Problem: completion: code can be improved
Solution: remove reposition_match() and use mergesort_list(),
for fuzzy completion, sort by fuzzy score immediately after
setting a new leader (Girish Palya)
closes: #17460
Co-authored-by: glepnir <glephunter@gmail.com>
Signed-off-by: Girish Palya <girishji@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: completion: various flaws in fuzzy completion
Solution: fix the issues (Girish Palya)
- Remove the brittle `qsort()` on `compl_match_array`.
- Add a stable, non-recursive `mergesort` for the internal doubly
linked list of matches.
- The sort now happens directly on the internal representation (`compl_T`),
preserving sync with external structures and making sorting stable.
- Update fuzzy match logic to enforce `max_matches` limits after
sorting.
- Remove `trim_compl_match_array()`, which is no longer necessary.
- Fixe test failures by correctly setting `selected` index and
maintaining match consistency.
- Introduce `mergesort_list()` in `misc2.c`, which operates generically
over doubly linked lists.
- Remove `pum_score` and `pum_idx` variables
fixes: #17387closes: #17430
Signed-off-by: Girish Palya <girishji@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: CTRL-X CTRL-R only completes individual words from registers,
making it difficult to insert complete register content.
Solution: Add consecutive CTRL-X CTRL-R support - first press completes
words, second press completes full register lines, similar to
CTRL-X CTRL-L and CTRL-X CTRL-P behavior (glepnir).
closes: #17395
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: PMenu selection broken with multi-line selection and limits
(Maxim Kim)
Solution: update completion match index when limiting the completion
sources (Girish Palya)
fixes: #17394closes: #17401
Signed-off-by: Girish Palya <girishji@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: scheduling of complete function can be improved
Solution: call user completion functions earlier when just determining
the insertion column (Girish Palya)
This change improves the scheduling behavior of async user-defined
completion functions (such as `F{func}`, `F`, or `'o'` values in the
`'complete'` option), particularly benefiting LSP clients.
Currently, these user functions are invoked twice:
1. First with `findstart = 1` to determine the completion start
position.
2. Then with `findstart = 0` to retrieve the actual matches.
Previously, both calls were executed back-to-back. With this change, the
first call (`findstart = 1`) is performed earlier—before any matches are
gathered from other sources.
This adjustment gives event-driven completion sources (e.g., LSP
clients) more time to send their requests while Vim concurrently
collects matches from other sources like the current buffer.
Not sure about the real-world performance gains, but this approach
should, in theory, improve responsiveness and reduce latency for
asynchronous completions.
To test, try using yegappan LSP client:
```vim
set cpt+=o^10
autocmd VimEnter * g:LspOptionsSet({ autoComplete: false, omniComplete: true })
```
If you prefer to use 'native' auto-completion (without plugins), try the
following configuration:
```vim
set cot=menuone,popup,noselect,nearest
autocmd TextChangedI * InsComplete()
def InsComplete()
if getcharstr(1) == '' && getline('.')->strpart(0, col('.') - 1) =~ '\k$'
SkipTextChangedI()
feedkeys("\<c-n>", "n")
endif
enddef
inoremap <silent> <c-e> <c-r>=<SID>SkipTextChangedI()<cr><c-e>
inoremap <silent> <c-y> <c-r>=<SID>SkipTextChangedI()<cr><c-y>
def SkipTextChangedI(): string
set eventignore+=TextChangedI
timer_start(1, (_) => {
set eventignore-=TextChangedI
})
return ''
enddef
inoremap <silent><expr> <tab> pumvisible() ? "\<c-n>" : "\<tab>"
inoremap <silent><expr> <s-tab> pumvisible() ? "\<c-p>" : "\<s-tab>"
```
closes: #17396
Signed-off-by: Girish Palya <girishji@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: out-of-bounds access with 'completefunc' (csetc)
Solution: check if it is safe to advance cpt_sources_index
(Girish Palya)
fixes: #17363closes: #17374
Co-authored-by: @csetc
Signed-off-by: Girish Palya <girishji@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: using f-flag in 'complete' conflicts with Neovims filename
completion (glepnir, after v9.1.1301).
