Bleve

Bleve is a full-text search and indexing library for Go, developed by Couchbase (Marty Schoch and team). With over 10,300 stars as of 2026, Bleve provides modern text search capabilities (similar to Lucene/Elasticsearch) as an embeddable Go library, eliminating the need to run a separate search server. Key features include: full-text indexing and search with configurable analyzers (standard, keyword, simple, web, whitespace, unicode, and custom), text analysis pipeline with character filters (mapping, regexp replace), tokenizer (unicode word segmentation, single token, letter, edge N-gram for autocomplete), token filters (lowercase, stop words for 30+ languages, stemmer for 20+ languages via snowball, porter, elision, length, truncate, trim, unique, ngram, shingle, apostrophe, camel case, keyword marker, and categorical stemmer), BM25 scoring algorithm (default, with configurable parameters b and k1), query DSL supporting boolean queries (must, should, must_not), match queries, phrase queries, prefix queries, regexp queries, wildcard queries, numeric range queries, date range queries, fuzzy queries (edit-distance-based), and query string parsing, faceting (terms, numeric, date histograms), search result highlighting (with configurable fragment size and tags), field sorting, search after (pagination via sort values), suggestions and spell correction (via the blevesuggest package), geospatial queries (bounding box, distance), index storage backends (boltdb, leveldb, badger, scorch for memory-optimized index), and Kubernetes deployment support via the Couchbase Full-Text Search service. Bleve powers search in Couchbase Server, Mattermost, and Caddy.