Merge pull request #71 from microsoft/master

Fetch Upstream fork
microsoft · Apr 19, 2023 · 17de6bb · 17de6bb
2 parents b03280c + 47ace20
commit 17de6bb
Show file tree

Hide file tree

Showing 513 changed files with 6,306 additions and 6,637 deletions.
diff --git a/.NET/Build.CI.cmd b/.NET/Build.CI.cmd
@@ -53,7 +53,7 @@ IF %ERRORLEVEL% NEQ 0 (
 
 ECHO.
 ECHO # Restoring NuGet dependencies
-CALL "buildtools\nuget" restore
+CALL !MSBuild! Microsoft.Recognizers.Text.sln /t:Restore
 
 set configuration=Release
 ECHO.

diff --git a/.NET/Microsoft.Recognizers.Definitions.Common/Chinese/NumbersDefinitions.cs b/.NET/Microsoft.Recognizers.Definitions.Common/Chinese/NumbersDefinitions.cs
@@ -256,7 +256,7 @@ public static class NumbersDefinitions
       public static readonly Dictionary<string, string> AmbiguityFiltersDict = new Dictionary<string, string>
         {
             { @"十", @"十足" },
-            { @"伍", @"队伍" },
+            { @"伍", @"队伍|入伍|退伍|伍仁" },
             { @"肆", @"放肆|肆意|肆无忌惮" },
             { @"陆", @"大陆|陆地|登陆|海陆" },
             { @"拾", @"拾取|拾起|收拾|拾到|朝花夕拾" }

diff --git a/.NET/Microsoft.Recognizers.Definitions.Common/English/DateTimeDefinitions.cs b/.NET/Microsoft.Recognizers.Definitions.Common/English/DateTimeDefinitions.cs
@@ -160,7 +160,7 @@ public static class DateTimeDefinitions
       public const string LunchRegex = @"\blunchtime\b";
       public const string NightRegex = @"\b(mid)?night\b";
       public const string CommonDatePrefixRegex = @"^[\.]";
-      public static readonly string LessThanOneHour = $@"(?<lth>(a\s+)?quarter|three quarter(s)?|half( an hour)?|{BaseDateTime.DeltaMinuteRegex}(\s+(minutes?|mins?))|{DeltaMinuteNumRegex}(\s+(minutes?|mins?)))";
+      public static readonly string LessThanOneHour = $@"(?<lth>(a\s+)?quarter|three quarter(s)?|half( an hour)?|{BaseDateTime.DeltaMinuteRegex}(\s+(minutes?|mins?)|(?=\s+past))|{DeltaMinuteNumRegex}(\s+(minutes?|mins?)|(?=\s+past)))";
       public static readonly string WrittenTimeRegex = $@"(?<writtentime>{HourNumRegex}\s+{MinuteNumRegex}(\s+(minutes?|mins?))?)";
       public static readonly string TimePrefix = $@"(?<prefix>{LessThanOneHour}\s+(past|to))";
       public static readonly string TimeSuffix = $@"(?<suffix>{AmRegex}|{PmRegex}|{OclockRegex})";

