예제 #1
0
        public void RegexCacheGetMethodOption()
        {
            RegexCache rc = new RegexCache();
            var        r  = rc.Get("pista", RegexOptions.Multiline);

            Assert.IsTrue(r.Options == RegexOptions.Multiline);
        }
예제 #2
0
        private IntegerExtractor(string placeholder = NumbersDefinitions.PlaceHolderDefault)
        {
            var regexes = new Dictionary <Regex, TypeTag>
            {
                {
                    RegexCache.Get(NumbersDefinitions.NumbersWithPlaceHolder(placeholder), RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
                },
                {
                    RegexCache.Get(NumbersDefinitions.IndianNumberingSystemRegex, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
                },
                {
                    RegexCache.Get(NumbersDefinitions.NumbersWithSuffix, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
                },
                {
                    RegexCache.Get(NumbersDefinitions.RoundNumberIntegerRegexWithLocks, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
                },
                {
                    RegexCache.Get(NumbersDefinitions.NumbersWithDozenSuffix, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
                },
                {
                    RegexCache.Get(NumbersDefinitions.AllIntRegexWithLocks, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.HINDI)
                },
                {
                    RegexCache.Get(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)
                },
                {
                    RegexCache.Get(NumbersDefinitions.NegativeHinglishRegex, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.HINDI)
                },
                {
                    RegexCache.Get(NumbersDefinitions.CompoundEnglishNumberRegex, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.HINDI)
                },
                {
                    RegexCache.Get(NumbersDefinitions.DecimalUnitsWithRoundNumberRegex, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.HINDI)
                },
            };

            Regexes = regexes.ToImmutableDictionary();
        }
        private Regex BuildTextNumberRegex()
        {
            var singleIntFrac = $"{this.Config.WordSeparatorToken}| -|" +
                                GetKeyRegex(this.Config.CardinalNumberMap.Keys) + "|" +
                                GetKeyRegex(this.Config.OrdinalNumberMap.Keys);

            // @TODO consider remodeling the creation of this regex
            // For Italian, we invert the order of Cardinal and Ordinal in singleIntFrac in order to correctly extract
            // ordinals that contain cardinals such as 'tredicesimo' (thirteenth) which starts with 'tre' (three).
            // With the standard order, the parser fails to return '13' since only the cardinal 'tre' (3) is extracted
            if (this.Config.CultureInfo.Name == "it-IT")
            {
                singleIntFrac = $"{this.Config.WordSeparatorToken}| -|" +
                                GetKeyRegex(this.Config.OrdinalNumberMap.Keys) + "|" +
                                GetKeyRegex(this.Config.CardinalNumberMap.Keys);
            }

            string textNumberPattern;

            // Checks for languages that use "compound numbers". I.e. written number parts are not separated by whitespaces or special characters (e.g., dreihundert in German).
            if (isCompoundNumberLanguage)
            {
                textNumberPattern = @"(" + singleIntFrac + @")";
            }
            else
            {
                // Default case, like in English.
                textNumberPattern = @"(?<=\b)(" + singleIntFrac + @")(?=\b)";
            }

            return(RegexCache.Get(textNumberPattern, RegexOptions.Singleline | RegexOptions.Compiled));
        }
예제 #4
0
        public NumberExtractor(BaseNumberOptionsConfiguration config, CJKNumberExtractorMode mode = CJKNumberExtractorMode.Default)
        {
            var builder = ImmutableDictionary.CreateBuilder<Regex, TypeTag>();

            // Add Cardinal
            var cardExtractChs = new CardinalExtractor(config, mode);
            builder.AddRange(cardExtractChs.Regexes);

            // Add Fraction
            var fracExtractChs = new FractionExtractor(config);
            builder.AddRange(fracExtractChs.Regexes);

            Regexes = builder.ToImmutable();

            var ambiguityBuilder = ImmutableDictionary.CreateBuilder<Regex, Regex>();

            // Do not filter the ambiguous number cases like 'that one' in NumberWithUnit, otherwise they can't be resolved.
            if (config.Mode != NumberMode.Unit)
            {
                foreach (var item in NumbersDefinitions.AmbiguityFiltersDict)
                {
                    ambiguityBuilder.Add(RegexCache.Get(item.Key, RegexFlags), RegexCache.Get(item.Value, RegexFlags));
                }
            }

            AmbiguityFiltersDict = ambiguityBuilder.ToImmutable();
        }
        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();
        }
        public BaseURLExtractor(URLConfiguration config)
        {
            this.config = config;

            var regexes = new Dictionary <Regex, string>
            {
                {
                    config.UrlRegex,
                    Constants.URL_REGEX
                },
                {
                    config.IpUrlRegex,
                    Constants.URL_REGEX
                },
                {
                    RegexCache.Get(BaseURL.UrlRegex2, RegexOptions.Compiled),
                    Constants.URL_REGEX
                },
            };

