Api
Nvim :help
pages, generated
from source
using the tree-sitter-vimdoc parser.
Nvim exposes a powerful API that can be used by plugins and external processes
via
RPC,
Lua and Vimscript (
eval-api).
Applications can also embed libnvim to work with the C API directly.
msgpack-rpc
RPC is the main way to control Nvim programmatically. Nvim implements the
MessagePack-RPC protocol with these extra (out-of-spec) constraints:
1. Responses must be given in reverse order of requests (like "unwinding
a stack").
2. Nvim processes all messages (requests and notifications) in the order they
are received.
Many clients use the API: user interfaces (GUIs), remote plugins, scripts like
"nvr" (
https://github.com/mhinz/neovim-remote). Even Nvim itself can control
other Nvim instances. API clients can:
Call any API function
Listen for events
Receive remote calls from Nvim
The RPC API is like a more powerful version of Vim's "clientserver" feature.
See
channel-intro for various ways to open a channel. Channel-opening
functions take an
rpc
key in the options dict. RPC channels can also be
opened by other processes connecting to TCP/IP sockets or named pipes listened
to by Nvim.
Nvim creates a default RPC socket at
startup, given by
v:servername. To
start with a TCP/IP socket instead, use
--listen with a TCP-style address:
nvim --listen 127.0.0.1:6666
More endpoints can be started with
serverstart().
Note that localhost TCP sockets are generally less secure than named pipes,
and can lead to vulnerabilities like remote code execution.
Connecting to the socket is the easiest way a programmer can test the API,
which can be done through any msgpack-rpc client library or full-featured
api-client. Here's a Ruby script that prints "hello world!" in the current
Nvim instance:
#!/usr/bin/env ruby
# Requires msgpack-rpc: gem install msgpack-rpc
#
# To run this script, execute it from a running Nvim instance (notice the
# trailing '&' which is required since Nvim won't process events while
# running a blocking command):
#
# :!./hello.rb &
#
# Or from another shell by setting NVIM_LISTEN_ADDRESS:
# $ NVIM_LISTEN_ADDRESS=[address] ./hello.rb
require 'msgpack/rpc'
require 'msgpack/rpc/transport/unix'
nvim = MessagePack::RPC::Client.new(MessagePack::RPC::UNIXTransport.new, ENV['NVIM_LISTEN_ADDRESS'])
result = nvim.call(:nvim_command, 'echo "hello world!"')
A better way is to use the Python REPL with the "pynvim" package, where API
functions can be called interactively:
>>> from pynvim import attach
>>> nvim = attach('socket', path='[address]')
>>> nvim.command('echo "hello world!"')
You can also embed Nvim via
jobstart(), and communicate using
rpcrequest()
and
rpcnotify():
let nvim = jobstart(['nvim', '--embed'], {'rpc': v:true})
echo rpcrequest(nvim, 'nvim_eval', '"Hello " . "world!"')
call jobstop(nvim)
api-types
The Nvim C API defines custom types for all function parameters. Some are just
typedefs around C99 standard types, others are Nvim-defined data structures.
Basic types
API Type C type
------------------------------------------------------------------------
Nil
Boolean bool
Integer (signed 64-bit integer) int64_t
Float (IEEE 754 double precision) double
String {char* data, size_t size} struct
Array kvec
Dict (msgpack: map) kvec
Object any of the above
Note:
Empty Array is accepted as a valid Dictionary parameter.
Functions cannot cross RPC boundaries. But API functions (e.g.
nvim_create_autocmd()) may support Lua function parameters for non-RPC
invocations.
Special types (msgpack EXT)
These are integer typedefs discriminated as separate Object subtypes. They
can be treated as opaque integers, but are mutually incompatible: Buffer may
be passed as an integer but not as Window or Tabpage.
The EXT object data is the (integer) object handle. The EXT type codes given
in the
api-metadata types
key are stable: they will not change and are
thus forward-compatible.
EXT Type C type Data
------------------------------------------------------------------------
Buffer enum value kObjectTypeBuffer |bufnr()|
Window enum value kObjectTypeWindow |window-ID|
Tabpage enum value kObjectTypeTabpage internal handle
api-indexing
Most of the API uses 0-based indices, and ranges are end-exclusive. For the
end of a range, -1 denotes the last line/column.
Exception: the following API functions use "mark-like" indexing (1-based
lines, 0-based columns):
Exception: the following API functions use
extmarks indexing (0-based
indices, end-inclusive):
api-fast
Most API functions are "deferred": they are queued on the main loop and
processed sequentially with normal input. So if the editor is waiting for
user input in a "modal" fashion (e.g. the
hit-enter-prompt), the request
will block. Non-deferred (fast) functions such as
nvim_get_mode() and
nvim_input() are served immediately (i.e. without waiting in the input
queue). Lua code can use
vim.in_fast_event() to detect a fast context.
The Nvim C API is automatically exposed to RPC by the build system, which
parses headers in src/nvim/api/* and generates dispatch-functions mapping RPC
API method names to public C API functions, converting/validating arguments
and return values.
Nvim exposes its API metadata as a Dictionary with these items:
version Nvim version, API level/compatibility
version.api_level API version integer
api-level
version.api_compatible API is backwards-compatible with this level
version.api_prerelease Declares the API as unstable/unreleased
(version.api_prerelease && fn.since == version.api_level)
functions API function signatures, containing
api-types info
describing the return value and parameters.
ui_events
UI event signatures
{fn}
.since API level where function {fn}
was introduced
{fn}
.deprecated_since API level where function {fn}
was deprecated
types Custom handle types defined by Nvim
error_types Possible error types returned by API functions
About the functions
map:
Container types may be decorated with type/size constraints, e.g.
ArrayOf(Buffer) or ArrayOf(Integer, 2).
Functions considered to be methods that operate on instances of Nvim
special types (msgpack EXT) have the "method=true" flag. The receiver type
is that of the first argument. Method names are prefixed with
nvim_
plus
a type name, e.g.
nvim_buf_get_lines
is the
get_lines
method of
a Buffer instance.
dev-api
Global functions have the "method=false" flag and are prefixed with just
nvim_
, e.g. nvim_list_bufs
.
api-mapping
External programs (clients) can use the metadata to discover the API, using
any of these approaches:
1. Connect to a running Nvim instance and call
nvim_get_api_info() via
msgpack-RPC. This is best for clients written in dynamic languages which
can define functions at runtime.
2. Start Nvim with
--api-info. Useful for statically-compiled clients.
Example (requires Python "pyyaml" and "msgpack-python" modules):
nvim --api-info | python -c 'import msgpack, sys, yaml; yaml.dump(msgpack.unpackb(sys.stdin.buffer.read()), sys.stdout)'
3. Use the
api_info() Vimscript function.
:lua vim.print(vim.fn.api_info())
Example using
filter() to exclude non-deprecated API functions:
:new|put =map(filter(api_info().functions, '!has_key(v:val,''deprecated_since'')'), 'v:val.name')
The Nvim API is composed of functions and events.
Clients call functions like those described at
api-global.
API function names are prefixed with "nvim_".
API event names are prefixed with "nvim_" and suffixed with "_event".
As Nvim evolves the API may change in compliance with this CONTRACT:
New functions and events may be added.
Any such extensions are OPTIONAL: old clients may ignore them.
Function signatures will NOT CHANGE (after release).
Functions introduced in the development (unreleased) version MAY CHANGE.
(Clients can dynamically check
api_prerelease
, etc.
api-metadata)
Event parameters will not be removed or reordered (after release).
Events may be EXTENDED: new parameters may be added.
New items may be ADDED to map/list parameters/results of functions and
events.
Any such new items are OPTIONAL: old clients may ignore them.
Existing items will not be removed (after release).
Deprecated functions will not be removed until Nvim version 2.0
"Private" interfaces are NOT covered by this contract:
Undocumented (not in :help) functions or events of any kind
nvim__x ("double underscore") functions
The idea is "versionless evolution", in the words of Rich Hickey:
Relaxing a requirement should be a compatible change.
Strengthening a promise should be a compatible change.
When a client invokes an API request as an async notification, it is not
possible for Nvim to send an error response. Instead, in case of error, the
following notification will be sent to the client:
{type}
is a numeric id as defined by api_info().error_types
, and {message}
is
a string with the error message.
API clients can "attach" to Nvim buffers to subscribe to buffer update events.
This is similar to
TextChanged but more powerful and granular.
nvim_buf_lines_event
nvim_buf_lines_event[
{buf}
,
{changedtick}
,
{firstline}
,
{lastline}
,
{linedata}
,
{more}
]
When the buffer text between {firstline}
and {lastline}
(end-exclusive,
zero-indexed) were changed to the new text in the {linedata}
list. The
granularity is a line, i.e. if a single character is changed in the
editor, the entire line is sent.
When
{changedtick}
is
v:null this means the screen lines (display)
changed but not the buffer contents.
{linedata}
contains the changed
screen lines. This happens when
'inccommand' shows a buffer preview.
Properties:
{buf}
API buffer handle (buffer number)
{changedtick}
value of
b:changedtick for the buffer. If you send an
API command back to nvim you can check the value of
b:changedtick as
part of your request to ensure that no other changes have been made.
{firstline}
integer line number of the first line that was replaced.
Zero-indexed: if line 1 was replaced then {firstline}
will be 0, not
1. {firstline}
is always less than or equal to the number of lines
that were in the buffer before the lines were replaced.
{lastline}
integer line number of the first line that was not replaced
(i.e. the range {firstline}
, {lastline}
is end-exclusive).
Zero-indexed: if line numbers 2 to 5 were replaced, this will be 5
instead of 6. {lastline}
is always be less than or equal to the number
of lines that were in the buffer before the lines were replaced.
{lastline}
will be -1 if the event is part of the initial update after
attaching.
{linedata}
list of strings containing the contents of the new buffer
lines. Newline characters are omitted; empty lines are sent as empty
strings.
{more}
boolean, true for a "multipart" change notification: the
current change was chunked into multiple
nvim_buf_lines_event
notifications (e.g. because it was too big).
When
b:changedtick was incremented but no text was changed. Relevant for
undo/redo.
Properties:
{buf}
API buffer handle (buffer number)
{changedtick}
new value of
b:changedtick for the buffer
When buffer is detached (i.e. updates are disabled). Triggered explicitly by
nvim_buf_detach() or implicitly in these cases:
Generally: whenever the buffer contents are unloaded from memory.
Properties:
{buf}
API buffer handle (buffer number)
Calling
nvim_buf_attach() with send_buffer=true on an empty buffer, emits:
nvim_buf_lines_event[{buf}, {changedtick}, 0, -1, [""], v:false]
User adds two lines to the buffer, emits:
nvim_buf_lines_event[{buf}, {changedtick}, 0, 0, ["line1", "line2"], v:false]
User moves to a line containing the text "Hello world" and inserts "!", emits:
nvim_buf_lines_event[{buf}, {changedtick}, {linenr}, {linenr} + 1,
["Hello world!"], v:false]
User moves to line 3 and deletes 20 lines using "20dd", emits:
nvim_buf_lines_event[{buf}, {changedtick}, 2, 22, [], v:false]
User selects lines 3-5 using
linewise-visual mode and then types "p" to
paste a block of 6 lines, emits:
nvim_buf_lines_event[{buf}, {changedtick}, 2, 5,
['pasted line 1', 'pasted line 2', 'pasted line 3', 'pasted line 4',
'pasted line 5', 'pasted line 6'],
v:false
]
User reloads the buffer with ":edit", emits:
nvim_buf_detach_event[{buf}]
LUA
api-buffer-updates-lua
In-process Lua plugins can receive buffer updates in the form of Lua
callbacks. These callbacks are called frequently in various contexts;
textlock prevents changing buffer contents and window layout (use
vim.schedule() to defer such operations to the main loop instead).
