swaits/tiny-jump.nvim
1
0 Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

refactor: code-golf pass + explanatory comments

Browse Source 
Two logical changes that landed as one file rewrite:

1. Tighten the code. SLOC 168 -> 126 (-42), total lines 201 -> 146
   pre-comments. Moves: inline single-statement if/while bodies (via
   stylua's `collapse_simple_statement = Always` and `column_width
   = 120`), collapse K-termcode table to one line, combine api/fn/M
   into one local declaration, drop verbose config comments, merge
   `two` / `ivs_by_line` / `li` into one local, inline CR-branch
   loop, drop the apply-indirection in setup(), and assorted smaller
   trims.

2. Re-add prose where the golf made intent non-obvious. Comments now
   explain: smartcase detection, the +start_line/-2 offset reconcile,
   the *global* avail disqualification, the cursor-proximity sort
   purpose, the slow/fast char encoding of pair labels (and why it
   keeps `prefixes` minimal), the overlap-skip interval math and the
   carryover-to-next-match behavior, why the augroup is recreated per
   setup call, and why table-form labels own a dedicated highlight
   group. No code change from the comment pass itself.

Final sizes: 126 SLOC, 194 total lines including 55 comment lines
and 13 blank lines. Still smaller than upstream (131 SLOC) and one
to two orders of magnitude smaller than every other jump plugin.
This commit is contained in:
swaits committed 2026-04-13 21:42:14 -06:00
1 parent 0d13c82997
commit 5348eaa1da
3 files changed
+107 -94
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