            Regexes           = regexes.ToImmutableDictionary();
            AmbiguousTimeTerm = RegexCache.Get(BaseURL.AmbiguousTimeTerm, RegexOptions.Compiled);

            TldMatcher = new StringMatcher();
            TldMatcher.Init(BaseURL.TldList);
        }
예제 #7
0
        public FrenchNumberParserConfiguration(INumberOptionsConfiguration config)
        {
            this.Config         = config;
            this.LanguageMarker = NumbersDefinitions.LangMarker;
            this.CultureInfo    = new CultureInfo(config.Culture);

            this.IsCompoundNumberLanguage       = NumbersDefinitions.CompoundNumberLanguage;
            this.IsMultiDecimalSeparatorCulture = NumbersDefinitions.MultiDecimalSeparatorCulture;

            this.DecimalSeparatorChar    = NumbersDefinitions.DecimalSeparatorChar;
            this.FractionMarkerToken     = NumbersDefinitions.FractionMarkerToken;
            this.NonDecimalSeparatorChar = NumbersDefinitions.NonDecimalSeparatorChar;
            this.HalfADozenText          = NumbersDefinitions.HalfADozenText;
            this.WordSeparatorToken      = NumbersDefinitions.WordSeparatorToken;

            this.WrittenDecimalSeparatorTexts  = NumbersDefinitions.WrittenDecimalSeparatorTexts;
            this.WrittenGroupSeparatorTexts    = NumbersDefinitions.WrittenGroupSeparatorTexts;
            this.WrittenIntegerSeparatorTexts  = NumbersDefinitions.WrittenIntegerSeparatorTexts;
            this.WrittenFractionSeparatorTexts = NumbersDefinitions.WrittenFractionSeparatorTexts;

            this.CardinalNumberMap              = NumbersDefinitions.CardinalNumberMap.ToImmutableDictionary();
            this.RelativeReferenceOffsetMap     = NumbersDefinitions.RelativeReferenceOffsetMap.ToImmutableDictionary();
            this.RelativeReferenceRelativeToMap = NumbersDefinitions.RelativeReferenceRelativeToMap.ToImmutableDictionary();
            this.OrdinalNumberMap = NumberMapGenerator.InitOrdinalNumberMap(NumbersDefinitions.OrdinalNumberMap, NumbersDefinitions.PrefixCardinalMap, NumbersDefinitions.SuffixOrdinalMap);
            this.RoundNumberMap   = NumbersDefinitions.RoundNumberMap.ToImmutableDictionary();

            // @TODO Change init to follow design in other languages
            this.HalfADozenRegex          = RegexCache.Get(NumbersDefinitions.HalfADozenRegex, RegexFlags);
            this.DigitalNumberRegex       = RegexCache.Get(NumbersDefinitions.DigitalNumberRegex, RegexFlags);
            this.NegativeNumberSignRegex  = RegexCache.Get(NumbersDefinitions.NegativeNumberSignRegex, RegexFlags);
            this.FractionPrepositionRegex = RegexCache.Get(NumbersDefinitions.FractionPrepositionRegex, RegexFlags);
        }
예제 #8
0
        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();
        }
        public HindiNumberParserConfiguration(INumberOptionsConfiguration config)
        {
            this.Config         = config;
            this.LanguageMarker = NumbersDefinitions.LangMarker;
            this.CultureInfo    = new CultureInfo(config.Culture);

            this.IsCompoundNumberLanguage       = NumbersDefinitions.CompoundNumberLanguage;
            this.IsMultiDecimalSeparatorCulture = NumbersDefinitions.MultiDecimalSeparatorCulture;

            this.DecimalSeparatorChar    = NumbersDefinitions.DecimalSeparatorChar;
            this.FractionMarkerToken     = NumbersDefinitions.FractionMarkerToken;
            this.NonDecimalSeparatorChar = NumbersDefinitions.NonDecimalSeparatorChar;
            this.HalfADozenText          = NumbersDefinitions.HalfADozenText;
            this.WordSeparatorToken      = NumbersDefinitions.WordSeparatorToken;

            this.WrittenDecimalSeparatorTexts  = NumbersDefinitions.WrittenDecimalSeparatorTexts;
            this.WrittenGroupSeparatorTexts    = NumbersDefinitions.WrittenGroupSeparatorTexts;
            this.WrittenIntegerSeparatorTexts  = NumbersDefinitions.WrittenIntegerSeparatorTexts;
            this.WrittenFractionSeparatorTexts = NumbersDefinitions.WrittenFractionSeparatorTexts;

            this.CardinalNumberMap              = NumbersDefinitions.CardinalNumberMap.ToImmutableDictionary();
            this.OrdinalNumberMap               = NumbersDefinitions.OrdinalNumberMap.ToImmutableDictionary();
            this.DecimalUnitsMap                = NumbersDefinitions.DecimalUnitsMap.ToImmutableDictionary();
            this.RelativeReferenceOffsetMap     = NumbersDefinitions.RelativeReferenceOffsetMap.ToImmutableDictionary();
            this.RelativeReferenceRelativeToMap = NumbersDefinitions.RelativeReferenceRelativeToMap.ToImmutableDictionary();
            this.RoundNumberMap = NumbersDefinitions.RoundNumberMap.ToImmutableDictionary();
            this.ZeroToNineMap  = NumbersDefinitions.ZeroToNineMap.ToImmutableDictionary();