Moving the cursor is allowed, but it is restored afterwards.
nvim_buf_attach() will take keyword args for the callbacks. "on_lines" will
receive parameters ("lines",
{buf}
,
{changedtick}
,
{firstline}
,
{lastline}
,
{new_lastline}
,
{old_byte_size}
[,
{old_utf32_size}
,
{old_utf16_size}
]).
Unlike remote channel events the text contents are not passed. The new text can
be accessed inside the callback as
vim.api.nvim_buf_get_lines(buf, firstline, new_lastline, true)
{old_byte_size}
is the total size of the replaced region
{firstline}
to
{lastline}
in bytes, including the final newline after
{lastline}
. if
utf_sizes
is set to true in
nvim_buf_attach() keyword args, then the
UTF-32 and UTF-16 sizes of the deleted region is also passed as additional
arguments
{old_utf32_size}
and
{old_utf16_size}
.
"on_changedtick" is invoked when
b:changedtick was incremented but no text
was changed. The parameters received are ("changedtick",
{buf}
,
{changedtick}
).
Nvim allows plugins to add position-based highlights to buffers. This is
similar to
matchaddpos() but with some key differences. The added highlights
are associated with a buffer and adapts to line insertions and deletions,
similar to signs. It is also possible to manage a set of highlights as a group
and delete or replace all at once.
The intended use case are linter or semantic highlighter plugins that monitor
a buffer for changes, and in the background compute highlights to the buffer.
Another use case are plugins that show output in an append-only buffer, and
want to add highlights to the outputs. Highlight data cannot be preserved
on writing and loading a buffer to file, nor in undo/redo cycles.
Highlights are registered using the
nvim_buf_add_highlight() function. If an
external highlighter plugin wants to add many highlights in a batch,
performance can be improved by calling
nvim_buf_add_highlight() as an
asynchronous notification, after first (synchronously) requesting a source id.
nvim_buf_add_highlight() adds highlights as
extmarks. If highlights need to
be tracked or manipulated after adding them, it is better to use
nvim_buf_set_extmark() directly, as this function returns the placed
extmark
id. Thus, instead of
vim.api.nvim_buf_add_highlight(buf, ns_id, hl_group, line, col_start, col_end)
use
-- create the highlight through an extmark
extid = vim.api.nvim_buf_set_extmark(buf, ns_id, line, col_start, {end_col = col_end, hl_group = hl_group})
-- example: modify the extmark's highlight group
vim.api.nvim_buf_set_extmark(buf, ns_id, line, col_start, {end_col = col_end, hl_group = NEW_HL_GROUP, id = extid})
-- example: change the highlight's position
vim.api.nvim_buf_set_extmark(buf, ns_id, NEW_LINE, col_start, {end_col = col_end, hl_group = NEW_HL_GROUP, id = extid})
Example using the Python API client (
pynvim):
src = vim.new_highlight_source()
buf = vim.current.buffer
for i in range(5):
buf.add_highlight("String",i,0,-1,src_id=src)
# some time later ...
buf.clear_namespace(src)
If the highlights don't need to be deleted or updated, just pass -1 as
src_id (this is the default in python). Use
nvim_buf_clear_namespace() to
clear highlights from a specific source, in a specific line range or the
entire buffer by passing in the line range 0, -1 (the latter is the default in
python as used above).
Example using the API from Vimscript:
call nvim_buf_set_lines(0, 0, 0, v:true, ["test text"])
let src = nvim_buf_add_highlight(0, 0, "String", 1, 0, 4)
call nvim_buf_add_highlight(0, src, "Identifier", 0, 5, -1)
" some time later ...
call nvim_buf_clear_namespace(0, src, 0, -1)
Floating windows ("floats") are displayed on top of normal windows. This is
useful to implement simple widgets, such as tooltips displayed next to the
cursor. Floats are fully functional windows supporting user editing, common
api-window calls, and most window options (except
'statusline').
Two ways to create a floating window:
To check whether a window is floating, check whether the
relative
option in
its config is non-empty:
if vim.api.nvim_win_get_config(window_id).relative ~= '' then
-- window with this window_id is floating
end
Buffer text can be highlighted by typical mechanisms (syntax highlighting,
api-highlights). The
hl-NormalFloat group highlights normal text;
'winhighlight' can be used as usual to override groups locally. Floats inherit
options from the current window; specify
style=minimal
in
nvim_open_win()
to disable various visual features such as the
'number' column.
Other highlight groups specific to floating windows:
Currently, floating windows don't support some widgets like scrollbar.
The output of
:mksession does not include commands for restoring floating
windows.
Example: create a float with scratch buffer:
let buf = nvim_create_buf(v:false, v:true)
call nvim_buf_set_lines(buf, 0, -1, v:true, ["test", "text"])
let opts = {'relative': 'cursor', 'width': 10, 'height': 2, 'col': 0,
\ 'row': 1, 'anchor': 'NW', 'style': 'minimal'}
let win = nvim_open_win(buf, 0, opts)
" optional: change highlight, otherwise Pmenu is used
call nvim_set_option_value('winhl', 'Normal:MyHighlight', {'win': win})
Extended marks (extmarks) represent buffer annotations that track text changes
in the buffer. They can represent cursors, folds, misspelled words, anything
that needs to track a logical location in the buffer over time.
api-indexing
Extmark position works like a "vertical bar" cursor: it exists between
characters. Thus, the maximum extmark index on a line is 1 more than the
character index:
f o o b a r line contents
0 1 2 3 4 5 character positions (0-based)
0 1 2 3 4 5 6 extmark positions (0-based)
Extmarks have "forward gravity": if you place the cursor directly on an
extmark position and enter some text, the extmark migrates forward.
f o o|b a r line (| = cursor)
3 extmark
f o o z|b a r line (| = cursor)
4 extmark (after typing "z")
If an extmark is on the last index of a line and you input a newline at that
point, the extmark will accordingly migrate to the next line:
f o o z b a r| line (| = cursor)
7 extmark
f o o z b a r first line
extmarks (none present)
| second line (| = cursor)
0 extmark (after typing <CR>)
Example:
Let's set an extmark at the first row (row=0) and third column (column=2).
api-indexing Passing id=0 creates a new mark and returns the id:
01 2345678
0 ex|ample..
^ extmark position
let g:mark_ns = nvim_create_namespace('myplugin')
let g:mark_id = nvim_buf_set_extmark(0, g:mark_ns, 0, 2, {})
We can get the mark by its id:
echo nvim_buf_get_extmark_by_id(0, g:mark_ns, g:mark_id, {})
" => [0, 2]
We can get all marks in a buffer by
namespace (or by a range):
echo nvim_buf_get_extmarks(0, g:mark_ns, 0, -1, {})
" => [[1, 0, 2]]
Deleting all surrounding text does NOT remove an extmark! To remove extmarks
use
nvim_buf_del_extmark(). Deleting "x" in our example:
0 12345678
0 e|ample..
^ extmark position
echo nvim_buf_get_extmark_by_id(0, g:mark_ns, g:mark_id, {})
" => [0, 1]
Namespaces allow any plugin to manage only its own extmarks, ignoring those
created by another plugin.
Extmark positions changed by an edit will be restored on undo/redo. Creating
and deleting extmarks is not a buffer change, thus new undo states are not
created for extmark changes.
nvim_chan_send(
{chan}
,
{data}
)
nvim_chan_send()
Send data to channel
id
. For a job, it writes it to the stdin of the
process. For the stdio channel
channel-stdio, it writes to Nvim's
stdout. For an internal terminal instance (
nvim_open_term()) it writes
directly to terminal output. See
channel-bytes for more information.
This function writes raw data, not RPC messages. If the channel was
created with
rpc=true
then the channel expects RPC messages, use
vim.rpcnotify() and
vim.rpcrequest() instead.
Parameters:
{chan}
id of the channel
{data}
data to write. 8-bit clean: can contain NUL bytes.
nvim_create_buf(
{listed}
,
{scratch}
)
nvim_create_buf()
Creates a new, empty, unnamed buffer.
Return:
Buffer handle, or 0 on error
See also:
buf_open_scratch
Attributes:
not allowed when
textlock is active
Note:
Lowercase name (or other buffer-local mark) is an error.
Parameters:
{name}
Mark name
Return:
true if the mark was deleted, else false.
Parameters:
{name}
Variable name
nvim_echo(
{chunks}
,
{history}
,
{opts}
)
nvim_echo()
Echo a message.
Parameters:
{chunks}
A list of [text, hl_group]
arrays, each representing a
text chunk with specified highlight. hl_group
element can
be omitted for no highlight.
{opts}
Optional parameters.
verbose: Message was printed as a result of
'verbose'
option if Nvim was invoked with -V3log_file, the message
will be redirected to the log_file and suppressed from
direct output.
nvim_err_write(
{str}
)
nvim_err_write()
Writes a message to the Vim error buffer. Does not append "\n", the
message is buffered (won't display) until a linefeed is written.
Parameters:
{str}
Message
nvim_err_writeln(
{str}
)
nvim_err_writeln()
Writes a message to the Vim error buffer. Appends "\n", so the buffer is
flushed (and displayed).
Parameters:
{str}
Message
See also:
nvim_err_write()
Parameters:
{opts}
Optional parameters.
winid: (number)
window-ID of the window to use as context
for statusline.
maxwidth: (number) Maximum width of statusline.
fillchar: (string) Character to fill blank spaces in the
statusline (see
'fillchars'). Treated as single-width even
if it isn't.
highlights: (boolean) Return highlight information.
use_winbar: (boolean) Evaluate winbar instead of statusline.
use_tabline: (boolean) Evaluate tabline instead of
statusline. When true, {winid}
is ignored. Mutually
exclusive with {use_winbar}
.
use_statuscol_lnum: (number) Evaluate statuscolumn for this
line number instead of statusline.
Return:
Dict containing statusline information, with these keys:
str: (string) Characters that will be displayed on the statusline.
width: (number) Display width of the statusline.
highlights: Array containing highlight information of the
statusline. Only included when the "highlights" key in
{opts}
is
true. Each element of the array is a
Dict with these keys:
start: (number) Byte index (0-based) of first character that uses
the highlight.
group: (string) Name of highlight group.
nvim_exec_lua(
{code}
,
{args}
)
nvim_exec_lua()
Execute Lua code. Parameters (if any) are available as
...
inside the
chunk. The chunk can return a value.
Only statements are executed. To evaluate an expression, prefix it with
return
: return my_function(...)
Parameters:
{code}
Lua code to execute
{args}
Arguments to the code
Return:
Return value of Lua code if present or NIL.
nvim_feedkeys(
{keys}
,
{mode}
,
{escape_ks}
)
nvim_feedkeys()
Sends input-keys to Nvim, subject to various quirks controlled by
mode
flags. This is a blocking call, unlike
nvim_input().
On execution error: does not fail, but updates v:errmsg.
Example:
:let key = nvim_replace_termcodes("<C-o>", v:true, v:false, v:true)
:call nvim_feedkeys(key, 'n', v:false)
Parameters:
{keys}
to be typed
{escape_ks}
If true, escape K_SPECIAL bytes in
keys
. This should be
false if you already used
nvim_replace_termcodes(), and
true otherwise.
See also:
feedkeys()
vim_strsave_escape_ks
Return:
2-tuple
[{channel-id}, {api-metadata}]
Parameters:
{chan}
channel_id, or 0 for current channel
Return:
Channel info dict with these keys:
"id" Channel id.
