public static Recognizer Create(System.Globalization.CultureInfo culture, int priority)
{
try
{
// TODO support non-blank languages for unit separation
MeasureRegexRecognizer result = new MeasureRegexRecognizer(100, "DEFAULT_MEASURE_RECOGNIZER", culture);
Core.CharacterSet first = null;
// augmentation doesn't change FIRST()
List <string> patterns = NumberRegexRecognizer.ComputeRXPatterns(culture, NumberSeparatorMode.CultureDefault, out first);
AugmentPatterns(patterns, culture);
foreach (string p in patterns)
{
// use the same first for all patterns (the number regex pattern computer returns just one pattern anyway)
result.Add(p, first, 2);
}
SeparatorCombination defaultSc = new SeparatorCombination(culture, false);
if (defaultSc.IsSwappable())
{
patterns = NumberRegexRecognizer.ComputeRXPatterns(culture, NumberSeparatorMode.Swapped, out first);
AugmentPatterns(patterns, culture);
foreach (string p in patterns)
{
result.Add(p, first, 1);
}
}
if (NumberPatternComputer.DoAddENUSSeparators(culture))
{
patterns = NumberRegexRecognizer.ComputeRXPatterns(culture, NumberSeparatorMode.EnUS, out first);
AugmentPatterns(patterns, culture);
foreach (string p in patterns)
{
result.Add(p, first, 0);
}
}
result.OnlyIfFollowedByNonwordCharacter
= true; // otherwise "123 ABC" will be recognized as "123 A" "BC" in Japanese
return(result);
}
catch // (System.Exception e)
{
return(null);
}
}
public static Recognizer Create(System.Globalization.CultureInfo culture, int priority)
{
try
{
NumberRegexRecognizer recog = new NumberRegexRecognizer(100, "DEFAULT_NUMBER_RECOGNIZER", culture.NumberFormat);
CharacterSet first = null;
foreach (string p in ComputeRXPatterns(culture, NumberSeparatorMode.CultureDefault, out first))
{
recog.Add(p, first, 2);
}
SeparatorCombination defaultSc = new SeparatorCombination(culture, false);
if (defaultSc.IsSwappable())
{
foreach (string p in ComputeRXPatterns(culture, NumberSeparatorMode.Swapped, out first))
{
recog.Add(p, first, 1);
}
}
if (NumberPatternComputer.DoAddENUSSeparators(culture))
{
foreach (string p in ComputeRXPatterns(culture, NumberSeparatorMode.EnUS, out first))
{
recog.Add(p, first, 1);
}
}
recog.OnlyIfFollowedByNonwordCharacter
= CultureInfoExtensions.UseBlankAsWordSeparator(culture);
recog.AdditionalTerminators = new CharacterSet();
recog.AdditionalTerminators.Add('-'); // TODO other math symbols?
recog.OverrideFallbackRecognizer = true;
return(recog);
}
catch // (System.Exception e)
{
return(null);
}
}
internal static LanguagePlatform.Lingua.FST.FST CreateFST(System.Globalization.CultureInfo culture,
bool appendWordTerminator)
{
NumberFormatData nfd
= NumberPatternComputer.GetNumberFormatData(culture, true, true);
string fstPattern = Lingua.Tokenization.NumberPatternComputer.ComputeFSTPattern(nfd,
true, appendWordTerminator);
LanguagePlatform.Lingua.FST.FST fst = LanguagePlatform.Lingua.FST.FST.Create(fstPattern);
fst.MakeDeterministic();
#if DEBUG
bool dump = false;
if (dump)
{
fst.Dump(String.Format("d:/temp/number-fst-{0}.txt", culture.Name));
}
#endif
return(fst);
}
internal static LanguagePlatform.Lingua.FST.FST CreateFST(System.Globalization.CultureInfo culture,
bool appendWordTerminator)
{
NumberFormatData nfd
= NumberPatternComputer.GetNumberFormatData(culture, true, true);
string numberPattern = Lingua.Tokenization.NumberPatternComputer.ComputeFSTPattern(nfd,
true, false);
System.Text.StringBuilder sb = new StringBuilder(numberPattern);
sb.Append("(");
bool first = true;
NumberPatternComputer.AppendDisjunction(sb, Core.CharacterProperties.Blanks, 'U', ref first);
sb.Append(")?(");
first = true;
Core.Wordlist units = Core.Tokenization.PhysicalUnit.GetUnits(culture, false);
foreach (string unit in units.Items)
{
if (first)
{
first = false;
}
else
{
sb.Append("|");
}
// append single unit, make sure that first char emits 'U' (in case no whitespace
// sep is in the input)
sb.AppendFormat("(<{0}:U>", FST.FST.EscapeSpecial(unit[0]));
string remainder = unit.Substring(1);
if (!String.IsNullOrEmpty(remainder))
{
sb.Append(FST.FST.EscapeSpecial(remainder));
}
sb.Append(")");
}
sb.Append(")");
if (appendWordTerminator)
{
// Append "word terminator"
sb.Append("#>");
}
LanguagePlatform.Lingua.FST.FST fst = LanguagePlatform.Lingua.FST.FST.Create(sb.ToString());
fst.MakeDeterministic();
#if DEBUG
bool dump = false;
if (dump)
{
fst.Dump(String.Format("d:/temp/measure-fst-{0}.txt", culture.Name));
}
#endif
return(fst);
}
internal static List <string> ComputeRXPatterns(System.Globalization.CultureInfo culture,
NumberSeparatorMode separatorMode,
out CharacterSet first)
{
first = null;
if (culture == null)
{
throw new ArgumentNullException("culture");
}
if (culture.IsNeutralCulture)
{
System.Diagnostics.Debug.Assert(false, "Cannot compute number pattern for neutral cultures");
return(null);
}
IList <string> digits = culture.NumberFormat.NativeDigits;
string decSep = culture.NumberFormat.NumberDecimalSeparator;
string grpSep = culture.NumberFormat.NumberGroupSeparator;
switch (separatorMode)
{
case NumberSeparatorMode.CultureDefault:
// nop
break;
case NumberSeparatorMode.Swapped:
{
string tmp = decSep;
decSep = grpSep;
grpSep = tmp;
}
break;
case NumberSeparatorMode.EnUS:
decSep = ".";
grpSep = ",";
break;
default:
throw new Exception("Unexpected");
}
List <string> result = ComputeRXPatterns(grpSep, decSep, digits, out first);
if (result == null)
{
return(null);
}
// add full-width variant digits for jp, zh, ko
if (NumberPatternComputer.DoAddFullwidthVariantDigits(culture))
{
Core.CharacterSet alternateFirst;
digits = "\uFF10|\uFF11|\uFF12|\uFF13|\uFF14|\uFF15|\uFF16|\uFF17|\uFF18|\uFF19".Split('|');
List <string> alternate = ComputeRXPatterns(grpSep, decSep, digits, out alternateFirst);
if (alternate != null)
{
result.AddRange(alternate);
first.Add(alternateFirst);
}
}
// add patters with default digits
if (!CultureInfoExtensions.UsesDefaultDigits(culture))
{
Core.CharacterSet alternateFirst;
digits = "0|1|2|3|4|5|6|7|8|9".Split('|');
List <string> alternate = ComputeRXPatterns(grpSep, decSep, digits, out alternateFirst);
if (alternate != null)
{
result.AddRange(alternate);
first.Add(alternateFirst);
}
}
return(result);
}
