Fuzzy matching for Emacs ... a la Sublime Text.
This project is more than a year old now. Lots of bugs have been worked out.
It appears some people use it on a regular basis.
Just drop all
.elfiles somewhere on your
load-path. Here's an example using the folder
(add-to-list 'load-path "~/emacs.d/vendor")
flx- matching engine
flx-ido- ido interface for flx
flx-idowill pull in
flxas a dependency.
If you're an Emacs 24 user or you have a recent version of
package.elyou can install
This version will always be up-to-date.
flx-idois also available on the Marmalade
flx-idois part of the Emacs Prelude. If you're a Prelude user -
flx-idois already properly configured and ready for action.
Users of Debian 9 or Ubuntu 16.04 or later may simply
apt-get install elpa-flx.
The sorting algorithm is a balance between word beginnings (abbreviation) and contiguous matches (substring).
The longer the substring match, the higher it scores. This maps well to how we think about matching.
In general, it's better form queries with only lowercase characters so the sorting algorithm can do something smart.
For example, if you have these files:
If the search term was pre-mode, you might expect "prelude-mode.el" to rank higher. However because the substring match "re-mode" is so long, "clojure-mode.el" actually scores higher.
Here, using premode would give the expected order. Notice that the "-" actually prevents the algorithm from helping you.
Flx always folds lowercase letters to match uppercase. However, you can use uppercase letters for force flx to only match uppercase.
This is similar to Emacs' case-folding. The difference is mixing in uppercase letters does not disable folding.
Matches within the basepath score higher.
Add this to your init file and flx match will be enabled for ido.
(require 'flx-ido) (ido-mode 1) (ido-everywhere 1) (flx-ido-mode 1) ;; disable ido faces to see flx highlights. (setq ido-enable-flex-matching t) (setq ido-use-faces nil)
If you don't want to use the
flx's highlights you can turn them off like this:
(setq flx-ido-use-faces nil)
flx-ido-thresholdto change the collection size above which flx will revert to flex matching.
As soon as the collection is narrowed below
flx-ido-threshold, flx will kick in again.
As a point of reference for a 2.3 GHz quad-core i7 processor, a value of
10000still provides a reasonable completion experience.
Helm is not supported yet. There is a demo showing how it could work, but I'm still working through how to integrate it into helm.
The Helm demo shows the score of the top 20 matches.
flxalgorithm willingly sacrifices memory usage for speed.
For 10k file names, about 10 MB of memory will be used to speed up future matching. This memory is never released to keep the match speed fast.
So far with modern computers, this feels like a reasonable design decision.
It may change in future.
Emacs's garbage collector is fairly primitive stop the world type. GC time can contribute significantly to the run-time of computation that allocates and frees a lot of memory.
Consider the following example:
(defun uuid () (format "%08x-%08x-%08x-%08x" (random (expt 16 4)) (random (expt 16 4)) (random (expt 16 4)) (random (expt 16 4))))
(benchmark-run 1 (let ((cache (flx-make-filename-cache))) (dolist (i (number-sequence 0 10000)) (flx-process-cache (uuid) cache)))) ;;; ⇒ (0.899678 9 0.33650300000000044)
This means that roughly 30% of time is spent just doing garbage-collection.
flxcan benefit significantly from garbage collection tuning.
By default Emacs will initiate GC every 0.76 MB allocated (
gc-cons-threshold== 800000). If we increase this to 20 MB (
gc-cons-threshold== 20000000) we get:
(benchmark-run 1 (setq gc-cons-threshold 20000000) (let ((cache (flx-make-filename-cache))) (dolist (i (number-sequence 0 10000)) (flx-process-cache (uuid) cache)))) ;;; ⇒ (0.62035 1 0.05461100000000041)
So if you have a modern machine, I encourage you to add the following:
(setq gc-cons-threshold 20000000)
to your init file.