"argv" (optional) Job arguments list.
"stream" Stream underlying the channel.
"stdio" stdin and stdout of this Nvim instance
"stderr" stderr of this Nvim instance
"socket" TCP/IP socket or named pipe
"job" Job with communication over its stdio.
"mode" How data received on the channel is interpreted.
"bytes" Send and receive raw bytes.
"terminal"
terminal instance interprets ASCII sequences.
"rpc"
RPC communication on the channel is active.
"pty" (optional) Name of pseudoterminal. On a POSIX system this is a
device path like "/dev/pts/1". If unknown, the key will still be
present if a pty is used (e.g. for conpty on Windows).
"buffer" (optional) Buffer connected to
terminal instance.
"client" (optional) Info about the peer (client on the other end of
the RPC channel), which it provided via
nvim_set_client_info().
Example:
:echo nvim_get_color_by_name("Pink")
:echo nvim_get_color_by_name("#cbcbcb")
Parameters:
{name}
Color name or "#rrggbb" string
Return:
24-bit RGB value, or -1 for invalid argument.
Keys are color names (e.g. "Aqua") and values are 24-bit RGB color values
(e.g. 65535).
Return:
Map of color names and RGB values.
Parameters:
{opts}
Optional parameters.
types: List of
context-types ("regs", "jumps", "bufs",
"gvars", …) to gather, or empty for "all".
Return:
Current line string
nvim_get_hl(
{ns_id}
,
{opts}
)
nvim_get_hl()
Gets all or specific highlight groups in a namespace.
Note:
When the
link
attribute is defined in the highlight definition map,
other attributes will not be taking effect (see
:hi-link).
Parameters:
{opts}
Options dict:
name: (string) Get a highlight definition by name.
id: (integer) Get a highlight definition by id.
link: (boolean, default true) Show linked group name
instead of effective definition
:hi-link.
create: (boolean, default true) When highlight group
doesn't exist create it.
Return:
Highlight groups as a map from group name to a highlight definition
map as in
nvim_set_hl(), or only a single highlight definition map
if requested by name or id.
similar to
hlID(), but allocates a new ID if not present.
Parameters:
{opts}
Optional parameters
winid: (number)
window-ID for retrieving a window's
highlight namespace. A value of -1 is returned when
nvim_win_set_hl_ns() has not been called for the window
(or was called with a namespace of -1).
Return:
Namespace id, or -1
Parameters:
{mode}
Mode short-name ("n", "i", "v", ...)
Return:
Array of
maparg()-like dictionaries describing mappings. The
"buffer" key is always zero.
nvim_get_mark(
{name}
,
{opts}
)
nvim_get_mark()
Returns a
(row, col, buffer, buffername)
tuple representing the position
of the uppercase/file named mark. "End of line" column position is
returned as
v:maxcol (big number). See
mark-motions.
Note:
Lowercase name (or other buffer-local mark) is an error.
Parameters:
{name}
Mark name
{opts}
Optional parameters. Reserved for future use.
Return:
4-tuple (row, col, buffer, buffername), (0, 0, 0, '') if the mark is
not set.
nvim_get_mode()
nvim_get_mode()
Gets the current mode.
mode() "blocking" is true if Nvim is waiting for
input.
Return:
Dict { "mode": String, "blocking": Boolean }
Return:
Map of process properties, or NIL if process not found.
Return:
Array of child process ids, empty if process not found.
"name" can contain wildcards. For example
nvim_get_runtime_file("colors/*.{vim,lua}", true)
will return all color
scheme files. Always use forward slashes (/) in the search pattern for
subdirectories regardless of platform.
It is not an error to not find any files. An empty array is returned then.
Parameters:
{name}
pattern of files to search for
{all}
whether to return all matches or only the first
Return:
list of absolute paths to the found files
Parameters:
{name}
Variable name
Parameters:
{name}
Variable name
nvim_input(
{keys}
)
nvim_input()
Queues raw user-input. Unlike
nvim_feedkeys(), this uses a low-level
input buffer and the call is non-blocking (input is processed
asynchronously by the eventloop).
To input blocks of text,
nvim_paste() is much faster and should be
preferred.
On execution error: does not fail, but updates v:errmsg.
Note:
keycodes like
<CR>
are translated, so "<" is special. To input a
literal "<", send
<LT>
.
Parameters:
{keys}
to be typed
Return:
Number of bytes actually written (can be fewer than requested if the
buffer becomes full).
nvim_input_mouse()
nvim_input_mouse(
{button}
,
{action}
,
{modifier}
,
{grid}
,
{row}
,
{col}
)
Send mouse event from GUI.
Non-blocking: does not wait on any result, but queues the event to be
processed soon by the event loop.
Note:
Currently this doesn't support "scripting" multiple mouse events by
calling it multiple times in a loop: the intermediate mouse positions
will be ignored. It should be used to implement real-time mouse input
in a GUI. The deprecated pseudokey form (
<LeftMouse><col,row>
) of
nvim_input() has the same limitation.
Parameters:
{button}
Mouse button: one of "left", "right", "middle", "wheel",
"move", "x1", "x2".
{action}
For ordinary buttons, one of "press", "drag", "release".
For the wheel, one of "up", "down", "left", "right".
Ignored for "move".
{modifier}
String of modifiers each represented by a single char. The
same specifiers are used as for a key press, except that
the "-" separator is optional, so "C-A-", "c-a" and "CA"
can all be used to specify Ctrl+Alt+click.
{row}
Mouse row-position (zero-based, like redraw events)
{col}
Mouse column-position (zero-based, like redraw events)
Includes unlisted (unloaded/deleted) buffers, like
:ls!
. Use
nvim_buf_is_loaded() to check if a buffer is loaded.
Return:
List of buffer handles
Return:
Array of Dictionaries, each describing a channel with the format
specified at
nvim_get_chan_info().
Return:
List of tabpage handles
nvim_list_uis()
nvim_list_uis()
Gets a list of dictionaries representing attached UIs.
Return:
Array of UI dictionaries, each with these keys:
"height" Requested height of the UI
"width" Requested width of the UI
"rgb" true if the UI uses RGB colors (false implies
cterm-colors)
"ext_..." Requested UI extensions, see
ui-option
Return:
List of window handles
nvim_notify(
{msg}
,
{log_level}
,
{opts}
)
nvim_notify()
Notify the user with a message
Relays the call to vim.notify . By default forwards your message in the
echo area but can be overridden to trigger desktop notifications.
Parameters:
{msg}
Message to display to the user
{log_level}
The log level
{opts}
Reserved for future use.
nvim_open_term(
{buffer}
,
{opts}
)
nvim_open_term()
Open a terminal instance in a buffer
By default (and currently the only option) the terminal will not be
connected to an external process. Instead, input send on the channel will
be echoed directly by the terminal. This is useful to display ANSI
terminal sequences returned as part of a rpc message, or similar.
Note: to directly initiate the terminal using the right size, display the
buffer in a configured window before calling this. For instance, for a
floating display, first create an empty buffer using
nvim_create_buf(),
then display it using
nvim_open_win(), and then call this function. Then
nvim_chan_send() can be called immediately to process sequences in a
virtual terminal having the intended size.
Attributes:
not allowed when
textlock is active
Parameters:
{buffer}
the buffer to use (expected to be empty)
{opts}
Optional parameters.
on_input: Lua callback for input sent, i e keypresses in
terminal mode.
Note: keypresses are sent raw as they would
be to the pty master end. For instance, a carriage return
is sent as a "\r", not as a "\n".
textlock applies. It
is possible to call
nvim_chan_send() directly in the
callback however.
["input", term, bufnr, data]
force_crlf: (boolean, default true) Convert "\n" to
"\r\n".
Return:
Channel id, or 0 on error
nvim_out_write(
{str}
)
nvim_out_write()
Writes a message to the Vim output buffer. Does not append "\n", the
message is buffered (won't display) until a linefeed is written.
Parameters:
{str}
Message
nvim_paste(
{data}
,
{crlf}
,
{phase}
)
nvim_paste()
Pastes at cursor (in any mode), and sets "redo" so dot (
.) will repeat
the input. UIs call this to implement "paste", but it's also intended for
use by scripts to input large, dot-repeatable blocks of text (as opposed
to
nvim_input() which is subject to mappings/events and is thus much
slower).
Invokes the
vim.paste() handler, which handles each mode appropriately.
Errors (
'nomodifiable',
vim.paste()
failure, …) are reflected in
err
but do not affect the return value (which is strictly decided by
vim.paste()
). On error or cancel, subsequent calls are ignored
("drained") until the next paste is initiated (phase 1 or -1).
Useful in mappings and scripts to insert multiline text. Example:
vim.keymap.set('n', 'x', function()
vim.api.nvim_paste([[
line1
line2
line3
]], false, -1)
end, { buffer = true })
Attributes:
not allowed when
textlock is active
Parameters:
{data}
Multiline input. Lines break at LF ("\n"). May be binary
(containing NUL bytes).
{crlf}
Also break lines at CR and CRLF.
{phase}
-1: paste in a single call (i.e. without streaming). To
"stream" a paste, call nvim_paste
sequentially with these
phase
values:
1: starts the paste (exactly once)
2: continues the paste (zero or more times)
3: ends the paste (exactly once)
Return:
true: Client may continue pasting.
false: Client should cancel the paste.
nvim_put(
{lines}
,
{type}
,
{after}
,
{follow}
)
nvim_put()
Puts text at cursor, in any mode. For dot-repeatable input, use
nvim_paste().
Compare
:put and
p which are always linewise.
Attributes:
not allowed when
textlock is active
Parameters:
{after}
If true insert after cursor (like
p), or before (like
P).
{follow}
If true place cursor at end of inserted text.
nvim_replace_termcodes()
nvim_replace_termcodes(
{str}
,
{from_part}
,
{do_lt}
,
{special}
)
Replaces terminal codes and
keycodes (
<CR>
,
<Esc>
, ...) in a string with
the internal representation.
Parameters:
{str}
String to be converted.
{from_part}
Legacy Vim parameter. Usually true.
{do_lt}
Also translate <lt>
. Ignored if special
is false.
{special}
Replace
keycodes, e.g.
<CR>
becomes a "\r" char.
See also:
replace_termcodes
cpoptions
nvim_select_popupmenu_item({item}
, {insert}
, {finish}
, {opts}
)
Selects an item in the completion popup menu.
If neither
ins-completion nor
cmdline-completion popup menu is active
this API call is silently ignored. Useful for an external UI using
ui-popupmenu to control the popup menu with the mouse. Can also be used
in a mapping; use
<Cmd>
:map-cmd or a Lua mapping to ensure the mapping
doesn't end completion mode.
Parameters:
{item}
Index (zero-based) of the item to select. Value of -1
selects nothing and restores the original text.
{finish}
Finish the completion and dismiss the popup menu. Implies
{insert}
.
{opts}
Optional parameters. Reserved for future use.
nvim_set_client_info()
nvim_set_client_info(
{name}
,
{version}
,
{type}
,
{methods}
,
{attributes}
)
Self-identifies the client.
The client/plugin/application should call this after connecting, to
provide hints about its identity and purpose, for debugging and
orchestration.
Can be called more than once; the caller should merge old info if
appropriate. Example: library first identifies the channel, then a plugin
using that library later identifies itself.
Note:
"Something is better than nothing". You don't need to include all the
fields.
Parameters:
{name}
Short name for the connected client
{version}
Dict describing the version, with these (optional) keys:
"major" major version (defaults to 0 if not set, for
no release yet)
"minor" minor version
"patch" patch number
"prerelease" string describing a prerelease, like
"dev" or "beta1"
"commit" hash or similar identifier of commit
{type}
Must be one of the following values. Client libraries
should default to "remote" unless overridden by the
user.