Solution: use upper-case "F" flag for completion functions
(Girish Palya).
fixes: #17347closes: #17378
Signed-off-by: Girish Palya <girishji@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: not easily possible to complete from register content
Solution: add register-completion submode using i_CTRL-X_CTRL-R
(glepnir)
closes: #17354
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: Cannot get completion startcol when space is not the first
trigger character (after v9.1.1383)
Solution: Detect the next comma followed by a space in the option string
and use in next compare loop (glepnir)
closes: #17311
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: When a space character is used as a trigger in 'isexpand' option
it doesn't get recognized because skip_to_option_part() skips
spaces after a comma, treating them as option separators
rather than option value (after v9.1.1341)
Solution: manually set the part to a space character (glepnir).
closes: #17305
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: Cannot return to the original text after selecting the next
item when the currently selected item is the last one.
Solution: When continuing to move down past the last item, locate the
original completion at the head/tail nodes of the completed
linked list (glepnir).
closes: #17300
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: Currently, 'smartcase' is respected when completing keywords
using <C-N>, <C-P>, <C-X><C-N>, and <C-X><C-P>. However, when
a user continues typing and the completion menu is filtered
using cached matches, 'smartcase' is not applied. This leads
to poor-quality or irrelevant completion suggestions, as shown
in the example below.
Solution: When filtering cached completion items after typing additional
characters, apply case-sensitive comparison if 'smartcase' is
enabled and the typed pattern includes uppercase characters.
This ensures consistent and expected completion behavior.
(Girish Palya)
closes: #17271
Signed-off-by: Girish Palya <girishji@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: Flag checking logic uses a temporary variable and multiple
bitwise operations in insexpand.c
Solution: Consolidate into a single equality check using bitwise OR and
comparison (glepnir)
closes: #17276
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: double free in f_complete_match() (after v9.1.1341)
Solution: remove additional free of trig pointer, correctly free
regmatch.regprog and before_cursor in the error case
closes: #17203
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: Cannot define completion triggers and act upon it
Solution: add the new option 'isexpand' and add the complete_match()
function to return the completion matches according to the
'isexpand' setting (glepnir)
Currently, completion trigger position is determined solely by the
'iskeyword' pattern (\k\+$), which causes issues when users need
different completion behaviors - such as triggering after '/' for
comments or '.' for methods. Modifying 'iskeyword' to include these
characters has undesirable side effects on other Vim functionality that
relies on keyword definitions.
Introduce a new buffer-local option 'isexpand' that allows specifying
different completion triggers and add the complete_match() function that
finds the appropriate start column for completion based on these
triggers, scanning backwards from cursor position.
This separation of concerns allows customized completion behavior
without affecting iskeyword-dependent features. The option's
buffer-local nature enables per-filetype completion triggers.
closes: #16716
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: Undo corrupted with 'completeopt' "preinsert" when switching
buffer or window.
Solution: Do not delete preinsert text when switching buffer or window.
(zeertzjq)
related: neovim/neovim#33581
closes: #17193
Signed-off-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: chain complete does not work when 'cot' includes fuzzy
and 'completefuzzycollect' collects wrong next word.
(Konfekt)
Solution: compl_startpos is not set correctly, remove next word check
in search_for_fuzzy_match (glepnir).
fixes#17131fixes#16942closes: #17136
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: compile warning about uninitialized value
(Tony Mechelynck, after v9.1.1311)
Solution: initialize variable on declaration
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: completion: not possible to limit number of matches
Solution: allow to limit the matches for 'complete' sources by using the
"{flag}^{limit}" notation (Girish Palya)
This change extends the 'complete' option to support limiting the
number of matches returned from individual completion sources.
