Skip to content

Commit

Permalink
[* Currency] Expanded recognition of currency value when using [curr…
Browse files Browse the repository at this point in the history 
…encyname]$ (e.g. USD$) (#2711) (#2756)

* Expanded recognition of currency value when using [currencyname]$ (e.g. USD$) (#2711)

* Propagated to NL, DE, HI, IT, SV, TR. PT, ES, FR to be treated in separate issue.

Co-authored-by: LionbridgeCS2 <[email protected]>
  • Loading branch information
@aitelint
aitelint and LionbridgeCS2 authored Nov 10, 2021
1 parent 52c16ac commit 4db5674
Show file tree
Hide file tree
Showing 22 changed files with 984 additions and 295 deletions.
2 changes: 1 addition & 1 deletion .NET/Microsoft.Recognizers.Definitions.Common/Dutch/NumbersWithUnitDefinitions.cs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -530,7 +530,7 @@ public static class NumbersWithUnitDefinitions
{ @"Rwandan franc", @"RWF" },
{ @"Russian ruble", @"RUB" },
{ @"Transnistrian ruble", @"PRB" },
{ @"Belarusian ruble", @"BYN" },
{ @"New Belarusian ruble", @"BYN" },
{ @"Algerian dinar", @"DZD" },
{ @"Bahraini dinar", @"BHD" },
{ @"Iraqi dinar", @"IQD" },
Expand Down
268 changes: 134 additions & 134 deletions .NET/Microsoft.Recognizers.Definitions.Common/German/NumbersWithUnitDefinitions.cs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -312,156 +312,156 @@ public static class NumbersWithUnitDefinitions
{ @"Afghan afghani", @"AFN" },
{ @"Euro", @"EUR" },
{ @"Albanian lek", @"ALL" },
{ @"Kwanza angolais", @"AOA" },
{ @"Angolan kwanza", @"AOA" },
{ @"Armenian dram", @"AMD" },
{ @"Florins d'Aruba", @"AWG" },
{ @"Aruban florin", @"AWG" },
{ @"Bangladeshi taka", @"BDT" },
{ @"Ngultrum bhoutanais", @"BTN" },
{ @"Boliviano bolivien", @"BOB" },
{ @"Bosnie-Herzégovine mark convertible", @"BAM" },
{ @"Pula", @"BWP" },
{ @"Réal brésilien", @"BRL" },
{ @"Lev bulgare", @"BGN" },
{ @"Riel cambodgien", @"KHR" },
{ @"Escudo du cap-vert", @"CVE" },
{ @"Colon du costa rica", @"CRC" },
{ @"Kuna croate", @"HRK" },
{ @"Couronne tchèque", @"CZK" },
{ @"Nakfas érythréens", @"ERN" },
{ @"Birr éthiopien", @"ETB" },
{ @"Dalasi gambienne", @"GMD" },
{ @"Lari géorgien", @"GEL" },
{ @"Cedi", @"GHS" },
{ @"Quetzal guatémaltèque", @"GTQ" },
{ @"Gourdes haïtiennes", @"HTG" },
{ @"Lempira hondurien", @"HNL" },
{ @"Forint hongrois", @"HUF" },
{ @"Rial iranien", @"IRR" },
{ @"Bhutanese ngultrum", @"BTN" },
{ @"Bolivian boliviano", @"BOB" },
{ @"Bosnia and Herzegovina convertible mark", @"BAM" },
{ @"Botswana pula", @"BWP" },
{ @"Brazilian real", @"BRL" },
{ @"Bulgarian lev", @"BGN" },
{ @"Cambodian riel", @"KHR" },
{ @"Cape Verdean escudo", @"CVE" },
{ @"Costa Rican colón", @"CRC" },
{ @"Croatian kuna", @"HRK" },
{ @"Czech koruna", @"CZK" },
{ @"Eritrean nakfa", @"ERN" },
{ @"Ethiopian birr", @"ETB" },
{ @"Gambian dalasi", @"GMD" },
{ @"Georgian lari", @"GEL" },
{ @"Ghanaian cedi", @"GHS" },
{ @"Guatemalan quetzal", @"GTQ" },
{ @"Haitian gourde", @"HTG" },
{ @"Honduran lempira", @"HNL" },
{ @"Hungarian forint", @"HUF" },
{ @"Iranian rial", @"IRR" },
{ @"Yemeni rial", @"YER" },
{ @"Israeli new shekel", @"ILS" },
{ @"Yen Japonais", @"JPY" },
{ @"Tenge kazakh", @"KZT" },
{ @"Shilling kényan", @"KES" },
{ @"Corée du nord won", @"KPW" },
{ @"Won sud-coréen", @"KRW" },
