[PhoneNumber] Add filter for "bank account" prefix (#1906)

.NET/Microsoft.Recognizers.Definitions.Common/English/PhoneNumbersDefinitions.cs

@@ -0,0 +1,27 @@
+//------------------------------------------------------------------------------
+// <auto-generated>
+//     This code was generated by a tool.
+//     Changes to this file may cause incorrect behavior and will be lost if
+//     the code is regenerated.
+//
+//     Generation parameters:
+//     - DataFilename: Patterns\English\English-PhoneNumbers.yaml
+//     - Language: English
+//     - ClassName: PhoneNumbersDefinitions
+// </auto-generated>
+//
+// Copyright (c) Microsoft Corporation. All rights reserved.
+// Licensed under the MIT License.
+// ------------------------------------------------------------------------------
+
+namespace Microsoft.Recognizers.Definitions.English
+{
+    using System;
+    using System.Collections.Generic;
+
+    public static class PhoneNumbersDefinitions
+    {
+      public const string NumberReplaceToken = @"@builtin.phonenumber";
+      public const string FalsePositivePrefixRegex = @"(account|card)(\s+(#|number))?(\s+is)?:?\s*$";
+    }
+}

.NET/Microsoft.Recognizers.Definitions.Common/English/PhoneNumbersDefinitions.tt

@@ -0,0 +1,7 @@
+<#@ template debug="true" hostspecific="true" language="C#" #>
+<#
+	this.DataFilename = @"Patterns\English\English-PhoneNumbers.yaml";
+	this.Language = "English";
+	this.ClassName = "PhoneNumbersDefinitions";
+#>
+<#@ include file="..\CommonDefinitions.ttinclude"#>

.NET/Microsoft.Recognizers.Definitions.Common/Microsoft.Recognizers.Definitions.Common.csproj

@@ -172,6 +172,11 @@
       <DesignTime>True</DesignTime>
       <AutoGen>True</AutoGen>
     </Compile>
+    <Compile Update="English\PhoneNumbersDefinitions.cs">
+      <DesignTime>True</DesignTime>
+      <AutoGen>True</AutoGen>
+      <DependentUpon>PhoneNumbersDefinitions.tt</DependentUpon>
+    </Compile>
     <Compile Update="English\TimeZoneDefinitions.cs">
       <DesignTime>True</DesignTime>
       <AutoGen>True</AutoGen>
@@ -297,6 +302,11 @@
       <DesignTime>True</DesignTime>
       <AutoGen>True</AutoGen>
     </Compile>
+    <Compile Update="Portuguese\PhoneNumbersDefinitions.cs">
+      <DesignTime>True</DesignTime>
+      <AutoGen>True</AutoGen>
+      <DependentUpon>PhoneNumbersDefinitions.tt</DependentUpon>
+    </Compile>
     <Compile Update="Spanish\ChoiceDefinitions.cs">
       <DesignTime>True</DesignTime>
       <AutoGen>True</AutoGen>
@@ -408,6 +418,8 @@
     </None>
     <None Update="English\NumbersWithUnitDefinitions.tt">
     </None>
+    <None Update="English\PhoneNumbersDefinitions.tt">
+    </None>
     <None Update="English\TimeZoneDefinitions.tt">
     </None>
     <None Update="French\ChoiceDefinitions.tt">
@@ -458,6 +470,8 @@
     </None>
     <None Update="Portuguese\NumbersWithUnitDefinitions.tt">
     </None>
+    <None Update="Portuguese\PhoneNumbersDefinitions.tt">
+    </None>
     <None Update="Spanish\ChoiceDefinitions.tt">
     </None>
     <None Update="Spanish\DateTimeDefinitions.tt">

.NET/Microsoft.Recognizers.Definitions.Common/Portuguese/PhoneNumbersDefinitions.cs