"remote" remote client connected "Nvim flavored"
MessagePack-RPC (responses must be in reverse order of
requests).
msgpack-rpc
"msgpack-rpc" remote client connected to Nvim via
fully MessagePack-RPC compliant protocol.
"ui" gui frontend
"embedder" application using Nvim as a component (for
example, IDE/editor implementing a vim mode).
"host" plugin host, typically started by nvim
"plugin" single plugin, started by nvim
{methods}
Builtin methods in the client. For a host, this does not
include plugin methods which will be discovered later.
The key should be the method name, the values are dicts
with these (optional) keys (more keys may be added in
future versions of Nvim, thus unknown keys are ignored.
Clients must only use keys defined in this or later
versions of Nvim):
"async" if true, send as a notification. If false or
unspecified, use a blocking request
"nargs" Number of arguments. Could be a single integer
or an array of two integers, minimum and maximum
inclusive.
{attributes}
Arbitrary string:string map of informal client
properties. Suggested keys:
"pid": Process id.
"website": Client homepage URL (e.g. GitHub
repository)
"license": License description ("Apache 2", "GPLv3",
"MIT", …)
"logo": URI or path to image, preferably small logo or
icon. .png or .svg format is preferred.
Parameters:
{buffer}
Buffer handle
Parameters:
{dir}
Directory path
Attributes:
not allowed when
textlock is active
Parameters:
{line}
Line contents
Parameters:
{tabpage}
Tabpage handle
Parameters:
{window}
Window handle
nvim_set_hl(
{ns_id}
,
{name}
,
{val}
)
nvim_set_hl()
Sets a highlight group.
Note:
Unlike the :highlight
command which can update a highlight group,
this function completely replaces the definition. For example:
nvim_set_hl(0, 'Visual', {})
will clear the highlight group
'Visual'.
The fg and bg keys also accept the string values "fg"
or "bg"
which act as aliases to the corresponding foreground and background
values of the Normal group. If the Normal group has not been defined,
using these values results in an error.
If
link
is used in combination with other attributes; only the
link
will take effect (see
:hi-link).
Parameters:
{name}
Highlight group name, e.g. "ErrorMsg"
{val}
Highlight definition map, accepts the following keys:
fg: color name or "#RRGGBB", see note.
bg: color name or "#RRGGBB", see note.
sp: color name or "#RRGGBB"
blend: integer between 0 and 100
bold: boolean
standout: boolean
underline: boolean
undercurl: boolean
underdouble: boolean
underdotted: boolean
underdashed: boolean
strikethrough: boolean
italic: boolean
reverse: boolean
nocombine: boolean
link: name of another highlight group to link to, see
:hi-link.
default: Don't override existing definition
:hi-default
ctermfg: Sets foreground of cterm color
ctermfg
ctermbg: Sets background of cterm color
ctermbg
cterm: cterm attribute map, like
highlight-args. If not
set, cterm attributes will match those from the attribute
map documented above.
force: if true force update the highlight group when it
exists.
Parameters:
{ns_id}
the namespace to use
This function meant to be called while redrawing, primarily from
nvim_set_decoration_provider() on_win and on_line callbacks, which are
allowed to change the namespace during a redraw cycle.
Parameters:
{ns_id}
the namespace to activate
Unlike
:map, leading/trailing whitespace is accepted as part of the
{lhs}
or
{rhs}
. Empty
{rhs}
is
<Nop>
.
keycodes are replaced as usual.
Example:
call nvim_set_keymap('n', ' <NL>', '', {'nowait': v:true})
is equivalent to:
nmap <nowait> <Space><NL> <Nop>
Parameters:
{mode}
Mode short-name (map command prefix: "n", "i", "v", "x", …)
or "!" for
:map!, or empty string for
:map. "ia", "ca" or
"!a" for abbreviation in Insert mode, Cmdline mode, or both,
respectively
{lhs}
Left-hand-side
{lhs} of the mapping.
{rhs}
Right-hand-side
{rhs} of the mapping.
{opts}
Optional parameters map: Accepts all
:map-arguments as keys
except
<buffer>
, values are booleans (default false). Also:
"desc" human-readable description.
"callback" Lua function called in place of {rhs}
.
"replace_keycodes" (boolean) When "expr" is true, replace
keycodes in the resulting string (see
nvim_replace_termcodes()). Returning nil from the Lua
"callback" is equivalent to returning an empty string.
nvim_set_var(
{name}
,
{value}
)
nvim_set_var()
Sets a global (g:) variable.
Parameters:
{name}
Variable name
{value}
Variable value
nvim_set_vvar(
{name}
,
{value}
)
nvim_set_vvar()
Sets a v: variable, if it is not readonly.
Parameters:
{name}
Variable name
{value}
Variable value
nvim_strwidth(
{text}
)
nvim_strwidth()
Calculates the number of display cells occupied by
text
. Control
characters including
<Tab>
count as one cell.
Parameters:
{text}
Some text
nvim__complete_set(
{index}
,
{opts}
)
nvim__complete_set()
EXPERIMENTAL: this API may change in the future.
Sets info for the completion item at the given index. If the info text was
shown in a window, returns the window and buffer ids, or empty dict if not
shown.
Parameters:
{index}
Completion candidate index
{opts}
Optional parameters.
info: (string) info text.
Return:
Dict containing these keys:
winid: (number) floating window id
bufnr: (number) buffer id in floating window
Parameters:
{pat}
pattern of files to search for
{all}
whether to return all matches or only the first
{opts}
is_lua: only search Lua subdirs
Return:
list of absolute paths to the found files
nvim__id(
{obj}
)
nvim__id()
Returns object given as argument.
This API function is used for testing. One should not rely on its presence
in plugins.
Parameters:
{obj}
Object to return.
This API function is used for testing. One should not rely on its presence
in plugins.
Parameters:
{arr}
Array to return.
This API function is used for testing. One should not rely on its presence
in plugins.
Parameters:
{dct}
Dict to return.
nvim__id_float(
{flt}
)
nvim__id_float()
Returns floating-point value given as argument.
This API function is used for testing. One should not rely on its presence
in plugins.
Parameters:
{flt}
Value to return.
nvim__inspect_cell(
{grid}
,
{row}
,
{col}
)
nvim__inspect_cell()
NB: if your UI doesn't use hlstate, this will not return hlstate first
time.
nvim__invalidate_glyph_cache()
nvim__invalidate_glyph_cache()
For testing. The condition in schar_cache_clear_if_full is hard to reach,
so this function can be used to force a cache clear in a test.
nvim__redraw(
{opts}
)
nvim__redraw()
EXPERIMENTAL: this API may change in the future.
Instruct Nvim to redraw various components.
Parameters:
{opts}
Optional parameters.
win: Target a specific
window-ID as described below.
buf: Target a specific buffer number as described below.
flush: Update the screen with pending updates.
valid: When present mark win
, buf
, or all windows for
redraw. When true
, only redraw changed lines (useful for
decoration providers). When false
, forcefully redraw.
range: Redraw a range in
buf
, the buffer in
win
or the
current buffer (useful for decoration providers). Expects a
tuple
[first, last]
with the first and last line number of
the range, 0-based end-exclusive
api-indexing.
cursor: Immediately update cursor position on the screen in
win
or the current window.
statusline: Redraw the
'statusline' in
buf
,
win
or all
windows.
winbar: Redraw the
'winbar' in
buf
,
win
or all windows.
Return:
Map of various internal stats.
On execution error: fails with Vimscript error, updates v:errmsg.
Parameters:
{dict}
Dict, or String evaluating to a Vimscript
self dict
{fn}
Name of the function defined on the Vimscript dict
{args}
Function arguments packed in an Array
Return:
Result of the function call
nvim_call_function(
{fn}
,
{args}
)
nvim_call_function()
Calls a Vimscript function with the given arguments.
On execution error: fails with Vimscript error, updates v:errmsg.
Parameters:
{fn}
Function to call
{args}
Function arguments packed in an Array
Return:
Result of the function call
On execution error: fails with Vimscript error, updates v:errmsg.
Parameters:
{command}
Ex command string
nvim_eval(
{expr}
)
nvim_eval()
Evaluates a Vimscript
expression. Dicts and Lists are recursively
expanded.
On execution error: fails with Vimscript error, updates v:errmsg.
Parameters:
{expr}
Vimscript expression string
Return:
Evaluation result or expanded object
nvim_exec2(
{src}
,
{opts}
)
nvim_exec2()
Executes Vimscript (multiline block of Ex commands), like anonymous
:source.
Unlike
nvim_command() this function supports heredocs, script-scope
(s:), etc.
On execution error: fails with Vimscript error, updates v:errmsg.
Parameters:
{src}
Vimscript code
{opts}
Optional parameters.
output: (boolean, default false) Whether to capture and
return all (non-error, non-shell
:!) output.
Return:
Dict containing information about execution, with these keys:
output: (string|nil) Output if opts.output
is true.
Parameters:
{expr}
Expression to parse. Always treated as a single line.
{flags}
Flags:
"m" if multiple expressions in a row are allowed (only
the first one will be parsed),
"E" if EOC tokens are not allowed (determines whether
they will stop parsing process or be recognized as an
operator/space, though also yielding an error).
"l" when needing to start parsing with lvalues for
":let" or ":for". Common flag sets:
"m" to parse like for ":echo"
.
"E" to parse like for "<C-r>="
.
empty string for ":call".
"lm" to parse for ":let".
{highlight}
If true, return value will also include "highlight" key
containing array of 4-tuples (arrays) (Integer, Integer,
Integer, String), where first three numbers define the
highlighted region and represent line, starting column
and ending column (latter exclusive: one should highlight
region [start_col, end_col)).
Return:
AST: top-level dict with these keys:
"error": Dict with error, present only if parser saw some error.
Contains the following keys:
"message": String, error message in printf format, translated.
Must contain exactly one "%.*s".
"arg": String, error message argument.
"len": Amount of bytes successfully parsed. With flags equal to ""
that should be equal to the length of expr string. ("Successfully
parsed" here means "participated in AST creation", not "till the
first error".)
"ast": AST, either nil or a dict with these keys:
"type": node type, one of the value names from ExprASTNodeType
stringified without "kExprNode" prefix.
"start": a pair [line, column]
describing where node is
"started" where "line" is always 0 (will not be 0 if you will be
using this API on e.g. ":let", but that is not present yet).
Both elements are Integers.
"len": “length” of the node. This and "start" are there for
debugging purposes primary (debugging parser and providing debug
information).
"children": a list of nodes described in top/"ast". There always
is zero, one or two children, key will not be present if node
has no children. Maximum number of children may be found in
node_maxchildren array.
Local values (present only for certain nodes):
"scope": a single Integer, specifies scope for "Option" and
"PlainIdentifier" nodes. For "Option" it is one of ExprOptScope
values, for "PlainIdentifier" it is one of ExprVarScope values.
"ident": identifier (without scope, if any), present for "Option",
"PlainIdentifier", "PlainKey" and "Environment" nodes.
"name": Integer, register name (one character) or -1. Only present
for "Register" nodes.
"cmp_type": String, comparison type, one of the value names from
ExprComparisonType, stringified without "kExprCmp" prefix. Only
present for "Comparison" nodes.
"ccs_strategy": String, case comparison strategy, one of the value
names from ExprCaseCompareStrategy, stringified without
"kCCStrategy" prefix. Only present for "Comparison" nodes.
"augmentation": String, augmentation type for "Assignment" nodes.
Is either an empty string, "Add", "Subtract" or "Concat" for "=",
"+=", "-=" or ".=" respectively.
