EF Core Implementing Fuzzy Search

Fuzzy search allows finding strings that match the pattern approximately. It is a feature needed in almost every app, but it can be a little problematic to implement. We will focus on EFCore PostgreSql provider as it is free and seems to be the most popular choice nowadays (for me at least).

First, our EFCore project needs this package: PostgreSQL.FuzzyStringMatch

Then, we need to register it with our DbContext, so let’s navigate to Startup.cs and edit our registration like so:

public void ConfigureServices(IServiceCollection services)
{
    services
        .AddEntityFrameworkNpgsql()
        .AddDbContext<ApplicationDbContext>(opt =>
            opt.UseNpgsql(Configuration.GetConnectionString(
                "DefaultConnection"), 
            n => n.UseFuzzyStringMatch()));

    // ...
}

After that, we need to create a new database migration with our new extension. In ApplicationDbContext.cs edit OnModelCreating method and register the extension:

protected override void OnModelCreating(ModelBuilder modelBuilder)
{
    base.OnModelCreating(modelBuilder);

    modelBuilder.HasPostgresExtension("fuzzystrmatch");

    // ...
}

With that out of the way, we can create new migration as usual with Package Manager Console:

PM> add-migration AddFuzzySearchExtension
PM> update-database

That’s it! We can now use EF.Functions to query our database, just like on the example:

public IQueryable<Example> GetAllWithMatchingName(string name)
{
    return _dbSet
        .Where(_ => EF.Functions.FuzzyStringMatchLevenshtein(
            _.Name, name) <= 5)
        .OrderBy(_ => EF.Functions.FuzzyStringMatchLevenshtein(
            _.Name, name));
}

This code will output an IQueryable of our Example entity with a matching name, ordered from the most similar to the least. The method also accepts a threshold acting as a cutoff point for matching, so that it won’t return the whole database for you. You will need to experiment with the values and of course, there are different types of functions to perform the search, but I’ll leave the research up to you as it might depend on your case.

With that you are ready to create a better search bar for your users!