diff --git a/.NET/Microsoft.Recognizers.Definitions.Common/Italian/DateTimeDefinitions.cs b/.NET/Microsoft.Recognizers.Definitions.Common/Italian/DateTimeDefinitions.cs
@@ -139,9 +139,9 @@ public static class DateTimeDefinitions
       public const string AmRegex = @"\b(?<am>((((la|alla|verso|per|della|di|in)\s+(la\s+)?)?(mattinata|mattina))|(((il|al|verso|per|del|di)\s+(il\s+)?)?(mattino))))\b";
       public const string LunchRegex = @"\b(ora di pranzo)\b";
       public const string NightRegex = @"\b(mezzanotte|notte)\b";
-      public static readonly string LessThanOneHour = $@"(?<lth>(un\s+quarto|tre quarti?|mezz[oa]|mezz'ora|{BaseDateTime.DeltaMinuteRegex}(\s+(minut[oi]|min))?|{DeltaMinuteNumRegex}(\s+(minut[oi]|min))|(?<=(e|meno)\s+){DeltaMinuteNumRegex}))";
+      public static readonly string LessThanOneHour = $@"(?<lth>(un\s+quarto|tre quarti?|mezz[oa]|mezz'ora|{BaseDateTime.DeltaMinuteRegex}|{DeltaMinuteNumRegex}))";
       public static readonly string EngTimeRegex = $@"(?<engtime>{HourNumRegex}\s+e\s+({MinuteNumRegex}|(?<tens>venti?|trenta?|quaranta?|cinquanta?){MinuteNumRegex}))";
-      public static readonly string TimePrefix = $@"(?<prefix>(e\s+{LessThanOneHour}|{LessThanOneHour}\s+(minut[oi]|min)\s+all[e']|meno {LessThanOneHour}))";
+      public static readonly string TimePrefix = $@"(?<prefix>((e|meno)\s+{LessThanOneHour}(\s+(minut[oi]|min))?|{LessThanOneHour}\s+(minut[oi]|min)\s+all[e']))";
       public static readonly string TimeSuffix = $@"(?<suffix>{AmRegex}|{PmRegex}|{OclockRegex})";
       public static readonly string BasicTime = $@"\b(?<basictime>{EngTimeRegex}|{HourNumRegex}|{BaseDateTime.HourRegex}:{BaseDateTime.MinuteRegex}(:{BaseDateTime.SecondRegex})?|{BaseDateTime.HourRegex})";
       public const string MidnightRegex = @"(?<midnight>mezzanotte|mezza notte)";

diff --git a/.NET/Microsoft.Recognizers.Definitions.Common/Spanish/DateTimeDefinitions.cs b/.NET/Microsoft.Recognizers.Definitions.Common/Spanish/DateTimeDefinitions.cs
@@ -124,7 +124,7 @@ public static class DateTimeDefinitions
       public static readonly string DateExtractor6 = $@"(?<=\b(en|el)\s+){MonthNumRegex}[\-\.]{DayRegex}{BaseDateTime.CheckDecimalRegex}\b(?!\s*[/\\\.]\s*\d+)";
       public static readonly string DateExtractor7 = $@"\b(?<!\d[.,]){MonthNumRegex}\s*/\s*{DayRegex}((\s+|\s*,\s*|\s+d[eo]\s+){DateYearRegex})?\b{BaseDateTime.CheckDecimalRegex}(?!\s*[/\\\.]\s*\d+)";
       public static readonly string DateExtractor8 = $@"(?<=\b(en|el)\s+){DayRegex}[\\\-]{MonthNumRegex}{BaseDateTime.CheckDecimalRegex}\b(?!\s*[/\\\.]\s*\d+)";
-      public static readonly string DateExtractor9 = $@"\b({WeekDayRegex}\s+)?(?<!\d[.,]){DayRegex}\s*/\s*{MonthNumRegex}((\s+|\s*,\s*|\s+d[eo]\s+){DateYearRegex})?\b{BaseDateTime.CheckDecimalRegex}(?!\s*[/\\\.]\s*\d+)";
+      public static readonly string DateExtractor9 = $@"\b({WeekDayRegex}\s+)?(?<!\d[.,]){DayRegex}\s*(/|\bdel\b)\s*{MonthNumRegex}((\s+|\s*,\s*|\s+d[eo]\s+){DateYearRegex})?\b{BaseDateTime.CheckDecimalRegex}(?!\s*[/\\\.]\s*\d+)";
       public static readonly string DateExtractor10 = $@"\b(?<!\d[.,])(({YearRegex}\s*[/\\\-\.]\s*({MonthNumRegex}|{MonthRegex})\s*[/\\\-\.]\s*{DayRegex}(?!\s*[/\\\-\.]\s*\d+))|({MonthRegex}\s*[/\\\-\.]\s*{BaseDateTime.FourDigitYearRegex}\s*[/\\\-\.]\s*{DayRegex})|({DayRegex}\s*[/\\\-\.]\s*{BaseDateTime.FourDigitYearRegex}\s*[/\\\-\.]\s*{MonthRegex}))";
       public const string HourRegex = @"\b(?<!\d[,.])(?<hour>2[0-4]|[0-1]?\d)";
       public const string HourNumRegex = @"\b(?<hournum>cero|una|dos|tres|cuatro|cinco|seis|siete|ocho|nueve|diez|once|doce)\b";

