Release and Version History¶

Backlog¶

Features and Improvements

Minor Improvements

Bugfixes

Miscellaneous

Bugfixes

fts() now returns a descriptive error Alfred item instead of raising an exception when the dataset setting file is missing, when index building fails, or when the search itself fails. The error item opens the setting file in Finder so the user can diagnose and fix the configuration directly from Alfred.

Features and Improvements

Replaced the whoosh search backend with sayt2, delivering faster indexing and more robust full-text search.
Introduced DataCatalog — a directory-backed registry that discovers and manages multiple named datasets under a single root directory.
Added ActionEnum (open_url / open_file) so each dataset can specify whether pressing Enter opens a URL or a local file path.
Added setup-sample-data CLI command that copies the bundled movie dataset into the workflow home, enabling a zero-configuration quick-start experience.
Added list-datasets-for-reset CLI command that lists all valid datasets with fuzzy filtering so users can rebuild a specific dataset index directly from Alfred.
Added rebuild-index CLI command to destroy and recreate the search index for a dataset, with optional re-download when data_url is configured.
The ? query now reveals the dataset’s setting file in Finder instead of returning search results.

Breaking Changes

Dataset is now a dataclass with a dir_root parameter; the old path_setting / path_data / dir_index / dir_icon attribute overrides are removed. Each dataset is expected to live under its own subdirectory ({root}/{name}/).
Setting has been rewritten using Pydantic and sayt2 field types; the old whoosh-based schema helpers (create_whoosh_schema, searchable_fields, sortable_fields) are gone.
Removed the diskcache-based cache.py module and the global cache object; query caching is now handled internally by sayt2.
Removed compat.py; functools.cached_property (Python 3.8+) is used directly.
Removed the requests dependency; data downloads now use the standard library urllib.

Miscellaneous

Migrated build and dependency management from poetry to uv.
Bundled the movie sample dataset inside the package (afwf_fts_anything/tests/data/movie/) so it is always available without a network call.
Restructured documentation into topic-based pages covering quick-start, data catalog, setting/data files, Alfred workflow setup, and building custom datasets.

Features and Improvements

Add ngram_words index type. It tokenizes the text into words before index. It is more accurate than ngram if you have delimiter in your text.

Minor Improvements

Features and Improvements

Minor Improvements

Miscellaneous

Features and Improvements

Features and Improvements