Skip to main content

Changelog

All notable changes to reclink will be documented here.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

0.1.1​

Fixed​

  • PyPI sdist now includes LICENSE file at package root
  • PyPI README logo renders correctly (absolute URL)
  • Type stubs: Scorer and CompositePreset accept str for custom metric support
  • register_metric has proper @overload signatures for all calling conventions
  • Clippy and ruff lint errors resolved across codebase
  • Mypy runs on test files with correct --python-executable for numpy resolution
  • WASM playground loads correctly under /reclink/ base path
  • Internal doc links updated for routeBasePath: '/'

Added​

  • Pre-commit hooks: cargo fmt, clippy, ruff check, ruff format, mypy
  • CI workflow with lint, rust-test, and cross-platform Python test matrix
  • GitHub Pages docs workflow with WASM build
  • PyPI release workflow with trusted publishing (OIDC) for 6 platform targets
  • PR cache cleanup workflow
  • Changelog page in docs site
  • PyPI metadata: badges, classifiers, keywords, project URLs

Changed​

  • Docs served at docs.byteveda.org/reclink/ via org-level GitHub Pages
  • Removed /docs prefix from doc routes (redundant with docs. subdomain)
  • Updated crate dependencies: unicode-normalization 0.1.25, ahash 0.8.12, thiserror 2.0, regex 1.12, memmap2 0.9.10, crossbeam-channel 0.5.15, wasm-bindgen 0.2.114, serde-wasm-bindgen 0.6.5, js-sys 0.3.91
  • Apache-2.0 license consistent across all crates

0.1.0​

Initial release.

Added​

  • 21 string metrics: Levenshtein, Damerau-Levenshtein, Hamming, Jaro, Jaro-Winkler, Cosine, Jaccard, Sorensen-Dice, Weighted Levenshtein, Token Sort, Token Set, Partial Ratio, LCS, Longest Common Substring, N-gram Similarity, Smith-Waterman, Phonetic Hybrid, Ratcliff-Obershelp, Needleman-Wunsch, Gotoh, Monge-Elkan
  • 10 phonetic algorithms: Soundex, Metaphone, Double Metaphone, NYSIIS, Caverphone, Cologne Phonetic, Beider-Morse, Phonex, MRA, Daitch-Mokotoff
  • 11 blocking strategies: Exact, Phonetic, Sorted Neighborhood, Trie, Q-gram, Numeric, Date, LSH, Canopy, Hybrid (union/intersection modes), Custom
  • 8 classifiers: Threshold, Weighted, Threshold Bands, Weighted Bands, Fellegi-Sunter, Fellegi-Sunter Auto (EM), Logistic Regression, Decision Tree
  • 5 clustering algorithms: Connected Components, Hierarchical, DBSCAN, OPTICS, Incremental
  • 7 index structures: BK-tree, VP-tree, N-gram Index, Memory-mapped N-gram Index, MinHash LSH, Bloom Filter, Inverted Index
  • Full record linkage pipeline with builder pattern: preprocessing, blocking, comparison, classification, clustering
  • Batch operations: cdist, match_best, match_batch with Rayon parallelism
  • Composite scoring with presets (name matching, address matching, general purpose)
  • Streaming matchers with backpressure support
  • Custom metric/blocker/comparator/classifier registration with decorator syntax
  • Pandas and Polars integration: DataFrame accessors, fuzzy join, Polars expression plugin
  • Text preprocessing: case folding, whitespace normalization, diacritics stripping, stop word removal, abbreviation expansion, domain-specific cleaners (name, address, company, email, URL), transliteration (Cyrillic, Greek, Arabic, Hebrew, Devanagari, Hangul)
  • Arrow-compatible batch operations for zero-copy interop
  • TF-IDF matcher for corpus-aware similarity
  • EM parameter estimation for Fellegi-Sunter models
  • Evaluation utilities: precision, recall, F1, confusion matrix
  • Export: CSV, JSON, NetworkX graph
  • CLI: reclink compare, reclink batch, reclink pipeline
  • WASM bindings for browser-based playground
  • Interactive docs site with live WASM-powered playground