Indexes: Analyzers

Lucene offers several Analyzers out of the box.
New customized analyzers can also be created.

Various Analyzers differ in the way they split the text stream ("tokenize"),
and in the way they process those tokens in post-tokenization.

The examples below use the following text:

The quick brown fox jumped over the lazy dogs, Bob@hotmail.com 123432.

Analyzers that remove common "Stop Words":

• StandardAnalyzer, which is Lucene's default, will produce the following tokens:

  [quick] [brown] [fox] [jumped] [over] [lazy] [dog] [bob@hotmail.com] [123432]

  Removes common "stop words".
  Separates on whitespace and punctuation that is followed by whitespace - a dot that is not followed by whitespace is considered part of the token.
  Converts to lower-case letters so that searches aren't case-sensitive.
  Email addresses and internet hostnames are one token.
  Splits words at hyphens, unless there's a number in the token, in which case the whole token is interpreted as a product number and is not split.

• StopAnalyzer will work similarly, but will not perform light stemming and will only tokenize on white space:

  [quick] [brown] [fox] [jumped] [over] [lazy] [dogs] [bob] [hotmail] [com]

  Removes numbers and symbols and separates tokens with them.
  This means that email and web addresses are separated.
  Removes common "stop words".
  Separates on white spaces.
  Converts to lower-case letters so that searches aren't case sensitive.

Analyzers that do not remove common "Stop Words"

• SimpleAnalyzer will tokenize on all non-alpha characters and will make all the tokens lowercase:

  [the] [quick] [brown] [fox] [jumped] [over] [the] [lazy] [dogs] [bob] [hotmail] [com]

  Includes common "stop words".
  Removes numbers and symbols and separates tokens with them.
  This means that email and web addresses are separated.
  Separates on white spaces.
  Converts to lower-case letters so that searches aren't case sensitive.

• WhitespaceAnalyzer will just tokenize on white spaces:

  [The] [quick] [brown] [fox] [jumped] [over] [the] [lazy] [dogs,] [Bob@hotmail.com] [123432.]

  Only separates on whitespaces.
  Preserves upper/lower cases in text, which means that searches will be case-sensitive.
  Email and web addresses, phone numbers, and other such forms of ID are kept whole.

• KeywordAnalyzer will perform no tokenization, and will consider the whole text as one token:

  [The quick brown fox jumped over the lazy dogs, bob@hotmail.com 123432.]

  Preserves upper/lower cases in text for case-sensitive searches.
  Useful in situations like IDs and codes where you do not want to separate into multiple tokens.

Analyzers that tokenize according to the defined number of characters

• NGramAnalyzer will tokenize on predefined token lengths, 2-6 chars long, which are defined by Indexing.Analyzers.NGram.MinGram and Indexing.Analyzers.NGram.MaxGram configuration options:

[.c] [.co] [.com] [12] [123] [1234] [12343] [123432] [23] [234] [2343] [23432] [32] [34] [343] [3432] [43] [432] [@h] [@ho] [@hot] [@hotm] [@hotma] [ai] [ail] [ail.] [ail.c] [ail.co] [az] [azy] [b@] [b@h] [b@ho] [b@hot] [b@hotm] [bo] [bob] [bob@] [bob@h] [bob@ho] [br] [bro] [brow] [brown] [ck] [co] [com] [do] [dog] [dogs] [ed] [er] [fo] [fox] [gs] [ho] [hot] [hotm] [hotma] [hotmai] [ic] [ick] [il] [il.] [il.c] [il.co] [il.com] [ju] [jum] [jump] [jumpe] [jumped] [l.] [l.c] [l.co] [l.com] [la] [laz] [lazy] [ma] [mai] [mail] [mail.] [mail.c] [mp] [mpe] [mped] [ob] [ob@] [ob@h] [ob@ho] [ob@hot] [og] [ogs] [om] [ot] [otm] [otma] [otmai] [otmail] [ov] [ove] [over] [ow] [own] [ox] [pe] [ped] [qu] [qui] [quic] [quick] [ro] [row] [rown] [tm] [tma] [tmai] [tmail] [tmail.] [ui] [uic] [uick] [um] [ump] [umpe] [umped] [ve] [ver] [wn] [zy]

