private static void ReadUnicodeCharacterDataEntry(BinaryReader reader, byte[] nameBuffer, out UnicodeCharacterData value)
{
var fields = (UcdFields)reader.ReadUInt16();
var codePointRange = (fields & UcdFields.CodePointRange) != 0 ? new UnicodeCodePointRange(ReadCodePoint(reader), ReadCodePoint(reader)) : new UnicodeCodePointRange(ReadCodePoint(reader));
string name = null;
UnicodeNameAlias[] nameAliases = UnicodeNameAlias.EmptyArray;
// Read all the official names of the character.
if ((fields & UcdFields.Name) != 0)
{
int length = reader.ReadByte();
byte @case = (byte)(length & 0xC0);
if (@case < 0x80) // Handles the case where only the name is present.
{
length = (length & 0x7F) + 1;
if (reader.Read(nameBuffer, 0, length) != length)
{
throw new EndOfStreamException();
}
name = Encoding.UTF8.GetString(nameBuffer, 0, length);
}
else
{
nameAliases = new UnicodeNameAlias[(length & 0x3F) + 1];
if ((@case & 0x40) != 0)
{
length = reader.ReadByte() + 1;
if (length > 128)
{
throw new InvalidDataException("Did not expect names longer than 128 bytes.");
}
if (reader.Read(nameBuffer, 0, length) != length)
{
throw new EndOfStreamException();
}
name = Encoding.UTF8.GetString(nameBuffer, 0, length);
}
for (int i = 0; i < nameAliases.Length; ++i)
{
nameAliases[i] = new UnicodeNameAlias(reader.ReadString(), (UnicodeNameAliasKind)(reader.ReadByte()));
}
}
}
var category = (fields & UcdFields.Category) != 0 ? (UnicodeCategory)reader.ReadByte() : UnicodeCategory.OtherNotAssigned;
var canonicalCombiningClass = (fields & UcdFields.CanonicalCombiningClass) != 0 ? (CanonicalCombiningClass)reader.ReadByte() : CanonicalCombiningClass.NotReordered;
var bidirectionalClass = (fields & UcdFields.BidirectionalClass) != 0 ? (BidirectionalClass)reader.ReadByte() : 0;
CompatibilityFormattingTag decompositionType = (fields & UcdFields.DecompositionMapping) != 0 ? (CompatibilityFormattingTag)reader.ReadByte() : CompatibilityFormattingTag.Canonical;
string decompositionMapping = (fields & UcdFields.DecompositionMapping) != 0 ? reader.ReadString() : null;
var numericType = (UnicodeNumericType)((int)(fields & UcdFields.NumericNumeric) >> 6);
UnicodeRationalNumber numericValue = numericType != UnicodeNumericType.None ?
new UnicodeRationalNumber(reader.ReadInt64(), reader.ReadByte()) :
default(UnicodeRationalNumber);
string oldName = (fields & UcdFields.OldName) != 0 ? reader.ReadString() : null;
string simpleUpperCaseMapping = (fields & UcdFields.SimpleUpperCaseMapping) != 0 ? reader.ReadString() : null;
string simpleLowerCaseMapping = (fields & UcdFields.SimpleLowerCaseMapping) != 0 ? reader.ReadString() : null;
string simpleTitleCaseMapping = (fields & UcdFields.SimpleTitleCaseMapping) != 0 ? reader.ReadString() : null;
ContributoryProperties contributoryProperties = (fields & UcdFields.ContributoryProperties) != 0 ? (ContributoryProperties)reader.ReadInt32() : 0;
int corePropertiesAndEmojiProperties = (fields & UcdFields.CorePropertiesAndEmojiProperties) != 0 ? ReadInt24(reader) : 0;
int[] crossReferences = (fields & UcdFields.CrossRerefences) != 0 ? new int[reader.ReadByte() + 1] : null;
if (crossReferences != null)
{
for (int i = 0; i < crossReferences.Length; ++i)
{
crossReferences[i] = ReadCodePoint(reader);
}
}
value = new UnicodeCharacterData
(
codePointRange,
name,
nameAliases,
category,
canonicalCombiningClass,
bidirectionalClass,
decompositionType,
decompositionMapping,
numericType,
numericValue,
(fields & UcdFields.BidirectionalMirrored) != 0,
oldName,
simpleUpperCaseMapping,
simpleLowerCaseMapping,
simpleTitleCaseMapping,
contributoryProperties,
corePropertiesAndEmojiProperties,
crossReferences
);
}
internal static void ReadFromStream(Stream stream, out Version unicodeVersion, out UnicodeCharacterData[] unicodeCharacterData, out UnihanCharacterData[] unihanCharacterData, out CjkRadicalData[] radicals, out UnicodeBlock[] blocks, out int maxContiguousIndex)
{
using (var reader = new BinaryReader(stream, Encoding.UTF8))
{
int i;
if (reader.ReadByte() != 'U'
| reader.ReadByte() != 'C'
| reader.ReadByte() != 'D')
{
throw new InvalidDataException();
}
byte formatVersion = reader.ReadByte();
if (formatVersion != 2)
{
throw new InvalidDataException();
}
var fileUnicodeVersion = new Version(reader.ReadUInt16(), reader.ReadByte(), reader.ReadByte());
var unicodeCharacterDataEntries = new UnicodeCharacterData[ReadCodePoint(reader)]; // Allocate one extra entry to act as a dummy entry.