"invert": Boolean, true if result of comparison needs to be
inverted. Only present for "Comparison" nodes.
"ivalue": Integer, integer value for "Integer" nodes.
"fvalue": Float, floating-point value for "Float" nodes.
"svalue": String, value for "SingleQuotedString" and
"DoubleQuotedString" nodes.
Parameters:
{buffer}
Buffer handle, or 0 for current buffer.
See also:
nvim_create_user_command
Parameters:
{buffer}
Buffer handle, or 0 for current buffer.
{name}
Name of the command to delete.
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{opts}
Optional parameters. Currently not used.
Return:
Map of maps describing commands.
nvim_cmd(
{cmd}
,
{opts}
)
nvim_cmd()
Executes an Ex command.
Unlike
nvim_command() this command takes a structured Dict instead of a
String. This allows for easier construction and manipulation of an Ex
command. This also allows for things such as having spaces inside a
command argument, expanding filenames in a command that otherwise doesn't
expand filenames, etc. Command arguments may also be Number, Boolean or
String.
The first argument may also be used instead of count for commands that
support it in order to make their usage simpler with
vim.cmd(). For
example, instead of
vim.cmd.bdelete{ count = 2 }
, you may do
vim.cmd.bdelete(2)
.
On execution error: fails with Vimscript error, updates v:errmsg.
Parameters:
{cmd}
Command to execute. Must be a Dict that can contain the same
values as the return value of
nvim_parse_cmd() except
"addr", "nargs" and "nextcmd" which are ignored if provided.
All values except for "cmd" are optional.
{opts}
Optional parameters.
output: (boolean, default false) Whether to return command
output.
Return:
Command output (non-error, non-shell
:!) if
output
is true, else
empty string.
Example:
:call nvim_create_user_command('SayHello', 'echo "Hello world!"', {'bang': v:true})
:SayHello
Hello world!
Parameters:
{name}
Name of the new user command. Must begin with an uppercase
letter.
{command}
Replacement command to execute when this user command is
executed. When called from Lua, the command can also be a
Lua function. The function is called with a single table
argument that contains the following keys:
name: (string) Command name
args: (string) The args passed to the command, if any
<args>
fargs: (table) The args split by unescaped whitespace
(when more than one argument is allowed), if any <f-args>
bang: (boolean) "true" if the command was executed with a
! modifier <bang>
line1: (number) The starting line of the command range
<line1>
line2: (number) The final line of the command range
<line2>
range: (number) The number of items in the command range:
0, 1, or 2 <range>
count: (number) Any count supplied <count>
reg: (string) The optional register, if specified <reg>
mods: (string) Command modifiers, if any <mods>
smods: (table) Command modifiers in a structured format.
Has the same structure as the "mods" key of
nvim_parse_cmd().
Other parameters:
desc: (string) Used for listing the command when a Lua
function is used for {command}
.
force: (boolean, default true) Override any previous
definition.
Parameters:
{name}
Name of the command to delete.
nvim_get_commands(
{opts}
)
nvim_get_commands()
Gets a map of global (non-buffer-local) Ex commands.
Currently only
user-commands are supported, not builtin Ex commands.
Parameters:
{opts}
Optional parameters. Currently only supports {"builtin":false}
Return:
Map of maps describing commands.
Doesn't check the validity of command arguments.
Parameters:
{str}
Command line string to parse. Cannot contain "\n".
{opts}
Optional parameters. Reserved for future use.
Return:
Dict containing command information, with these keys:
cmd: (string) Command name.
range: (array) (optional) Command range (<line1>
<line2>
). Omitted
if command doesn't accept a range. Otherwise, has no elements if no
range was specified, one element if only a single range item was
specified, or two elements if both range items were specified.
count: (number) (optional) Command <count>
. Omitted if command
cannot take a count.
reg: (string) (optional) Command <register>
. Omitted if command
cannot take a register.
bang: (boolean) Whether command contains a <bang>
(!) modifier.
args: (array) Command arguments.
addr: (string) Value of
:command-addr. Uses short name or "line"
for -addr=lines.
nextcmd: (string) Next command if there are multiple commands
separated by a
:bar. Empty if there isn't a next command.
magic: (dict) Which characters have special meaning in the command
arguments.
file: (boolean) The command expands filenames. Which means
characters such as "%", "#" and wildcards are expanded.
bar: (boolean) The "|" character is treated as a command separator
and the double quote character (") is treated as the start of a
comment.
pattern: (string) Filter pattern. Empty string if there is no
filter.
force: (boolean) Whether filter is inverted or not.
tab: (integer)
:tab. -1 when omitted.
verbose: (integer)
:verbose. -1 when omitted.
split: (string) Split modifier string, is an empty string when
there's no split modifier. If there is a split modifier it can be
one of:
Return:
dict of all options
nvim_get_option_info2(
{name}
,
{opts}
)
nvim_get_option_info2()
Gets the option information for one option from arbitrary buffer or window
Resulting dict has keys:
shortname: Shortened name of the option (like
'ft')
type: type of option ("string", "number" or "boolean")
default: The default value for the option
was_set: Whether the option was set.
last_set_sid: Last set script id (if any)
last_set_linenr: line number where option was set
last_set_chan: Channel where option was set (0 for local)
scope: one of "global", "win", or "buf"
global_local: whether win or buf option has a global value
commalist: List of comma separated values
flaglist: List of single char flags
When {scope}
is not provided, the last set information applies to the
local value in the current buffer or window if it is available, otherwise
the global value information is returned. This behavior can be disabled by
explicitly specifying {scope}
in the {opts}
table.
Parameters:
{name}
Option name
{opts}
Optional parameters
win:
window-ID. Used for getting window local options.
buf: Buffer number. Used for getting buffer local options.
Implies {scope}
is "local".
Return:
Option Information
nvim_get_option_value(
{name}
,
{opts}
)
nvim_get_option_value()
Gets the value of an option. The behavior of this function matches that of
:set: the local value of an option is returned if it exists; otherwise,
the global value is returned. Local values always correspond to the
current buffer or window, unless "buf" or "win" is set in
{opts}
.
Parameters:
{name}
Option name
{opts}
Optional parameters
win:
window-ID. Used for getting window local options.
buf: Buffer number. Used for getting buffer local options.
Implies {scope}
is "local".
filetype:
filetype. Used to get the default option for a
specific filetype. Cannot be used with any other option.
Note: this will trigger
ftplugin and all
FileType
autocommands for the corresponding filetype.
nvim_set_option_value()
nvim_set_option_value(
{name}
,
{value}
,
{opts}
)
Sets the value of an option. The behavior of this function matches that of
:set: for global-local options, both the global and local value are set
unless otherwise specified with
{scope}
.
Note the options {win}
and {buf}
cannot be used together.
Parameters:
{name}
Option name
{value}
New option value
{opts}
Optional parameters
win:
window-ID. Used for setting window local option.
buf: Buffer number. Used for setting buffer local option.
For more information on buffers, see
buffers.
Buffers may be unloaded by the
:bunload command or the buffer's
'bufhidden' option. When a buffer is unloaded its file contents are freed
from memory and vim cannot operate on the buffer lines until it is reloaded
(usually by opening the buffer again in a new window). API methods such as
nvim_buf_get_lines() and
nvim_buf_line_count() will be affected.
nvim_buf_attach(
{buffer}
,
{send_buffer}
,
{opts}
)
nvim_buf_attach()
Activates buffer-update events on a channel, or as Lua callbacks.
Example (Lua): capture buffer updates in a global
events
variable (use
"vim.print(events)" to see its contents):
events = {}
vim.api.nvim_buf_attach(0, false, {
on_lines = function(...)
table.insert(events, {...})
end,
})
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{send_buffer}
True if the initial notification should contain the
whole buffer: first notification will be
nvim_buf_lines_event
. Else the first notification
will be nvim_buf_changedtick_event
. Not for Lua
callbacks.
{opts}
Optional parameters.
on_lines: Lua callback invoked on change. Return a
truthy value (not false
or nil
) to detach. Args:
the string "lines"
buffer handle
b:changedtick
first line that changed (zero-indexed)
last line that was changed
last line in the updated range
byte count of previous contents
deleted_codepoints (if utf_sizes
is true)
deleted_codeunits (if utf_sizes
is true)
on_bytes: Lua callback invoked on change. This
callback receives more granular information about the
change compared to on_lines. Return a truthy value
(not false
or nil
) to detach. Args:
the string "bytes"
buffer handle
b:changedtick
start row of the changed text (zero-indexed)
start column of the changed text
byte offset of the changed text (from the start of
the buffer)
old end row of the changed text (offset from start
row)
old end column of the changed text (if old end row
= 0, offset from start column)
old end byte length of the changed text
new end row of the changed text (offset from start
row)
new end column of the changed text (if new end row
= 0, offset from start column)
new end byte length of the changed text
on_changedtick: Lua callback invoked on changedtick
increment without text change. Args:
the string "changedtick"
buffer handle
b:changedtick
on_detach: Lua callback invoked on detach. Args:
the string "detach"
buffer handle
on_reload: Lua callback invoked on reload. The entire
buffer content should be considered changed. Args:
the string "reload"
buffer handle
utf_sizes: include UTF-32 and UTF-16 size of the
replaced region, as args to on_lines
.
preview: also attach to command preview (i.e.
'inccommand') events.
Return:
False if attach failed (invalid parameter, or buffer isn't loaded);
otherwise True. TODO: LUA_API_NO_EVAL
nvim_buf_call(
{buffer}
,
{fun}
)
nvim_buf_call()
Call a function with buffer as temporary current buffer.
This temporarily switches current buffer to "buffer". If the current
window already shows "buffer", the window is not switched. If a window
inside the current tabpage (including a float) already shows the buffer,
then one of these windows will be set as current window temporarily.
Otherwise a temporary scratch window (called the "autocmd window" for
historical reasons) will be used.
This is useful e.g. to call Vimscript functions that only work with the
current buffer/window currently, like
termopen().
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{fun}
Function to call inside the buffer (currently Lua callable
only)
Return:
Return value of function.
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
Note:
only deletes marks set in the buffer, if the mark is not set in the
buffer it will return false.
Parameters:
{buffer}
Buffer to set the mark on
{name}
Mark name
Return:
true if the mark was deleted, else false.
nvim_buf_del_var(
{buffer}
,
{name}
)
nvim_buf_del_var()
Removes a buffer-scoped (b:) variable
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{name}
Variable name
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{opts}
Optional parameters. Keys:
force: Force deletion and ignore unsaved changes.
unload: Unloaded only, do not delete. See
:bunload
nvim_buf_detach(
{buffer}
)
nvim_buf_detach()
Deactivates buffer-update events on the channel.
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
Return:
False if detach failed (because the buffer isn't loaded); otherwise
True.
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
Return:
b:changedtick
value.
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{mode}
Mode short-name ("n", "i", "v", ...)
Return:
Array of
maparg()-like dictionaries describing mappings. The
"buffer" key holds the associated buffer handle.
nvim_buf_get_lines()
nvim_buf_get_lines(
{buffer}
,
{start}
,
{end}
,
{strict_indexing}
)
Gets a line-range from the buffer.
Indexing is zero-based, end-exclusive. Negative indices are interpreted as
length+1+index: -1 refers to the index past the end. So to get the last
element use start=-2 and end=-1.
Out-of-bounds indices are clamped to the nearest valid value, unless
strict_indexing
is set.
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{start}
First line index
{end}
Last line index, exclusive
{strict_indexing}
Whether out-of-bounds should be an error.
