Redis
redis-vl-dotnet Documentation redis-vl-dotnet

redis-vl-dotnet

Partial Updates

SearchIndex supports targeted document updates for both JSON and HASH storage, but the two storage models have different constraints.

Update by id versus update by key

Each storage type exposes two entry points:

  • UpdateJsonByIdAsync and UpdateHashByIdAsync

  • UpdateJsonByKeyAsync and UpdateHashByKeyAsync

Update-by-id resolves the Redis key as schema.Index.Prefix + id by using DocumentKeyResolver.ResolveKeyFromId(…​). Update-by-key expects the full Redis key and trims surrounding whitespace before issuing Redis commands.

If the target document does not exist, the update methods return false instead of throwing.

JSON partial updates

JSON updates accept one or more JsonPartialUpdate values:

var updated = await index.UpdateJsonByIdAsync(
    "movie-2",
    [
        new JsonPartialUpdate("$.summary", "A linguist helps decode the intent of alien visitors."),
        new JsonPartialUpdate("$.year", 2017)
    ]);

Behavior and limitations:

  • each path must be an absolute JSONPath such as $.title or $.metadata.rating

  • root replacement with $ is rejected

  • duplicate paths in one call are rejected

  • the library first checks that the document exists with JSON.GET

  • each update path is then written with a separate JSON.SET

Nested JSON updates work when the path already addresses a valid RedisJSON location, as shown in the integration tests with $.metadata.rating and $.metadata.director.

HASH partial updates

HASH updates accept one or more HashPartialUpdate values:

var updated = await index.UpdateHashByIdAsync(
    "arrival",
    [
        new HashPartialUpdate("summary", "A linguist decodes messages from alien visitors."),
        new HashPartialUpdate("genre", "sci-fi")
    ]);

Behavior and limitations:

  • field names are trimmed before execution

  • duplicate fields in one call are rejected

  • null values are rejected

  • updates write a single HashSetAsync call after the document existence check

  • HASH updates only set top-level field values; there is no nested-path syntax

Complex values are serialized into a single hash field value. For example, array values are stored as JSON text in the hash entry.

Example coverage

Use these runnable examples when validating the documented behavior:

  • /examples/JsonStorageExample demonstrates JSON load, fetch, update-by-id, update-by-key, clear, and drop flows

  • /examples/VectorSearchExample demonstrates HASH update-by-id plus final cleanup with DropAsync(deleteDocuments: true)

The matching source-backed behavior is also covered in:

  • tests/RedisVL.Tests/Indexes/SearchIndexIntegrationTests.cs

  • tests/RedisVL.Tests/Indexes/SearchIndexAsyncTests.cs

Related pages