swcs/NetUnicodeInfo
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Exposed cross-references, and added them to the UI.

Browse Source
This commit is contained in:
GoldenCrystal
2014-11-24 00:24:55 +01:00
parent 95fd83cb6d
commit 4f03ba85d9
23 changed files with 169 additions and 55 deletions
Download Patch File Download Diff File Expand all files Collapse all files
UnicodeInformation.Builder/Program.cs
+3 -1
View File
@@ -21,8 +21,10 @@ namespace System.Unicode.Builder
"UnicodeData.txt",
"PropList.txt",
"DerivedCoreProperties.txt",
//"Jamo.txt", // Not used right now, as the hangul syllable algorithm implementation takes care of this.
"NameAliases.txt",
"NamesList.txt",
"Blocks.txt",
"Jamo.txt",
};
public static readonly string[] unihanRequiredFiles = new[]
UnicodeInformation.Builder/UnicodeCharacterDataBuilder.cs
+8 -8
View File
@@ -27,7 +27,7 @@ namespace System.Unicode.Builder
private CoreProperties coreProperties;
private readonly List<UnicodeNameAlias> nameAliases = new List<UnicodeNameAlias>();
private readonly List<int> relatedCodePoints = new List<int>();
private readonly List<int> crossRerefences = new List<int>();
public UnicodeCharacterRange CodePointRange { get { return codePointRange; } }
@@ -124,7 +124,7 @@ namespace System.Unicode.Builder
set { coreProperties = value; }
}
public ICollection<int> RelatedCodePoints { get { return relatedCodePoints; } }
public ICollection<int> CrossRerefences { get { return crossRerefences; } }
public UnicodeCharacterDataBuilder(int codePoint)
: this(new UnicodeCharacterRange(codePoint))
@@ -158,7 +158,7 @@ namespace System.Unicode.Builder
SimpleTitleCaseMapping,
ContributoryProperties,
CoreProperties,
RelatedCodePoints.Count > 0 ? RelatedCodePoints.ToArray() : null
CrossRerefences.Count > 0 ? CrossRerefences.ToArray() : null
);
}
@@ -184,7 +184,7 @@ namespace System.Unicode.Builder
if (simpleTitleCaseMapping != null) fields |= UcdFields.SimpleTitleCaseMapping;
if (contributoryProperties != 0) fields |= UcdFields.ContributoryProperties;
if (coreProperties != 0) fields |= UcdFields.CoreProperties;
if (relatedCodePoints.Count > 0) fields |= UcdFields.RelatedCodePoints;
if (crossRerefences.Count > 0) fields |= UcdFields.CrossRerefences;
writer.Write((ushort)fields);
@@ -232,11 +232,11 @@ namespace System.Unicode.Builder
if ((fields & UcdFields.SimpleTitleCaseMapping) != 0) writer.Write(simpleTitleCaseMapping);
if ((fields & UcdFields.ContributoryProperties) != 0) writer.Write((int)contributoryProperties);
if ((fields & UcdFields.CoreProperties) != 0) writer.WriteUInt24((int)coreProperties);
if ((fields & UcdFields.RelatedCodePoints) != 0)
if ((fields & UcdFields.CrossRerefences) != 0)
{
writer.Write(checked((byte)(relatedCodePoints.Count - 1)));
foreach (int relatedCodePoint in relatedCodePoints)
writer.WriteCodePoint(relatedCodePoint);
writer.Write(checked((byte)(crossRerefences.Count - 1)));
foreach (int crossReference in crossRerefences)
writer.WriteCodePoint(crossReference);
}
}
}
UnicodeInformation.Builder/UnicodeDataProcessor.cs
+4 -4
View File
@@ -41,7 +41,7 @@ namespace System.Unicode.Builder
await ProcessDerivedCorePropertiesFile(ucdSource, builder).ConfigureAwait(false);
await ProcessNameAliasesFile(ucdSource, builder).ConfigureAwait(false);
await ProcessNamesListFile(ucdSource, builder).ConfigureAwait(false);
await ProcessBlocksFile(ucdSource, builder).ConfigureAwait(false);
await ProcessBlocksFile(ucdSource, builder).ConfigureAwait(false);
await ProcessUnihanReadings(unihanSource, builder).ConfigureAwait(false);
await ProcessUnihanVariants(unihanSource, builder).ConfigureAwait(false);
await ProcessUnihanNumericValues(unihanSource, builder).ConfigureAwait(false);
@@ -255,12 +255,12 @@ namespace System.Unicode.Builder
if (length < 0) length = line.Length;
length -= 3;
characterData.RelatedCodePoints.Add(int.Parse(line.Substring(3, length), NumberStyles.HexNumber));
characterData.CrossRerefences.Add(int.Parse(line.Substring(3, length), NumberStyles.HexNumber));
}
else if (line[3] == '(')
{
bool hasBrackets = line[4] == '<';
int codePointOffset = line.IndexOf(hasBrackets ? "> - " : "- ", 4);
int codePointOffset = line.IndexOf(hasBrackets ? "> - " : "- ", 4);
if (codePointOffset < 0) throw new InvalidDataException();
codePointOffset += hasBrackets ? 4 : 2;
@@ -269,7 +269,7 @@ namespace System.Unicode.Builder
if (length < 0) throw new InvalidDataException();
length -= codePointOffset;
characterData.RelatedCodePoints.Add(int.Parse(line.Substring(codePointOffset, length), NumberStyles.HexNumber));
characterData.CrossRerefences.Add(int.Parse(line.Substring(codePointOffset, length), NumberStyles.HexNumber));
}
else throw new InvalidDataException();
}