Return:
Array of lines, or empty array for unloaded buffer.
nvim_buf_get_mark(
{buffer}
,
{name}
)
nvim_buf_get_mark()
Returns a
(row,col)
tuple representing the position of the named mark.
"End of line" column position is returned as
v:maxcol (big number). See
mark-motions.
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{name}
Mark name
Return:
(row, col) tuple, (0, 0) if the mark is not set, or is an
uppercase/file mark set in another buffer.
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
Line 1 (index=0) has offset 0. UTF-8 bytes are counted. EOL is one byte.
'fileformat' and
'fileencoding' are ignored. The line index just after the
last line gives the total byte-count of the buffer. A final EOL byte is
counted if it would be written, see
'eol'.
Unlike
line2byte(), throws error for out-of-bounds indexing. Returns -1
for unloaded buffer.
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{index}
Line index
Return:
Integer byte offset, or -1 for unloaded buffer.
nvim_buf_get_text()
nvim_buf_get_text(
{buffer}
,
{start_row}
,
{start_col}
,
{end_row}
,
{end_col}
,
{opts}
)
Gets a range from the buffer.
Indexing is zero-based. Row indices are end-inclusive, and column indices
are end-exclusive.
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{start_row}
First line index
{start_col}
Starting column (byte offset) on first line
{end_row}
Last line index, inclusive
{end_col}
Ending column (byte offset) on last line, exclusive
{opts}
Optional parameters. Currently unused.
Return:
Array of lines, or empty array for unloaded buffer.
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{name}
Variable name
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
Return:
true if the buffer is valid and loaded, false otherwise.
Note:
Even if a buffer is valid it may have been unloaded. See
api-buffer
for more info about unloaded buffers.
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
Return:
true if the buffer is valid, false otherwise.
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
Return:
Line count, or 0 for unloaded buffer.
api-buffer
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
nvim_buf_set_lines()
nvim_buf_set_lines(
{buffer}
,
{start}
,
{end}
,
{strict_indexing}
,
{replacement}
)
Sets (replaces) a line-range in the buffer.
Indexing is zero-based, end-exclusive. Negative indices are interpreted as
length+1+index: -1 refers to the index past the end. So to change or
delete the last element use start=-2 and end=-1.
To insert lines at a given index, set start
and end
to the same index.
To delete a range of lines, set replacement
to an empty array.
Out-of-bounds indices are clamped to the nearest valid value, unless
strict_indexing
is set.
Attributes:
not allowed when
textlock is active
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{start}
First line index
{end}
Last line index, exclusive
{strict_indexing}
Whether out-of-bounds should be an error.
{replacement}
Array of lines to use as replacement
nvim_buf_set_mark()
nvim_buf_set_mark(
{buffer}
,
{name}
,
{line}
,
{col}
,
{opts}
)
Sets a named mark in the given buffer, all marks are allowed
file/uppercase, visual, last change, etc. See
mark-motions.
Note:
Passing 0 as line deletes the mark
Parameters:
{buffer}
Buffer to set the mark on
{name}
Mark name
{line}
Line number
{col}
Column/row number
{opts}
Optional parameters. Reserved for future use.
Return:
true if the mark was set, else false.
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{name}
Buffer name
nvim_buf_set_text()
nvim_buf_set_text(
{buffer}
,
{start_row}
,
{start_col}
,
{end_row}
,
{end_col}
,
{replacement}
)
Sets (replaces) a range in the buffer
This is recommended over
nvim_buf_set_lines() when only modifying parts
of a line, as extmarks will be preserved on non-modified parts of the
touched lines.
Indexing is zero-based. Row indices are end-inclusive, and column indices
are end-exclusive.
To insert text at a given (row, column)
location, use
start_row = end_row = row
and start_col = end_col = col
. To delete the
text in a range, use replacement = {}
.
Attributes:
not allowed when
textlock is active
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{start_row}
First line index
{start_col}
Starting column (byte offset) on first line
{end_row}
Last line index, inclusive
{end_col}
Ending column (byte offset) on last line, exclusive
{replacement}
Array of lines to use as replacement
nvim_buf_set_var(
{buffer}
,
{name}
,
{value}
)
nvim_buf_set_var()
Sets a buffer-scoped (b:) variable
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{name}
Variable name
{value}
Variable value
nvim_buf_add_highlight()
nvim_buf_add_highlight(
{buffer}
,
{ns_id}
,
{hl_group}
,
{line}
,
{col_start}
,
{col_end}
)
Adds a highlight to buffer.
Useful for plugins that dynamically generate highlights to a buffer (like
a semantic highlighter or linter). The function adds a single highlight to
a buffer. Unlike
matchaddpos() highlights follow changes to line
numbering (as lines are inserted/removed above the highlighted line), like
signs and marks do.
Namespaces are used for batch deletion/updating of a set of highlights. To
create a namespace, use
nvim_create_namespace() which returns a
namespace id. Pass it in to this function as
ns_id
to add highlights to
the namespace. All highlights in the same namespace can then be cleared
with single call to
nvim_buf_clear_namespace(). If the highlight never
will be deleted by an API call, pass
ns_id = -1
.
As a shorthand,
ns_id = 0
can be used to create a new namespace for the
highlight, the allocated id is then returned. If
hl_group
is the empty
string no highlight is added, but a new
ns_id
is still returned. This is
supported for backwards compatibility, new code should use
nvim_create_namespace() to create a new empty namespace.
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{ns_id}
namespace to use or -1 for ungrouped highlight
{hl_group}
Name of the highlight group to use
{line}
Line to highlight (zero-indexed)
{col_start}
Start of (byte-indexed) column range to highlight
{col_end}
End of (byte-indexed) column range to highlight, or -1 to
highlight to end of line
Return:
The ns_id that was used
Lines are 0-indexed.
api-indexing To clear the namespace in the entire
buffer, specify line_start=0 and line_end=-1.
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{ns_id}
Namespace to clear, or -1 to clear all namespaces.
{line_start}
Start of range of lines to clear
{line_end}
End of range of lines to clear (exclusive) or -1 to
clear to end of buffer.
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{id}
Extmark id
Return:
true if the extmark was found, else false
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{id}
Extmark id
{opts}
Optional parameters. Keys:
details: Whether to include the details dict
hl_name: Whether to include highlight group name instead
of id, true if omitted
Return:
0-indexed (row, col) tuple or empty list () if extmark id was absent
Region can be given as (row,col) tuples, or valid extmark ids (whose
positions define the bounds). 0 and -1 are understood as (0,0) and (-1,-1)
respectively, thus the following are equivalent:
vim.api.nvim_buf_get_extmarks(0, my_ns, 0, -1, {})
vim.api.nvim_buf_get_extmarks(0, my_ns, {0,0}, {-1,-1}, {})
If end
is less than start
, traversal works backwards. (Useful with
limit
, to get the first marks prior to a given position.)
Note: when using extmark ranges (marks with a end_row/end_col position)
the overlap
option might be useful. Otherwise only the start position of
an extmark will be considered.
Note: legacy signs placed through the
:sign commands are implemented as
extmarks and will show up here. Their details array will contain a
sign_name
field.
Example:
local api = vim.api
local pos = api.nvim_win_get_cursor(0)
local ns = api.nvim_create_namespace('my-plugin')
-- Create new extmark at line 1, column 1.
local m1 = api.nvim_buf_set_extmark(0, ns, 0, 0, {})
-- Create new extmark at line 3, column 1.
local m2 = api.nvim_buf_set_extmark(0, ns, 2, 0, {})
-- Get extmarks only from line 3.
local ms = api.nvim_buf_get_extmarks(0, ns, {2,0}, {2,0}, {})
-- Get all marks in this buffer + namespace.
local all = api.nvim_buf_get_extmarks(0, ns, 0, -1, {})
vim.print(ms)
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{start}
Start of range: a 0-indexed (row, col) or valid extmark id
(whose position defines the bound).
api-indexing
{end}
End of range (inclusive): a 0-indexed (row, col) or valid
extmark id (whose position defines the bound).
api-indexing
{opts}
Optional parameters. Keys:
limit: Maximum number of marks to return
details: Whether to include the details dict
hl_name: Whether to include highlight group name instead
of id, true if omitted
overlap: Also include marks which overlap the range, even
if their start position is less than start
type: Filter marks by type: "highlight", "sign",
"virt_text" and "virt_lines"
Return:
List of
[extmark_id, row, col]
tuples in "traversal order".
By default a new extmark is created when no id is passed in, but it is
also possible to create a new mark by passing in a previously unused id or
move an existing mark by passing in its id. The caller must then keep
track of existing and unused ids itself. (Useful over RPC, to avoid
waiting for the return value.)
Using the optional arguments, it is possible to use this to highlight a
range of text, and also to associate virtual text to the mark.
If present, the position defined by end_col
and end_row
should be
after the start position in order for the extmark to cover a range. An
earlier end position is not an error, but then it behaves like an empty
range (no highlighting).
Parameters:
{buffer}
Buffer handle, or 0 for current buffer
{opts}
Optional parameters.
id : id of the extmark to edit.
end_row : ending line of the mark, 0-based inclusive.
end_col : ending col of the mark, 0-based exclusive.
hl_group : name of the highlight group used to highlight
this mark.
hl_eol : when true, for a multiline highlight covering the
EOL of a line, continue the highlight for the rest of the
screen line (just like for diff and cursorline highlight).
virt_text : virtual text to link to this mark. A list of
[text, highlight]
tuples, each representing a text chunk
with specified highlight.
highlight
element can either
be a single highlight group, or an array of multiple
highlight groups that will be stacked (highest priority
last). A highlight group can be supplied either as a
string or as an integer, the latter which can be obtained
using
nvim_get_hl_id_by_name().
virt_text_pos : position of virtual text. Possible values:
"eol": right after eol character (default).
"overlay": display over the specified column, without
shifting the underlying text.
"right_align": display right aligned in the window.
"inline": display at the specified column, and shift the
buffer text to the right as needed.
virt_text_win_col : position the virtual text at a fixed
window column (starting from the first text column of the
screen line) instead of "virt_text_pos".
virt_text_hide : hide the virtual text when the background
text is selected or hidden because of scrolling with
'nowrap' or
'smoothscroll'. Currently only affects
"overlay" virt_text.
virt_text_repeat_linebreak : repeat the virtual text on
wrapped lines.
hl_mode : control how highlights are combined with the
highlights of the text. Currently only affects virt_text
highlights, but might affect hl_group
in later versions.
"replace": only show the virt_text color. This is the
default.
"combine": combine with background text color.
"blend": blend with background text color. Not supported
for "inline" virt_text.
virt_lines : virtual lines to add next to this mark This
should be an array over lines, where each line in turn is
an array over
[text, highlight]
tuples. In general,
buffer and window options do not affect the display of the
text. In particular
'wrap' and
'linebreak' options do not
take effect, so the number of extra screen lines will
always match the size of the array. However the
'tabstop'
buffer option is still used for hard tabs. By default
lines are placed below the buffer line containing the
mark.
virt_lines_above: place virtual lines above instead.
virt_lines_leftcol: Place extmarks in the leftmost column
of the window, bypassing sign and number columns.
ephemeral : for use with
nvim_set_decoration_provider()
callbacks. The mark will only be used for the current
redraw cycle, and not be permantently stored in the
buffer.
right_gravity : boolean that indicates the direction the
extmark will be shifted in when new text is inserted (true
for right, false for left). Defaults to true.
end_right_gravity : boolean that indicates the direction
the extmark end position (if it exists) will be shifted in
when new text is inserted (true for right, false for
left). Defaults to false.
undo_restore : Restore the exact position of the mark if
text around the mark was deleted and then restored by
undo. Defaults to true.
invalidate : boolean that indicates whether to hide the
extmark if the entirety of its range is deleted. For
hidden marks, an "invalid" key is added to the "details"
array of
nvim_buf_get_extmarks() and family. If
"undo_restore" is false, the extmark is deleted instead.
priority: a priority value for the highlight group, sign
attribute or virtual text. For virtual text, item with
highest priority is drawn last. For example treesitter
highlighting uses a value of 100.
strict: boolean that indicates extmark should not be
placed if the line or column value is past the end of the
buffer or end of the line respectively. Defaults to true.
sign_text: string of length 1-2 used to display in the
sign column.
sign_hl_group: name of the highlight group used to
highlight the sign column text.
number_hl_group: name of the highlight group used to
highlight the number column.
line_hl_group: name of the highlight group used to
highlight the whole line.
cursorline_hl_group: name of the highlight group used to
highlight the sign column text when the cursor is on the
same line as the mark and
'cursorline' is enabled.
conceal: string which should be either empty or a single
character. Enable concealing similar to
:syn-conceal.