@@ -0,0 +1,27 @@
+//------------------------------------------------------------------------------
+// <auto-generated>
+//     This code was generated by a tool.
+//     Changes to this file may cause incorrect behavior and will be lost if
+//     the code is regenerated.
+//
+//     Generation parameters:
+//     - DataFilename: Patterns\Portuguese\Portuguese-PhoneNumbers.yaml
+//     - Language: Portuguese
+//     - ClassName: PhoneNumbersDefinitions
+// </auto-generated>
+//
+// Copyright (c) Microsoft Corporation. All rights reserved.
+// Licensed under the MIT License.
+// ------------------------------------------------------------------------------
+
+namespace Microsoft.Recognizers.Definitions.Portuguese
+{
+    using System;
+    using System.Collections.Generic;
+
+    public static class PhoneNumbersDefinitions
+    {
+      public const string NumberReplaceToken = @"@builtin.phonenumber";
+      public const string FalsePositivePrefixRegex = @"conta(\s+(número|bancária(\s+número)?))?(\s+é)?:?\s*$";
+    }
+}

.NET/Microsoft.Recognizers.Definitions.Common/Portuguese/PhoneNumbersDefinitions.tt

@@ -0,0 +1,7 @@
+<#@ template debug="true" hostspecific="true" language="C#" #>
+<#
+	this.DataFilename = @"Patterns\Portuguese\Portuguese-PhoneNumbers.yaml";
+	this.Language = "Portuguese";
+	this.ClassName = "PhoneNumbersDefinitions";
+#>
+<#@ include file="..\CommonDefinitions.ttinclude"#>

.NET/Microsoft.Recognizers.Definitions/Microsoft.Recognizers.Definitions.csproj

@@ -42,6 +42,7 @@
     <Compile Include="..\Microsoft.Recognizers.Definitions.Common\BasePhoneNumbers.cs" Link="BasePhoneNumbers.cs" />
     <Compile Include="..\Microsoft.Recognizers.Definitions.Common\BaseUnits.cs" Link="BaseUnits.cs" />
     <Compile Include="..\Microsoft.Recognizers.Definitions.Common\BaseURL.cs" Link="BaseURL.cs" />
+    <Compile Include="..\Microsoft.Recognizers.Definitions.Common\English\PhoneNumbersDefinitions.cs" Link="English\PhoneNumbersDefinitions.cs" />
     <Compile Include="..\Microsoft.Recognizers.Definitions.Common\Chinese\ChoiceDefinitions.cs" Link="Chinese\ChoiceDefinitions.cs" />
     <Compile Include="..\Microsoft.Recognizers.Definitions.Common\Chinese\DateTimeDefinitions.cs" Link="Chinese\DateTimeDefinitions.cs" />
     <Compile Include="..\Microsoft.Recognizers.Definitions.Common\Chinese\NumbersDefinitions.cs" Link="Chinese\NumbersDefinitions.cs" />
@@ -83,6 +84,7 @@
     <Compile Include="..\Microsoft.Recognizers.Definitions.Common\Portuguese\DateTimeDefinitions.cs" Link="Portuguese\DateTimeDefinitions.cs" />
     <Compile Include="..\Microsoft.Recognizers.Definitions.Common\Portuguese\NumbersDefinitions.cs" Link="Portuguese\NumbersDefinitions.cs" />
     <Compile Include="..\Microsoft.Recognizers.Definitions.Common\Portuguese\NumbersWithUnitDefinitions.cs" Link="Portuguese\NumbersWithUnitDefinitions.cs" />
+    <Compile Include="..\Microsoft.Recognizers.Definitions.Common\Portuguese\PhoneNumbersDefinitions.cs" Link="Portuguese\PhoneNumbersDefinitions.cs" />
     <Compile Include="..\Microsoft.Recognizers.Definitions.Common\Spanish\ChoiceDefinitions.cs" Link="Spanish\ChoiceDefinitions.cs" />
     <Compile Include="..\Microsoft.Recognizers.Definitions.Common\Spanish\DateTimeDefinitions.cs" Link="Spanish\DateTimeDefinitions.cs" />
     <Compile Include="..\Microsoft.Recognizers.Definitions.Common\Spanish\NumbersDefinitions.cs" Link="Spanish\NumbersDefinitions.cs" />

