Support custom token chars in (edge)ngram tokenizer (#4384)

Relates: #4341

This commit adds support for custom token characters for
edgengram and ngram tokenizers.

Author: russcam

src/Nest/Analysis/Tokenizers/NGram/EdgeNGramTokenizer.cs

@@ -29,6 +29,18 @@ public interface IEdgeNGramTokenizer : ITokenizer
 		/// </summary>
 		[DataMember(Name ="token_chars")]
 		IEnumerable<TokenChar> TokenChars { get; set; }
+
+		/// <summary>
+		/// Custom characters that should be treated as part of a token. For example,
+		/// setting this to +-_ will make the tokenizer treat the plus, minus and
+		/// underscore sign as part of a token.
+		/// <para />
+		/// Requires setting <see cref="TokenChar.Custom"/> as part of <see cref="TokenChars"/>
+		/// <para />
+		/// Available in Elasticsearch 7.6.0+.
+		/// </summary>
+		[DataMember(Name = "custom_token_chars")]
+		string CustomTokenChars { get; set; }
 	}
 
 	/// <inheritdoc />
@@ -44,6 +56,9 @@ public class EdgeNGramTokenizer : TokenizerBase, IEdgeNGramTokenizer
 
 		/// <inheritdoc />
 		public IEnumerable<TokenChar> TokenChars { get; set; }
+
+		/// <inheritdoc />
+		public string CustomTokenChars { get; set; }
 	}
 
 	/// <inheritdoc />
@@ -52,22 +67,27 @@ public class EdgeNGramTokenizerDescriptor
 	{
 		protected override string Type => "edge_ngram";
 		int? IEdgeNGramTokenizer.MaxGram { get; set; }
-
 		int? IEdgeNGramTokenizer.MinGram { get; set; }
 		IEnumerable<TokenChar> IEdgeNGramTokenizer.TokenChars { get; set; }
 
-		/// <inheritdoc />
+		string IEdgeNGramTokenizer.CustomTokenChars { get; set; }
+
+		/// <inheritdoc cref="IEdgeNGramTokenizer.MinGram" />
 		public EdgeNGramTokenizerDescriptor MinGram(int? minGram) => Assign(minGram, (a, v) => a.MinGram = v);
 
-		/// <inheritdoc />
+		/// <inheritdoc cref="IEdgeNGramTokenizer.MaxGram" />
 		public EdgeNGramTokenizerDescriptor MaxGram(int? maxGram) => Assign(maxGram, (a, v) => a.MaxGram = v);
 
-		/// <inheritdoc />
+		/// <inheritdoc cref="IEdgeNGramTokenizer.TokenChars" />
 		public EdgeNGramTokenizerDescriptor TokenChars(IEnumerable<TokenChar> tokenChars) =>
 			Assign(tokenChars, (a, v) => a.TokenChars = v);
 
-		/// <inheritdoc />
+		/// <inheritdoc cref="IEdgeNGramTokenizer.TokenChars" />
 		public EdgeNGramTokenizerDescriptor TokenChars(params TokenChar[] tokenChars) =>
 			Assign(tokenChars, (a, v) => a.TokenChars = v);
+
+		/// <inheritdoc cref="IEdgeNGramTokenizer.CustomTokenChars" />
+		public EdgeNGramTokenizerDescriptor CustomTokenChars(string customTokenChars) =>
+			Assign(customTokenChars, (a, v) => a.CustomTokenChars = v);
 	}
 }

src/Nest/Analysis/Tokenizers/NGram/NGramTokenizer.cs

@@ -29,6 +29,18 @@ public interface INGramTokenizer : ITokenizer
 		/// </summary>
 		[DataMember(Name ="token_chars")]
 		IEnumerable<TokenChar> TokenChars { get; set; }
+
+		/// <summary>
+		/// Custom characters that should be treated as part of a token. For example,
+		/// setting this to +-_ will make the tokenizer treat the plus, minus and
+		/// underscore sign as part of a token.
+		/// <para />
+		/// Requires setting <see cref="TokenChar.Custom"/> as part of <see cref="TokenChars"/>
+		/// <para />
+		/// Available in Elasticsearch 7.6.0+.
+		/// </summary>
+		[DataMember(Name = "custom_token_chars")]
+		string CustomTokenChars { get; set; }
 	}
 
 	/// <inheritdoc />
@@ -44,6 +56,9 @@ public class NGramTokenizer : TokenizerBase, INGramTokenizer
 
 		/// <inheritdoc />
 		public IEnumerable<TokenChar> TokenChars { get; set; }
+
+		/// <inheritdoc />
+		public string CustomTokenChars { get; set; }
 	}
 
 	/// <inheritdoc />
@@ -52,21 +67,26 @@ public class NGramTokenizerDescriptor
 	{
 		protected override string Type => "ngram";
 		int? INGramTokenizer.MaxGram { get; set; }
-
 		int? INGramTokenizer.MinGram { get; set; }
 		IEnumerable<TokenChar> INGramTokenizer.TokenChars { get; set; }
 
