Loading

Field attributes

Field attributes on your document class properties control which Elasticsearch field type the source generator emits. When no attribute is present, the generator infers a type from CLR conventions.

CLR type Default ES field type
string text with .keyword sub-field
int, long long
float, double, decimal double
bool boolean
DateTime, DateTimeOffset date
Guid keyword
Enum types keyword (serialized by name)
Complex types (classes) object (recursively mapped)
IEnumerable<T> Array of inferred type for T

Use explicit attributes to override the default or set field-specific options.

Text and keyword fields

public class Article
{
    [Text(Analyzer = "english", SearchAnalyzer = "english")]
    public string Body { get; set; }

    [Keyword(Normalizer = "lowercase", IgnoreAbove = 256)]
    public string Category { get; set; }

    [Completion(Analyzer = "simple")]
    public string Suggest { get; set; }
}
[Text] options
Property Type Default Description
Analyzer string? null Analyzer used at index time
SearchAnalyzer string? null Analyzer used at search time (falls back to Analyzer)
Norms bool true Store field length norms for relevance scoring
Index bool true Whether the field is searchable

When no [Text] or [Keyword] attribute is present on a string property, the generator defaults to text with an automatic .keyword sub-field. Use [Text] explicitly when you need to set analyzer options.

[Keyword] options
Property Type Default Description
Normalizer string? null Normalizer for case-insensitive matching
IgnoreAbove int 0 (no limit) Values longer than this are not indexed (still stored)
DocValues bool true Store doc values for sorting and aggregations
Index bool true Whether the field is searchable
[Completion] options
Property Type Default Description
Analyzer string? null Analyzer for indexing
SearchAnalyzer string? null Analyzer for search 
public class Order
{
    [Long]
    public int Quantity { get; set; }

    [Double]
    public decimal Price { get; set; }

    [Date(Format = "strict_date_optional_time||epoch_millis")]
    public DateTimeOffset CreatedAt { get; set; }

    [Boolean]
    public bool IsActive { get; set; }
}
Options for numeric and date fields

All of [Long], [Double], [Date], and [Boolean] support:

Property Type Default Description
DocValues bool true Store doc values for sorting and aggregations
Index bool true Whether the field is searchable

[Date] additionally supports:

Property Type Default Description
Format string? null Date format string 
public class Store
{
    [GeoPoint]
    public GeoLocation Location { get; set; }

    [GeoShape]
    public object DeliveryZone { get; set; }

    [Ip]
    public string ClientIp { get; set; }
}

These attributes have no additional options.

public class Document
{
    [DenseVector(Dims = 384, Similarity = "cosine")]
    public float[] Embedding { get; set; }

    [SemanticText(InferenceId = "my-elser-endpoint")]
    public string Content { get; set; }
}
[DenseVector] and [SemanticText] options

[DenseVector]

Property Type Default Description
Dims int 0 Number of dimensions
Similarity string? null Similarity function (cosine, dot_product, l2_norm)

[SemanticText]

Property Type Default Description
InferenceId string? null Inference endpoint ID 
public class Order
{
    [Object]
    public Address ShippingAddress { get; set; }

    [Nested(IncludeInParent = true)]
    public List<LineItem> Items { get; set; }
}
[Object] and [Nested] options

[Object]

Property Type Default Description
Enabled bool true When false, the object is stored but not indexed

[Nested]

Property Type Default Description
IncludeInParent bool false Include nested documents in the parent document
IncludeInRoot bool false Include nested documents in the root document

Use [Nested] when you need to query array elements independently (e.g. "find orders where item.name = X AND item.price < Y" without cross-matching).

Note

These attributes do not produce Elasticsearch field type mappings. They tell the source generator to produce accessor delegates that the ingest channel uses at runtime for bulk operations, index naming, and deduplication.

Attribute Purpose Effect on ingest
[Id] Marks the document _id field Bulk operations use index (upsert) instead of create
[Timestamp] Marks the timestamp field Required for data streams; used for date-based index naming
[ContentHash] Marks a content hash field Enables hash-based index reuse (skip recreating unchanged indices)
[BatchIndexDate] Auto-stamped batch date All documents in a batch share one timestamp for index naming
[LastUpdated] Auto-stamped current time Each document gets the write time 
public class Product
{
    [Id]
    [Keyword]
    public string Sku { get; set; }

    [Timestamp]
    public DateTimeOffset CreatedAt { get; set; }

    [ContentHash]
    [Keyword]
    public string Hash { get; set; }

    [BatchIndexDate]
    public DateTimeOffset IndexedAt { get; set; }

    [LastUpdated]
    public DateTimeOffset UpdatedAt { get; set; }
}
Note

These attributes mark fields for the AI enrichment pipeline. They do not affect Elasticsearch field type mappings.

Attribute Purpose
[AiInput] Field content is sent to the LLM as input
[AiField] Field is populated by LLM output

Use [Dimension] on keyword fields when DataStreamMode = DataStreamMode.Tsdb:

public class Metric
{
    [Dimension]
    [Keyword]
    public string Host { get; set; }

    [Dimension]
    [Keyword]
    public string Service { get; set; }

    [Long]
    public long CpuPercent { get; set; }
}

Field attributes can be combined with ingest metadata attributes on the same property:

public class Product
{
    [Id]
    [Keyword]
    public string Sku { get; set; }

    [Timestamp]
    [Date]
    public DateTimeOffset CreatedAt { get; set; }
}
  1. Ingest: use as _id
  2. Mapping: keyword field type
  3. Ingest: use for date-pattern naming
  4. Mapping: date field type