public void Should_Generate_Data()
{
if (!Directory.Exists("Generated"))
{
Directory.CreateDirectory("Generated");
}
var unicodeDataTrie = UnicodeDataGenerator.GenerateUnicodeDataTrie(out var unicodeDataEntries, out var unicodeData);
foreach (var value in unicodeData.Values)
{
var data = unicodeDataTrie.Get(value.Codepoint);
Assert.Equal(value.GeneralCategory, GetValue(data, 0, UnicodeData.CATEGORY_MASK));
Assert.Equal(value.Script, GetValue(data, UnicodeData.SCRIPT_SHIFT, UnicodeData.SCRIPT_MASK));
Assert.Equal(value.LineBreakClass, GetValue(data, UnicodeData.LINEBREAK_SHIFT, UnicodeData.LINEBREAK_MASK));
}
var biDiTrie = UnicodeDataGenerator.GenerateBiDiTrie(out var biDiDataEntries, out var biDiData);
foreach (var value in biDiData.Values)
{
var data = biDiTrie.Get(value.Codepoint);
Assert.Equal(value.Bracket, GetValue(data, 0, UnicodeData.BIDIPAIREDBRACKED_MASK));
Assert.Equal(value.BracketType, GetValue(data, UnicodeData.BIDIPAIREDBRACKEDTYPE_SHIFT, UnicodeData.BIDIPAIREDBRACKEDTYPE_MASK));
Assert.Equal(value.BiDiClass, GetValue(data, UnicodeData.BIDICLASS_SHIFT, UnicodeData.BIDICLASS_MASK));
}
UnicodeEnumsGenerator.CreatePropertyValueAliasHelper(unicodeDataEntries, biDiDataEntries);
}
public static UnicodeTrie GenerateBiDiTrie(out BiDiDataEntries biDiDataEntries, out Dictionary <int, BiDiDataItem> biDiData)
{
biDiData = new Dictionary <int, BiDiDataItem>();
var biDiClassEntries =
UnicodeEnumsGenerator.CreateBiDiClassEnum();
var biDiClassMappings = CreateTagToIndexMappings(biDiClassEntries);
var biDiClassData = ReadBiDiData();
foreach (var(range, name) in biDiClassData)
{
var biDiClass = biDiClassMappings[name];
AddBiDiClassRange(biDiData, range, biDiClass);
}
var biDiPairedBracketTypeEntries = UnicodeEnumsGenerator.CreateBiDiPairedBracketTypeEnum();
var biDiPairedBracketTypeMappings = CreateTagToIndexMappings(biDiPairedBracketTypeEntries);
var biDiPairedBracketData = ReadBiDiPairedBracketData();
foreach (var(range, name) in biDiPairedBracketData)
{
var bracketType = biDiPairedBracketTypeMappings[name];
AddBiDiBracket(biDiData, range, bracketType);
}
var biDiTrieBuilder = new UnicodeTrieBuilder(/*initialValue*/);
foreach (var properties in biDiData.Values)
{
//[bracket]|[bracketType]|[biDiClass]
var value = (properties.BiDiClass << UnicodeData.BIDICLASS_SHIFT) |
(properties.BracketType << UnicodeData.BIDIPAIREDBRACKEDTYPE_SHIFT) | properties.Bracket;
biDiTrieBuilder.Set(properties.Codepoint, (uint)value);
}
biDiDataEntries = new BiDiDataEntries()
{
PairedBracketTypes = biDiPairedBracketTypeEntries, BiDiClasses = biDiClassEntries
};
using (var stream = File.Create("Generated\\BiDi.trie"))
{
var trie = biDiTrieBuilder.Freeze();
trie.Save(stream);
return(trie);
}
}
private static UnicodeTrie GenerateBreakTypeTrie()
{
var graphemeBreakClassValues = UnicodeEnumsGenerator.GetPropertyValueAliases("# Grapheme_Cluster_Break (GCB)");
var graphemeBreakClassMapping = graphemeBreakClassValues.Select(x => x.name).ToList();
var trieBuilder = new UnicodeTrieBuilder();
var graphemeBreakData = ReadBreakData(
"https://www.unicode.org/Public/UCD/latest/ucd/auxiliary/GraphemeBreakProperty.txt");
foreach (var(start, end, graphemeBreakType) in graphemeBreakData)
{
if (!graphemeBreakClassMapping.Contains(graphemeBreakType))
{
continue;
}
if (start == end)
{
trieBuilder.Set(start, (uint)graphemeBreakClassMapping.IndexOf(graphemeBreakType));
}
else
{
trieBuilder.SetRange(start, end, (uint)graphemeBreakClassMapping.IndexOf(graphemeBreakType));
}
}
var emojiBreakData = ReadBreakData("https://unicode.org/Public/emoji/12.0/emoji-data.txt");
foreach (var(start, end, graphemeBreakType) in emojiBreakData)
{
if (!graphemeBreakClassMapping.Contains(graphemeBreakType))
{
continue;
}
if (start == end)
{