{ @"Som Kirghizie", @"KGS" },
{ @"Kip laotien", @"LAK" },
{ @"Loti", @"LSL" },
{ @"Rand sud-africain", @"ZAR" },
{ @"Pataca macanais", @"MOP" },
{ @"Dinar macédonien", @"MKD" },
{ @"Ariary malagache", @"MGA" },
{ @"Kwacha malawien", @"MWK" },
{ @"Ringitt malaisien", @"MYR" },
{ @"Ouguiya mauritanienne", @"MRO" },
{ @"Togrogs mongoles", @"MNT" },
{ @"Metical mozambique", @"MZN" },
{ @"Kyat birmanie", @"MMK" },
{ @"Cordoba nicaraguayen", @"NIO" },
{ @"Japanese yen", @"JPY" },
{ @"Kazakhstani tenge", @"KZT" },
{ @"Kenyan shilling", @"KES" },
{ @"North Korean won", @"KPW" },
{ @"South Korean won", @"KRW" },
{ @"Kyrgyzstani som", @"KGS" },
{ @"Lao kip", @"LAK" },
{ @"Lesotho loti", @"LSL" },
{ @"South African rand", @"ZAR" },
{ @"Macanese pataca", @"MOP" },
{ @"Macedonian denar", @"MKD" },
{ @"Malagasy ariary", @"MGA" },
{ @"Malawian kwacha", @"MWK" },
{ @"Malaysian ringgit", @"MYR" },
{ @"Mauritanian ouguiya", @"MRO" },
{ @"Mongolian tögrög", @"MNT" },
{ @"Mozambican metical", @"MZN" },
{ @"Burmese kyat", @"MMK" },
{ @"Nicaraguan córdoba", @"NIO" },
{ @"Nigerian naira", @"NGN" },
{ @"Livre turque", @"TRY" },
{ @"Rials omanais", @"OMR" },
{ @"Balboa panaméennes", @"PAB" },
{ @"Kina", @"PGK" },
{ @"Guaraní paraguayen", @"PYG" },
{ @"Turkish lira", @"TRY" },
{ @"Omani rial", @"OMR" },
{ @"Panamanian balboa", @"PAB" },
{ @"Papua New Guinean kina", @"PGK" },
{ @"Paraguayan guaraní", @"PYG" },
{ @"Peruvian sol", @"PEN" },
{ @"Złoty polonais", @"PLN" },
{ @"Riyal qatari", @"QAR" },
{ @"Riyal saudi", @"SAR" },
{ @"Tala", @"WST" },
{ @"Polish złoty", @"PLN" },
{ @"Qatari riyal", @"QAR" },
{ @"Saudi riyal", @"SAR" },
{ @"Samoan tālā", @"WST" },
{ @"São Tomé and Príncipe dobra", @"STN" },
{ @"Leone", @"SLL" },
{ @"Lilangeni", @"SZL" },
{ @"Somoni tadjikistan", @"TJS" },
{ @"Baht thaïlandais", @"THB" },
{ @"Hryvnia ukrainien", @"UAH" },
{ @"Sierra Leonean leone", @"SLL" },
{ @"Swazi lilangeni", @"SZL" },
{ @"Tajikistani somoni", @"TJS" },
{ @"Thai baht", @"THB" },
{ @"Ukrainian hryvnia", @"UAH" },
{ @"Vanuatu vatu", @"VUV" },
{ @"Bolívar vénézuélien", @"VEF" },
{ @"Kwacha de Zambie", @"ZMW" },
{ @"Dirham marocain", @"MAD" },
{ @"Dirham des Émirats arabes unis", @"AED" },
{ @"Manat azerbaïdjanais", @"AZN" },
{ @"Manat turkmène", @"TMT" },
{ @"Shilling somalien", @"SOS" },
{ @"Shilling tanzanien", @"TZS" },
{ @"Shilling ougandais", @"UGX" },
{ @"Leu roumain", @"RON" },
{ @"Leu moldave", @"MDL" },
{ @"Roupie népalaise", @"NPR" },
{ @"Roupie pakistanaise", @"PKR" },
{ @"Roupie indienne", @"INR" },
{ @"Roupie seychelloise", @"SCR" },
{ @"Roupie mauricienne", @"MUR" },
{ @"Rufiyaa maldives", @"MVR" },
{ @"Venezuelan bolívar", @"VEF" },
{ @"Zambian kwacha", @"ZMW" },
{ @"Moroccan dirham", @"MAD" },
{ @"United Arab Emirates dirham", @"AED" },
{ @"Azerbaijani manat", @"AZN" },
{ @"Turkmenistan manat", @"TMT" },
{ @"Somali shilling", @"SOS" },
{ @"Tanzanian shilling", @"TZS" },
{ @"Ugandan shilling", @"UGX" },
{ @"Romanian leu", @"RON" },
{ @"Moldovan leu", @"MDL" },
{ @"Nepalese rupee", @"NPR" },
{ @"Pakistani rupee", @"PKR" },
{ @"Indian rupee", @"INR" },
{ @"Seychellois rupee", @"SCR" },
{ @"Mauritian rupee", @"MUR" },
{ @"Maldivian rufiyaa", @"MVR" },
{ @"Sri Lankan rupee", @"LKR" },
{ @"Rupiah Indonésie", @"IDR" },
{ @"Couronne danoise", @"DKK" },