You can override NGram analyzer default token lengths by configuring Indexing.Analyzers.NGram.MinGram and Indexing.Analyzers.NGram.MaxGram per index e.g. setting them to 3 and 4 accordingly will generate:

[.co] [.com] [123] [1234] [234] [2343] [343] [3432] [432] [@ho] [@hot] [ail] [ail.] [azy] [b@h] [b@ho] [bob] [bob@] [bro] [brow] [com] [dog] [dogs] [fox] [hot] [hotm] [ick] [il.] [il.c] [jum] [jump] [l.c] [l.co] [laz] [lazy] [mai] [mail] [mpe] [mped] [ob@] [ob@h] [ogs] [otm] [otma] [ove] [over] [own] [ped] [qui] [quic] [row] [rown] [tma] [tmai] [uic] [uick] [ump] [umpe] [ver]

To allow full-text search on the text fields, you can use the analyzers provided out of the box with Lucene.
These are available as part of the Lucene library which ships with RavenDB.

For most cases, Lucene's StandardAnalyzer would be your analyzer of choice. As shown above, this analyzer is aware of e-mail and network addresses when tokenizing. It normalizes cases, filters out common English words, and does some basic English stemming as well.

For languages other than English, or if you need a custom analysis process, you can roll your own Analyzer. It is quite simple and may already be available as a contrib package for Lucene.
There are also Collation analyzers available (you can read more about them here).

To index a document field using a specific analyzer, all you need to do is to match it with the field's name:

public class BlogPosts_ByTagsAndContent : AbstractIndexCreationTask<BlogPost>
{
    public BlogPosts_ByTagsAndContent()
    {
        Map = posts => from post in posts
                       select new
                       {
                           post.Tags,
                           post.Content
                       };

        // Field Tags will be tokenized by the SimpleAnalyzer
        Analyzers.Add(x => x.Tags, "SimpleAnalyzer");

        // Field Content will be tokenized by the custom analyzer SnowballAnalyzer
        Analyzers.Add(x => x.Content, typeof(SnowballAnalyzer).AssemblyQualifiedName);
    }
}

store.Maintenance.Send(new PutIndexesOperation(new IndexDefinitionBuilder<BlogPost>("BlogPosts/ByTagsAndContent")
{
    Map = posts => from post in posts
                   select new
                   {
                       post.Tags,
                       post.Content
                   },
    Analyzers =
    {
        {x => x.Tags, "SimpleAnalyzer"},
        {x => x.Content, typeof(SnowballAnalyzer).AssemblyQualifiedName}
    }
}.ToIndexDefinition(store.Conventions)));

You can write your own custom analyzers as a .cs file. Custom analyzers can be defined as:

• Database Custom Analyzers - can only be used by the indexes of the database where they are defined.
• Server-Wide Custom Analyzers - can be used by indexes on all databases on all servers in the cluster.

A database analyzer can have the same name as a server-wide analyzer. In this situation, the indexes of that database will use the database version of the analyzer. So you can think of database analyzers as overriding the server-wide analyzers with the same names.

There are a few ways to create a custom analyzer and add it to your server:
1. Using the Studio
2. Using the Client API
3. Adding it directly to RavenDB's binaries

Using the Client API

First, create a class that inherits from abstract Lucene.Net.Analysis.Analyzer (you need to reference Lucene.Net.dll, which is supplied with RavenDB Server package). For example:

public class MyAnalyzer : Lucene.Net.Analysis.Analyzer
{
    public override TokenStream TokenStream(string fieldName, TextReader reader)
    {
        throw new CodeOmitted();
    }
}

Next, define the analyzer for a specific database using the operation PutAnalyzersOperation. Or, to make it a server-wide analyzer, use PutServerWideOperation. These operations are very similar in how they work. Both of them take one parameter: either an AnalyzerDefinition, or an array of AnalyzerDefinition's.

public class PutAnalyzersOperation
{
    private readonly AnalyzerDefinition[] _analyzersToAdd;
}

public class PutServerWideAnalyzersOperation
{
    private readonly AnalyzerDefinition[] _analyzersToAdd;
}

By default, the PutAnalyzersOperation will apply to the default database of the document store you're using. To target a different database, use the ForDatabase() method - read more here.

