private FractionExtractor()
{
var regexes = new Dictionary <Regex, TypeTag>
{
{
new Regex(NumbersDefinitions.FractionNotationWithSpacesRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.FRACTION_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.FractionNotationRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.FRACTION_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.FractionNounRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.FRACTION_PREFIX, Constants.GERMAN)
},
{
new Regex(NumbersDefinitions.FractionNounWithArticleRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.FRACTION_PREFIX, Constants.GERMAN)
},
{
new Regex(NumbersDefinitions.FractionPrepositionRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.FRACTION_PREFIX, Constants.GERMAN)
},
};
Regexes = regexes.ToImmutableDictionary();
}
} = Constants.SYS_NUM_INTEGER; // "Integer";
public IntegerExtractor(string placeholder = NumbersDefinitions.PlaceHolderDefault)
{
this.Regexes = new Dictionary <Regex, string>
{
{
new Regex(NumbersDefinitions.NumbersWithPlaceHolder(placeholder), RegexOptions.IgnoreCase | RegexOptions.Singleline)
, RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.NumbersWithSuffix, RegexOptions.Singleline), RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
GenerateLongFormatNumberRegexes(LongFormatType.IntegerNumDot, placeholder), RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.RoundNumberIntegerRegexWithLocks, RegexOptions.IgnoreCase | RegexOptions.Singleline), RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.NumbersWithDozenSuffix, RegexOptions.IgnoreCase | RegexOptions.Singleline), RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.AllIntRegexWithLocks, RegexOptions.IgnoreCase | RegexOptions.Singleline), RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.FRENCH)
},
{
new Regex(NumbersDefinitions.AllIntRegexWithDozenSuffixLocks, RegexOptions.IgnoreCase | RegexOptions.Singleline), RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.FRENCH)
},
{
GenerateLongFormatNumberRegexes(LongFormatType.IntegerNumBlank, placeholder), RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
GenerateLongFormatNumberRegexes(LongFormatType.IntegerNumNoBreakSpace, placeholder), RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
}
}.ToImmutableDictionary();
}
private IntegerExtractor(string placeholder = NumbersDefinitions.PlaceHolderDefault)
{
var regexes = new Dictionary <Regex, TypeTag>
{
{
new Regex(NumbersDefinitions.NumbersWithPlaceHolder(placeholder), RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.IndianNumberingSystemRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.NumbersWithSuffix, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.RoundNumberIntegerRegexWithLocks, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.NumbersWithDozenSuffix, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.AllIntRegexWithLocks, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.HINDI)
},
{
new Regex(NumbersDefinitions.AllIntRegexWithDozenSuffixLocks, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.HINDI)
},
{
GenerateLongFormatNumberRegexes(LongFormatType.IntegerNumComma, placeholder, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
GenerateLongFormatNumberRegexes(LongFormatType.IntegerNumBlank, placeholder, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
GenerateLongFormatNumberRegexes(LongFormatType.IntegerNumNoBreakSpace, placeholder, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.NegativeHinglishRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.HINDI)
},
{
new Regex(NumbersDefinitions.CompoundEnglishNumberRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.HINDI)
},
{
new Regex(NumbersDefinitions.DecimalUnitsWithRoundNumberRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.HINDI)
},
};
Regexes = regexes.ToImmutableDictionary();
}
private OrdinalExtractor(BaseNumberOptionsConfiguration config)
: base(config.Options)
{
keyPrefix = string.Intern(ExtractType + "_" + config.Options.ToString() + "_" + config.Culture);
this.Regexes = new Dictionary <Regex, TypeTag>
{
{