{ @"Couronne norvégienne", @"NOK" },
{ @"Indonesian rupiah", @"IDR" },
{ @"Danish krone", @"DKK" },
{ @"Norwegian krone", @"NOK" },
{ @"Icelandic króna", @"ISK" },
{ @"Couronne suédoise", @"SEK" },
{ @"Franc CFA de l'Afrique de l'Ouest", @"XOF" },
{ @"Franc CFA d'Afrique centrale", @"XAF" },
{ @"Franc comorien", @"KMF" },
{ @"Franc congolais", @"CDF" },
{ @"Franc burundais", @"BIF" },
{ @"Franc djiboutienne", @"DJF" },
{ @"Franc CFP", @"XPF" },
{ @"Franc guinéen", @"GNF" },
{ @"Franc Suisse", @"CHF" },
{ @"Franc rwandais", @"RWF" },
{ @"Rouble russe", @"RUB" },
{ @"Rouble transnistriens", @"PRB" },
{ @"Nouveau rouble biélorusse", @"BYN" },
{ @"Dinar algérien", @"DZD" },
{ @"Dinar de bahreïn", @"BHD" },
{ @"Dinar iraquien", @"IQD" },
{ @"Dinar jordanien", @"JOD" },
{ @"Dinar koweïtien", @"KWD" },
{ @"Dinar libyen", @"LYD" },
{ @"Dinar serbe", @"RSD" },
{ @"Dinar tunisien", @"TND" },
{ @"Peso argentin", @"ARS" },
{ @"Peso chilien", @"CLP" },
{ @"Peso colombien", @"COP" },
{ @"Peso cubains convertibles", @"CUC" },
{ @"Peso cubains", @"CUP" },
{ @"Peso dominicain", @"DOP" },
{ @"Swedish krona", @"SEK" },
{ @"West African CFA franc", @"XOF" },
{ @"Central African CFA franc", @"XAF" },
{ @"Comorian franc", @"KMF" },
{ @"Congolese franc", @"CDF" },
{ @"Burundian franc", @"BIF" },
{ @"Djiboutian franc", @"DJF" },
{ @"CFP franc", @"XPF" },
{ @"Guinean franc", @"GNF" },
{ @"Swiss franc", @"CHF" },
{ @"Rwandan franc", @"RWF" },
{ @"Russian ruble", @"RUB" },
{ @"Transnistrian ruble", @"PRB" },
{ @"New Belarusian ruble", @"BYN" },
{ @"Algerian dinar", @"DZD" },
{ @"Bahraini dinar", @"BHD" },
{ @"Iraqi dinar", @"IQD" },
{ @"Jordanian dinar", @"JOD" },
{ @"Kuwaiti dinar", @"KWD" },
{ @"Libyan dinar", @"LYD" },
{ @"Serbian dinar", @"RSD" },
{ @"Tunisian dinar", @"TND" },
{ @"Argentine peso", @"ARS" },
{ @"Chilean peso", @"CLP" },
{ @"Colombian peso", @"COP" },
{ @"Cuban convertible peso", @"CUC" },
{ @"Cuban peso", @"CUP" },
{ @"Dominican peso", @"DOP" },
{ @"Mexican peso", @"MXN" },
{ @"Peso uruguayen", @"UYU" },
{ @"Livre britannique", @"GBP" },
{ @"Livre sainte-hélène", @"SHP" },
{ @"Livre égyptienne", @"EGP" },
{ @"Livre des îles falkland", @"FKP" },
{ @"Livre gibraltar", @"GIP" },
{ @"Livre manx", @"IMP" },
{ @"Livre jersey", @"JEP" },
{ @"Livre libanaise", @"LBP" },
{ @"Uruguayan peso", @"UYU" },
{ @"British pound", @"GBP" },
{ @"Saint Helena pound", @"SHP" },
{ @"Egyptian pound", @"EGP" },
{ @"Falkland Islands pound", @"FKP" },
{ @"Gibraltar pound", @"GIP" },
{ @"Manx pound", @"IMP" },
{ @"Jersey pound", @"JEP" },
{ @"Lebanese pound", @"LBP" },
{ @"South Sudanese pound", @"SSP" },
{ @"Livre soudanaise", @"SDG" },
{ @"Livre syrienne", @"SYP" },
{ @"Dollar États-Unis", @"USD" },
{ @"Dollar Australien", @"AUD" },
{ @"Dollar des bahamas", @"BSD" },
{ @"Sudanese pound", @"SDG" },
{ @"Syrian pound", @"SYP" },
{ @"United States dollar", @"USD" },
{ @"Australian dollar", @"AUD" },
{ @"Bahamian dollar", @"BSD" },
{ @"Barbadian dollar", @"BBD" },
{ @"Dollar de belize", @"BZD" },
{ @"Dollar des bermudes", @"BMD" },
{ @"Dollar de brunei", @"BND" },
{ @"Dollar de Singapour", @"SGD" },
{ @"Dollar Canadien", @"CAD" },
{ @"Dollar des îles Caïmans", @"KYD" },
{ @"Belize dollar", @"BZD" },
{ @"Bermudian dollar", @"BMD" },