**Rationale:** In large files, certain sources (such as the current
buffer) can generate an overwhelming number of matches, which may cause
more relevant results from other sources (e.g., LSP or tags) to be
pushed out of view. By specifying per-source match limits, the
completion menu remains balanced and diverse, improving visibility and
relevance of suggestions.
A caret (`^`) followed by a number can be appended to a source flag to
specify the maximum number of matches for that source. For example:
```
:set complete=.^9,w,u,t^5
```
In this configuration:
- The current buffer (`.`) will return up to 9 matches.
- The tag completion (`t`) will return up to 5 matches.
- Other sources (`w`, `u`) are not limited.
This feature is fully backward-compatible and does not affect behavior
when the `^count` suffix is not used.
The caret (`^`) was chosen as the delimiter because it is least likely
to appear in file names.
closes: #17087
Signed-off-by: Girish Palya <girishji@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: Duplicate check for preinsert effect, particularly for Ctrl_w
and Ctrl_U.
Solution: Remove the specific check for Ctrl_w and Ctrl_U to eliminate
redundancy (glepnir).
closes: #17129
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: During insert-mode completion, the most relevant match is often
the one closest to the cursor—frequently just above the current line.
However, both `<C-N>` and `<C-P>` tend to rank candidates from the
current buffer that appear above the cursor near the bottom of the
completion menu, rather than near the top. This ordering can feel
unintuitive, especially when `noselect` is active, as it doesn't
prioritize the most contextually relevant suggestions.
Solution: This change introduces a new sub-option value "nearest" for the
'completeopt' setting. When enabled, matches from the current buffer
are prioritized based on their proximity to the cursor position,
improving the relevance of suggestions during completion
(Girish Palya).
Key Details:
- Option: "nearest" added to 'completeopt'
- Applies to: Matches from the current buffer only
- Effect: Sorts completion candidates by their distance from the cursor
- Interaction with other options:
- Has no effect if the `fuzzy` option is also present
This feature is helpful especially when working within large buffers where
multiple similar matches may exist at different locations.
You can test this feature with auto-completion using the snippet below. Try it
in a large file like `vim/src/insexpand.c`, where you'll encounter many
potential matches. You'll notice that the popup menu now typically surfaces the
most relevant matches—those closest to the cursor—at the top. Sorting by
spatial proximity (i.e., contextual relevance) often produces more useful
matches than sorting purely by lexical distance ("fuzzy").
Another way to sort matches is by recency, using an LRU (Least Recently Used)
cache—essentially ranking candidates based on how recently they were used.
However, this is often overkill in practice, as spatial proximity (as provided
by the "nearest" option) is usually sufficient to surface the most relevant
matches.
```vim
set cot=menuone,popup,noselect,nearest inf
def SkipTextChangedIEvent(): string
# Suppress next event caused by <c-e> (or <c-n> when no matches found)
set eventignore+=TextChangedI
timer_start(1, (_) => {
set eventignore-=TextChangedI
})
return ''
enddef
autocmd TextChangedI * InsComplete()
def InsComplete()
if getcharstr(1) == '' && getline('.')->strpart(0, col('.') - 1) =~ '\k$'
SkipTextChangedIEvent()
feedkeys("\<c-n>", "n")
endif
enddef
inoremap <silent> <c-e> <c-r>=<SID>SkipTextChangedIEvent()<cr><c-e>
inoremap <silent><expr> <tab> pumvisible() ? "\<c-n>" : "\<tab>"
inoremap <silent><expr> <s-tab> pumvisible() ? "\<c-p>" : "\<s-tab>"
```
closes: #17076
Signed-off-by: Girish Palya <girishji@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: When switching to another window or tab page while the
completion menu is active, the menu stays visible, although it
belongs to the previous window/tab page context (Evgeni
Chasnovski).