            this.AdditionTermsRegex              = RegexCache.Get(NumbersDefinitions.AdditionTermsRegex, RegexFlags);
            this.HalfADozenRegex                 = RegexCache.Get(NumbersDefinitions.HalfADozenRegex, RegexFlags);
            this.DigitalNumberRegex              = RegexCache.Get(NumbersDefinitions.DigitalNumberRegex, RegexFlags);
            this.NegativeNumberSignRegex         = RegexCache.Get(NumbersDefinitions.NegativeNumberSignRegex, RegexFlags);
            this.FractionPrepositionRegex        = RegexCache.Get(NumbersDefinitions.FractionPrepositionRegex, RegexFlags);
            this.FractionPrepositionInverseRegex = RegexCache.Get(NumbersDefinitions.FractionPrepositionInverseRegex, RegexFlags);
        }
        public GermanDateTimeParserConfiguration(ICommonDateTimeParserConfiguration config)
            : base(config)
        {
            TokenBeforeDate = DateTimeDefinitions.TokenBeforeDate;
            TokenBeforeTime = DateTimeDefinitions.TokenBeforeTime;

            DateExtractor = config.DateExtractor;
            TimeExtractor = config.TimeExtractor;
            DateParser    = config.DateParser;
            TimeParser    = config.TimeParser;

            NowRegex = GermanDateTimeExtractorConfiguration.NowRegex;

            AMTimeRegex = RegexCache.Get(DateTimeDefinitions.AMTimeRegex, RegexFlags);
            PMTimeRegex = RegexCache.Get(DateTimeDefinitions.PMTimeRegex, RegexFlags);

            SimpleTimeOfTodayAfterRegex  = GermanDateTimeExtractorConfiguration.SimpleTimeOfTodayAfterRegex;
            SimpleTimeOfTodayBeforeRegex = GermanDateTimeExtractorConfiguration.SimpleTimeOfTodayBeforeRegex;
            SpecificTimeOfDayRegex       = GermanDateTimeExtractorConfiguration.SpecificTimeOfDayRegex;
            SpecificEndOfRegex           = GermanDateTimeExtractorConfiguration.SpecificEndOfRegex;
            UnspecificEndOfRegex         = GermanDateTimeExtractorConfiguration.UnspecificEndOfRegex;
            UnitRegex = GermanTimeExtractorConfiguration.TimeUnitRegex;
            DateNumberConnectorRegex = GermanDateTimeExtractorConfiguration.DateNumberConnectorRegex;
            YearRegex = GermanDateTimeExtractorConfiguration.YearRegex;

            Numbers              = config.Numbers;
            CardinalExtractor    = config.CardinalExtractor;
            IntegerExtractor     = config.IntegerExtractor;
            NumberParser         = config.NumberParser;
            DurationExtractor    = config.DurationExtractor;
            DurationParser       = config.DurationParser;
            UnitMap              = config.UnitMap;
            UtilityConfiguration = config.UtilityConfiguration;
        }
        public FrenchDateExtractorConfiguration(IDateTimeOptionsConfiguration config)
            : base(config)
        {
            var numOptions = NumberOptions.None;

            if ((config.Options & DateTimeOptions.NoProtoCache) != 0)
            {
                numOptions = NumberOptions.NoProtoCache;
            }

            var numConfig = new BaseNumberOptionsConfiguration(config.Culture, numOptions);

            IntegerExtractor = Number.French.IntegerExtractor.GetInstance();
            OrdinalExtractor = Number.French.OrdinalExtractor.GetInstance();
            NumberParser     = new BaseNumberParser(new FrenchNumberParserConfiguration(numConfig));

            DurationExtractor    = new BaseDurationExtractor(new FrenchDurationExtractorConfiguration(this));
            HolidayExtractor     = new BaseHolidayExtractor(new FrenchHolidayExtractorConfiguration(this));
            UtilityConfiguration = new FrenchDatetimeUtilityConfiguration();

            // 3-23-2017
            var dateRegex4 = RegexCache.Get(DateTimeDefinitions.DateExtractor4, RegexFlags);

            // 23-3-2015
            var dateRegex5 = RegexCache.Get(DateTimeDefinitions.DateExtractor5, RegexFlags);

            // on 1.3
            var dateRegex6 = RegexCache.Get(DateTimeDefinitions.DateExtractor6, RegexFlags);

            // on 24-12
            var dateRegex8 = RegexCache.Get(DateTimeDefinitions.DateExtractor8, RegexFlags);

            // 7/23
            var dateRegex7 = RegexCache.Get(DateTimeDefinitions.DateExtractor7, RegexFlags);