byte[] nameBuffer = new byte[128];
int mci = 0;
for (i = 0; i < unicodeCharacterDataEntries.Length; ++i)
{
ReadUnicodeCharacterDataEntry(reader, nameBuffer, out unicodeCharacterDataEntries[i]);
if (unicodeCharacterDataEntries[i].CodePointRange.Contains(i))
{
mci = i;
}
else
{
++i;
break;
}
}
maxContiguousIndex = mci;
for (; i < unicodeCharacterDataEntries.Length; ++i)
{
ReadUnicodeCharacterDataEntry(reader, nameBuffer, out unicodeCharacterDataEntries[i]);
}
var blockEntries = new UnicodeBlock[reader.ReadUInt16()];
for (i = 0; i < blockEntries.Length; ++i)
{
ReadBlockEntry(reader, out blockEntries[i]);
}
var cjkRadicalEntries = new CjkRadicalData[reader.ReadByte()];
for (i = 0; i < cjkRadicalEntries.Length; ++i)
{
ReadCjkRadicalInfo(reader, out cjkRadicalEntries[i]);
}
var unihanCharacterDataEntries = new UnihanCharacterData[ReadCodePoint(reader)];
for (i = 0; i < unihanCharacterDataEntries.Length; ++i)
{
ReadUnihanCharacterDataEntry(reader, out unihanCharacterDataEntries[i]);
}
unicodeVersion = fileUnicodeVersion;
unicodeCharacterData = unicodeCharacterDataEntries;
unihanCharacterData = unihanCharacterDataEntries;
radicals = cjkRadicalEntries;
blocks = blockEntries;
}
}
public static UnicodeData ReadFromStream(Stream stream)
{
using (var reader = new BinaryReader(stream, Encoding.UTF8))
{
int i;
if (reader.ReadByte() != 'U'
| reader.ReadByte() != 'C'
| reader.ReadByte() != 'D')
{
throw new InvalidDataException();
}
byte formatVersion = reader.ReadByte();
if (formatVersion != 2)
{
throw new InvalidDataException();
}
var fileUnicodeVersion = new Version(reader.ReadUInt16(), reader.ReadByte(), reader.ReadByte());
var unicodeCharacterDataEntries = new UnicodeCharacterData[ReadCodePoint(reader)]; // Allocate one extra entry to act as a dummy entry.
byte[] nameBuffer = new byte[128];
int maxContiguousIndex = 0;
for (i = 0; i < unicodeCharacterDataEntries.Length; ++i)
{
ReadUnicodeCharacterDataEntry(reader, nameBuffer, out unicodeCharacterDataEntries[i]);
if (unicodeCharacterDataEntries[i].CodePointRange.Contains(i))
{
maxContiguousIndex = i;
}
else
{
++i;
break;
}
}
for (; i < unicodeCharacterDataEntries.Length; ++i)
{
ReadUnicodeCharacterDataEntry(reader, nameBuffer, out unicodeCharacterDataEntries[i]);
}
var blockEntries = new UnicodeBlock[reader.ReadUInt16()];
for (i = 0; i < blockEntries.Length; ++i)
{
ReadBlockEntry(reader, out blockEntries[i]);
}
var cjkRadicalEntries = new CjkRadicalData[reader.ReadByte()];
for (i = 0; i < cjkRadicalEntries.Length; ++i)
{
ReadCjkRadicalInfo(reader, out cjkRadicalEntries[i]);
}
var unihanCharacterDataEntries = new UnihanCharacterData[ReadCodePoint(reader)];
for (i = 0; i < unihanCharacterDataEntries.Length; ++i)
{
ReadUnihanCharacterDataEntry(reader, out unihanCharacterDataEntries[i]);
}
return(new UnicodeData
(
fileUnicodeVersion,
unicodeCharacterDataEntries,
unihanCharacterDataEntries,
blockEntries,
cjkRadicalEntries,
maxContiguousIndex
));
}
}