{ @"Brunei dollar", @"BND" },
{ @"Singapore dollar", @"SGD" },
{ @"Canadian dollar", @"CAD" },
{ @"Cayman Islands dollar", @"KYD" },
{ @"New Zealand dollar", @"NZD" },
{ @"Fijian dollar", @"FJD" },
{ @"Dollar guyanien", @"GYD" },
{ @"Dollar de Hong Kong", @"HKD" },
{ @"Dollar jamaïcain", @"JMD" },
{ @"Dollar libérien", @"LRD" },
{ @"Dollar namibien", @"NAD" },
{ @"Dollar des îles Salomon", @"SBD" },
{ @"Dollar du suriname", @"SRD" },
{ @"Nouveau dollar de Taïwan", @"TWD" },
{ @"Dollar trinidadien", @"TTD" },
{ @"Guyanese dollar", @"GYD" },
{ @"Hong Kong dollar", @"HKD" },
{ @"Jamaican dollar", @"JMD" },
{ @"Liberian dollar", @"LRD" },
{ @"Namibian dollar", @"NAD" },
{ @"Solomon Islands dollar", @"SBD" },
{ @"Surinamese dollar", @"SRD" },
{ @"New Taiwan dollar", @"TWD" },
{ @"Trinidad and Tobago dollar", @"TTD" },
{ @"Tuvaluan dollar", @"TVD" },
{ @"Yuan Chinois", @"CNY" },
{ @"Chinese yuan", @"CNY" },
{ @"Rial", @"__RI" },
{ @"Shiling", @"__S" },
{ @"Som", @"__SO" },
Expand Down
2 changes: 1 addition & 1 deletion .NET/Microsoft.Recognizers.Definitions.Common/Hindi/NumbersWithUnitDefinitions.cs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -401,7 +401,7 @@ public static class NumbersWithUnitDefinitions
{ @"Rwandan franc", @"RWF" },
{ @"Russian ruble", @"RUB" },
{ @"Transnistrian ruble", @"PRB" },
{ @"Belarusian ruble", @"BYN" },
{ @"New Belarusian ruble", @"BYN" },
{ @"Algerian dinar", @"DZD" },
{ @"Bahraini dinar", @"BHD" },
{ @"Iraqi dinar", @"IQD" },
Expand Down
4 changes: 2 additions & 2 deletions .NET/Microsoft.Recognizers.Definitions.Common/Turkish/NumbersWithUnitDefinitions.cs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -70,7 +70,7 @@ public static class NumbersWithUnitDefinitions
};
public static readonly Dictionary<string, string> CurrencySuffixList = new Dictionary<string, string>
{
{ @"Afganistan afganisi", @"afganistan afganisi|afgani|؋|afn|af|afs|afganistan afganisine|afganistan afganisinde|afganistan afganisinden|afganistan afganisini|afganistan afganisinin|afganistan afganisidir|afganistan afganisiydi|afganistan afganisiymiş" },
{ @"Afghan afghani", @"afganistan afganisi|afgani|؋|afn|af|afs|afganistan afganisine|afganistan afganisinde|afganistan afganisinden|afganistan afganisini|afganistan afganisinin|afganistan afganisidir|afganistan afganisiydi|afganistan afganisiymiş" },
{ @"Pul", @"pul|pula|pulda|puldan|pulu|pulun|puldur|puldu|pulmuş|pulluk" },
{ @"Euro", @"euro|€|eur|avro|avroya|avroda|avrodan|avroyu|avronun|avrodur|avroydu|avroymuş|avroluk" },
{ @"Cent", @"sent|s|sentte|sentten|senti|sentin|senttir|sentti|sentmiş|sentlik" },
Expand Down Expand Up @@ -417,7 +417,7 @@ public static class NumbersWithUnitDefinitions
{ @"Rwandan franc", @"RWF" },
{ @"Russian ruble", @"RUB" },
{ @"Transnistrian ruble", @"PRB" },
{ @"Belarusian ruble", @"BYN" },
{ @"New Belarusian ruble", @"BYN" },
{ @"Algerian dinar", @"DZD" },
{ @"Bahraini dinar", @"BHD" },
{ @"Iraqi dinar", @"IQD" },
Expand Down
24 changes: 22 additions & 2 deletions ...rosoft.Recognizers.Text.NumberWithUnit/Dutch/Extractors/CurrencyExtractorConfiguration.cs
View file
Original file line number Diff line number Diff line change
@@ -1,8 +1,11 @@
// Copyright (c) Microsoft Corporation. All rights reserved.
// Licensed under the MIT License.