new Regex(NumbersDefinitions.OrdinalSuffixRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.OrdinalNumericRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.OrdinalGermanRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.GERMAN)
},
{
new Regex(NumbersDefinitions.OrdinalRoundNumberRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.GERMAN)
},
}.ToImmutableDictionary();
}
private OrdinalExtractor(BaseNumberOptionsConfiguration config)
: base(config.Options)
{
keyPrefix = string.Intern(ExtractType + "_" + config.Options.ToString() + "_" + config.Culture);
AmbiguousFractionConnectorsRegex = new Regex(NumbersDefinitions.AmbiguousFractionConnectorsRegex, RegexFlags);
RelativeReferenceRegex = new Regex(NumbersDefinitions.RelativeOrdinalRegex, RegexFlags);
var regexes = new Dictionary <Regex, TypeTag>
{
{
new Regex(NumbersDefinitions.OrdinalSuffixRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.OrdinalNumericRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.OrdinalSwedishRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.SWEDISH)
},
{
new Regex(NumbersDefinitions.OrdinalRoundNumberRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.SWEDISH)
},
};
Regexes = regexes.ToImmutableDictionary();
}
public NumberExtractor(NumberMode mode = NumberMode.Default)
{
var builder = ImmutableDictionary.CreateBuilder <Regex, TypeTag>();
CardinalExtractor cardExtract = null;
switch (mode)
{
case NumberMode.PureNumber:
cardExtract = new CardinalExtractor(NumbersDefinitions.PlaceHolderPureNumber);
break;
case NumberMode.Currency:
builder.Add(new Regex(NumbersDefinitions.CurrencyRegex, RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX));
break;
case NumberMode.Default:
break;
}
if (cardExtract == null)
{
cardExtract = new CardinalExtractor();
}
builder.AddRange(cardExtract.Regexes);
var fracExtract = new FractionExtractor();
builder.AddRange(fracExtract.Regexes);
Regexes = builder.ToImmutable();
}
} = Constants.SYS_NUM_FRACTION; // "Fraction";
public FractionExtractor()
{
var regexes = new Dictionary <Regex, TypeTag>
{
{
new Regex(NumbersDefinitions.FractionNotationRegex,
RegexOptions.IgnoreCase | RegexOptions.Singleline)
, RegexTagGenerator.GenerateRegexTag(Constants.FRACTION_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.FractionNotationWithSpacesRegex,
RegexOptions.IgnoreCase | RegexOptions.Singleline)
, RegexTagGenerator.GenerateRegexTag(Constants.FRACTION_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.FractionNounRegex,
RegexOptions.IgnoreCase | RegexOptions.Singleline)
, RegexTagGenerator.GenerateRegexTag(Constants.FRACTION_PREFIX, Constants.SPANISH)
},
{
new Regex(NumbersDefinitions.FractionNounWithArticleRegex,
RegexOptions.IgnoreCase | RegexOptions.Singleline)
, RegexTagGenerator.GenerateRegexTag(Constants.FRACTION_PREFIX, Constants.SPANISH)
},
{
new Regex(NumbersDefinitions.FractionPrepositionRegex,
RegexOptions.IgnoreCase | RegexOptions.Singleline)
, RegexTagGenerator.GenerateRegexTag(Constants.FRACTION_PREFIX, Constants.SPANISH)
},
};
this.Regexes = regexes.ToImmutableDictionary();
}
private OrdinalExtractor(NumberOptions options)
: base(options)
{
RelativeReferenceRegex = RegexCache.Get(NumbersDefinitions.RelativeOrdinalRegex, RegexFlags);
var regexes = new Dictionary <Regex, TypeTag>
{
{
RegexCache.Get(NumbersDefinitions.OrdinalSuffixRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.NUMBER_SUFFIX)
},
{
RegexCache.Get(NumbersDefinitions.OrdinalNumericRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.NUMBER_SUFFIX)
},
{
RegexCache.Get(NumbersDefinitions.OrdinalTurkishRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.TURKISH)
},
{
RegexCache.Get(NumbersDefinitions.RoundNumberOrdinalRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.TURKISH)
},
};
Regexes = regexes.ToImmutableDictionary();
}
private OrdinalExtractor(BaseNumberOptionsConfiguration config)
: base(config.Options)
{
keyPrefix = string.Intern(ExtractType + "_" + config.Options.ToString() + "_" + config.Culture);
var regexes = new Dictionary <Regex, TypeTag>
{
{
RegexCache.Get(NumbersDefinitions.OrdinalSuffixRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.NUMBER_SUFFIX)
},
{
