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();
}
// CardinalExtractor = Int + Double
public CardinalExtractor(ChineseNumberExtractorMode mode = ChineseNumberExtractorMode.Default)
{
var builder = ImmutableDictionary.CreateBuilder <Regex, TypeTag>();
var intExtractChs = new IntegerExtractor(mode);
builder.AddRange(intExtractChs.Regexes);
var douExtractorChs = new DoubleExtractor();
builder.AddRange(douExtractorChs.Regexes);
Regexes = builder.ToImmutable();
}
public IntegerExtractor(ChineseNumberExtractorMode mode = ChineseNumberExtractorMode.Default)
{
var regexes = new Dictionary <Regex, string>()
{
{
// 123456, －１２３４５６
new Regex(NumbersDefinitions.NumbersSpecialsChars, RegexOptions.IgnoreCase | RegexOptions.Singleline),
"IntegerNum"
},
{
//15k, 16 G
new Regex(NumbersDefinitions.NumbersSpecialsCharsWithSuffix, RegexOptions.Singleline),
"IntegerNum"
},
{
//1,234, ２，３３２，１１１
new Regex(NumbersDefinitions.DottedNumbersSpecialsChar, RegexOptions.IgnoreCase | RegexOptions.Singleline),
"IntegerNum"
},
{
//半百 半打
new Regex(NumbersDefinitions.NumbersWithHalfDozen, RegexOptions.Singleline),
"IntegerChs"
},
{
//一打 五十打
new Regex(NumbersDefinitions.NumbersWithDozen, RegexOptions.Singleline),
"IntegerChs"
}
};
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, RegexOptions.Singleline),
"IntegerChs");
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, RegexOptions.Singleline),
"IntegerChs");
break;
}
Regexes = regexes.ToImmutableDictionary();
}
public NumberExtractor(ChineseNumberExtractorMode mode = ChineseNumberExtractorMode.Default)
{
var builder = ImmutableDictionary.CreateBuilder <Regex, TypeTag>();
// Add Cardinal
var cardExtractChs = new CardinalExtractor(mode);
builder.AddRange(cardExtractChs.Regexes);
// Add Fraction
var fracExtractChs = new FractionExtractor();
builder.AddRange(fracExtractChs.Regexes);
Regexes = builder.ToImmutable();
}