diff --git a/.NET/Microsoft.Recognizers.Definitions/Microsoft.Recognizers.Definitions.csproj b/.NET/Microsoft.Recognizers.Definitions/Microsoft.Recognizers.Definitions.csproj
@@ -1,7 +1,7 @@
 <Project Sdk="Microsoft.NET.Sdk">
 
   <PropertyGroup>
-    <TargetFrameworks>netstandard2.1;netstandard2.0;net462</TargetFrameworks>
+    <TargetFrameworks>netstandard2.1;netstandard2.0;net462;net6.0</TargetFrameworks>
     <LangVersion>9</LangVersion>
     <!-- Disable GenerateAssemblyInfo to use the existing AssemblyInfo.cs -->
     <GenerateAssemblyInfo>false</GenerateAssemblyInfo>

diff --git a/...icrosoft.Recognizers.Text.Choice/Arabic/Extractors/ArabicBooleanExtractorConfiguration.cs b/...icrosoft.Recognizers.Text.Choice/Arabic/Extractors/ArabicBooleanExtractorConfiguration.cs
@@ -1,47 +1,24 @@
 // Copyright (c) Microsoft Corporation. All rights reserved.
 // Licensed under the MIT License.
 
-using System.Collections.Generic;
 using System.Text.RegularExpressions;
-
 using Microsoft.Recognizers.Definitions.Arabic;
+using Microsoft.Recognizers.Text.Choice.Extractors;
 
 namespace Microsoft.Recognizers.Text.Choice.Arabic
 {
-    public class ArabicBooleanExtractorConfiguration : IBooleanExtractorConfiguration
+    public class ArabicBooleanExtractorConfiguration : BaseBooleanExtractorConfiguration
     {
-        public static readonly Regex TrueRegex =
-            new Regex(ChoiceDefinitions.TrueRegex, RegexOptions.Singleline | RegexOptions.RightToLeft);
-
-        public static readonly Regex FalseRegex =
-            new Regex(ChoiceDefinitions.FalseRegex, RegexOptions.Singleline | RegexOptions.RightToLeft);
-
-        public static readonly Regex TokenRegex =
-            new Regex(ChoiceDefinitions.TokenizerRegex, RegexOptions.Singleline | RegexOptions.RightToLeft);
-
-        public static readonly IDictionary<Regex, string> MapRegexes = new Dictionary<Regex, string>()
-        {
-            { TrueRegex, Constants.SYS_BOOLEAN_TRUE },
-            { FalseRegex, Constants.SYS_BOOLEAN_FALSE },
-        };
-
         public ArabicBooleanExtractorConfiguration(bool onlyTopMatch = true)
+            : base(
+                  trueRegex: ChoiceDefinitions.TrueRegex,
+                  falseRegex: ChoiceDefinitions.FalseRegex,
+                  tokenRegex: ChoiceDefinitions.TokenizerRegex,
+                  options: RegexOptions.Singleline | RegexOptions.RightToLeft,
+                  allowPartialMatch: false,
+                  maxDistance: 2,
+                  onlyTopMatch)
         {
-            this.OnlyTopMatch = onlyTopMatch;
         }
-
-        Regex IBooleanExtractorConfiguration.TrueRegex => TrueRegex;
-
-        Regex IBooleanExtractorConfiguration.FalseRegex => FalseRegex;
-
-        IDictionary<Regex, string> IChoiceExtractorConfiguration.MapRegexes => MapRegexes;
-
-        Regex IChoiceExtractorConfiguration.TokenRegex => TokenRegex;
-
-        public bool AllowPartialMatch => false;
-
-        public int MaxDistance => 2;
-
-        public bool OnlyTopMatch { get; }
     }
 }
diff --git a/...ft.Recognizers.Text.Choice/Bulgarian/Extractors/BulgarianBooleanExtractorConfiguration.cs b/...ft.Recognizers.Text.Choice/Bulgarian/Extractors/BulgarianBooleanExtractorConfiguration.cs
@@ -1,47 +1,24 @@
 // Copyright (c) Microsoft Corporation. All rights reserved.
 // Licensed under the MIT License.
 
