You Decide What Happens

Every search result is an action waiting to trigger. When you press Enter on a result, afwf_fts_anything can:

• Open a URL — jump straight to a web page (IMDB movie page, AWS docs, GitHub issue, anything with a link).

• Open a file — open a local file in its default application, or reveal it in Finder.

You configure the action once in the Alfred Script Filter:

# open a URL on Enter
~/.local/bin/uvx --from "afwf-fts-anything==2.0.2" afwf-fts-anything fts \
    --dataset-name 'movie' --query '{query}' --action open_url

# open a local file on Enter
~/.local/bin/uvx --from "afwf-fts-anything==2.0.2" afwf-fts-anything fts \
    --dataset-name 'movie' --query '{query}' --action open_file

No installation, no virtual environment. uvx handles everything.

Your Data, Your Schema

The dataset is a plain JSON file — a list of objects, any fields you want:

[
    {
        "movie_id": 1,
        "title": "The Shawshank Redemption",
        "description": "Two imprisoned men bond over a number of years, finding solace and eventual redemption through acts of common decency.",
        "genres": "Drama",
        "rating": 9.2,
        "url": "https://www.imdb.com/title/tt0111161"
    },
    {
        "movie_id": 2,
        "title": "The Godfather",
        "description": "The aging patriarch of an organized crime dynasty transfers control of his clandestine empire to his reluctant son.",
        "genres": "Crime, Drama",
        "rating": 9.2,
        "url": "https://www.imdb.com/title/tt0068646"
    },
    ...
]

It could be movies, bookmarks, API references, Terraform resources, local files — anything you can put in a JSON array.

Your Config, Your Search Behavior

A single JSON setting file controls how fields are indexed, how results are sorted, and what Alfred displays:

{
    "fields": [
        // store only — not searchable, but available in display templates
        {"type": "stored",  "name": "movie_id"},
        // n-gram: typing "god" already matches "godfather"
        {"type": "ngram",   "name": "title",       "min_gram": 2, "max_gram": 10, "boost": 2.0},
        // full-word phrase search on description
        {"type": "text",    "name": "description"},
        // full-word text search on genres with a relevance boost
        {"type": "text",    "name": "genres",       "boost": 1.5},
        // numeric field, sortable by rating descending
        {"type": "numeric", "name": "rating",       "kind": "f64", "indexed": true, "fast": true},
        // stored for use as the action argument (URL or file path)
        {"type": "stored",  "name": "url"}
    ],
    // default sort order: highest rated first
    "sort": [{"name": "rating", "descending": true}],
    // optional: auto-download data on first run / rebuild
    "data_url": "https://github.com/MacHu-GWU/afwf_fts_anything-project/releases/download/1.1.1/movie-data.json.zip",
    // Alfred display templates — {field_name} is replaced per result
    "title_field":        "{title} ({genres}) rate {rating}",
    "subtitle_field":     "{description}",
    "arg_field":          "{url}",
    "autocomplete_field": "{title}",
    "icon_field":         "movie-icon.png"
}

Comments (//) are supported and stripped automatically.

Alfred Owns the UI

There is no custom UI code to write. Everything visible in Alfred — the dropdown list, keyboard navigation, icons, subtitles, clipboard copy (CMD+C), tab-autocomplete — is standard Alfred behavior. You wire it together with one uvx command in a Script Filter:

~/.local/bin/uvx --from "afwf-fts-anything==2.0.2" afwf-fts-anything fts --dataset-name 'movie' --action 'open_url' --query {query}

The only thing you touch is the Script field. Everything else is Alfred.

Also a Python Library

afwf_fts_anything is built on sayt2 (powered by Tantivy, written in Rust) and works as a standalone Python full-text search library with no Alfred dependency. Build an index from any JSON dataset and query it directly:

from pathlib import Path
from afwf_fts_anything.api import DataCatalog

# point to the directory that holds your dataset folders
catalog = DataCatalog(
    dir_root=Path("~/.alfred-afwf/afwf_fts_anything").expanduser()
)

# get a dataset by name — reads movie-setting.json automatically
dataset = catalog.get_dataset("movie")

# build the index on first use (skipped if already built)
dataset.build_index()

# search and get plain dicts back
results = dataset.search("godfather", limit=10)
for doc in results:
    print(doc["title"], doc["rating"])

Documentation

Full documentation — Quick Start, Setting File Reference, Alfred Workflow Setup, and a step-by-step guide to building your own dataset.

Projects Built on afwf_fts_anything

• AWS CloudFormation Resource & Property Reference — jump to official CloudFormation docs

• Terraform AWS Resource Reference — jump to official Terraform docs

• AWS Python Boto3 API Reference — jump to Boto3 service/method docs