wellle / targets.vim Goto Github PK

Targets.vim is a Vim plugin that adds various text objects to give you more targets to operate on. It expands on the idea of simple commands like di' (delete inside the single quotes around the cursor) to give you more opportunities to craft powerful commands that can be repeated reliably. One major goal is to handle all corner cases correctly.

The following examples are displayed as three lines each. The top line denotes cursor positions from where the presented command works. The middle line shows the contents of the example line that we're working on. The last line shows the part of the line that the command will operate on.

To change the text in the next pair of parentheses, use the cin) command

cursor position │    .....................
buffer line     │    This is example text (with a pair of parentheses).
selection       │                          └───────── cin) ─────────┘

To delete the item in a comma separated list under the cursor, use da,

cursor position │                                  .........
buffer line     │    Shopping list: oranges, apples, bananas, tomatoes
selection       │                                  └─ da, ─┘

Notice how the selection includes exactly one of the surrounding commas to leave a proper comma separated list behind.

Targets.vim comes with five kinds for text objects:

• Pair text objects
• Quote text objects
• Separator text objects
• Argument text objects
• Tag text objects

Each of those kinds is implemented by a targets source. Third party plugins can provide additional sources to add even more text objects which behave like the built in ones. See plugins for details on how to implement your own targets source.

These text objects are similar to the built in text objects such as i). Supported trigger characters:

• ( ) (work on parentheses)
• { } B (work on curly braces)
• [ ] (work on square brackets)
• < > (work on angle brackets)
• t (work on tags)

Pair text objects work over multiple lines and support seeking. See below for details about seeking.

The following examples will use parentheses, but they all work for each listed trigger character accordingly.

i( i) i{ i} iB i[ i] i< i> it

• Select inside of pair characters.
• This overrides Vim's default text object to allow seeking for the next pair in the current line to the right or left when the cursor is not inside a pair. This behavior is similar to Vim's seeking behavior of di' when not inside of quotes, but it works both ways.
• Accepts a count to select multiple blocks.

      ............
a ( b ( cccccccc ) d ) e
   │   └── i) ──┘   │
   └───── 2i) ──────┘

a( a) a{ a} aB a[ a] a< a> at

• Select a pair including pair characters.
• Overrides Vim's default text object to allow seeking.
• Accepts a count.

      ............
a ( b ( cccccccc ) d ) e
  │   └─── a) ───┘   │
  └────── 2a) ───────┘

I( I) I{ I} IB I[ I] I< I> It

• Select contents of pair characters.
• Like inside of parentheses, but exclude whitespace at both ends. Useful for changing contents while preserving spacing.
• Accepts a count.

      ............
a ( b ( cccccccc ) d ) e
    │   └─ I) ─┘   │
    └──── 2I) ─────┘

A( A) A{ A} AB A[ A] A< A> At

• Select around pair characters.
• Like a pair, but include whitespace at one side of the pair. Prefers to select trailing whitespace, falls back to select leading whitespace.
• Accepts a count.

      ............
a ( b ( cccccccc ) d ) e
  │   └─── A) ────┘   │
  └────── 2A) ────────┘

in( an( In( An( il( al( Il( Al( ...

Work directly on distant pairs without moving there separately.

All the above pair text objects can be shifted to the next pair by including the letter n. The command in) selects inside of the next pair. Use the letter l instead to work on the previous (last) pair. Uses a count to skip multiple pairs. Skipping works over multiple lines.

See our Cheat Sheet for two charts summarizing all pair mappings.

If any of the normal pair commands (not containing n or l) is executed when the cursor is not positioned inside a pair, it seeks for pairs before or after the cursor by searching for the appropriate delimiter on the current line. This is similar to using the explicit version containing n or l, but in only seeks on the current line.

These text objects are similar to the built in text objects such as i'. Supported trigger characters:

• ' (work on single quotes)
• " (work on double quotes)
• ` (work on back ticks)

These quote text objects try to be smarter than the default ones. They count the quotation marks from the beginning of the line to decide which of these are the beginning of a quote and which ones are the end.

If you type ci" on the , in the example below, it will automatically skip and change world instead of changing , between hello and world.

buffer │ join("hello", "world")
proper │      └─────┘  └─────┘
false  │            └──┘

Quote text objects work over multiple lines and support seeking. See below for details about seeking.