-		/// <inheritdoc />
+		string INGramTokenizer.CustomTokenChars { get; set; }
+
+		/// <inheritdoc cref="INGramTokenizer.MinGram" />
 		public NGramTokenizerDescriptor MinGram(int? minGram) => Assign(minGram, (a, v) => a.MinGram = v);
 
-		/// <inheritdoc />
+		/// <inheritdoc cref="INGramTokenizer.MaxGram" />
 		public NGramTokenizerDescriptor MaxGram(int? minGram) => Assign(minGram, (a, v) => a.MaxGram = v);
 
-		/// <inheritdoc />
+		/// <inheritdoc cref="INGramTokenizer.TokenChars" />
 		public NGramTokenizerDescriptor TokenChars(IEnumerable<TokenChar> tokenChars) =>
 			Assign(tokenChars, (a, v) => a.TokenChars = v);
 
-		/// <inheritdoc />
+		/// <inheritdoc cref="INGramTokenizer.TokenChars" />
 		public NGramTokenizerDescriptor TokenChars(params TokenChar[] tokenChars) => Assign(tokenChars, (a, v) => a.TokenChars = v);
+
+		/// <inheritdoc cref="INGramTokenizer.CustomTokenChars" />
+		public NGramTokenizerDescriptor CustomTokenChars(string customTokenChars) =>
+			Assign(customTokenChars, (a, v) => a.CustomTokenChars = v);
 	}
 }

src/Nest/Analysis/Tokenizers/NGram/TokenChar.cs

@@ -20,5 +20,13 @@ public enum TokenChar
 
 		[EnumMember(Value = "symbol")]
 		Symbol,
+
+		/// <summary>
+		/// Custom token characters.
+		/// <para></para>
+		/// Available in Elasticsearch 7.6.0+
+		/// </summary>
+		[EnumMember(Value = "custom")]
+		Custom,
 	}
 }

tests/Tests/Analysis/Tokenizers/TokenizerTests.cs

@@ -34,6 +34,36 @@ public class EdgeNGramTests : TokenizerAssertionBase<EdgeNGramTests>
 			public override string Name => "endgen";
 		}
 
+		[SkipVersion("<7.6.0", "CustomTokenChars introduced in 7.6.0")]
+		public class EdgeNGramCustomTokenCharsTests : TokenizerAssertionBase<EdgeNGramCustomTokenCharsTests>
+		{
+			public override FuncTokenizer Fluent => (n, t) => t.EdgeNGram(n, e => e
+				.MaxGram(2)
+				.MinGram(1)
+				.TokenChars(TokenChar.Custom)
+				.CustomTokenChars("+-_")
+			);
+
+			public override ITokenizer Initializer => new EdgeNGramTokenizer
+			{
+				MaxGram = 2,
+				MinGram = 1,
+				TokenChars = new[] { TokenChar.Custom },
+				CustomTokenChars = "+-_"
+			};
+
+			public override object Json => new
+			{
+				min_gram = 1,
+				max_gram = 2,
+				token_chars = new[] { "custom" },
+				custom_token_chars = "+-_",
+ 				type = "edge_ngram"
+			};
+
+			public override string Name => "endgen_custom";
+		}
+
 		public class NGramTests : TokenizerAssertionBase<NGramTests>
 		{
 			public override FuncTokenizer Fluent => (n, t) => t.NGram(n, e => e
@@ -60,6 +90,36 @@ public class NGramTests : TokenizerAssertionBase<NGramTests>
 			public override string Name => "ng";
 		}
 
+		[SkipVersion("<7.6.0", "CustomTokenChars introduced in 7.6.0")]
+		public class NGramCustomTokenCharsTests : TokenizerAssertionBase<NGramCustomTokenCharsTests>
+		{
+			public override FuncTokenizer Fluent => (n, t) => t.NGram(n, e => e
+				.MaxGram(2)
+				.MinGram(1)
+				.TokenChars(TokenChar.Custom)
+				.CustomTokenChars("+-_")
+			);
+
+			public override ITokenizer Initializer => new NGramTokenizer
+			{
+				MaxGram = 2,
+				MinGram = 1,
+				TokenChars = new[] { TokenChar.Custom },
+				CustomTokenChars = "+-_"
+			};
+
+			public override object Json => new
+			{
+				min_gram = 1,
+				max_gram = 2,
+				token_chars = new[] { "custom" },
+				custom_token_chars = "+-_",
+				type = "ngram"
+			};
+
+			public override string Name => "ngram_custom";
+		}
+
 		public class PathHierarchyTests : TokenizerAssertionBase<PathHierarchyTests>
 		{
 			public override FuncTokenizer Fluent => (n, t) => t.PathHierarchy(n, e => e