When a character is supplied it is used as
:syn-cchar.
"hl_group" is used as highlight for the cchar if provided,
otherwise it defaults to
hl-Conceal.
spell: boolean indicating that spell checking should be
performed within this extmark
ui_watched: boolean that indicates the mark should be
drawn by a UI. When set, the UI will receive win_extmark
events. Note: the mark is positioned by virt_text
attributes. Can be used together with virt_text.
url: A URL to associate with this extmark. In the TUI, the
OSC 8 control sequence is used to generate a clickable
hyperlink to this URL.
Return:
Id of the created/updated extmark
Namespaces can be named or anonymous. If name
matches an existing
namespace, the associated id is returned. If name
is an empty string a
new, anonymous namespace is created.
Parameters:
{name}
Namespace name or empty string
Return:
dict that maps from names to namespace ids.
This is a very general purpose interface for having Lua callbacks being
triggered during the redraw code.
The expected usage is to set
extmarks for the currently redrawn buffer.
nvim_buf_set_extmark() can be called to add marks on a per-window or
per-lines basis. Use the
ephemeral
key to only use the mark for the
current screen redraw (the callback will be called again for the next
redraw).
Note: this function should not be called often. Rather, the callbacks
themselves can be used to throttle unneeded callbacks. the on_start
callback can return false
to disable the provider until the next redraw.
Similarly, return false
in on_win
will skip the on_line
calls for
that window (but any extmarks set in on_win
will still be used). A
plugin managing multiple sources of decoration should ideally only set one
provider, and merge the sources internally. You can use multiple ns_id
for the extmarks set/modified inside the callback anyway.
Note: doing anything other than setting extmarks is considered
experimental. Doing things like changing options are not explicitly
forbidden, but is likely to have unexpected consequences (such as 100% CPU
consumption). Doing vim.rpcnotify
should be OK, but vim.rpcrequest
is
quite dubious for the moment.
Note: It is not allowed to remove or update extmarks in on_line
callbacks.
Parameters:
{opts}
Table of callbacks:
on_start: called first on each screen redraw
["start", tick]
on_buf: called for each buffer being redrawn (before window
callbacks)
["buf", bufnr, tick]
on_win: called when starting to redraw a specific window.
["win", winid, bufnr, toprow, botrow]
on_line: called for each buffer line being redrawn. (The
interaction with fold lines is subject to change)
["line", winid, bufnr, row]
on_end: called at the end of a redraw cycle
["end", tick]
nvim__ns_get(
{ns_id}
)
nvim__ns_get()
EXPERIMENTAL: this API will change in the future.
Get the properties for namespace
Parameters:
{ns_id}
Namespace
nvim__ns_set(
{ns_id}
,
{opts}
)
nvim__ns_set()
EXPERIMENTAL: this API will change in the future.
Set some properties for namespace
Parameters:
{ns_id}
Namespace
{opts}
Optional parameters to set:
wins: a list of windows to be scoped in
nvim_win_call(
{window}
,
{fun}
)
nvim_win_call()
Calls a function with window as temporary current window.
Parameters:
{window}
Window handle, or 0 for current window
{fun}
Function to call inside the window (currently Lua callable
only)
Return:
Return value of function.
Attributes:
not allowed when
textlock is active
Parameters:
{window}
Window handle, or 0 for current window
{force}
Behave like
:close!
The last window of a buffer with
unwritten changes can be closed. The buffer will become
hidden, even if
'hidden' is not set.
nvim_win_del_var(
{window}
,
{name}
)
nvim_win_del_var()
Removes a window-scoped (w:) variable
Parameters:
{window}
Window handle, or 0 for current window
{name}
Variable name
Parameters:
{window}
Window handle, or 0 for current window
nvim_win_get_cursor(
{window}
)
nvim_win_get_cursor()
Gets the (1,0)-indexed, buffer-relative cursor position for a given window
(different windows showing the same buffer have independent cursor
positions).
api-indexing
Parameters:
{window}
Window handle, or 0 for current window
Parameters:
{window}
Window handle, or 0 for current window
Return:
Height as a count of rows
Parameters:
{window}
Window handle, or 0 for current window
nvim_win_get_position(
{window}
)
nvim_win_get_position()
Gets the window position in display cells. First position is zero.
Parameters:
{window}
Window handle, or 0 for current window
Return:
(row, col) tuple with the window position
Parameters:
{window}
Window handle, or 0 for current window
Return:
Tabpage that contains the window
Parameters:
{window}
Window handle, or 0 for current window
{name}
Variable name
Parameters:
{window}
Window handle, or 0 for current window
Return:
Width as a count of columns
Attributes:
not allowed when
textlock is active
Parameters:
{window}
Window handle, or 0 for current window
Parameters:
{window}
Window handle, or 0 for current window
Return:
true if the window is valid, false otherwise
nvim_win_set_buf(
{window}
,
{buffer}
)
nvim_win_set_buf()
Sets the current buffer in a window, without side effects
Attributes:
not allowed when
textlock is active
Parameters:
{window}
Window handle, or 0 for current window
{buffer}
Buffer handle
nvim_win_set_cursor(
{window}
,
{pos}
)
nvim_win_set_cursor()
Sets the (1,0)-indexed cursor position in the window.
api-indexing This
scrolls the window even if it is not the current one.
Parameters:
{window}
Window handle, or 0 for current window
{pos}
(row, col) tuple representing the new position
Parameters:
{window}
Window handle, or 0 for current window
{height}
Height as a count of rows
nvim_win_set_hl_ns(
{window}
,
{ns_id}
)
nvim_win_set_hl_ns()
Set highlight namespace for a window. This will use highlights defined
with
nvim_set_hl() for this namespace, but fall back to global
highlights (ns=0) when missing.
Parameters:
{ns_id}
the namespace to use
nvim_win_set_var(
{window}
,
{name}
,
{value}
)
nvim_win_set_var()
Sets a window-scoped (w:) variable
Parameters:
{window}
Window handle, or 0 for current window
{name}
Variable name
{value}
Variable value
nvim_win_set_width(
{window}
,
{width}
)
nvim_win_set_width()
Sets the window width. This will only succeed if the screen is split
vertically.
Parameters:
{window}
Window handle, or 0 for current window
{width}
Width as a count of columns
nvim_win_text_height(
{window}
,
{opts}
)
nvim_win_text_height()
Computes the number of screen lines occupied by a range of text in a given
window. Works for off-screen text and takes folds into account.
Diff filler or virtual lines above a line are counted as a part of that
line, unless the line is on "start_row" and "start_vcol" is specified.
Diff filler or virtual lines below the last buffer line are counted in the
result when "end_row" is omitted.
Parameters:
{window}
Window handle, or 0 for current window.
{opts}
Optional parameters:
start_row: Starting line index, 0-based inclusive. When
omitted start at the very top.
end_row: Ending line index, 0-based inclusive. When
omitted end at the very bottom.
start_vcol: Starting virtual column index on "start_row",
0-based inclusive, rounded down to full screen lines. When
omitted include the whole line.
end_vcol: Ending virtual column index on "end_row",
0-based exclusive, rounded up to full screen lines. When
omitted include the whole line.
Return:
Dict containing text height information, with these keys:
all: The total number of screen lines occupied by the range.
fill: The number of diff filler or virtual lines among them.
nvim_open_win(
{buffer}
,
{enter}
,
{config}
)
nvim_open_win()
Opens a new split window, or a floating window if
relative
is specified,
or an external window (managed by the UI) if
external
is specified.
Floats are windows that are drawn above the split layout, at some anchor
position in some other window. Floats can be drawn internally or by
external GUI with the
ui-multigrid extension. External windows are only
supported with multigrid GUIs, and are displayed as separate top-level
windows.
The width
and height
of the new window must be specified when opening
a floating window, but are optional for normal windows.
If
relative
and
external
are omitted, a normal "split" window is
created. The
win
property determines which window will be split. If no
win
is provided or
win == 0
, a window will be created adjacent to the
current window. If -1 is provided, a top-level split will be created.
vertical
and
split
are only valid for normal windows, and are used to
control split direction. For
vertical
, the exact direction is determined
by
'splitright' and
'splitbelow'. Split windows cannot have
bufpos
/`row`/`col`/`border`/`title`/`footer` properties.
With relative=editor (row=0,col=0) refers to the top-left corner of the
screen-grid and (row=Lines-1,col=Columns-1) refers to the bottom-right
corner. Fractional values are allowed, but the builtin implementation
(used by non-multigrid UIs) will always round down to nearest integer.
Out-of-bounds values, and configurations that make the float not fit
inside the main editor, are allowed. The builtin implementation truncates
values so floats are fully within the main screen grid. External GUIs
could let floats hover outside of the main window like a tooltip, but this
should not be used to specify arbitrary WM screen positions.
Example (Lua): window-relative float
vim.api.nvim_open_win(0, false,
{relative='win', row=3, col=3, width=12, height=3})
Example (Lua): buffer-relative float (travels as buffer is scrolled)
vim.api.nvim_open_win(0, false,
{relative='win', width=12, height=3, bufpos={100,10}})
Example (Lua): vertical split left of the current window
vim.api.nvim_open_win(0, false, {
split = 'left',
win = 0
})
Attributes:
not allowed when
textlock is active
Parameters:
{buffer}
Buffer to display, or 0 for current buffer
{enter}
Enter the window (make it the current window)
{config}
Map defining the window configuration. Keys:
relative: Sets the window layout to "floating", placed at
(row,col) coordinates relative to:
"editor" The global editor grid
"win" Window given by the win
field, or current
window.
"cursor" Cursor position in current window.
"mouse" Mouse position
win:
window-ID window to split, or relative window when
creating a float (relative="win").
anchor: Decides which corner of the float to place at
(row,col):
"NW" northwest (default)
"NE" northeast
"SW" southwest
"SE" southeast
width: Window width (in character cells). Minimum of 1.
height: Window height (in character cells). Minimum of 1.
bufpos: Places float relative to buffer text (only when
relative="win"). Takes a tuple of zero-indexed
[line, column]
. row
and col
if given are applied
relative to this position, else they default to:
row=1
and col=0
if anchor
is "NW" or "NE"
row=0
and col=0
if anchor
is "SW" or "SE" (thus
like a tooltip near the buffer text).
row: Row position in units of "screen cell height", may be
fractional.
col: Column position in units of "screen cell width", may
be fractional.
focusable: Enable focus by user actions (wincmds, mouse
events). Defaults to true. Non-focusable windows can be
entered by
nvim_set_current_win(), or, when the
mouse
field is set to true, by mouse events.
mouse: Specify how this window interacts with mouse
events. Defaults to focusable
value.