The following examples will use single quotes, but they all work for each mentioned separator character accordingly.

i' i" i`

• Select inside quote.
• This overrides Vim's default text object to allow seeking in both directions.

  ............
a ' bbbbbbbb ' c ' d ' e
   └── i' ──┘

a' a" a`

• Select a quote.
• This overrides Vim's default text object to support seeking.
• Unlike Vim's quote text objects, this incudes no surrounding whitespace.

  ............
a ' bbbbbbbb ' c ' d ' e
  └─── a' ───┘

I' I" I`

• Select contents of a quote.
• Like inside quote, but exclude whitespace at both ends. Useful for changing contents while preserving spacing.

  ............
a ' bbbbbbbb ' c ' d ' e
    └─ I' ─┘

A' A" A`

• Select around a quote.
• Like a quote, but include whitespace in one direction. Prefers to select trailing whitespace, falls back to select leading whitespace.

  ............
a ' bbbbbbbb ' c ' d ' e
  └─── A' ────┘

in' In' An' il' Il' Al' ...

Work directly on distant quotes without moving there separately.

All the above pair text objects can be shifted to the next quote by including the letter n. The command in' selects inside of the next single quotes. Use the letter l instead to work on the previous (last) quote. Uses a count to skip multiple quotation characters.

See our Cheat Sheet for a chart summarizing all quote mappings.

If any of the normal quote commands (not containing n or l) is executed when the cursor is not positioned inside a quote, it seeks for quotes before or after the cursor by searching for the appropriate delimiter on the current line. This is similar to using the explicit version containing n or l.

These text objects are based on single separator characters like the comma in one of our examples above. The text between two instances of the separator character can be operated on with these targets.

Supported separators:

, . ; : + - = ~ _ * # / | \ & $

Separator text objects work over multiple lines and support seeking.

The following examples will use commas, but they all work for each listed separator character accordingly.

i, i. i; i: i+ i- i= i~ i_ i* i# i/ i| i\ i& i$

• Select inside separators. Similar to in quote.

      ...........
a , b , cccccccc , d , e
       └── i, ──┘

a, a. a; a: a+ a- a= a~ a_ a* a# a/ a| a\ a& a$

• Select an item in a list separated by the separator character.
• Includes the leading separator, but excludes the trailing one. This leaves a proper list separated by the separator character after deletion. See the examples above.

      ...........
a , b , cccccccc , d , e
      └─── a, ──┘

I, I. I; I: I+ I- I= I~ I_ I* I# I/ I| I\ I& I$

• Select contents between separators.
• Like inside separators, but exclude whitespace at both ends. Useful for changing contents while preserving spacing.

      ...........
a , b , cccccccc , d , e
        └─ I, ─┘

A, A. A; A: A+ A- A= A~ A_ A* A# A/ A| A\ A& A$

• Select around a pair of separators.
• Includes both separators and a surrounding whitespace, similar to a' and A(.

      ...........
a , b , cccccccc , d , e
      └─── A, ────┘

in, an, In, An, il, al, Il, Al, ...

Work directly on distant separators without moving there separately.

All the above separator text objects can be shifted to the next separator by including the letter n. The command in, selects inside of the next commas. Use the letter l instead to work on the previous (last) separators. Uses the count to skip multiple separator characters.

See our Cheat Sheet for a chart summarizing all separator mappings.

Like quote seeking. If any of the normal separator commands (not containing n or l) is executed when the cursor is not positioned inside a pair of separators, it seeks for the separator before or after the cursor. This is similar to using the explicit version containing n or l.

These text objects are similar to separator text objects, but are specialized for arguments surrounded by braces and commas. They also take matching braces into account to capture only valid arguments.

Argument text objects work over multiple lines and support seeking.

ia

• Select inside arguments. Similar to in quote.
• Accepts a count.

      ...........
a , b ( cccccccc , d ) e
       └── ia ──┘

aa

• Select an argument in a list of arguments.
• Includes a separator if preset, but excludes surrounding braces. This leaves a proper argument list after deletion.
• Accepts a count.

      ...........
a , b ( cccccccc , d ) e
        └─── aa ──┘

Ia

• Select content of an argument.
• Like inside separators, but exclude whitespace at both ends. Useful for changing contents while preserving spacing.
• Accepts a count.