The AnalyzerDefinition object has two properties, Name and Code:

public class AnalyzerDefinition
{
    public string Name { get; set; }
    public string Code { get; set; }
}

Client API Example

Now let's see how everything fits together.

store.Maintenance.ForDatabase("MyDatabase").Send(new PutAnalyzersOperation(new AnalyzerDefinition
{
    Name = "MyAnalyzer", // The name must be same as the analyzer's class name
    Code = @"
        using System.IO;
        using Lucene.Net.Analysis; 
        using Lucene.Net.Analysis.Standard;

        namespace MyAnalyzer
        {
            public class MyAnalyzer : Lucene.Net.Analysis.Analyzer
            {
                public override TokenStream TokenStream(string fieldName, TextReader reader)
                {
                    throw new CodeOmitted();
                }
            }
        }"
}));

Adding an Analyzer to the Binaries

Another way of adding custom analyzers to RavenDB is to place them next to RavenDB's binaries. Note that it needs to be compatible with .NET Core 2.0 (e.g. .NET Standard 2.0 assembly). The fully qualified name needs to be specified for an indexing field that is going to be tokenized by the analyzer. This is the only way to add custom analyzers in RavenDB versions older than 5.2.

RavenDB has three default analyzers that it uses to index text when no other analyzer was specified:

• Default Analyzer - LowerCaseKeywordAnalyzer
• Default Exact Analyzer - KeywordAnalyzer
• Default Search Analyzer - RavenStandardAnalyzer

You can choose other analyzers to serve as your default analyzers by modifying the indexing configuration.

Default Analyzer

For regular text fields, RavenDB uses a custom analyzer called LowerCaseKeywordAnalyzer. Its implementation behaves like Lucene's KeywordAnalyzer, but it also performs case normalization by converting all characters to lower case. That is - RavenDB stores the entire text field as a single token, in a lowercased form. Given the same sample text above, LowerCaseKeywordAnalyzer will produce a single token:

[the quick brown fox jumped over the lazy dogs, bob@hotmail.com 123432.]

Default Exact Analyzer

For 'exact case' text fields, RavenDB uses Lucene's KeywordAnalyzer, which treats the entire text field as one token and does not change the case of the original text. To make an index store text with the exact case, see the section on changing field indexing behavior below.

Default Search Analyzer

For full-text search text fields, RavenDB uses RavenStandardAnalyzer, which is just an optimized version of Lucene's StandardAnalyzer. To make an index that allows full-text search, see the section on changing field indexing behavior below.

By default, each indexed field is analyzed using the LowerCaseKeywordAnalyzer which indexes a field as a single, lowercased term.

This behavior can be changed by setting the FieldIndexing option for a particular field. The possible values are:

• FieldIndexing.Exact
• FieldIndexing.Search
• FieldIndexing.No

Setting the FieldIndexing option for this field to Exact turns off the field analysis. This causes all the properties to be treated as a single token and the matches must be exact (case sensitive), using the KeywordAnalyzer behind the scenes.

public class Employees_ByFirstAndLastName : AbstractIndexCreationTask<Employee>
{
    public Employees_ByFirstAndLastName()
    {
        Map = employees => from employee in employees
                           select new
                           {
                               LastName = employee.LastName,
                               FirstName = employee.FirstName
                           };

        Indexes.Add(x => x.FirstName, FieldIndexing.Exact);
    }
}

FieldIndexing.Search allows performing full-text search operations against the field using the StandardAnalyzer by default:

public class BlogPosts_ByContent : AbstractIndexCreationTask<BlogPost>
{
    public BlogPosts_ByContent()
    {
        Map = posts => from post in posts
                       select new
                       {
                           Title = post.Title,
                           Content = post.Content
                       };

        Indexes.Add(x => x.Content, FieldIndexing.Search);
    }
}

If you want to disable indexing on a particular field, use the FieldIndexing.No option. This can be useful when you want to store field data in the index, but don't want to make it available for querying. However, it will still be available for extraction by projections:

public class BlogPosts_ByTitle : AbstractIndexCreationTask<BlogPost>
{
    public BlogPosts_ByTitle()
    {
        Map = posts => from post in posts
                       select new
                       {
                           Title = post.Title,
                           Content = post.Content
                       };

        Indexes.Add(x => x.Content, FieldIndexing.No);
        Stores.Add(x => x.Content, FieldStorage.Yes);
    }
}

When a field is marked as Search, sorting must be done using an additional field. More here.