Bug fixesedit

Breaking changesedit

In the course of fixing the code-generation of types used on IndexSettingsAnalysis, several breaking changes were introduced. Some of these were necessary to make the types usable, while others fixed the consistency of the generated code.

IndexSettingsAnalysisedit

Code-generation has been updated to apply transforms to fix the specification of the IndexSettingsAnalysis type. As a result, all properties have been renamed, and some property types have been changed.

Before

public sealed partial class IndexSettingsAnalysis
{
   public Elastic.Clients.Elasticsearch.Analysis.Analyzers? Analyzer { get; set; }
   public IDictionary<string, Elastic.Clients.Elasticsearch.Analysis.CharFilter>? CharFilter { get; set; }
   public IDictionary<string, Elastic.Clients.Elasticsearch.Analysis.TokenFilter>? Filter { get; set; }
   public Elastic.Clients.Elasticsearch.Analysis.Normalizers? Normalizer { get; set; }
   public IDictionary<string, Elastic.Clients.Elasticsearch.Analysis.Tokenizer>? Tokenizer { get; set; }
}

After

public sealed partial class IndexSettingsAnalysis
{
   public Elastic.Clients.Elasticsearch.Analysis.Analyzers? Analyzers { get; set; }
   public Elastic.Clients.Elasticsearch.Analysis.CharFilters? CharFilters { get; set; }
   public Elastic.Clients.Elasticsearch.Analysis.TokenFilters? TokenFilters { get; set; }
   public Elastic.Clients.Elasticsearch.Analysis.Normalizers? Normalizers { get; set; }
   public Elastic.Clients.Elasticsearch.Analysis.Tokenizers? Tokenizers { get; set; }
}

The IndexSettingsAnalysisDescriptor type has been updated accordingly to apply the above changes. It now supports a more convenient syntax to easily define the filters, normalizers and tokenizers that apply to the settings for indices.

Example usage of updated fluent syntax:edit

var descriptor = new CreateIndexRequestDescriptor("test")
   .Settings(s => s
      .Analysis(a => a
         .Analyzers(a => a
            .Stop("stop-name", stop => stop.StopwordsPath("analysis/path.txt"))
            .Pattern("pattern-name", pattern => pattern.Version("version"))
            .Custom("my-custom-analyzer", c => c
               .Filter(new[] { "stop", "synonym" })
               .Tokenizer("standard")))
         .TokenFilters(f => f
            .Synonym("synonym", synonym => synonym
               .SynonymsPath("analysis/synonym.txt")))));

Token Filtersedit

Token filter types now implement the ITokenFilter interface, rather than ITokenFilterDefinition.

The TokenFilter union type has been renamed to CategorizationTokenFilter to clearly signify it’s use only within ML categorization contexts.

A TokenFilters type has been introduced, which derives from IsADictionary and supports convenient addition of known token filters via the fluent API.

Character Filtersedit

Character filter types now implement the ICharFilter interface, rather than ICharFilterDefinition.

The CharFilter union type has been renamed to CategorizationCharFilter to clearly signify it’s use only within ML categorization contexts.

A CharFilters type has been introduced, which derives from IsADictionary and supports convenient addition of known character filters via the fluent API.

Tokenizersedit

Tokenizer types now implement the ITokenizer interface, rather than ITokenizerDefinition.

The Tokenizer union type has been renamed to CategorizationTokenizer to clearly signify it’s use only within ML categorization contexts.

A Tokenizers type has been introduced, which derives from IsADictionary and supports convenient addition of known tokenizers via the fluent API.

IndexManagement.StorageTypeedit

The 8.6 specification fixed this type to mark is as a non-exhaustive enum, since it supports additional values besides those coded into the specification. As a result the code-generation for this type causes some breaking changes. The type is no longer generated as an enum and is not a custom readonly struct.