      ...........
a , b ( cccccccc , d ) e
        └─ Ia ─┘

Aa

• Select around an argument.
• Includes both delimiters and a surrounding whitespace, similar to a' and A(.
• Accepts a count.

      ...........
a , b ( cccccccc , d ) e
      └─── Aa ────┘

ina ana Ina Ana ila ala Ila Ala

Work directly on distant arguments without moving there separately.

All the above argument text objects can be shifted to the next argument by including the letter n. The command ina selects inside of the next argument. Use the letter l instead to work on the previous (last) argument. Uses a [count] to skip multiple argument characters. The order is determined by the nearest surrounding argument delimiter.

See our Cheat Sheet for a chart summarizing all argument mappings.

Like separator seeking. If any of the normal argument commands (not containing n or l) is executed when the cursor is not positioned inside an argument, it seeks for the argument before or after the cursor. This is similar to using the explicit version containing n or l.

Two multi text objects are included in default settings. See the section on settings below to see how to set up other similar multi text objects or customize the built in ones.

inb anb Inb Anb ilb alb Ilb Alb

Similar to pair text objects, if you type dib within () it will delete in these. If you do the same within {} it will delete in those. If you type d2inb it will skip one next pair (any kind) and delete in the one after (any kind). If you're within () nested in {}, type d2ib to delete in {}. All of the usual seeking, growing and skipping works.

inq anq Inq Anq ilq alq Ilq Alq

Similar to quote text objects, if you type diq within "" it will delete in these. If you do the same within '' it will delete in those. If you type d2inq it will skip one next quote text object (any kind) and delete in the one after (any kind). If you're within "" nested in '', type d2iq to delete in ''. All of the usual seeking, growing and skipping works.

You can customize the mappings and text objects with the settings described here.

Default:

let g:targets_aiAI = 'aiAI'

Controls the normal mode operator mode maps that get created for In Pair (i), A Pair (a), Inside Pair (I), and Around Pair (A). Required to be either a string or a list with 4 characters/elements.

Use a space to deactivate a mode. If you want to use multiple keys, for example <Space>a instead of A, you must use a list.

In contrast to g:targets_nl, special keys must not be escaped with a backslash. For example, use "<Space>" or '<Space>', not "\<Space>". Example for configuring g:targets_aiAI:

let g:targets_aiAI = ['<Space>a', '<Space>i', '<Space>A', '<Space>I']

Default:

let g:targets_mapped_aiAI = g:targets_aiAI

If you can't get your g:targets_aiAI settings to work because they conflict with other mappings you have, you might need to use g:targets_mapped_aiAI. For example if you want to map k to i and use k as i in targets mappings, you need to NOT map k to i in operator pending mode, and set g:targets_aiAI = 'akAI' and g:targets_mapped_aiAI = 'aiAI'.

Has the same format as g:targets_aiAI.

For more details see issue #213 and don't hesitate to comment there or open a new issue if you need assistance.

Default:

let g:targets_nl = 'nl'

Controls the keys used in maps for seeking next and last text objects. For example, if you want n to always search for the next object and N to search for the last, you could set:

let g:targets_nl = 'nN'

Required to be either a string or a list with 2 characters/elements.

Use a space to deactivate a mode. If you want to use multiple keys, for example <Space>n instead of n, you must use a list.

In contrast to g:targets_aiAI, special keys must be escaped with a backslash. For example, use "\<Space>", not "<Space>" nor '<Space>'. Example for configuring g:targets_nl:

let g:targets_nl = ["\<Space>n", "\<Space>l"]

Default:

let g:targets_seekRanges = 'cc cr cb cB lc ac Ac lr rr ll lb ar ab lB Ar aB Ab AB rb al rB Al bb aa bB Aa BB AA'

Defines a priority ordered, space separated list of range types which can be used to customize seeking behavior.

The default setting generally prefers targets around the cursor, with one exception: If the target around the cursor is not contained in the current cursor line, but the next or last target are, then prefer those. Targets beginning or ending on the cursor are preferred over everything else.