If false, mouse events pass through this window.
If true, mouse events interact with this window
normally.
external: GUI should display the window as an external
top-level window. Currently accepts no other positioning
configuration together with this.
zindex: Stacking order. floats with higher zindex
go on
top on floats with lower indices. Must be larger than
zero. The following screen elements have hard-coded
z-indices:
100: insert completion popupmenu
200: message scrollback
250: cmdline completion popupmenu (when
wildoptions+=pum) The default value for floats are 50.
In general, values below 100 are recommended, unless
there is a good reason to overshadow builtin elements.
style: (optional) Configure the appearance of the window.
Currently only supports one value:
border: Style of (optional) window border. This can either
be a string or an array. The string values are
"none": No border (default).
"single": A single line box.
"double": A double line box.
"rounded": Like "single", but with rounded corners
("╭" etc.).
"solid": Adds padding by a single whitespace cell.
"shadow": A drop shadow effect by blending with the
background.
If it is an array, it should have a length of eight or
any divisor of eight. The array will specify the eight
chars building up the border in a clockwise fashion
starting with the top-left corner. As an example, the
double box style could be specified as:
[ "╔", "═" ,"╗", "║", "╝", "═", "╚", "║" ].
If the number of chars are less than eight, they will be
repeated. Thus an ASCII border could be specified as
[ "/", "-", \"\\\\\", "|" ],
or all chars the same as
[ "x" ].
An empty string can be used to turn off a specific border,
for instance,
[ "", "", "", ">", "", "", "", "<" ]
will only make vertical borders but not horizontal ones.
By default,
FloatBorder
highlight is used, which links
to
WinSeparator
when not defined. It could also be
specified by character:
[ ["+", "MyCorner"], ["x", "MyBorder"] ].
title: Title (optional) in window border, string or list.
List should consist of [text, highlight]
tuples. If
string, or a tuple lacks a highlight, the default
highlight group is FloatTitle
.
title_pos: Title position. Must be set with title
option. Value can be one of "left", "center", or "right".
Default is "left"
.
footer: Footer (optional) in window border, string or
list. List should consist of [text, highlight]
tuples.
If string, or a tuple lacks a highlight, the default
highlight group is FloatFooter
.
footer_pos: Footer position. Must be set with footer
option. Value can be one of "left", "center", or "right".
Default is "left"
.
noautocmd: If true then all autocommands are blocked for
the duration of the call.
fixed: If true when anchor is NW or SW, the float window
would be kept fixed even if the window would be truncated.
hide: If true the floating window will be hidden.
split: Split direction: "left", "right", "above", "below".
Return:
Window handle, or 0 on error
relative
is empty for normal windows.
Parameters:
{window}
Window handle, or 0 for current window
nvim_win_set_config(
{window}
,
{config}
)
nvim_win_set_config()
Configures window layout. Cannot be used to move the last window in a
tabpage to a different one.
When reconfiguring a window, absent option keys will not be changed.
row
/`col` and relative
must be reconfigured together.
Parameters:
{window}
Window handle, or 0 for current window
Parameters:
{tabpage}
Tabpage handle, or 0 for current tabpage
{name}
Variable name
Parameters:
{tabpage}
Tabpage handle, or 0 for current tabpage
Parameters:
{tabpage}
Tabpage handle, or 0 for current tabpage
{name}
Variable name
Parameters:
{tabpage}
Tabpage handle, or 0 for current tabpage
Parameters:
{tabpage}
Tabpage handle, or 0 for current tabpage
Return:
true if the tabpage is valid, false otherwise
Parameters:
{tabpage}
Tabpage handle, or 0 for current tabpage
Return:
List of windows in
tabpage
Parameters:
{tabpage}
Tabpage handle, or 0 for current tabpage
{name}
Variable name
{value}
Variable value
Parameters:
{tabpage}
Tabpage handle, or 0 for current tabpage
{win}
Window handle, must already belong to {tabpage}
Parameters:
{opts}
Parameters
event: (string|table) Examples:
event: "pat1"
event: { "pat1" }
event: { "pat1", "pat2", "pat3" }
pattern: (string|table)
pattern or patterns to match exactly.
For example, if you have *.py
as that pattern for the
autocmd, you must pass *.py
exactly to clear it.
test.py
will not match the pattern.
defaults to clearing all patterns.
NOTE: Cannot be used with {buffer}
buffer: (bufnr)
NOTE: Cannot be used with {pattern}
group: (string|int) The augroup name or id.
NOTE: If not passed, will only delete autocmds not in any
group.
To get an existing group id, do:
local id = vim.api.nvim_create_augroup("MyGroup", {
clear = false
})
Parameters:
{name}
String: The name of the group
{opts}
Dict Parameters
clear (bool) optional: defaults to true. Clear existing
commands if the group already exists
autocmd-groups.
Return:
Integer id of the created group.
nvim_create_autocmd(
{event}
,
{opts}
)
nvim_create_autocmd()
Creates an
autocommand event handler, defined by
callback
(Lua
function or Vimscript function name string) or
command
(Ex command
string).
Example using Lua callback:
vim.api.nvim_create_autocmd({"BufEnter", "BufWinEnter"}, {
pattern = {"*.c", "*.h"},
callback = function(ev)
print(string.format('event fired: %s', vim.inspect(ev)))
end
})
Example using an Ex command as the handler:
vim.api.nvim_create_autocmd({"BufEnter", "BufWinEnter"}, {
pattern = {"*.c", "*.h"},
command = "echo 'Entering a C or C++ file'",
})
Note: pattern
is NOT automatically expanded (unlike with
:autocmd),
thus names like "$HOME" and "~" must be expanded explicitly:
pattern = vim.fn.expand("~") .. "/some/path/*.py"
Parameters:
{event}
(string|array) Event(s) that will trigger the handler
(callback
or command
).
{opts}
Options dict:
group (string|integer) optional: autocommand group name or
id to match against.
pattern (string|array) optional: pattern(s) to match
literally
autocmd-pattern.
buffer (integer) optional: buffer number for buffer-local
autocommands
autocmd-buflocal. Cannot be used with
{pattern}
.
desc (string) optional: description (for documentation and
troubleshooting).
callback (function|string) optional: Lua function (or
Vimscript function name, if string) called when the
event(s) is triggered. Lua callback can return a truthy
value (not
false
or
nil
) to delete the autocommand.
Receives one argument, a table with these keys:
event-args
id: (number) autocommand id
group: (number|nil) autocommand group id, if any
match: (string) expanded value of <amatch>
buf: (number) expanded value of <abuf>
file: (string) expanded value of <afile>
command (string) optional: Vim command to execute on event.
Cannot be used with {callback}
once (boolean) optional: defaults to false. Run the
autocommand only once
autocmd-once.
nested (boolean) optional: defaults to false. Run nested
autocommands
autocmd-nested.
Return:
Autocommand id (number)
NOTE: behavior differs from
:augroup-delete. When deleting a group,
autocommands contained in this group will also be deleted and cleared.
This group will no longer exist.
Parameters:
{id}
Integer The id of the group.
NOTE: behavior differs from
:augroup-delete. When deleting a group,
autocommands contained in this group will also be deleted and cleared.
This group will no longer exist.
Parameters:
{name}
String The name of the group.
Parameters:
{event}
(String|Array) The event or events to execute
{opts}
Dict of autocommand options:
group (string|integer) optional: the autocommand group name
or id to match against.
autocmd-groups.
pattern (string|array) optional: defaults to "*"
autocmd-pattern. Cannot be used with
{buffer}
.
buffer (integer) optional: buffer number
autocmd-buflocal. Cannot be used with
{pattern}
.
modeline (bool) optional: defaults to true. Process the
modeline after the autocommands <nomodeline>
.
nvim_get_autocmds(
{opts}
)
nvim_get_autocmds()
Get all autocommands that match the corresponding
{opts}
.
These examples will get autocommands matching ALL the given criteria:
-- Matches all criteria
autocommands = vim.api.nvim_get_autocmds({
group = "MyGroup",
event = {"BufEnter", "BufWinEnter"},
pattern = {"*.c", "*.h"}
})
-- All commands from one group
autocommands = vim.api.nvim_get_autocmds({
group = "MyGroup",
})
NOTE: When multiple patterns or events are provided, it will find all the
autocommands that match any combination of them.
Parameters:
{opts}
Dict with at least one of the following:
group (string|integer): the autocommand group name or id to
match against.
event (string|array): event or events to match against
autocmd-events.
pattern (string|array): pattern or patterns to match against
autocmd-pattern. Cannot be used with
{buffer}
buffer: Buffer number or list of buffer numbers for buffer
local autocommands
autocmd-buflocal. Cannot be used with
{pattern}
Return:
Array of autocommands matching the criteria, with each item containing
the following fields:
id (number): the autocommand id (only when defined with the API).
group (integer): the autocommand group id.
group_name (string): the autocommand group name.
desc (string): the autocommand description.
event (string): the autocommand event.
command (string): the autocommand command. Note: this will be empty
if a callback is set.
callback (function|string|nil): Lua function or name of a Vim script
function which is executed when this autocommand is triggered.
once (boolean): whether the autocommand is only run once.
buflocal (boolean): true if the autocommand is buffer local.
buffer (number): the buffer number.
nvim_ui_attach(
{width}
,
{height}
,
{options}
)
nvim_ui_attach()
Activates UI events on the channel.
Entry point of all UI clients. Allows
--embed to continue startup.
Implies that the client is ready to show the UI. Adds the client to the
list of UIs.
nvim_list_uis()
Note:
If multiple UI clients are attached, the global screen dimensions
degrade to the smallest client. E.g. if client A requests 80x40 but
client B requests 200x100, the global screen has size 80x40.
Parameters:
{width}
Requested screen columns
{height}
Requested screen rows
nvim_ui_pum_set_bounds()
nvim_ui_pum_set_bounds(
{width}
,
{height}
,
{row}
,
{col}
)
Tells Nvim the geometry of the popupmenu, to align floating windows with
an external popup menu.
Note that this method is not to be confused with
nvim_ui_pum_set_height(), which sets the number of visible items in the
popup menu, while this function sets the bounding box of the popup menu,
including visual elements such as borders and sliders. Floats need not use
the same font size, nor be anchored to exact grid corners, so one can set
floating-point numbers to the popup menu geometry.
Parameters:
{width}
Popupmenu width.
{height}
Popupmenu height.
{row}
Popupmenu row.
{col}
Popupmenu height.
nvim_ui_pum_set_height(
{height}
)
nvim_ui_pum_set_height()
Tells Nvim the number of elements displaying in the popupmenu, to decide
<PageUp>
and
<PageDown>
movement.
Parameters:
{height}
Popupmenu height, must be greater than zero.
nvim_ui_set_focus(
{gained}
)
nvim_ui_set_focus()
Tells the nvim server if focus was gained or lost by the GUI
nvim_ui_term_event(
{event}
,
{value}
)
nvim_ui_term_event()
Tells Nvim when a terminal event has occurred
The following terminal events are supported:
"termresponse": The terminal sent an OSC or DCS response sequence to
Nvim. The payload is the received response. Sets
v:termresponse and
fires
TermResponse.
Parameters:
{event}
Event name
{value}
Event payload
nvim_ui_try_resize_grid()
nvim_ui_try_resize_grid(
{grid}
,
{width}
,
{height}
)
Tell Nvim to resize a grid. Triggers a grid_resize event with the
requested grid size or the maximum size if it exceeds size limits.
On invalid grid handle, fails with error.
Parameters:
{grid}
The handle of the grid to be changed.
{width}
The new requested width.
{height}
The new requested height.