trieBuilder.Set(start, (uint)graphemeBreakClassMapping.IndexOf(graphemeBreakType));
}
else
{
trieBuilder.SetRange(start, end, (uint)graphemeBreakClassMapping.IndexOf(graphemeBreakType));
}
}
return(trieBuilder.Freeze());
}
public static UnicodeTrie GenerateUnicodeDataTrie(out UnicodeDataEntries dataEntries, out Dictionary <int, UnicodeDataItem> unicodeData)
{
var generalCategoryEntries =
UnicodeEnumsGenerator.CreateGeneralCategoryEnum();
var generalCategoryMappings = CreateTagToIndexMappings(generalCategoryEntries);
var scriptEntries = UnicodeEnumsGenerator.CreateScriptEnum();
var scriptMappings = CreateNameToIndexMappings(scriptEntries);
var lineBreakClassEntries =
UnicodeEnumsGenerator.CreateLineBreakClassEnum();
var lineBreakClassMappings = CreateTagToIndexMappings(lineBreakClassEntries);
unicodeData = GetUnicodeData(generalCategoryMappings, scriptMappings, lineBreakClassMappings);
var unicodeDataTrieBuilder = new UnicodeTrieBuilder(/*initialValue*/);
foreach (var properties in unicodeData.Values)
{
//[line break]|[biDi]|[script]|[category]
var value = (properties.LineBreakClass << UnicodeData.LINEBREAK_SHIFT) |
(properties.Script << UnicodeData.SCRIPT_SHIFT) | properties.GeneralCategory;
unicodeDataTrieBuilder.Set(properties.Codepoint, (uint)value);
}
dataEntries = new UnicodeDataEntries
{
Scripts = scriptEntries,
GeneralCategories = generalCategoryEntries,
LineBreakClasses = lineBreakClassEntries
};
using (var stream = File.Create("Generated\\UnicodeData.trie"))
{
var trie = unicodeDataTrieBuilder.Freeze();
trie.Save(stream);
return(trie);
}
}
public static void Execute()
{
var codepoints = new Dictionary <int, UnicodeDataItem>();
var generalCategoryEntries =
UnicodeEnumsGenerator.CreateGeneralCategoryEnum();
var generalCategoryMappings = CreateTagToIndexMappings(generalCategoryEntries);
var generalCategoryData = ReadGeneralCategoryData();
foreach (var(range, name) in generalCategoryData)
{
var generalCategory = generalCategoryMappings[name];
AddGeneralCategoryRange(codepoints, range, generalCategory);
}
var scriptEntries = UnicodeEnumsGenerator.CreateScriptEnum();
var scriptMappings = CreateNameToIndexMappings(scriptEntries);
var scriptData = ReadScriptData();
foreach (var(range, name) in scriptData)
{
var script = scriptMappings[name];
AddScriptRange(codepoints, range, script);
}
var biDiClassEntries =
UnicodeEnumsGenerator.CreateBiDiClassEnum();
var biDiClassMappings = CreateTagToIndexMappings(biDiClassEntries);
var biDiData = ReadBiDiData();
foreach (var(range, name) in biDiData)
{
var biDiClass = biDiClassMappings[name];
AddBiDiClassRange(codepoints, range, biDiClass);
}
var lineBreakClassEntries =
UnicodeEnumsGenerator.CreateLineBreakClassEnum();
var lineBreakClassMappings = CreateTagToIndexMappings(lineBreakClassEntries);
var lineBreakClassData = ReadLineBreakClassData();
foreach (var(range, name) in lineBreakClassData)
{
var lineBreakClass = lineBreakClassMappings[name];
AddLineBreakClassRange(codepoints, range, lineBreakClass);
}
//const int initialValue = (0 << UnicodeData.LINEBREAK_SHIFT) |
// (0 << UnicodeData.BIDI_SHIFT) |
// (0 << UnicodeData.SCRIPT_SHIFT) | (int)GeneralCategory.Other;
var builder = new UnicodeTrieBuilder(/*initialValue*/);
foreach (var properties in codepoints.Values)
{
//[line break]|[biDi]|[script]|[category]
var value = (properties.LineBreakClass << UnicodeData.LINEBREAK_SHIFT) |
(properties.BiDiClass << UnicodeData.BIDI_SHIFT) |
(properties.Script << UnicodeData.SCRIPT_SHIFT) | properties.GeneralCategory;
builder.Set(properties.Codepoint, (uint)value);
}
using (var stream = File.Create("Generated\\UnicodeData.trie"))
{
var trie = builder.Freeze();
trie.Save(stream);
}
UnicodeEnumsGenerator.CreatePropertyValueAliasHelper(scriptEntries, generalCategoryEntries,
biDiClassEntries, lineBreakClassEntries);
}