using System;
using System.Collections.Generic;
using System.Collections.Immutable;
using System.Globalization;
using System.Linq;

using Microsoft.Recognizers.Definitions.Dutch;

Expand All @@ -13,8 +16,25 @@ public class CurrencyExtractorConfiguration : DutchNumberWithUnitExtractorConfig
public static readonly ImmutableDictionary<string, string> CurrencySuffixList =
NumbersWithUnitDefinitions.CurrencySuffixList.ToImmutableDictionary();

public static readonly ImmutableDictionary<string, string> CurrencyPrefixList =
NumbersWithUnitDefinitions.CurrencyPrefixList.ToImmutableDictionary();
// CurrencyNameToIsoCodeMap dictionary (excluding fake and unofficial Iso codes starting with underscore)
public static readonly Dictionary<string, string> IsoCodeDict =
NumbersWithUnitDefinitions.CurrencyNameToIsoCodeMap.Where(x => !x.Value.StartsWith("_", StringComparison.Ordinal))
.ToDictionary(x => x.Key, x => x.Value.ToLower(CultureInfo.InvariantCulture));

// CurrencyNameToIsoCodeMap followed by '$' symbol (e.g. 'AUD$')
public static readonly Dictionary<string, string> IsoCodeWithSymbolDict =
NumbersWithUnitDefinitions.CurrencyNameToIsoCodeMap.Where(x => !x.Value.StartsWith("_", StringComparison.Ordinal))
.ToDictionary(x => x.Key, x => x.Value.ToLower(CultureInfo.InvariantCulture) + "$");

// Merge IsoCodeDict and IsoCodeWithSymbolDict
public static readonly Dictionary<string, string> IsoCodeCombinedDict = IsoCodeDict.Concat(IsoCodeWithSymbolDict)
.GroupBy(x => x.Key).ToDictionary(x => x.Key, y => y.Count() > 1 ? string.Join("|", new string[] { y.First().Value, y.Last().Value }) : y.First().Value);

// Merge IsoCodeCombinedDict with CurrencyPrefixList (excluding fake and unofficial Iso codes starting with underscore)
public static readonly Dictionary<string, string> CurrencyPrefixDict = NumbersWithUnitDefinitions.CurrencyPrefixList.Concat(IsoCodeCombinedDict)
.GroupBy(x => x.Key).ToDictionary(x => x.Key, y => y.Count() > 1 ? string.Join("|", new string[] { y.First().Value, y.Last().Value }) : y.First().Value);

public static readonly ImmutableDictionary<string, string> CurrencyPrefixList = CurrencyPrefixDict.ToImmutableDictionary();

public static readonly ImmutableDictionary<string, string> FractionalUnitNameToCodeMap =
NumbersWithUnitDefinitions.FractionalUnitNameToCodeMap.ToImmutableDictionary();
Expand Down
Loading

0 comments on commit 4db5674

Please sign in to comment.