lucaong/minisearch

Features

• Trie-Based Inverted Indexes - Stores term-to-document mappings in a compact in-memory trie for fast full-text lookups without external dependencies.
• Client-Side Search Engines - Adds fast, offline search to web applications by indexing documents in the browser with exact, fuzzy, or prefix matching.
• Auto-Complete Libraries - Generates ranked query completions from the index as the user types, supporting fuzzy suggestions.
• Dynamic Document Operations - Inserts or deletes documents from the index at any time without rebuilding the entire index.
• Full Text Search - Indexes documents in memory and returns ranked results for full-text queries with multiple matching modes.
• Fuzzy Matching - Finds documents matching a query even when terms are misspelled using configurable edit distance.
• BM25 Full-Text Indices - An in-memory search library that indexes documents and ranks results using BM25+ relevance scoring with fuzzy and prefix matching.
• Fuzzy Trie Searches - Performs fuzzy and prefix searches by traversing a trie with Levenshtein-aware branching for configurable edit distance.
• Fuzzy Search Engines - Tolerates typos and misspellings by allowing configurable edit distance between query terms and indexed terms.
• In-Memory Indexing - Indexes documents in memory and makes them searchable by textual content.
• Search Suggestions - Generates ranked query completions from the index as the user types, supporting fuzzy suggestions.
• Microtask-Based Batch Indexing - Ships microtask-scheduled batch indexing that keeps the UI responsive during bulk document inserts.
• Relevance Scoring - Scores and sorts search results using the BM25+ algorithm for relevance.
• BM25+ Scoring - Implements BM25+ relevance scoring with term frequency, inverse document frequency, and field-length normalization.
• Incremental Updates - Supports incremental document addition, removal, and replacement with automatic cleanup of stale index entries.
• Customizable Pipelines - Accepts custom tokenizers and term processors for controlling text splitting and normalization during indexing and search.
• Customizable Tokenizers - Allows replacing the default tokenizer with custom logic for splitting text into tokens during indexing and search.
• Separate Indexing and Search Pipelines - Controls tokenization and term processing with separate rules for indexing and searching.
• Search Query Composition - Provides boolean query composition with AND, OR, and AND_NOT operators for building complex search expressions.
• Document Indexing - Deletes or updates documents in the index so only the current version appears in search results.
• Document Lifecycle Management - Discards or replaces documents by ID with automatic index cleanup.
• Field Value Stores - Keeps arbitrary field values alongside indexed terms and retrieves them by document ID without re-indexing.
• Index Serialization - Exports the full search index to JSON for storage and later restoration without re-indexing.
• Custom Preprocessing Functions - Accepts user-provided functions for stemming, stop-word removal, or other text preprocessing instead of imposing a built-in locale.
• Multi-Field Indexing - Stores multiple fields per document in a single compact index using short numeric IDs.
• Field Value Extraction - Extracts field values from documents using custom functions, supporting nested fields and non-string data.
• Index Serialization - Exports the full index to a JSON string for storage or transfer and rebuilds it later.
• JSON Export and Restore - Exports the full index to JSON for persistence and restores it later without re-indexing documents.
• Prefix Searches - Finds all indexed terms that start with a given prefix for autocomplete and type-ahead.
• Serializable Indexes - A search index that can be exported to JSON for storage and later restored to resume searching without re-indexing.
• Logical Search Operators - Joins multiple search terms using AND, OR, and AND_NOT operators to refine result sets.
• Custom Term Processors - Applies custom normalization, filtering, or stemming to terms during indexing and search via a configurable processor function.
• Exact Term Matches - Looks up indexed terms that exactly match the search query, returning associated documents.
• Custom Scoring Multipliers - Applies custom scoring multipliers to specific query terms or documents to influence ranking.
• Search Result Filtering - Restricts search results to documents satisfying custom predicates like date ranges or categories.
• Fuzzy Query Suggestions - Suggests complete search queries from partial input using fuzzy matching and relevance ranking.
• Ranked Query Completions - Generates ranked query completions from partial input to guide users toward valid searches.
• Asynchronous Indexing - Adds many documents to the index in chunks without blocking the main thread.
• Combinator Logic - Builds complex search expressions by combining subqueries with AND, OR, and AND_NOT operators.
• Relevance Score Boosting - Assigns higher weight to matches in specified document fields to influence result ranking.
• Databases and Search - Tiny full-text search engine for browser and Node.

Star history