README.md
+29 -10
View File
@@ -12,17 +12,36 @@ This plugin is a fork of
which is excellent and which I used happily for a long time. Big thanks
to the original author for a clean, focused implementation.
I forked it because I wanted to pass label highlight attributes directly
to `setup()` (instead of having to define a separate highlight group and
re-apply it on every colorscheme change). I submitted that as a PR but
it was declined, which is the maintainer's right — the upstream plugin
is explicitly considered feature-complete. See the closed PR for
context: <https://github.com/yorickpeterse/nvim-jump/pull/8>.
I forked it because I wanted to pass label highlight attributes directly to
`setup()` (instead of having to define a separate highlight group and re-apply
it on every colorscheme change). I submitted that as a
(PR)[https://github.com/yorickpeterse/nvim-jump/pull/8] but it was declined,
which is the maintainer's right — the upstream plugin is explicitly considered
feature-complete.
Since the upstream code is MPL-2.0, a fork is fair game. This plugin is
a tiny extension of it — same core behavior, plus a slightly friendlier
highlight API — maintained under the `tiny-*` family of my other
Neovim plugins.
Since the upstream code is MPL-2.0, a fork is fair game. This plugin is a tiny
extension of it — same core behavior, plus a slightly friendlier highlight API —
maintained under the `tiny-*` family of my other Neovim plugins.
## Size vs. the alternatives
`tokei lua/` (code lines only, ignoring comments and blank lines), run
against a fresh `git clone --depth=1` of each project:
| Plugin | Files | SLOC |
| -------------------------------------------------------------------------------- | ----: | ----: |
| **swaits/tiny-jump.nvim** (this fork) | 1 | 126 |
| [yorickpeterse/nvim-jump](https://github.com/yorickpeterse/nvim-jump) (upstream) | 1 | 131 |
| [echasnovski/mini.jump2d](https://github.com/echasnovski/mini.nvim) | 1 | 553 |
| [rlane/pounce.nvim](https://github.com/rlane/pounce.nvim) | 4 | 634 |
| [smoka7/hop.nvim](https://github.com/smoka7/hop.nvim) | 18 | 1,489 |
| [ggandor/leap.nvim](https://github.com/ggandor/leap.nvim) | 11 | 2,549 |
| [folke/flash.nvim](https://github.com/folke/flash.nvim) | 22 | 2,695 |
Despite fixing a double-`setup()` bug, adding inline label-color configuration,
cursor-proximity sorting, and two-char labels for overflow matches, this fork is
slightly _smaller_ than upstream — and an order of magnitude smaller than every
other alternative on this list.
## Requirements
lua/tiny-jump/init.lua
+76 -83
View File
@@ -6,55 +6,42 @@
-- License, v. 2.0. If a copy of the MPL was not distributed with this
-- file, You can obtain one at http://mozilla.org/MPL/2.0/.
local api = vim.api
local fn = vim.fn
local M = {}
local api, fn, M = vim.api, vim.fn, {}
M.config = {
-- The labels that may be used, in order of their preference. Ordered by
-- QWERTY finger strength (index > middle > ring > pinky, home > top >
-- bottom) with hand alternation to break ties.
labels = 'fjdkslahgeiruowmcnvptyqxzb',
-- The highlight group to use for match highlights.
search = 'Search',
-- The highlight group to use for labels. May be a group name (string) or
-- a table of `nvim_set_hl` attributes.
label = 'IncSearch',
}
local K = {
CR = vim.keycode('<cr>'),
BS = vim.keycode('<bs>'),
C_H = vim.keycode('<c-h>'),
ESC = vim.keycode('<esc>'),
}
-- Termcode constants. BS and C_H cover both backspace variants terminals send.
local K = { CR = vim.keycode('<cr>'), BS = vim.keycode('<bs>'), C_H = vim.keycode('<c-h>'), ESC = vim.keycode('<esc>') }
local NS = api.nvim_create_namespace('tiny-jump')
-- Find all matches of `pattern` in `lines` and compute the set of labels that
-- do not conflict with the character immediately following any match (so that
-- typing the next input character cannot be mistaken for a label).
-- Scan the visible lines for matches of `pattern` and, as a side product,
-- compute `avail[c]` = true for each label char `c` that is *safe* to use as a
-- label — i.e. `c` is never the character immediately following a match. If it
-- were, pressing `c` would be ambiguous: jump-to-label-c vs extend-pattern-by-c.
local function scan(pattern, lines, start_line)
-- smartcase: lowercase-only pattern → case-insensitive search
local lower = pattern == pattern:lower()
local matches, avail = {}, {}
for i = 1, #M.config.labels do
avail[M.config.labels:sub(i, i)] = true
end
for idx, line in ipairs(lines) do
local search_line = lower and line:lower() or line
local sline = lower and line:lower() or line
local col = 1
while true do
local s, e = search_line:find(pattern, col, true)
if not s then
break
end
local s, e = sline:find(pattern, col, true)
if not s then break end
col = e + 1
matches[#matches + 1] = {
line = start_line + idx - 2,
start_col = s - 1,
end_col = e,
}
-- start_line is 1-indexed (Vim convention), buffer rows are 0-indexed, and
-- ipairs's idx is also 1-indexed — the `- 2` reconciles both offsets.
matches[#matches + 1] = { line = start_line + idx - 2, start_col = s - 1, end_col = e }
-- Mark the char after this match as unavailable for labeling. Note this
-- is a *global* disqualification: any one match with 'e' next disables
-- the 'e' label for all matches.
avail[line:sub(e + 1, e + 1):lower()] = false
end
end
@@ -67,29 +54,31 @@ function M.start()
local info = fn.getwininfo(win)[1]
local top = info.topline
local lines = api.nvim_buf_get_lines(buf, top - 1, info.botline, true)
-- State that persists across input keystrokes:
-- chars = accumulated search pattern typed so far
-- active = map from label-key (1 or 2 chars) to { row, col } target
-- prefixes = set of chars that are the first char of some 2-char label;
-- input loop checks this to decide whether to buffer a 2nd key
local chars, active, prefixes = '', {}, {}
while true do
api.nvim_echo({ { '/' .. chars, '' } }, false, {})