-using System.Collections.Generic;
 using System.Text.RegularExpressions;
-
 using Microsoft.Recognizers.Definitions.Bulgarian;
+using Microsoft.Recognizers.Text.Choice.Extractors;
 
 namespace Microsoft.Recognizers.Text.Choice.Bulgarian
 {
-    public class BulgarianBooleanExtractorConfiguration : IBooleanExtractorConfiguration
+    public class BulgarianBooleanExtractorConfiguration : BaseBooleanExtractorConfiguration
     {
-        public static readonly Regex TrueRegex =
-            new Regex(ChoiceDefinitions.TrueRegex, RegexOptions.Singleline);
-
-        public static readonly Regex FalseRegex =
-            new Regex(ChoiceDefinitions.FalseRegex, RegexOptions.Singleline);
-
-        public static readonly Regex TokenRegex =
-            new Regex(ChoiceDefinitions.TokenizerRegex, RegexOptions.Singleline);
-
-        public static readonly IDictionary<Regex, string> MapRegexes = new Dictionary<Regex, string>()
-        {
-            { TrueRegex, Constants.SYS_BOOLEAN_TRUE },
-            { FalseRegex, Constants.SYS_BOOLEAN_FALSE },
-        };
-
         public BulgarianBooleanExtractorConfiguration(bool onlyTopMatch = true)
+            : base(
+                  trueRegex: ChoiceDefinitions.TrueRegex,
+                  falseRegex: ChoiceDefinitions.FalseRegex,
+                  tokenRegex: ChoiceDefinitions.TokenizerRegex,
+                  options: RegexOptions.Singleline,
+                  allowPartialMatch: false,
+                  maxDistance: 2,
+                  onlyTopMatch)
         {
-            this.OnlyTopMatch = onlyTopMatch;
         }
-
-        Regex IBooleanExtractorConfiguration.TrueRegex => TrueRegex;
-
-        Regex IBooleanExtractorConfiguration.FalseRegex => FalseRegex;
-
-        IDictionary<Regex, string> IChoiceExtractorConfiguration.MapRegexes => MapRegexes;
-
-        Regex IChoiceExtractorConfiguration.TokenRegex => TokenRegex;
-
-        public bool AllowPartialMatch => false;
-
-        public int MaxDistance => 2;
-
-        public bool OnlyTopMatch { get; }
     }
 }
diff --git a/...rosoft.Recognizers.Text.Choice/Chinese/Extractors/ChineseBooleanExtractorConfiguration.cs b/...rosoft.Recognizers.Text.Choice/Chinese/Extractors/ChineseBooleanExtractorConfiguration.cs
@@ -1,47 +1,24 @@
 // Copyright (c) Microsoft Corporation. All rights reserved.
 // Licensed under the MIT License.
 