RegexCache.Get(NumbersDefinitions.OrdinalNumericRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.NUMBER_SUFFIX)
},
{
RegexCache.Get(NumbersDefinitions.OrdinalDutchRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.DUTCH)
},
{
RegexCache.Get(NumbersDefinitions.OrdinalRoundNumberRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.DUTCH)
},
};
Regexes = regexes.ToImmutableDictionary();
}
private DoubleExtractor(string placeholder = NumbersDefinitions.PlaceHolderDefault)
{
var regexes = new Dictionary <Regex, TypeTag> {
{
new Regex(NumbersDefinitions.DoubleDecimalPointRegex(placeholder),
RegexOptions.IgnoreCase | RegexOptions.Singleline), RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
}, {
new Regex(NumbersDefinitions.DoubleWithoutIntegralRegex(placeholder),
RegexOptions.IgnoreCase | RegexOptions.Singleline), RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
}, {
new Regex(NumbersDefinitions.DoubleWithMultiplierRegex, RegexOptions.Singleline), RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
}, {
new Regex(NumbersDefinitions.DoubleWithRoundNumber, RegexOptions.IgnoreCase | RegexOptions.Singleline), RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
}, {
new Regex(NumbersDefinitions.DoubleAllFloatRegex, RegexOptions.IgnoreCase | RegexOptions.Singleline), RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.GERMAN)
}, {
new Regex(NumbersDefinitions.DoubleExponentialNotationRegex,
RegexOptions.IgnoreCase | RegexOptions.Singleline), RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.POWER_SUFFIX)
}, {
new Regex(NumbersDefinitions.DoubleCaretExponentialNotationRegex,
RegexOptions.IgnoreCase | RegexOptions.Singleline), RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.POWER_SUFFIX)
}, {
GenerateLongFormatNumberRegexes(LongFormatType.DoubleNumDotComma, placeholder), RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
}, {
GenerateLongFormatNumberRegexes(LongFormatType.DoubleNumNoBreakSpaceComma, placeholder), RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
}
};
Regexes = regexes.ToImmutableDictionary();
}
private OrdinalExtractor(NumberOptions options)
: base(options)
{
AmbiguousFractionConnectorsRegex = new Regex(NumbersDefinitions.AmbiguousFractionConnectorsRegex, RegexFlags);
RelativeReferenceRegex = new Regex(NumbersDefinitions.RelativeOrdinalRegex, RegexFlags);
var regexes = new Dictionary <Regex, TypeTag>
{
{
new Regex(NumbersDefinitions.OrdinalSuffixRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.OrdinalNumericRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.OrdinalEnglishRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.ENGLISH)
},
{
new Regex(NumbersDefinitions.OrdinalRoundNumberRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.ENGLISH)
},
};
Regexes = regexes.ToImmutableDictionary();
}
public IntegerExtractor(ChineseNumberExtractorMode mode = ChineseNumberExtractorMode.Default)
{
var regexes = new Dictionary <Regex, TypeTag>()
{
{
// 123456, －１２３４５６
new Regex(NumbersDefinitions.NumbersSpecialsChars, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
// 15k, 16 G
new Regex(NumbersDefinitions.NumbersSpecialsCharsWithSuffix, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
// 1,234, ２，３３２，１１１
new Regex(NumbersDefinitions.DottedNumbersSpecialsChar, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
// 半百 半打
new Regex(NumbersDefinitions.NumbersWithHalfDozen, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.CHINESE)
},
{
// 半
new Regex(NumbersDefinitions.HalfUnitRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.CHINESE)
},
{
// 一打 五十打
new Regex(NumbersDefinitions.NumbersWithDozen, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.CHINESE)
},
};
switch (mode)
{
case ChineseNumberExtractorMode.Default:
// 一百五十五, 负一亿三百二十二.
// Uses an allow list to avoid extracting "四" from "四川"
regexes.Add(
new Regex(NumbersDefinitions.NumbersWithAllowListRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.CHINESE));
break;
case ChineseNumberExtractorMode.ExtractAll:
// 一百五十五, 负一亿三百二十二, "四" from "四川".
// Uses no allow lists and extracts all potential integers (useful in Units, for example).