.NET/Microsoft.Recognizers.Text.DataDrivenTests/Sequence/TestSequence_Dutch.cs

@@ -0,0 +1,16 @@
+using Microsoft.Recognizers.Text.DataDrivenTests;
+using Microsoft.VisualStudio.TestTools.UnitTesting;
+
+namespace Microsoft.Recognizers.Text.Sequence.Tests
+{
+    [TestClass]
+    public class TestSequence_Dutch : TestBase
+    {
+        [NetCoreTestDataSource]
+        [TestMethod]
+        public void PhoneNumberModel(TestModel testSpec)
+        {
+            TestPhoneNumber(testSpec);
+        }
+    }
+}

.NET/Microsoft.Recognizers.Text.DataDrivenTests/Sequence/TestSequence_French.cs

@@ -0,0 +1,16 @@
+using Microsoft.Recognizers.Text.DataDrivenTests;
+using Microsoft.VisualStudio.TestTools.UnitTesting;
+
+namespace Microsoft.Recognizers.Text.Sequence.Tests
+{
+    [TestClass]
+    public class TestSequence_French : TestBase
+    {
+        [NetCoreTestDataSource]
+        [TestMethod]
+        public void PhoneNumberModel(TestModel testSpec)
+        {
+            TestPhoneNumber(testSpec);
+        }
+    }
+}

.NET/Microsoft.Recognizers.Text.DataDrivenTests/Sequence/TestSequence_German.cs

@@ -0,0 +1,16 @@
+using Microsoft.Recognizers.Text.DataDrivenTests;
+using Microsoft.VisualStudio.TestTools.UnitTesting;
+
+namespace Microsoft.Recognizers.Text.Sequence.Tests
+{
+    [TestClass]
+    public class TestSequence_German : TestBase
+    {
+        [NetCoreTestDataSource]
+        [TestMethod]
+        public void PhoneNumberModel(TestModel testSpec)
+        {
+            TestPhoneNumber(testSpec);
+        }
+    }
+}

.NET/Microsoft.Recognizers.Text.DataDrivenTests/Sequence/TestSequence_Hindi.cs

@@ -0,0 +1,16 @@
+using Microsoft.Recognizers.Text.DataDrivenTests;
+using Microsoft.VisualStudio.TestTools.UnitTesting;
+
+namespace Microsoft.Recognizers.Text.Sequence.Tests
+{
+    [TestClass]
+    public class TestSequence_Hindi : TestBase
+    {
+        [NetCoreTestDataSource]
+        [TestMethod]
+        public void PhoneNumberModel(TestModel testSpec)
+        {
+            TestPhoneNumber(testSpec);
+        }
+    }
+}

.NET/Microsoft.Recognizers.Text.DataDrivenTests/Sequence/TestSequence_Italian.cs

@@ -0,0 +1,16 @@
+using Microsoft.Recognizers.Text.DataDrivenTests;
+using Microsoft.VisualStudio.TestTools.UnitTesting;
+
+namespace Microsoft.Recognizers.Text.Sequence.Tests
+{
+    [TestClass]
+    public class TestSequence_Italian : TestBase
+    {
+        [NetCoreTestDataSource]
+        [TestMethod]
+        public void PhoneNumberModel(TestModel testSpec)
+        {
+            TestPhoneNumber(testSpec);
+        }
+    }
+}

.NET/Microsoft.Recognizers.Text.DataDrivenTests/Sequence/TestSequence_Korean.cs

@@ -0,0 +1,16 @@
+using Microsoft.Recognizers.Text.DataDrivenTests;
+using Microsoft.VisualStudio.TestTools.UnitTesting;
+
+namespace Microsoft.Recognizers.Text.Sequence.Tests
+{
+    [TestClass]
+    public class TestSequence_Korean : TestBase
+    {
+        [NetCoreTestDataSource]
+        [TestMethod]
+        public void PhoneNumberModel(TestModel testSpec)
+        {
+            TestPhoneNumber(testSpec);
+        }
+    }
+}