-using System.Collections.Generic;
 using System.Text.RegularExpressions;
-
 using Microsoft.Recognizers.Definitions.Chinese;
+using Microsoft.Recognizers.Text.Choice.Extractors;
 
 namespace Microsoft.Recognizers.Text.Choice.Chinese
 {
-    public class ChineseBooleanExtractorConfiguration : IBooleanExtractorConfiguration
+    public class ChineseBooleanExtractorConfiguration : BaseBooleanExtractorConfiguration
     {
-        public static readonly Regex TrueRegex =
-            new Regex(ChoiceDefinitions.TrueRegex, RegexOptions.Singleline);
-
-        public static readonly Regex FalseRegex =
-            new Regex(ChoiceDefinitions.FalseRegex, RegexOptions.Singleline);
-
-        public static readonly Regex TokenRegex =
-            new Regex(ChoiceDefinitions.TokenizerRegex, RegexOptions.Singleline);
-
-        public static readonly IDictionary<Regex, string> MapRegexes = new Dictionary<Regex, string>()
-        {
-            { TrueRegex, Constants.SYS_BOOLEAN_TRUE },
-            { FalseRegex, Constants.SYS_BOOLEAN_FALSE },
-        };
-
         public ChineseBooleanExtractorConfiguration(bool onlyTopMatch = true)
+            : base(
+                  trueRegex: ChoiceDefinitions.TrueRegex,
+                  falseRegex: ChoiceDefinitions.FalseRegex,
+                  tokenRegex: ChoiceDefinitions.TokenizerRegex,
+                  options: RegexOptions.Singleline,
+                  allowPartialMatch: false,
+                  maxDistance: 2,
+                  onlyTopMatch)
         {
-            this.OnlyTopMatch = onlyTopMatch;
         }
-
-        Regex IBooleanExtractorConfiguration.TrueRegex => TrueRegex;
-
-        Regex IBooleanExtractorConfiguration.FalseRegex => FalseRegex;
-
-        IDictionary<Regex, string> IChoiceExtractorConfiguration.MapRegexes => MapRegexes;
-
-        Regex IChoiceExtractorConfiguration.TokenRegex => TokenRegex;
-
-        public bool AllowPartialMatch => false;
-
-        public int MaxDistance => 2;
-
-        public bool OnlyTopMatch { get; }
     }
 }
diff --git a/.NET/Microsoft.Recognizers.Text.Choice/ChoiceRecognizer.cs b/.NET/Microsoft.Recognizers.Text.Choice/ChoiceRecognizer.cs
@@ -1,12 +1,14 @@
 // Copyright (c) Microsoft Corporation. All rights reserved.
 // Licensed under the MIT License.
 
+using System;
 using System.Collections.Generic;
 using Microsoft.Recognizers.Text.Choice.Arabic;
 using Microsoft.Recognizers.Text.Choice.Bulgarian;
 using Microsoft.Recognizers.Text.Choice.Chinese;
 using Microsoft.Recognizers.Text.Choice.Dutch;
 using Microsoft.Recognizers.Text.Choice.English;
+using Microsoft.Recognizers.Text.Choice.Extractors;
 using Microsoft.Recognizers.Text.Choice.French;
 using Microsoft.Recognizers.Text.Choice.German;
 using Microsoft.Recognizers.Text.Choice.Hindi;
@@ -21,23 +23,23 @@ namespace Microsoft.Recognizers.Text.Choice
 {
     public class ChoiceRecognizer : Recognizer<ChoiceOptions>
     {
-        public ChoiceRecognizer(string targetCulture, ChoiceOptions options = ChoiceOptions.None, bool lazyInitialization = false)
-            : base(targetCulture, options, lazyInitialization)
+        public ChoiceRecognizer(string targetCulture, ChoiceOptions options = ChoiceOptions.None, bool lazyInitialization = false, int timeoutInSeconds = 0)
+            : base(targetCulture, options, lazyInitialization, timeoutInSeconds)
         {
         }
 
-        public ChoiceRecognizer(string targetCulture, int options, bool lazyInitialization = false)
-            : this(targetCulture, GetOptions(options), lazyInitialization)
+        public ChoiceRecognizer(string targetCulture, int options, bool lazyInitialization = false, int timeoutInSeconds = 0)
+            : this(targetCulture, GetOptions(options), lazyInitialization, timeoutInSeconds)
         {
         }
 