regexes.Add(
new Regex(NumbersDefinitions.NumbersAggressiveRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.CHINESE));
break;
}
Regexes = regexes.ToImmutableDictionary();
}
private IntegerExtractor(BaseNumberOptionsConfiguration config)
: base(config.Options)
{
keyPrefix = string.Intern(ExtractType + "_" + config.Options + "_" + config.Placeholder + "_" + config.Culture);
var regexes = new Dictionary <Regex, TypeTag>
{
{
new Regex(NumbersDefinitions.NumbersWithPlaceHolder(config.Placeholder), RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.NumbersWithSpace(config.Placeholder), RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.NumbersWithSuffix, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.RoundNumberIntegerRegexWithLocks, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.NumbersWithDozenSuffix, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.AllIntRegexWithLocks, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.ARABIC)
},
{
new Regex(NumbersDefinitions.AllIntRegexWithDozenSuffixLocks, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.ARABIC)
},
{
GenerateLongFormatNumberRegexes(LongFormatType.IntegerNumComma, config.Placeholder, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
GenerateLongFormatNumberRegexes(LongFormatType.IntegerNumInvertedComma, config.Placeholder, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
GenerateLongFormatNumberRegexes(LongFormatType.IntegerNumBlank, config.Placeholder, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
GenerateLongFormatNumberRegexes(LongFormatType.IntegerNumNoBreakSpace, config.Placeholder, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
};
Regexes = regexes.ToImmutableDictionary();
}
public OrdinalExtractor()
{
var regexes = new Dictionary <Regex, TypeTag>
{
{
new Regex(NumbersDefinitions.OrdinalKoreanRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.NUMBER_SUFFIX)
},
};
Regexes = regexes.ToImmutableDictionary();
}
private DoubleExtractor(BaseNumberOptionsConfiguration config)
: base(config.Options)
{
var regexes = new Dictionary <Regex, TypeTag>
{
{
new Regex(NumbersDefinitions.DoubleDecimalPointRegex(config.Placeholder), RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.DoubleIndianDecimalPointRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.DoubleWithoutIntegralRegex(config.Placeholder), RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.DoubleWithMultiplierRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.DoubleWithRoundNumber, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.DoubleAllFloatRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.ENGLISH)
},
{
new Regex(NumbersDefinitions.DoubleExponentialNotationRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.POWER_SUFFIX)
},
{
new Regex(NumbersDefinitions.DoubleCaretExponentialNotationRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.POWER_SUFFIX)
},
{
GenerateLongFormatNumberRegexes(LongFormatType.DoubleNumCommaDot, config.Placeholder, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
GenerateLongFormatNumberRegexes(LongFormatType.DoubleNumBlankDot, config.Placeholder, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
GenerateLongFormatNumberRegexes(LongFormatType.DoubleNumNoBreakSpaceDot, config.Placeholder, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
};
Regexes = regexes.ToImmutableDictionary();
}
public IntegerExtractor(CJKNumberExtractorMode mode = CJKNumberExtractorMode.Default)
{
var regexes = new Dictionary <Regex, TypeTag>
{
{
// 123456, －１２３４５６
RegexCache.Get(NumbersDefinitions.NumbersSpecialsChars, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
// 15k, 16 G
RegexCache.Get(NumbersDefinitions.NumbersSpecialsCharsWithSuffix, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
// 1,234, ２，３３２，１１１
RegexCache.Get(NumbersDefinitions.DottedNumbersSpecialsChar, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
// 半百 半ダース
RegexCache.Get(NumbersDefinitions.NumbersWithHalfDozen, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.JAPANESE)
},
{
// 一ダース 五十ダース
RegexCache.Get(NumbersDefinitions.NumbersWithDozen, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.JAPANESE)
},
};
switch (mode)
{
case CJKNumberExtractorMode.Default:
// 一百五十五, 负一亿三百二十二.
// Uses an allow list to avoid extracting "西九条" from "九"
regexes.Add(
RegexCache.Get(NumbersDefinitions.NumbersWithAllowListRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.JAPANESE));
break;
case CJKNumberExtractorMode.ExtractAll:
// 一百五十五, 负一亿三百二十二, "西九条" from "九"
// Uses no allow lists and extracts all potential integers (useful in Units, for example).
regexes.Add(
RegexCache.Get(NumbersDefinitions.NumbersAggressiveRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.JAPANESE));
break;
}
Regexes = regexes.ToImmutableDictionary();
}
private OrdinalExtractor(BaseNumberOptionsConfiguration config)
: base(config.Options)
{
var regexes = new Dictionary <Regex, TypeTag>
{
{
new Regex(NumbersDefinitions.OrdinalKoreanRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.KOREAN)
},
};
Regexes = regexes.ToImmutableDictionary();
}
public OrdinalExtractor()