            // 23/7
            var dateRegex9 = RegexCache.Get(DateTimeDefinitions.DateExtractor9, RegexFlags);

            // 2015-12-23
            var dateRegexA = RegexCache.Get(DateTimeDefinitions.DateExtractorA, RegexFlags);

            DateRegexList = new List <Regex>
            {
                // (Sunday,)? April 5
                RegexCache.Get(DateTimeDefinitions.DateExtractor1, RegexFlags),

                // (Sunday,)? April 5, 2016
                RegexCache.Get(DateTimeDefinitions.DateExtractor2, RegexFlags),

                // (Sunday,)? 6th of April
                RegexCache.Get(DateTimeDefinitions.DateExtractor3, RegexFlags),
            };

            var enableDmy = DmyDateFormat ||
                            DateTimeDefinitions.DefaultLanguageFallback == Constants.DefaultLanguageFallback_DMY;

            DateRegexList = DateRegexList.Concat(enableDmy ?
                                                 new[] { dateRegex5, dateRegex8, dateRegex9, dateRegex4, dateRegex6, dateRegex7, dateRegexA } :
                                                 new[] { dateRegex4, dateRegex6, dateRegex7, dateRegex5, dateRegex8, dateRegex9, dateRegexA });
        }
예제 #12
0
        private OrdinalExtractor(BaseNumberOptionsConfiguration config)
            : base(config.Options)
        {
            keyPrefix = string.Intern(ExtractType + "_" + config.Options.ToString() + "_" + config.Culture);

            AmbiguousFractionConnectorsRegex = RegexCache.Get(NumbersDefinitions.AmbiguousFractionConnectorsRegex, RegexFlags);

            RelativeReferenceRegex = RegexCache.Get(NumbersDefinitions.RelativeOrdinalRegex, RegexFlags);

            var regexes = new Dictionary <Regex, TypeTag>
            {
                {
                    RegexCache.Get(NumbersDefinitions.OrdinalNumericRegex, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.NUMBER_SUFFIX)
                },
                {
                    RegexCache.Get(NumbersDefinitions.OrdinalEnglishRegex, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.ARABIC)
                },
                {
                    RegexCache.Get(NumbersDefinitions.OrdinalRoundNumberRegex, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.ARABIC)
                },
            };

            Regexes = regexes.ToImmutableDictionary();
        }
예제 #13
0
        private Regex BuildTextNumberRegex()
        {
            // For Hindi, there is a need for another NumberMap of the type double to handle values like 1.5.
            // As this cannot be included in either Cardinal or Ordinal NumberMap as they are of the type long,
            // DecimalUnitsList (type double) takes care of these entries and it needs to be added to the singleIntFrac
            // for extraction
            var singleIntFrac = $"{this.Config.WordSeparatorToken}| -|" +
                                GetKeyRegex(this.Config.OrdinalNumberMap.Keys) + "|" +
                                GetKeyRegex(this.Config.CardinalNumberMap.Keys) + "|" +
                                GetKeyRegex(this.Config.DecimalUnitsMap.Keys);

            string textNumberPattern;

            // Checks for languages that use "compound numbers". I.e. written number parts are not separated by whitespaces or special characters (e.g., dreihundert in German).
            if (isCompoundNumberLanguage)
            {
                textNumberPattern = @"(" + singleIntFrac + @")";
            }
            else
            {
                // Default case, like in English.
                textNumberPattern = @"(?<=\b)(" + singleIntFrac + @")(?=\b)";
            }

            return(RegexCache.Get(textNumberPattern, RegexOptions.Singleline | RegexOptions.Compiled));
        }
예제 #14
0
        protected Regex BuildSeparateRegexFromSet(bool ignoreCase = false)
        {
            var separateWords = new HashSet <string>();

            if (config.PrefixList?.Count > 0)
            {
                foreach (var addWord in config.PrefixList.Values)
                {
                    foreach (var word in addWord.Split(separators, StringSplitOptions.RemoveEmptyEntries))
                    {
                        if (ValidateUnit(word))
                        {
                            separateWords.Add(word);
                        }
                    }
                }
            }

            if (config.SuffixList?.Count > 0)
            {
                foreach (var addWord in config.SuffixList.Values)
                {
                    foreach (var word in addWord.Split(separators, StringSplitOptions.RemoveEmptyEntries))
                    {
                        if (ValidateUnit(word))
                        {
                            separateWords.Add(word);
                        }
                    }
                }
            }

            if (config.AmbiguousUnitList?.Count > 0)
            {
                var abandonWords = config.AmbiguousUnitList;
                foreach (var abandonWord in abandonWords)
                {
                    if (separateWords.Contains(abandonWord))
                    {
                        separateWords.Remove(abandonWord);
                    }
                }
            }

