What is searchcraft?

SearchCraft is a search engine written in Python from scratch, built as a learning project rather than a production tool. The author's stated goal is to understand how search engines actually work internally by writing the core pieces themselves instead of using an existing search library that would hide those details. The project is explicitly a work in progress, released in small pieces as each part of the search pipeline gets built. The current version includes the basics needed for a simple search tool: it can load documents, break their text into individual words through a process called tokenization, build what is called an inverted index, which is a data structure that maps each word to the documents it appears in, and run basic keyword searches. It also comes with a command line interface, so you can build the index and run searches directly from a terminal. To use it, you install the package with pip, then run a command to build the search index over your documents, followed by a search command with the words you are looking for. Right now the search is limited to straightforward keyword matching. Looking ahead, the roadmap lists several features not yet built, including posting lists, which are a more efficient way of storing where words appear, search result snippets, the ability to search for exact phrases, ranking methods such as TF-IDF or BM25 that would sort results by relevance rather than just matching keywords, fuzzy search for handling typos, and eventually a web interface so it does not need to be run from the command line. This project is best suited for someone who wants to learn how search engines work by reading and following along with a real implementation, not someone looking for a ready made search tool for their own application. It is released under the GPL 3.0 license, a copyleft license requiring that projects built on top of it also share their source code.