Indexing Data

Indexing text

As shown in this example, if all you have is a key and text to be indexed for it, you can just make use one of the AddAsync overloads that operate on strings - there is no need to construct any artificial objects in order to populate the index.

Each of these methods will use the default text extraction and tokenization configured for the index and the default field id, IFullTextIndex.FieldLookup.DefaultField.

Task AddAsync(TKey key, string text)

Adds a document to the index treating the single string as the complete document text.

Task AddAsync(TKey key, IEnumerable<string> text)

Adds a document to the index with multiple text fragments. Each fragment is considered to be fragments of the same text, i.e. the index and offset of tokens increments across the elements in the IEnumerable<string>.

It’s worth noting that each fragment is processed independently, so an implicit word break exists between element. This means that AddAsync(1, new[ "a", "b" ]) will index two words (a and b) whereas if the two strings had been naively concatenated together, only one (ab) would, as there was no word break between them.

Indexing Objects

In order to index an object against the index it needs to have first been configured when the index was built, using WithObjectTokenization. This allows for multiple properties of a single object to be extracted as different fields of the same document.

See the Indexing objects quick start for an example of this in action.

Task AddAsync<TObject>(TObject item)

Adds a single document to the index.

Task AddRangeAsync<TObject>(IEnumerable<TObject> items)

Adds a set of documents to the index in a single mutation. This is more efficient than making multiple calls to AddAsync<TObject(TObject item)> unless a batch mutation has already been manually started.