Some other useful example settings:

Prefer multiline targets around cursor over distant targets within cursor line:

let g:targets_seekRanges = 'cc cr cb cB lc ac Ac lr lb ar ab lB Ar aB Ab AB rr ll rb al rB Al bb aa bB Aa BB AA'

Never seek backwards:

let g:targets_seekRanges = 'cc cr cb cB lc ac Ac lr rr lb ar ab lB Ar aB Ab AB rb rB bb bB BB'

Only seek if next/last targets touch current line:

let g:targets_seekRanges = 'cc cr cb cB lc ac Ac lr rr ll lb ar ab lB Ar aB Ab AB rb rB al Al'

Only consider targets fully visible on screen:

let g:targets_seekRanges = 'cc cr cb cB lc ac Ac lr lb ar ab rr rb bb ll al aa'

Only consider targets around cursor:

let g:targets_seekRanges = 'cc cr cb cB lc ac Ac lr lb ar ab lB Ar aB Ab AB'

Only consider targets fully contained in current line:

let g:targets_seekRanges = 'cc cr cb cB lc ac Ac lr rr ll'

If you want to build your own, or are just curious what those cryptic letters mean, check out the full documentation in our Cheat Sheet.

Default:

let g:targets_jumpRanges = 'bb bB BB aa Aa AA'

Defines an unordered, space separated list of range types which can be used to customize the jumplist behavior (see documentation on seek ranges). It controls whether or not to add the cursor position prior to selecting the text object to the jumplist.

The default setting adds the previous cursor position to the jumplist if the target that was operated on doesn't intersect the cursor line. That means it adds a jumplist entry if the target ends above the cursor line or starts below the cursor line.

Some other useful example settings (or build your own!):

Never add cursor position to jumplist:

let g:targets_jumpRanges = ''

Always add cursor position to jumplist:

let g:targets_jumpRanges = 'cr cb cB lc ac Ac lr rr ll lb ar ab lB Ar aB Ab AB rb al rB Al bb aa bB Aa BB AA'

Only add to jumplist if cursor was not inside the target:

let g:targets_jumpRanges = 'rr rb rB bb bB BB ll al Al aa Aa AA'

Default:

let g:targets_gracious = 0

If enabled (set to 1) , both growing and seeking will work on the largest available count if a too large count is given. For example:

• v100ab will select the most outer block around the cursor
• v100inq will select the most distant quote to the right/down (the last one in the file)

This function can be used to modify an internal dictionary used to control the mappings. The default value of that dictionary is:

{
    \ '(': {'pair': [{'o': '(', 'c': ')'}]},
    \ ')': {'pair': [{'o': '(', 'c': ')'}]},
    \ '{': {'pair': [{'o': '{', 'c': '}'}]},
    \ '}': {'pair': [{'o': '{', 'c': '}'}]},
    \ 'B': {'pair': [{'o': '{', 'c': '}'}]},
    \ '[': {'pair': [{'o': '[', 'c': ']'}]},
    \ ']': {'pair': [{'o': '[', 'c': ']'}]},
    \ '<': {'pair': [{'o': '<', 'c': '>'}]},
    \ '>': {'pair': [{'o': '<', 'c': '>'}]},
    \ '"': {'quote': [{'d': '"'}]},
    \ "'": {'quote': [{'d': "'"}]},
    \ '`': {'quote': [{'d': '`'}]},
    \ ',': {'separator': [{'d': ','}]},
    \ '.': {'separator': [{'d': '.'}]},
    \ ';': {'separator': [{'d': ';'}]},
    \ ':': {'separator': [{'d': ':'}]},
    \ '+': {'separator': [{'d': '+'}]},
    \ '-': {'separator': [{'d': '-'}]},
    \ '=': {'separator': [{'d': '='}]},
    \ '~': {'separator': [{'d': '~'}]},
    \ '_': {'separator': [{'d': '_'}]},
    \ '*': {'separator': [{'d': '*'}]},
    \ '#': {'separator': [{'d': '#'}]},
    \ '/': {'separator': [{'d': '/'}]},
    \ '\': {'separator': [{'d': '\'}]},
    \ '|': {'separator': [{'d': '|'}]},
    \ '&': {'separator': [{'d': '&'}]},
    \ '$': {'separator': [{'d': '$'}]},
    \ 't': {'tag': [{}]},
    \ 'a': {'argument': [{'o': '[([]', 'c': '[])]', 's': ','}]},
    \ 'b': {'pair': [{'o':'(', 'c':')'}, {'o':'[', 'c':']'}, {'o':'{', 'c':'}'}]},
    \ 'q': {'quote': [{'d':"'"}, {'d':'"'}, {'d':'`'}]},
    \ }