            // Sort separateWords using descending length.
            var regexTokens = separateWords.Select(Regex.Escape).ToList();

            if (regexTokens.Count == 0)
            {
                return(null);
            }

            regexTokens.Sort(new StringComparer());
            var pattern = $@"{this.config.BuildPrefix}({string.Join("|", regexTokens)}){this.config.BuildSuffix}";
            var options = RegexOptions.Singleline | RegexOptions.ExplicitCapture | (ignoreCase ? RegexOptions.IgnoreCase : RegexOptions.None);

            var regex = RegexCache.Get(pattern, options);

            return(regex);
        }
        private static string SanitizeGroups(string source)
        {
            Regex matchGroup = RegexCache.Get(@"\?< (?<name>\w +) >");

            var result = Regex.Replace(source, matchGroup.ToString(), ReplaceMatchGroup);

            return(result);
        }
        public IntegerExtractor(BaseNumberOptionsConfiguration config, 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  半打
                    RegexCache.Get(NumbersDefinitions.NumbersWithHalfDozen, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.CHINESE)
                },
                {
                    // 半
                    RegexCache.Get(NumbersDefinitions.HalfUnitRegex, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.CHINESE)
                },
                {
                    // 一打  五十打
                    RegexCache.Get(NumbersDefinitions.NumbersWithDozen, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.CHINESE)
                },
            };

            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 "四" from "四川"
                regexes.Add(
                    RegexCache.Get(NumbersDefinitions.NumbersWithAllowListRegex, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.CHINESE));
                break;

            case CJKNumberExtractorMode.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(
                    RegexCache.Get(NumbersDefinitions.NumbersAggressiveRegex, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.CHINESE));
                break;
            }

            Regexes = regexes.ToImmutableDictionary();
        }
예제 #17
0
 public ChinesePhoneNumberExtractorConfiguration(SequenceOptions options)
     : base(options)
 {
     WordBoundariesRegex    = PhoneNumbersDefinitions.WordBoundariesRegex;
     NonWordBoundariesRegex = PhoneNumbersDefinitions.NonWordBoundariesRegex;
     EndWordBoundariesRegex = PhoneNumbersDefinitions.EndWordBoundariesRegex;
     ColonPrefixCheckRegex  = RegexCache.Get(PhoneNumbersDefinitions.ColonPrefixCheckRegex);
     ForbiddenPrefixMarkers = (List <char>)PhoneNumbersDefinitions.ForbiddenPrefixMarkers;
 }
예제 #18
0
 public DutchHolidayParserConfiguration(IDateTimeOptionsConfiguration config)
     : base(config)
 {
     ThisPrefixRegex       = RegexCache.Get(DateTimeDefinitions.ThisPrefixRegex, RegexFlags);
     NextPrefixRegex       = RegexCache.Get(DateTimeDefinitions.NextPrefixRegex, RegexFlags);
     PreviousPrefixRegex   = RegexCache.Get(DateTimeDefinitions.PreviousPrefixRegex, RegexFlags);
     this.HolidayRegexList = DutchHolidayExtractorConfiguration.HolidayRegexList;
     this.HolidayNames     = DateTimeDefinitions.HolidayNames.ToImmutableDictionary();
 }
예제 #19
0
        public ItalianDateParserConfiguration(ICommonDateTimeParserConfiguration config)
            : base(config)
        {
            DateTokenPrefix          = DateTimeDefinitions.DateTokenPrefix;
            IntegerExtractor         = config.IntegerExtractor;
            OrdinalExtractor         = config.OrdinalExtractor;
            CardinalExtractor        = config.CardinalExtractor;
            NumberParser             = config.NumberParser;
            DurationExtractor        = config.DurationExtractor;
            DateExtractor            = config.DateExtractor;
            DurationParser           = config.DurationParser;
            HolidayParser            = new BaseHolidayParser(new ItalianHolidayParserConfiguration(this));
            DateRegexes              = new ItalianDateExtractorConfiguration(this).DateRegexList;
            OnRegex                  = ItalianDateExtractorConfiguration.OnRegex;
            SpecialDayRegex          = ItalianDateExtractorConfiguration.SpecialDayRegex;
            SpecialDayWithNumRegex   = ItalianDateExtractorConfiguration.SpecialDayWithNumRegex;
            NextRegex                = ItalianDateExtractorConfiguration.NextDateRegex;
            ThisRegex                = ItalianDateExtractorConfiguration.ThisRegex;
            LastRegex                = ItalianDateExtractorConfiguration.LastDateRegex;
            UnitRegex                = ItalianDateExtractorConfiguration.DateUnitRegex;
            WeekDayRegex             = ItalianDateExtractorConfiguration.WeekDayRegex;
            StrictWeekDay            = ItalianDateExtractorConfiguration.StrictWeekDay;
            MonthRegex               = ItalianDateExtractorConfiguration.MonthRegex;
            WeekDayOfMonthRegex      = ItalianDateExtractorConfiguration.WeekDayOfMonthRegex;
            ForTheRegex              = ItalianDateExtractorConfiguration.ForTheRegex;
            WeekDayAndDayOfMothRegex = ItalianDateExtractorConfiguration.WeekDayAndDayOfMothRegex;
            WeekDayAndDayRegex       = ItalianDateExtractorConfiguration.WeekDayAndDayRegex;
            RelativeMonthRegex       = ItalianDateExtractorConfiguration.RelativeMonthRegex;
            StrictRelativeRegex      = ItalianDateExtractorConfiguration.StrictRelativeRegex;
            YearSuffix               = ItalianDateExtractorConfiguration.YearSuffix;
            RelativeWeekDayRegex     = ItalianDateExtractorConfiguration.RelativeWeekDayRegex;
            BeforeAfterRegex         = ItalianDateExtractorConfiguration.BeforeAfterRegex;

