfoWrapper>());
}
if (search.Length == 1)
{
return new[] { new CharInfoWrapper(UnicodeInfo.GetCharInfo(search[0])) }
}
;
if (search.Length == 2 && char.IsHighSurrogate(search[0]) && char.IsLowSurrogate(search[1]))
{
return new[] { new CharInfoWrapper(UnicodeInfo.GetCharInfo(char.ConvertToUtf32(search[0], search[1]))) }
}
;
int code;
if (search.StartsWith("\\u", StringComparison.OrdinalIgnoreCase) || search.StartsWith("U+", StringComparison.OrdinalIgnoreCase) || search.StartsWith("&#", StringComparison.OrdinalIgnoreCase))
{
var value = search.Substring(2);
if (int.TryParse(value, NumberStyles.AllowHexSpecifier, CultureInfo.InvariantCulture, out code))
{
return new[] { new CharInfoWrapper(UnicodeInfo.GetCharInfo(code)) }
}
;
}
if (int.TryParse(search, NumberStyles.AllowHexSpecifier, CultureInfo.InvariantCulture, out code))
{
return new[] { new CharInfoWrapper(UnicodeInfo.GetCharInfo(code)) }
}
;
// Search by description
if (s_index == null)
{
s_index = BuildUnicodeIndex();
}
var result = new List <CharInfoWrapper>();
search = search.ToUpperInvariant();
foreach (var entry in s_index)
{
if (entry.Text.Contains(search, StringComparison.Ordinal))
{
result.Add(new CharInfoWrapper(entry.CharInfo));
if (result.Count > 100)
{
break;
}
}
}
return(result);
}
public void ControlCharactersShouldHaveSpecificDisplayText()
{
for (int i = 0; i <= 0x20; ++i)
{
Assert.Equal(char.ConvertFromUtf32(0x2400 + i), UnicodeInfo.GetDisplayText(i));
Assert.Equal(char.ConvertFromUtf32(0x2400 + i), UnicodeInfo.GetDisplayText(UnicodeInfo.GetCharInfo(i)));
}
}
[InlineData(0xE007F, EmojiProperties.ExtendedPictographic)] // CANCEL TAG public void CodePointShouldNotHaveEmojiProperties(int codePoint, EmojiProperties emojiProperties) => Assert.Equal((EmojiProperties)0, UnicodeInfo.GetCharInfo(codePoint).EmojiProperties & emojiProperties);
[InlineData(0xE01EF, CoreProperties.ExtendedIdentifierContinue)] // VARIATION SELECTOR-256 public void CodePointShouldHaveCoreProperties(int codePoint, CoreProperties coreProperties) => Assert.Equal(coreProperties, UnicodeInfo.GetCharInfo(codePoint).CoreProperties & coreProperties);