The keys in this dictionary correspond to the trigger character. For example if you type di(, ( is the trigger and gets mapped to the pair target source with arguments 'o':'(' (opening) and 'c':')' (closing). Sources quote and separator have argument 'd' (delimiter), tag has no arguments and argument text objects take 'o' (opening), 'c' (closing) and 's' (separator). Notably the b (any block) and q (any quote) triggers map to one source with three sets of pair and quote argument dictionaries respectively. That means if you type dib each of those sources get taken into account to pick the proper target. Also note that it's even possible to have one target mapped to multiple different sources, so you can select any of those different text objects (see example below).

You can use the targets#mappings#extend() function to modify these internal mappings. For example if you wanted to switch b back to the Vim default behavior of operating on parentheses only, you can add this to your vimrc:

autocmd User targets#mappings#user call targets#mappings#extend({
    \ 'b': {'pair': [{'o':'(', 'c':')'}]}
    \ })

Note that you should always use that autocmd prefix to make sure your modifications get applied at the right time. There's a similar autogroup for plugins which can add other sources and default mappings, which gets triggered before this #user one. That way the user mappings always take precedence over the plugins default mappings

If you want to remove a mapping from the defaults, just set it to an empty list of sources:

autocmd User targets#mappings#user call targets#mappings#extend({
    \ 'q': {},
    \ })

That way targets.vim will ignore it and fall back to Vim default behavior, which for the case of q does nothing.

Finally here's a more complex example which adds two triggers s (any separator text object) and @ (anything at all). So you could type das to delete the closest separator text object near the cursor, or da@ to operate on the closest text object available via targets.vim. All of those support seeking and counts like d3ins.

autocmd User targets#mappings#user call targets#mappings#extend({
    \ 's': { 'separator': [{'d':','}, {'d':'.'}, {'d':';'}, {'d':':'}, {'d':'+'}, {'d':'-'},
    \                      {'d':'='}, {'d':'~'}, {'d':'_'}, {'d':'*'}, {'d':'#'}, {'d':'/'},
    \                      {'d':'\'}, {'d':'|'}, {'d':'&'}, {'d':'$'}] },
    \ '@': {
    \     'separator': [{'d':','}, {'d':'.'}, {'d':';'}, {'d':':'}, {'d':'+'}, {'d':'-'},
    \                   {'d':'='}, {'d':'~'}, {'d':'_'}, {'d':'*'}, {'d':'#'}, {'d':'/'},
    \                   {'d':'\'}, {'d':'|'}, {'d':'&'}, {'d':'$'}],
    \     'pair':      [{'o':'(', 'c':')'}, {'o':'[', 'c':']'}, {'o':'{', 'c':'}'}, {'o':'<', 'c':'>'}],
    \     'quote':     [{'d':"'"}, {'d':'"'}, {'d':'`'}],
    \     'tag':       [{}],
    \     },
    \ })

Also note how this example shows that you can set multiple triggers in a single targets#mappings#extend() call. To keep the autocmd overhead minimal I'd recommend to keep all your mappings setup in a single such call.

If you have set any of the following settings in your vimrc, they will still be respected when creating the default mappings dictionary. But it's not possible to set up any multi source targets (like any block or any quote) this way. It's recommended to retire those legacy settings and use targets#mappings#extend() as described above.

g:targets_pairs
g:targets_quotes
g:targets_separators
g:targets_tagTrigger
g:targets_argClosing
g:targets_argOpening
g:targets_argSeparator
g:targets_argTrigger

However, those new mappings settings will only be respected when targets.vim can use expression mappings, which need Neovim or Vim with version 7.3.338 or later. If you are using an older Vim version, these legacy settings are still the only way to do any customization. Please refer to an older version of this README (before October 2018) for details. Or open an issue for me to describe those legacy settings somewhere still.

• Repeating an operator-pending mapping forgets its last count. Works since Vim 7.4.160

• Empty matches can't be selected because it is not possible to visually select zero-character ranges.
• Forcing motion to work linewise by inserting V in dVan( doesn't work for operator-pending mappings. See :h o_v.
• Report issues or submit pull requests to github.com/wellle/targets.vim.

Create more mappings to support commands like danw or danp to delete the next word or paragraph.