.NET/Microsoft.Recognizers.Text.DataDrivenTests/Sequence/TestSequence_Spanish.cs

@@ -0,0 +1,16 @@
+using Microsoft.Recognizers.Text.DataDrivenTests;
+using Microsoft.VisualStudio.TestTools.UnitTesting;
+
+namespace Microsoft.Recognizers.Text.Sequence.Tests
+{
+    [TestClass]
+    public class TestSequence_Spanish : TestBase
+    {
+        [NetCoreTestDataSource]
+        [TestMethod]
+        public void PhoneNumberModel(TestModel testSpec)
+        {
+            TestPhoneNumber(testSpec);
+        }
+    }
+}

.NET/Microsoft.Recognizers.Text.DataDrivenTests/Sequence/TestSequence_Turkish.cs

@@ -0,0 +1,16 @@
+using Microsoft.Recognizers.Text.DataDrivenTests;
+using Microsoft.VisualStudio.TestTools.UnitTesting;
+
+namespace Microsoft.Recognizers.Text.Sequence.Tests
+{
+    [TestClass]
+    public class TestSequence_Turkish : TestBase
+    {
+        [NetCoreTestDataSource]
+        [TestMethod]
+        public void PhoneNumberModel(TestModel testSpec)
+        {
+            TestPhoneNumber(testSpec);
+        }
+    }
+}

.NET/Microsoft.Recognizers.Text.Sequence/Config/PhoneNumberConfiguration.cs

@@ -20,6 +20,8 @@ public PhoneNumberConfiguration(SequenceOptions options = SequenceOptions.None)
 
         public Regex ColonPrefixCheckRegex { get; set; }
 
+        public Regex FalsePositivePrefixRegex { get; set; }
+
         public List<char> ColonMarkers { get; set; }
 
         public List<char> ForbiddenPrefixMarkers { get; set; }

.NET/Microsoft.Recognizers.Text.Sequence/Dutch/DutchPhoneNumberExtractorConfiguration.cs

@@ -0,0 +1,14 @@
+using System.Text.RegularExpressions;
+using Microsoft.Recognizers.Definitions.Dutch;
+
+namespace Microsoft.Recognizers.Text.Sequence.Dutch
+{
+    public class DutchPhoneNumberExtractorConfiguration : BasePhoneNumberExtractorConfiguration
+    {
+        public DutchPhoneNumberExtractorConfiguration(SequenceOptions options)
+            : base(options)
+        {
+            FalsePositivePrefixRegex = null;
+        }
+    }
+}

.NET/Microsoft.Recognizers.Text.Sequence/English/Extractors/EnglishPhoneNumberExtractorConfiguration.cs

@@ -1,21 +1,14 @@
-using System.Collections.Generic;
-using System.Text.RegularExpressions;
-using Microsoft.Recognizers.Definitions;
+using System.Text.RegularExpressions;
+using Microsoft.Recognizers.Definitions.English;
 
 namespace Microsoft.Recognizers.Text.Sequence.English
 {
-    public class EnglishPhoneNumberExtractorConfiguration : PhoneNumberConfiguration
+    public class EnglishPhoneNumberExtractorConfiguration : BasePhoneNumberExtractorConfiguration
     {
         public EnglishPhoneNumberExtractorConfiguration(SequenceOptions options)
             : base(options)
         {
-            WordBoundariesRegex = BasePhoneNumbers.WordBoundariesRegex;
-            NonWordBoundariesRegex = BasePhoneNumbers.NonWordBoundariesRegex;
-            EndWordBoundariesRegex = BasePhoneNumbers.EndWordBoundariesRegex;
-            ColonPrefixCheckRegex = new Regex(BasePhoneNumbers.ColonPrefixCheckRegex);
-            ColonMarkers = (List<char>)BasePhoneNumbers.ColonMarkers;
-            ForbiddenPrefixMarkers = (List<char>)BasePhoneNumbers.ForbiddenPrefixMarkers;
-            ForbiddenSuffixMarkers = (List<char>)BasePhoneNumbers.ForbiddenSuffixMarkers;
+            FalsePositivePrefixRegex = new Regex(PhoneNumbersDefinitions.FalsePositivePrefixRegex);
         }
     }
 }