            // @TODO move to config
            RelativeDayRegex    = RegexCache.Get(DateTimeDefinitions.RelativeDayRegex, RegexFlags);
            NextPrefixRegex     = RegexCache.Get(DateTimeDefinitions.NextPrefixRegex, RegexFlags);
            PreviousPrefixRegex = RegexCache.Get(DateTimeDefinitions.PreviousPrefixRegex, RegexFlags);
            UpcomingPrefixRegex = RegexCache.Get(DateTimeDefinitions.UpcomingPrefixRegex, RegexFlags);
            PastPrefixRegex     = RegexCache.Get(DateTimeDefinitions.PastPrefixRegex, RegexFlags);

            DayOfMonth           = config.DayOfMonth;
            DayOfWeek            = config.DayOfWeek;
            MonthOfYear          = config.MonthOfYear;
            Numbers              = config.Numbers;
            CardinalMap          = config.CardinalMap;
            UnitMap              = config.UnitMap;
            UtilityConfiguration = config.UtilityConfiguration;

            SameDayTerms     = DateTimeDefinitions.SameDayTerms.ToImmutableList();
            PlusOneDayTerms  = DateTimeDefinitions.PlusOneDayTerms.ToImmutableList();
            PlusTwoDayTerms  = DateTimeDefinitions.PlusTwoDayTerms.ToImmutableList();
            MinusOneDayTerms = DateTimeDefinitions.MinusOneDayTerms.ToImmutableList();
            MinusTwoDayTerms = DateTimeDefinitions.MinusTwoDayTerms.ToImmutableList();
        }
 public LongFormTestConfiguration(char decimalSep, char nonDecimalSep)
 {
     this.DecimalSeparatorChar    = decimalSep;
     this.NonDecimalSeparatorChar = nonDecimalSep;
     this.CultureInfo             = new CultureInfo(Culture.English);
     this.CardinalNumberMap       = ImmutableDictionary <string, long> .Empty;
     this.OrdinalNumberMap        = ImmutableDictionary <string, long> .Empty;
     this.RoundNumberMap          = ImmutableDictionary <string, long> .Empty;
     this.DigitalNumberRegex      = RegexCache.Get(
         @"((?<=\b)(hundred|thousand|million|billion|trillion|dozen(s)?)(?=\b))|((?<=(\d|\b))(k|t|m|g|b)(?=\b))", RegexOptions.Singleline);
 }
        public BasePhoneNumberExtractor(PhoneNumberConfiguration config)
        {
            this.config = config;

            var wordBoundariesRegex    = config.WordBoundariesRegex;
            var nonWordBoundariesRegex = config.NonWordBoundariesRegex;
            var endWordBoundariesRegex = config.EndWordBoundariesRegex;

            var regexes = new Dictionary <Regex, string>
            {
                {
                    RegexCache.Get(BasePhoneNumbers.GeneralPhoneNumberRegex(wordBoundariesRegex, endWordBoundariesRegex), RegexOptions.Compiled),
                    Constants.PHONE_NUMBER_REGEX_GENERAL
                },
                {
                    RegexCache.Get(BasePhoneNumbers.BRPhoneNumberRegex(wordBoundariesRegex, nonWordBoundariesRegex, endWordBoundariesRegex), RegexOptions.Compiled),
                    Constants.PHONE_NUMBER_REGEX_BR
                },
                {
                    RegexCache.Get(BasePhoneNumbers.UKPhoneNumberRegex(wordBoundariesRegex, nonWordBoundariesRegex, endWordBoundariesRegex), RegexOptions.Compiled),
                    Constants.PHONE_NUMBER_REGEX_UK
                },
                {
                    RegexCache.Get(BasePhoneNumbers.DEPhoneNumberRegex(wordBoundariesRegex, endWordBoundariesRegex), RegexOptions.Compiled),
                    Constants.PHONE_NUMBER_REGEX_DE
                },
                {
                    RegexCache.Get(BasePhoneNumbers.USPhoneNumberRegex(wordBoundariesRegex, nonWordBoundariesRegex, endWordBoundariesRegex), RegexOptions.Compiled),
                    Constants.PHONE_NUMBER_REGEX_US
                },
                {
                    RegexCache.Get(BasePhoneNumbers.CNPhoneNumberRegex(wordBoundariesRegex, endWordBoundariesRegex), RegexOptions.Compiled),
                    Constants.PHONE_NUMBER_REGEX_CN
                },
                {
                    RegexCache.Get(BasePhoneNumbers.DKPhoneNumberRegex(wordBoundariesRegex, endWordBoundariesRegex), RegexOptions.Compiled),
                    Constants.PHONE_NUMBER_REGEX_DK
                },
                {
                    RegexCache.Get(BasePhoneNumbers.ITPhoneNumberRegex(wordBoundariesRegex, endWordBoundariesRegex), RegexOptions.Compiled),
                    Constants.PHONE_NUMBER_REGEX_IT
                },
                {
                    RegexCache.Get(BasePhoneNumbers.NLPhoneNumberRegex(wordBoundariesRegex, endWordBoundariesRegex), RegexOptions.Compiled),
                    Constants.PHONE_NUMBER_REGEX_NL
                },
                {
                    RegexCache.Get(BasePhoneNumbers.SpecialPhoneNumberRegex(wordBoundariesRegex, endWordBoundariesRegex), RegexOptions.Compiled),
                    Constants.PHONE_NUMBER_REGEX_SPECIAL
                },
            };