-        public ChoiceRecognizer(ChoiceOptions options = ChoiceOptions.None, bool lazyInitialization = true)
-            : base(null, options, lazyInitialization)
+        public ChoiceRecognizer(ChoiceOptions options = ChoiceOptions.None, bool lazyInitialization = true, int timeoutInSeconds = 0)
+            : base(null, options, lazyInitialization, timeoutInSeconds)
         {
         }
 
-        public ChoiceRecognizer(int options, bool lazyInitialization = true)
-            : this(null, GetOptions(options), lazyInitialization)
+        public ChoiceRecognizer(int options, bool lazyInitialization = true, int timeoutInSeconds = 0)
+            : this(null, GetOptions(options), lazyInitialization, timeoutInSeconds)
         {
         }
 
@@ -53,6 +55,14 @@ public IModel GetBooleanModel(string culture = null, bool fallbackToDefaultCultu
             return GetModel<BooleanModel>(culture, fallbackToDefaultCulture);
         }
 
+        protected override List<Type> GetRelatedTypes()
+        {
+            return new List<Type>()
+            {
+                typeof(BaseBooleanExtractorConfiguration),
+            };
+        }
+
         protected override void InitializeConfiguration()
         {
             RegisterModel<BooleanModel>(

diff --git a/.../Microsoft.Recognizers.Text.Choice/Dutch/Extractors/DutchBooleanExtractorConfiguration.cs b/.../Microsoft.Recognizers.Text.Choice/Dutch/Extractors/DutchBooleanExtractorConfiguration.cs
@@ -1,47 +1,24 @@
 // Copyright (c) Microsoft Corporation. All rights reserved.
 // Licensed under the MIT License.
 
-using System.Collections.Generic;
 using System.Text.RegularExpressions;
-
 using Microsoft.Recognizers.Definitions.Dutch;
+using Microsoft.Recognizers.Text.Choice.Extractors;
 
 namespace Microsoft.Recognizers.Text.Choice.Dutch
 {
-    public class DutchBooleanExtractorConfiguration : IBooleanExtractorConfiguration
+    public class DutchBooleanExtractorConfiguration : BaseBooleanExtractorConfiguration
     {
-        public static readonly Regex TrueRegex =
-            new Regex(ChoiceDefinitions.TrueRegex, RegexOptions.Singleline);
-
-        public static readonly Regex FalseRegex =
-            new Regex(ChoiceDefinitions.FalseRegex, RegexOptions.Singleline);
-
-        public static readonly Regex TokenRegex =
-            new Regex(ChoiceDefinitions.TokenizerRegex, RegexOptions.Singleline);
-
-        public static readonly IDictionary<Regex, string> MapRegexes = new Dictionary<Regex, string>()
-        {
-            { TrueRegex, Constants.SYS_BOOLEAN_TRUE },
-            { FalseRegex, Constants.SYS_BOOLEAN_FALSE },
-        };
-
         public DutchBooleanExtractorConfiguration(bool onlyTopMatch = true)
+            : base(
+                  trueRegex: ChoiceDefinitions.TrueRegex,
+                  falseRegex: ChoiceDefinitions.FalseRegex,
+                  tokenRegex: ChoiceDefinitions.TokenizerRegex,
+                  options: RegexOptions.Singleline,
+                  allowPartialMatch: false,
+                  maxDistance: 2,
+                  onlyTopMatch)
         {
-            this.OnlyTopMatch = onlyTopMatch;
         }
-
-        Regex IBooleanExtractorConfiguration.TrueRegex => TrueRegex;
-
-        Regex IBooleanExtractorConfiguration.FalseRegex => FalseRegex;
-
-        IDictionary<Regex, string> IChoiceExtractorConfiguration.MapRegexes => MapRegexes;
-
-        Regex IChoiceExtractorConfiguration.TokenRegex => TokenRegex;
-
-        public bool AllowPartialMatch => false;
-
-        public int MaxDistance => 2;
-
-        public bool OnlyTopMatch { get; }
     }
 }