{
var regexes = new Dictionary <Regex, TypeTag>
{
{
// だい一百五十四
RegexCache.Get(NumbersDefinitions.AllOrdinalRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.JAPANESE)
},
};
Regexes = regexes.ToImmutableDictionary();
}
public DoubleExtractor()
{
var regexes = new Dictionary <Regex, TypeTag>
{
{
new Regex(NumbersDefinitions.DoubleSpecialsChars, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
// (-)2.5, can avoid cases like ip address xx.xx.xx.xx
new Regex(NumbersDefinitions.DoubleSpecialsCharsWithNegatives, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
// (-).2
new Regex(NumbersDefinitions.SimpleDoubleSpecialsChars, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
// えは九・二三二一三一二
new Regex(NumbersDefinitions.DoubleRoundNumberSpecialsChars, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
// 1.0 K
new Regex(NumbersDefinitions.DoubleWithMultiplierRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
// １５.２万
new Regex(NumbersDefinitions.DoubleWithThousandsRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.JAPANESE)
},
{
// 2e6, 21.2e0
new Regex(NumbersDefinitions.DoubleExponentialNotationRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.POWER_SUFFIX)
},
{
new Regex(NumbersDefinitions.DoubleExponentialNotationKanjiRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.POWER_SUFFIX)
},
{
// 2^5
new Regex(NumbersDefinitions.DoubleScientificNotationRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.POWER_SUFFIX)
},
};
Regexes = regexes.ToImmutableDictionary();
}
private OrdinalExtractor()
{
this.Regexes = new Dictionary <Regex, TypeTag>
{
{
new Regex(NumbersDefinitions.OrdinalSuffixRegex, RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.OrdinalFrenchRegex, RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.FRENCH)
}
}.ToImmutableDictionary();
}
private OrdinalExtractor()
{
this.Regexes = new Dictionary <Regex, TypeTag>
{
{
RegexCache.Get(NumbersDefinitions.OrdinalSuffixRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.NUMBER_SUFFIX)
},
{
RegexCache.Get(NumbersDefinitions.OrdinalFrenchRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.FRENCH)
},
}.ToImmutableDictionary();
}
} = Constants.SYS_NUM_ORDINAL; // "Ordinal";
public OrdinalExtractor()
{
this.Regexes = new Dictionary <Regex, TypeTag>
{
{
new Regex(NumbersDefinitions.OrdinalSuffixRegex, RegexOptions.IgnoreCase | RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.OrdinalItalianRegex, RegexOptions.IgnoreCase | RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.ITALIAN)
}
}.ToImmutableDictionary();
}
public IntegerExtractor(JapaneseNumberExtractorMode mode = JapaneseNumberExtractorMode.Default)
{
var regexes = new Dictionary <Regex, string>()
{
{
// 123456, －１２３４５６
new Regex(NumbersDefinitions.NumbersSpecialsChars, RegexOptions.IgnoreCase | RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
//15k, 16 G
new Regex(NumbersDefinitions.NumbersSpecialsCharsWithSuffix, RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
//1,234, ２，３３２，１１１
new Regex(NumbersDefinitions.DottedNumbersSpecialsChar, RegexOptions.IgnoreCase | RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
//半百 半ダース
new Regex(NumbersDefinitions.NumbersWithHalfDozen, RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.JAPANESE)
},
{
//一ダース 五十ダース
new Regex(NumbersDefinitions.NumbersWithDozen, RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.JAPANESE)
}
};
switch (mode)
{
case JapaneseNumberExtractorMode.Default:
// 一百五十五, 负一亿三百二十二.
// Uses an allow list to avoid extracting "西九条" from "九"
regexes.Add(new Regex(NumbersDefinitions.NumbersWithAllowListRegex, RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.JAPANESE));
break;
case JapaneseNumberExtractorMode.ExtractAll:
// 一百五十五, 负一亿三百二十二, "西九条" from "九"
// Uses no allow lists and extracts all potential integers (useful in Units, for example).
regexes.Add(new Regex(NumbersDefinitions.NumbersAggressiveRegex, RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.JAPANESE));
break;
}
Regexes = regexes.ToImmutableDictionary();
}
private OrdinalExtractor()
{
this.Regexes = new Dictionary <Regex, TypeTag>
{
{
new Regex(NumbersDefinitions.OrdinalSuffixRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.OrdinalItalianRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.ITALIAN)
},
}.ToImmutableDictionary();
}
private DoubleExtractor(string placeholder = NumbersDefinitions.PlaceHolderDefault)
{
var regexes = new Dictionary <Regex, TypeTag>
{
{
RegexCache.Get(NumbersDefinitions.DoubleDecimalPointRegex(placeholder), RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
RegexCache.Get(NumbersDefinitions.DoubleWithoutIntegralRegex(placeholder), RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
RegexCache.Get(NumbersDefinitions.DoubleWithMultiplierRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
RegexCache.Get(NumbersDefinitions.DoubleWithRoundNumber, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
RegexCache.Get(NumbersDefinitions.DoubleAllFloatRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.HINDI)
},
{
