LIFTI

A lightweight full text indexer for .NET

Documentation

Read the documentation - there's lots of useful information and examples there.

Check out some sample code - the repo contains examples that can be run as a console application.

Try it out!

Use LIFTI in a Blazor app - try out various queries against Wikipedia content

Simplest possible quick starts

Install-Package Lifti.Core

Indexing loose text

Indexing objects

Contribute

It would be great to have more people contributing to LIFTI - how can you help?

• Create issues for bugs you find - level 1
• Create feature suggestions - level 2
• Create pull requests for documentation changes - level 3
• Create pull requests for bug fixes or features - boss level

Goals

1. Provide a very simple way to build an in-memory full text index.
2. netstandard2 compatibility
3. Persistance via explicit serialization/deserialization points. Automatic persistance to a backing store upon changes to the index is out of scope.

Non-goals

1. Distributed transactions are out of scope and don't make much sense anyway given automatic persistance is not in scope. (the old version did support this but I don't think anyone actually used it though - I know didn't :))

Significant differences between new and old implementations

The old implementation followed a very simple tree structure where every node in the tree linked to its descendants via relevant chars, e.g.:

A
╚P
 ╠E (matches APE)
 ╚P
  ╚L
   ╚E (matches APPLE)

The new implementation expands upon this by allowing a node to also contain a series of "intra-node" characters - this reduces the number of nodes in the tree, reducing the overall memory footprint and allows for potential gains in search times. (less object traversal required):

AP
 ╠E (matches APE)
 ╚PLE (matches APPLE)