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);
}