static JdkBreakIterator()
{
using (Stream @is =
typeof(JdkBreakIterator).FindAndGetManifestResourceStream("jdksent.brk"))
{
SentenceInstance = RuleBasedBreakIterator.GetInstanceFromCompiledRules(@is);
}
using (Stream @is =
typeof(JdkBreakIterator).FindAndGetManifestResourceStream("jdkword.brk"))
{
WordInstance = RuleBasedBreakIterator.GetInstanceFromCompiledRules(@is);
}
}
private static RuleBasedBreakIterator ReadBreakIterator(string filename)
{
using Stream @is = typeof(DefaultICUTokenizerConfig).FindAndGetManifestResourceStream(filename);
try
{
RuleBasedBreakIterator bi =
RuleBasedBreakIterator.GetInstanceFromCompiledRules(@is);
return(bi);
}
catch (IOException e)
{
throw new Exception(e.ToString(), e);
}
}
private static BreakIterator CreateBreakInstance(ULocale locale, int kind)
{
RuleBasedBreakIterator iter = null;
ICUResourceBundle rb = ICUResourceBundle.
GetBundleInstance(ICUData.ICU_BRKITR_BASE_NAME, locale,
ICUResourceBundle.OpenType.LOCALE_ROOT);
//
// Get the binary rules.
//
ByteBuffer bytes = null;
string typeKeyExt = null;
if (kind == BreakIterator.KIND_LINE)
{
string lbKeyValue = locale.GetKeywordValue("lb");
if (lbKeyValue != null && (lbKeyValue.Equals("strict") || lbKeyValue.Equals("normal") || lbKeyValue.Equals("loose")))
{
typeKeyExt = "_" + lbKeyValue;
}
}
try
{
string typeKey = (typeKeyExt == null) ? KIND_NAMES[kind] : KIND_NAMES[kind] + typeKeyExt;
string brkfname = rb.GetStringWithFallback("boundaries/" + typeKey);
string rulesFileName = ICUData.ICU_BRKITR_NAME + '/' + brkfname;
bytes = ICUBinary.GetData(rulesFileName);
}
catch (Exception e)
{
throw new MissingManifestResourceException(e.ToString(), e /*, "", ""*/);
}
//
// Create a normal RuleBasedBreakIterator.
//
try
{
#pragma warning disable 612, 618
iter = RuleBasedBreakIterator.GetInstanceFromCompiledRules(bytes);
#pragma warning restore 612, 618
}
catch (IOException e)
{
// Shouldn't be possible to get here.
// If it happens, the compiled rules are probably corrupted in some way.
Assert.Fail(e);
}
// TODO: Determine valid and actual locale correctly.
ULocale uloc = ULocale.ForLocale(rb.GetLocale());
iter.SetLocale(uloc, uloc);
iter.BreakType = kind;
// filtered break
if (kind == BreakIterator.KIND_SENTENCE)
{
string ssKeyword = locale.GetKeywordValue("ss");
if (ssKeyword != null && ssKeyword.Equals("standard"))
{
ULocale @base = new ULocale(locale.GetBaseName());
return(FilteredBreakIteratorBuilder.GetInstance(@base).WrapIteratorWithFilter(iter));
}
}
return(iter);
}
private static readonly RuleBasedBreakIterator WordInstance = LoadBreakRules("jdkword.brk"); // LUCENENET: CA1810: Initialize reference type static fields inline
private static RuleBasedBreakIterator LoadBreakRules(string fileName)
{
using Stream @is = typeof(JdkBreakIterator).FindAndGetManifestResourceStream(fileName);
return(RuleBasedBreakIterator.GetInstanceFromCompiledRules(@is));
}