RegexCache.Get(NumbersDefinitions.DoubleExponentialNotationRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.POWER_SUFFIX)
},
{
RegexCache.Get(NumbersDefinitions.DoubleCaretExponentialNotationRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.POWER_SUFFIX)
},
{
GenerateLongFormatNumberRegexes(LongFormatType.DoubleNumCommaDot, placeholder, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
GenerateLongFormatNumberRegexes(LongFormatType.DoubleNumBlankDot, placeholder, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
GenerateLongFormatNumberRegexes(LongFormatType.DoubleNumNoBreakSpaceDot, placeholder, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
};
Regexes = regexes.ToImmutableDictionary();
}
private IntegerExtractor(string placeholder = NumbersDefinitions.PlaceHolderDefault)
{
var regexes = new Dictionary <Regex, TypeTag>
{
{
new Regex(NumbersDefinitions.NumbersWithPlaceHolder(placeholder), RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.NumbersWithSuffix, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.RoundNumberIntegerRegexWithLocks, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.NumbersWithDozenSuffix, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.AllIntRegexWithLocks, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.TURKISH)
},
{
new Regex(NumbersDefinitions.NegativeAllIntRegexWithLocks, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.TURKISH)
},
{
new Regex(NumbersDefinitions.AllIntRegexWithDozenSuffixLocks, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.TURKISH)
},
{
GenerateLongFormatNumberRegexes(LongFormatType.IntegerNumDot, placeholder),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
GenerateLongFormatNumberRegexes(LongFormatType.IntegerNumBlank, placeholder),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
{
GenerateLongFormatNumberRegexes(LongFormatType.IntegerNumNoBreakSpace, placeholder),
RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
},
};
Regexes = regexes.ToImmutableDictionary();
}
private OrdinalExtractor()
{
var regexes = new Dictionary <Regex, TypeTag>
{
{
new Regex(NumbersDefinitions.OrdinalSuffixRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.OrdinalNounRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.SPANISH)
},
};
this.Regexes = regexes.ToImmutableDictionary();
}
private OrdinalExtractor()
{
var regexes = new Dictionary <Regex, TypeTag>
{
{
new Regex(NumbersDefinitions.OrdinalSuffixRegex, RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.NUMBER_SUFFIX)
},
{
new Regex(NumbersDefinitions.OrdinalEnglishRegex, RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.PORTUGUESE)
}
};
this.Regexes = regexes.ToImmutableDictionary();
}
public DoubleExtractor(BaseNumberOptionsConfiguration config)
{
var regexes = new Dictionary <Regex, TypeTag>
{
{
RegexCache.Get(NumbersDefinitions.DoubleSpecialsChars, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
// (-)2.5, can avoid cases like ip address xx.xx.xx.xx
RegexCache.Get(NumbersDefinitions.DoubleSpecialsCharsWithNegatives, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
// (-).2
RegexCache.Get(NumbersDefinitions.SimpleDoubleSpecialsChars, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
// 1.0 K
RegexCache.Get(NumbersDefinitions.DoubleWithMultiplierRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
// １５.２万
RegexCache.Get(NumbersDefinitions.DoubleWithThousandsRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.CHINESE)
},
{
// 四十五点三三
RegexCache.Get(NumbersDefinitions.DoubleAllFloatRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.CHINESE)
},
{
// 2e6, 21.2e0
RegexCache.Get(NumbersDefinitions.DoubleExponentialNotationRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.POWER_SUFFIX)
},
{
// 2^5
RegexCache.Get(NumbersDefinitions.DoubleScientificNotationRegex, RegexFlags),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.POWER_SUFFIX)
},
};
Regexes = regexes.ToImmutableDictionary();
}
public DoubleExtractor()
{
var regexes = new Dictionary <Regex, TypeTag>
{
{
new Regex(NumbersDefinitions.DoubleSpecialsChars, RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
// (-)2.5, can avoid cases like ip address xx.xx.xx.xx
new Regex(NumbersDefinitions.DoubleSpecialsCharsWithNegatives, RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
//(-).2
new Regex(NumbersDefinitions.SimpleDoubleSpecialsChars, RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
// 1.0 K
new Regex(NumbersDefinitions.DoubleWithMultiplierRegex, RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
},
{
//１５.２万
new Regex(NumbersDefinitions.DoubleWithThousandsRegex, RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.CHINESE)
},
{
//四十五点三三
new Regex(NumbersDefinitions.DoubleAllFloatRegex, RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.CHINESE)
},
{
// 2e6, 21.2e0
new Regex(NumbersDefinitions.DoubleExponentialNotationRegex, RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.POWER_SUFFIX)
},
{
//2^5
new Regex(NumbersDefinitions.DoubleScientificNotationRegex, RegexOptions.Singleline),
RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.POWER_SUFFIX)
}
};
Regexes = regexes.ToImmutableDictionary();
}