            Regexes = regexes.ToImmutableDictionary();
        }
        public BaseGUIDExtractor()
        {
            var regexes = new Dictionary <Regex, string>
            {
                {
                    RegexCache.Get(BaseGUID.GUIDRegex),
                    Constants.GUID_REGEX
                },
            };

            Regexes = regexes.ToImmutableDictionary();
        }
        public BaseHashtagExtractor()
        {
            var regexes = new Dictionary <Regex, string>
            {
                {
                    RegexCache.Get(BaseHashtag.HashtagRegex),
                    Constants.HASHTAG_REGEX
                },
            };

            Regexes = regexes.ToImmutableDictionary();
        }
        public BaseMentionExtractor()
        {
            var regexes = new Dictionary <Regex, string>
            {
                {
                    RegexCache.Get(BaseMention.MentionRegex),
                    Constants.MENTION_REGEX
                },
            };

            Regexes = regexes.ToImmutableDictionary();
        }
        private DoubleExtractor(BaseNumberOptionsConfiguration config)
            : base(config.Options)
        {
            var regexes = new Dictionary <Regex, TypeTag>
            {
                {
                    RegexCache.Get(NumbersDefinitions.DoubleDecimalPointRegex(config.Placeholder), RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
                },
                {
                    RegexCache.Get(NumbersDefinitions.DoubleIndianDecimalPointRegex, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.DOUBLE_PREFIX, Constants.NUMBER_SUFFIX)
                },
                {
                    RegexCache.Get(NumbersDefinitions.DoubleWithoutIntegralRegex(config.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.ENGLISH)
                },
                {
                    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, 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();
        }
예제 #26
0
        public OrdinalExtractor()
        {
            var regexes = new Dictionary <Regex, TypeTag>
            {
                {
                    RegexCache.Get(NumbersDefinitions.OrdinalKoreanRegex, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.ORDINAL_PREFIX, Constants.NUMBER_SUFFIX)
                },
            };

            Regexes = regexes.ToImmutableDictionary();
        }
예제 #27
0
        protected static Regex GenerateLongFormatNumberRegexes(LongFormatType type, string placeholder = BaseNumbers.PlaceHolderDefault,
                                                               RegexOptions flags = RegexOptions.Singleline)
        {
            var thousandsMark = Regex.Escape(type.ThousandsMark.ToString(CultureInfo.InvariantCulture));
            var decimalsMark  = Regex.Escape(type.DecimalsMark.ToString(CultureInfo.InvariantCulture));

            var regexDefinition = type.DecimalsMark.Equals('\0') ?
                                  BaseNumbers.IntegerRegexDefinition(placeholder, thousandsMark) :
                                  BaseNumbers.DoubleRegexDefinition(placeholder, thousandsMark, decimalsMark);

            return(RegexCache.Get(regexDefinition, flags));
        }
        public ChineseNumberParserConfiguration(INumberOptionsConfiguration config)
        {
            this.LanguageMarker = NumbersDefinitions.LangMarker;
            this.CultureInfo    = new CultureInfo(config.Culture);
            this.Config         = config;

            this.IsCompoundNumberLanguage       = NumbersDefinitions.CompoundNumberLanguage;
            this.IsMultiDecimalSeparatorCulture = NumbersDefinitions.MultiDecimalSeparatorCulture;

            this.DecimalSeparatorChar    = NumbersDefinitions.DecimalSeparatorChar;
            this.FractionMarkerToken     = NumbersDefinitions.FractionMarkerToken;
            this.NonDecimalSeparatorChar = NumbersDefinitions.NonDecimalSeparatorChar;
            this.HalfADozenText          = NumbersDefinitions.HalfADozenText;
            this.WordSeparatorToken      = NumbersDefinitions.WordSeparatorToken;
            this.ZeroChar = NumbersDefinitions.ZeroChar;
            this.PairChar = NumbersDefinitions.PairChar;

