Full-Text Search

Build powerful search features with native full-text search, ranking, highlighting, and fuzzy matching.

Overview

Prax provides a unified API for full-text search across all supported databases, from PostgreSQL's powerful tsvector to MongoDB Atlas Search.

Feature PostgreSQL MySQL SQLite MSSQL MongoDB
Full-Text Index ✅ tsvector/GIN ✅ FULLTEXT ✅ FTS5 ✅ Full-Text ✅ Atlas Search
Relevance Ranking ✅ ts_rank ✅ ✅ bm25 ✅ RANK ✅ score
Highlighting ✅ ts_headline ❌ ✅ ❌ ✅
Fuzzy Search ✅ pg_trgm ❌ ❌ ✅ SOUNDEX ✅
Faceted Search ✅ ❌ ❌ ❌ ✅

Basic Search

Use SearchQuery for simple text search.

Relevance Ranking

Weight columns differently and sort by relevance score.

Search Highlighting

Show users where their search terms matched in the results.

Fuzzy Search

Handle typos and spelling variations with fuzzy matching.

Phrase & Boolean Search

Search for exact phrases or use boolean operators.

Creating Search Indexes

Full-text search requires proper indexes for performance.

MongoDB Atlas Search

Powerful full-text search with Atlas Search, including facets, fuzzy matching, and scoring.

Note: Atlas Search requires MongoDB Atlas (not available on self-hosted). Create indexes in the Atlas UI or via the Admin API.

Schema & Migrations

Define full-text indexes in your schema and migrations.

Best Practices

Use GIN Indexes (PostgreSQL)

GIN indexes are faster for lookups but slower to update. Use GiST indexes if you have frequent writes and can tolerate slower searches.

Pre-compute tsvector

Store a computed tsvector column instead of generating it at query time. Update it via triggers when the source columns change.

Limit Result Sets

Full-text search can return many results. Always use LIMIT and implement pagination to avoid performance issues.

Consider External Search

For complex search requirements, consider Elasticsearch, Meilisearch, or Typesense. Database full-text search is great for basic to moderate needs.