Solution: Track the window and tab page where completion started. Detect
changes in the main editing loop and cancel completion mode if
the current window or tab page differs from where completion
started.
fixes: #17090closes: #17101
Co-authored-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: Coverity warns about using uninitialized value
(Coverity, Tony Mechelynck, after v9.1.1301)
Solution: initialize callback pointer to NULL
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: completion: cannot configure completion functions with
'complete'
Solution: add support for setting completion functions using the f and o
flag for 'complete' (Girish Palya)
This change adds two new values to the `'complete'` (`'cpt'`) option:
- `f` – invokes the function specified by the `'completefunc'` option
- `f{func}` – invokes a specific function `{func}` (can be a string or `Funcref`)
These new flags extend keyword completion behavior (e.g., via `<C-N>` /
`<C-P>`) by allowing function-based sources to participate in standard keyword
completion.
**Key behaviors:**
- Multiple `f{func}` values can be specified, and all will be called in order.
- Functions should follow the interface defined in `:help complete-functions`.
- When using `f{func}`, escaping is required for spaces (with `\`) and commas
(with `\\`) in `Funcref` names.
- If a function sets `'refresh'` to `'always'`, it will be re-invoked on every
change to the input text. Otherwise, Vim will attempt to reuse and filter
existing matches as the input changes, which matches the default behavior of
other completion sources.
- Matches are inserted at the keyword boundary for consistency with other completion methods.
- If finding matches is time-consuming, `complete_check()` can be used to
maintain responsiveness.
- Completion matches are gathered in the sequence defined by the `'cpt'`
option, preserving source priority.
This feature increases flexibility of standard completion mechanism and may
reduce the need for external completion plugins for many users.
**Examples:**
Complete matches from [LSP](https://github.com/yegappan/lsp) client. Notice the use of `refresh: always` and `function()`.
```vim
set cpt+=ffunction("g:LspCompletor"\\,\ [5]). # maxitems = 5
def! g:LspCompletor(maxitems: number, findstart: number, base: string): any
if findstart == 1
return g:LspOmniFunc(findstart, base)
endif
return {words: g:LspOmniFunc(findstart, base)->slice(0, maxitems), refresh: 'always'}
enddef
autocmd VimEnter * g:LspOptionsSet({ autoComplete: false, omniComplete: true })
```
Complete matches from `:iabbrev`.
```vim
set cpt+=fAbbrevCompletor
def! g:AbbrevCompletor(findstart: number, base: string): any
if findstart > 0
var prefix = getline('.')->strpart(0, col('.') - 1)->matchstr('\S\+$')
if prefix->empty()
return -2
endif
return col('.') - prefix->len() - 1
endif
var lines = execute('ia', 'silent!')
if lines =~? gettext('No abbreviation found')
return v:none # Suppresses warning message
endif
var items = []
for line in lines->split("\n")
var m = line->matchlist('\v^i\s+\zs(\S+)\s+(.*)$')
if m->len() > 2 && m[1]->stridx(base) == 0
items->add({ word: m[1], info: m[2], dup: 1 })
endif
endfor
return items->empty() ? v:none :
items->sort((v1, v2) => v1.word < v2.word ? -1 : v1.word ==# v2.word ? 0 : 1)
enddef
```
**Auto-completion:**
Vim's standard completion frequently checks for user input while searching for
new matches. It is responsive irrespective of file size. This makes it
well-suited for smooth auto-completion. You can try with above examples:
```vim
set cot=menuone,popup,noselect inf
autocmd TextChangedI * InsComplete()
def InsComplete()
if getcharstr(1) == '' && getline('.')->strpart(0, col('.') - 1) =~ '\k$'
SkipTextChangedIEvent()
feedkeys("\<c-n>", "n")
endif
enddef
inoremap <silent> <c-e> <c-r>=<SID>SkipTextChangedIEvent()<cr><c-e>
def SkipTextChangedIEvent(): string
# Suppress next event caused by <c-e> (or <c-n> when no matches found)
set eventignore+=TextChangedI
timer_start(1, (_) => {
set eventignore-=TextChangedI
})
return ''
enddef
```
closes: #17065
Co-authored-by: Christian Brabandt <cb@256bit.org>
Co-authored-by: zeertzjq <zeertzjq@outlook.com>
Co-authored-by: glepnir <glephunter@gmail.com>
Signed-off-by: Girish Palya <girishji@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: completion: in keyword completion Ctrl_P cannot go back after
Ctrl_N
Solution: in find_compl_when_fuzzy() always return first match of array, after Ctrl_P
use compl_shown_match->cp_next instead of compl_first_match.