            this.WrittenDecimalSeparatorTexts  = Enumerable.Empty <string>();
            this.WrittenGroupSeparatorTexts    = Enumerable.Empty <string>();
            this.WrittenIntegerSeparatorTexts  = Enumerable.Empty <string>();
            this.WrittenFractionSeparatorTexts = Enumerable.Empty <string>();

            this.CardinalNumberMap              = new Dictionary <string, long>().ToImmutableDictionary();
            this.OrdinalNumberMap               = new Dictionary <string, long>().ToImmutableDictionary();
            this.RelativeReferenceOffsetMap     = NumbersDefinitions.RelativeReferenceOffsetMap.ToImmutableDictionary();
            this.RelativeReferenceRelativeToMap = NumbersDefinitions.RelativeReferenceRelativeToMap.ToImmutableDictionary();
            this.RoundNumberMap     = NumbersDefinitions.RoundNumberMap.ToImmutableDictionary();
            this.ZeroToNineMap      = NumbersDefinitions.ZeroToNineMap.ToImmutableDictionary();
            this.RoundNumberMapChar = NumbersDefinitions.RoundNumberMapChar.ToImmutableDictionary();
            this.FullToHalfMap      = NumbersDefinitions.FullToHalfMap.ToImmutableDictionary();
            this.TratoSimMap        = NumbersDefinitions.TratoSimMap.ToImmutableDictionary();
            this.UnitMap            = NumbersDefinitions.UnitMap.OrderBy(o => o.Key.Length).ToImmutableDictionary(o => o.Key, p => p.Value);
            this.RoundDirectList    = NumbersDefinitions.RoundDirectList.ToImmutableList();
            this.TenChars           = NumbersDefinitions.TenChars.ToImmutableList();

            this.HalfADozenRegex = null;

            // @TODO Change init to follow design in other languages
            this.DigitalNumberRegex      = RegexCache.Get(NumbersDefinitions.DigitalNumberRegex, RegexFlags);
            this.DigitNumRegex           = RegexCache.Get(NumbersDefinitions.DigitNumRegex, RegexFlags);
            this.DozenRegex              = RegexCache.Get(NumbersDefinitions.DozenRegex, RegexFlags);
            this.PercentageRegex         = RegexCache.Get(NumbersDefinitions.PercentageRegex, RegexFlags);
            this.DoubleAndRoundRegex     = RegexCache.Get(NumbersDefinitions.DoubleAndRoundRegex, RegexFlags);
            this.FracSplitRegex          = RegexCache.Get(NumbersDefinitions.FracSplitRegex, RegexFlags);
            this.NegativeNumberSignRegex = RegexCache.Get(NumbersDefinitions.NegativeNumberSignRegex, RegexFlags);
            this.PointRegex              = RegexCache.Get(NumbersDefinitions.PointRegex, RegexFlags);
            this.SpeGetNumberRegex       = RegexCache.Get(NumbersDefinitions.SpeGetNumberRegex, RegexFlags);
            this.PairRegex = RegexCache.Get(NumbersDefinitions.PairRegex, RegexFlags);
            this.RoundNumberIntegerRegex  = RegexCache.Get(NumbersDefinitions.RoundNumberIntegerRegex, RegexFlags);
            this.PercentageNumRegex       = RegexCache.Get(NumbersDefinitions.PercentageNumRegex, RegexFlags);
            this.FractionPrepositionRegex = null;
        }
예제 #29
0
        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();
        }
예제 #30
0
        private IntegerExtractor(BaseNumberOptionsConfiguration config)
            : base(config.Options)
        {
            keyPrefix = string.Intern(ExtractType + "_" + config.Options + "_" + config.Placeholder + "_" + config.Culture);

            var regexes = new Dictionary <Regex, TypeTag>
            {
                {
                    RegexCache.Get(NumbersDefinitions.NumbersWithPlaceHolder(config.Placeholder), RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
                },
                {
                    RegexCache.Get(NumbersDefinitions.NumbersWithSuffix, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
                },
                {
                    GenerateLongFormatNumberRegexes(LongFormatType.IntegerNumDot, 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)
                },
                {
                    RegexCache.Get(NumbersDefinitions.RoundNumberIntegerRegexWithLocks, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
                },
                {
                    RegexCache.Get(NumbersDefinitions.NumbersWithDozen2Suffix, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
                },
                {
                    RegexCache.Get(NumbersDefinitions.NumbersWithDozenSuffix, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.NUMBER_SUFFIX)
                },
                {
                    RegexCache.Get(NumbersDefinitions.AllIntRegexWithLocks, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.PORTUGUESE)
                },
                {
                    RegexCache.Get(NumbersDefinitions.AllIntRegexWithDozenSuffixLocks, RegexFlags),
                    RegexTagGenerator.GenerateRegexTag(Constants.INTEGER_PREFIX, Constants.PORTUGUESE)
                },
            };

            this.Regexes = regexes.ToImmutableDictionary();
        }