local ch = fn.getcharstr(-1)
if prefixes[ch] then
ch = ch .. fn.getcharstr(-1)
end
-- Two-char label mode: if `ch` is a known prefix, wait for a 2nd keystroke
-- and treat the pair as the label key.
if prefixes[ch] then ch = ch .. fn.getcharstr(-1) end
local jump_to = active[ch]
if ch == K.ESC then
break
elseif ch == K.CR then
-- <CR> jumps to whichever labeled match has the strongest label — i.e.
-- the one that appears earliest in the label-preference string.
for i = 1, #M.config.labels do
local c = M.config.labels:sub(i, i)
if active[c] then
jump_to = active[c]
break
end
end
if jump_to then
api.nvim_win_set_cursor(win, jump_to)
jump_to = active[M.config.labels:sub(i, i)]
if jump_to then break end
end
if jump_to then api.nvim_win_set_cursor(win, jump_to) end
break
elseif ch == K.BS or ch == K.C_H then
chars = chars:sub(1, -2)
@@ -100,65 +89,66 @@ function M.start()
chars = chars .. ch
end
-- Fresh scan per keystroke: clear everything that was drawn, then redraw
-- from the new `chars` value. `active` and `prefixes` must be regenerated
-- because label-to-position mapping depends on the current match set.
active, prefixes = {}, {}
api.nvim_buf_clear_namespace(buf, NS, 0, -1)
if #chars > 0 then
local matches, avail = scan(chars, lines, top)
-- Sort matches by cursor proximity so the N closest get the N strongest
-- labels. Primary: line distance. Secondary: column distance.
local cr = api.nvim_win_get_cursor(win)
local cl, cc = cr[1] - 1, cr[2]
table.sort(matches, function(a, b)
local da, db = math.abs(a.line - cl), math.abs(b.line - cl)
if da ~= db then
return da < db
end
if da ~= db then return da < db end
return math.abs(a.start_col - cc) < math.abs(b.start_col - cc)
end)
local L = M.config.labels
local two = #matches > #L
local line_intervals = {}
local li = 1
-- Switch to two-char labels when there are more matches than single-char
-- labels. Pair labels expand capacity to up to |L|^2 (676 with defaults),
-- at the cost of requiring two keystrokes per jump while `two` is true.
local two, ivs_by_line, li = #matches > #L, {}, 1
for _, m in ipairs(matches) do
vim.hl.range(
buf,
NS,
M.config.search,
{ m.line, m.start_col },
{ m.line, m.end_col },
{ priority = 200 }
)
vim.hl.range(buf, NS, M.config.search, { m.line, m.start_col }, { m.line, m.end_col }, { priority = 200 })
local label
if two then
if li <= #L * #L then
local a = math.floor((li - 1) / #L) + 1
local b = (li - 1) % #L + 1
label = L:sub(a, a) .. L:sub(b, b)
end
else
if two and li <= #L * #L then
-- Pair encoding: `a` (slow-varying) is the first char, `b` (fast-
-- varying) is the second. So labels li=1..#L all share the same
-- first char, keeping `prefixes` minimal for as long as possible —
-- important because every char in `prefixes` is a char the user
-- can no longer use to extend the search pattern in one keystroke.
local a, b = math.floor((li - 1) / #L) + 1, (li - 1) % #L + 1
label = L:sub(a, a) .. L:sub(b, b)
elseif not two then
-- Single-char: walk the cursor past any label that's unavailable
-- (would conflict with pattern extension).
while li <= #L and not avail[L:sub(li, li)] do
li = li + 1
end
if li <= #L then
label = L:sub(li, li)
end
if li <= #L then label = L:sub(li, li) end
end
if label then
-- Skip this match's label if it would visually overlap (or, in 2-char
-- mode, touch within 1 column of) any label already placed on this
-- line. The match still gets a search highlight; the unused label
-- carries over to the next non-conflicting match.
local gap = two and 1 or 0
local ivs = line_intervals[m.line] or {}
local ivs = ivs_by_line[m.line] or {}
local fits = true
for _, iv in ipairs(ivs) do
if
m.start_col < iv[2] + gap
and iv[1] < m.start_col + #label + gap
then
-- Standard [a, b) interval-overlap test with `gap` padding on each side.
if m.start_col < iv[2] + gap and iv[1] < m.start_col + #label + gap then
fits = false
break
end
end
if fits then
if two then
prefixes[label:sub(1, 1)] = true
end
-- Record the 2-char label's first char so the input loop knows
-- to buffer a second keystroke when it sees that char.
if two then prefixes[label:sub(1, 1)] = true end
active[label] = { m.line + 1, m.start_col }
api.nvim_buf_set_extmark(buf, NS, m.line, m.start_col, {
virt_text = { { label, M.config.label } },
@@ -166,7 +156,7 @@ function M.start()
priority = 201,
})
ivs[#ivs + 1] = { m.start_col, m.start_col + #label }
line_intervals[m.line] = ivs
ivs_by_line[m.line] = ivs
li = li + 1
end
end
@@ -183,18 +173,21 @@ end
function M.setup(opts)
M.config = vim.tbl_extend('force', M.config, opts or {})
local group =
api.nvim_create_augroup('tiny-jump.highlights', { clear = true })
-- Recreate the augroup on every setup() call (clear = true drops any prior
-- autocmds). Without this, repeated setup calls with different label configs
-- would accumulate stale ColorScheme autocmds.
local group = api.nvim_create_augroup('tiny-jump.highlights', { clear = true })
if type(M.config.label) == 'table' then
-- User passed highlight attrs directly. Own a dedicated highlight group
-- (TinyJumpLabel) and re-apply it on ColorScheme so their colors survive
-- theme changes — colorschemes clear user-defined highlights.
local attrs = M.config.label
M.config.label = 'TinyJumpLabel'
-- Apply now (setup may run after the colorscheme is already loaded) and
-- on every ColorScheme event (themes clear user-defined highlights).
local function apply()
api.nvim_set_hl(0, 'TinyJumpLabel', attrs)
end
apply()
api.nvim_create_autocmd('ColorScheme', { group = group, callback = apply })
api.nvim_set_hl(0, 'TinyJumpLabel', attrs)
api.nvim_create_autocmd('ColorScheme', {
group = group,
callback = function() api.nvim_set_hl(0, 'TinyJumpLabel', attrs) end,
})
end
end
stylua.toml
+2 -1
View File
@@ -1,4 +1,5 @@
column_width = 80
column_width = 120
indent_type = 'Spaces'
indent_width = 2
quote_style = 'AutoPreferSingle'
collapse_simple_statement = 'Always'