(glepnir)
closes: #17043
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: compl_shown_match is updated when starting keyword completion
and does not include fuzzy matching.
Solution: Do not update compl_shown_match when starting keyword
completion, since it is the one already selected by the
keyword completion direction. (glepnir)
closes: #17033
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: completion: preinserted text not removed when closing pum
Solution: delete preinsert text inside in ins_compl_stop() (glepnir).
closes: #16891
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: endless loop with completefuzzycollect and no match found
Solution: move pointer to line end and break loop
closes: #16820
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: Unnecessary STRLEN() calls in insexpand.c (after 9.1.1178).
Solution: Use the already available length (zeertzjq).
closes: #16814
Signed-off-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: not possible to generate completion candidates using fuzzy
matching
Solution: add the 'completefuzzycollect' option for (some) ins-completion
modes (glepnir)
fixes#15296fixes#15295fixes#15294closes: #16032
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: preinsert requires bot "menu" and "menuone" to be set,
but "menu" is redundant (after v9.1.1160)
Solution: preinsert only requires menuone (glepnir)
closes: #16763
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: The 'preinsert' feature requires Ctrl-Y to confirm insertion,
but Ctrl-Y only works when the popup menu (pum) is displayed.
Without enforcing this dependency, it could lead to confusing
behavior or non-functional features.
Solution: Modify ins_compl_has_preinsert() to check for both 'menu' and
'menuone' flags when 'preinsert' is set. Update documentation
to clarify this requirement. This avoids adding complex
conditional behaviors. (glepnir)
fixes: #16728closes: #16753
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: When expanding omni completion items with newlines (e.g.
`then\n\t\nend`), the end statement gets wrong indentation.
Solution: Add OPENLINE_FORCE_INDENT flag to make open_line() use
second_line_indent directly (glepnir)
closes: #16614
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Justin M. Keyes <justinkz@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: ins_str() is inefficient by calling STRLLEN()
Solution: refactor ins_str() to take a length argument
and let all callers provide the correct length
when calling ins_str() (John Marriott)
closes: #16711
Signed-off-by: John Marriott <basilisk@internode.on.net>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: patch 9.1.1121 used a wrong way to handle enter
Solution: compl_enter_selects also needs to consider the selected item
in ins_compl_new_leader() (glepnir)
closes: #16673
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: cannot loop through pum menu with multiline items with
fuzzy and noselect in 'completeopt' (Tomasz N)
Solution: remove unnecessary compl_no_select condition (glepnir)
fixes: #16641closes: #16674
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: Enter does not insert newline with "noselect" when the pum is
visible (lifepillar)
Solution: When Enter is pressed and no complete-item is selected,
ins_compl_prep returns false, and the edit function continues
processing Enter to insert a new line. (glepnir)
fixes: #1653closes: #16653
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: cannot loop through completion menu with fuzzy and nosort in
'completeopt'
(Tomasz N)
Solution: Reset cur to zero and update compl_shown_match when
'completeopt' contains "nosort" but not "noselect"
(glepnir)
fixes: #16624closes: #16629
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: insexpand.c hard to read
Solution: refactor slightly to make it better readable
(glepnir)
Problem:
- Complex while loops with nested conditions
- Redundant if branches
- Hard to understand and maintain
Solution:
- Restructure using while(true) with clear break conditions
- Using ternary to replace some if conditions
- Add descriptive comments for each step
closes: #16600
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