.NET/Microsoft.Recognizers.Text.Sequence/Extractors/BasePhoneNumberExtractor.cs

@@ -111,6 +111,16 @@ public override List<ExtractResult> Extract(string text)
                 if (er.Start != 0)
                 {
                     var ch = text[(int)(er.Start - 1)];
+                    var front = text.Substring(0, (int)(er.Start - 1));
+
+                    if (this.config.FalsePositivePrefixRegex != null &&
+                            this.config.FalsePositivePrefixRegex.IsMatch(front))
+                    {
+                        ers.Remove(er);
+                        i--;
+                        continue;
+                    }
+
                     if (BasePhoneNumbers.BoundaryMarkers.Contains(ch))
                     {
                         if (SpecialBoundaryMarkers.Contains(ch) &&
@@ -131,7 +141,6 @@ public override List<ExtractResult> Extract(string text)
                             }
 
                             // check the international dialing prefix
-                            var front = text.Substring(0, (int)(er.Start - 1));
                             if (InternationDialingPrefixRegex.IsMatch(front))
                             {
                                 var moveOffset = InternationDialingPrefixRegex.Match(front).Length + 1;
@@ -152,7 +161,6 @@ public override List<ExtractResult> Extract(string text)
                         // Handle "tel:123456".
                         if (BasePhoneNumbers.ColonMarkers.Contains(ch))
                         {
-                            var front = text.Substring(0, (int)(er.Start - 1));
                             if (this.config.ColonPrefixCheckRegex.IsMatch(front))
                             {
                                 continue;

.NET/Microsoft.Recognizers.Text.Sequence/Extractors/BasePhoneNumberExtractorConfiguration.cs

@@ -0,0 +1,21 @@
+using System.Collections.Generic;
+using System.Text.RegularExpressions;
+using Microsoft.Recognizers.Definitions;
+
+namespace Microsoft.Recognizers.Text.Sequence
+{
+    public class BasePhoneNumberExtractorConfiguration : PhoneNumberConfiguration
+    {
+        public BasePhoneNumberExtractorConfiguration(SequenceOptions options)
+            : base(options)
+        {
+            WordBoundariesRegex = BasePhoneNumbers.WordBoundariesRegex;
+            NonWordBoundariesRegex = BasePhoneNumbers.NonWordBoundariesRegex;
+            EndWordBoundariesRegex = BasePhoneNumbers.EndWordBoundariesRegex;
+            ColonPrefixCheckRegex = new Regex(BasePhoneNumbers.ColonPrefixCheckRegex);
+            ColonMarkers = (List<char>)BasePhoneNumbers.ColonMarkers;
+            ForbiddenPrefixMarkers = (List<char>)BasePhoneNumbers.ForbiddenPrefixMarkers;
+            ForbiddenSuffixMarkers = (List<char>)BasePhoneNumbers.ForbiddenSuffixMarkers;
+        }
+    }
+}

.NET/Microsoft.Recognizers.Text.Sequence/French/FrenchPhoneNumberExtractorConfiguration.cs

@@ -0,0 +1,14 @@
+using System.Text.RegularExpressions;
+using Microsoft.Recognizers.Definitions.French;
+
+namespace Microsoft.Recognizers.Text.Sequence.French
+{
+    public class FrenchPhoneNumberExtractorConfiguration : BasePhoneNumberExtractorConfiguration
+    {
+        public FrenchPhoneNumberExtractorConfiguration(SequenceOptions options)
+            : base(options)
+        {
+            FalsePositivePrefixRegex = null;
+        }
+    }
+}