A Filetype plugin for csv files
Vim script Makefile
Latest commit a771e04 Feb 8, 2017 @chrisbra updated vimball



This plugin is used for handling column separated data with Vim. Usually those files are called csv files and use the ',' as delimiter, though sometimes they use e.g. the '|' or ';' as delimiter and there also exists fixedwidth columns. The aim of this plugin is to ease handling these kinds of files.

This is a filetype plugin for CSV files. It was heavily influenced by the Vim Wiki Tip667, though it works differently.


Table of Contents


In order to have vim automatically detect csv files, you need to have ftplugins enabled (e.g. by having this line in your .vimrc file:

:filetype plugin on

The plugin already sets up some logic to detect CSV files. By default, the plugin recognizes *.csv and *.dat files as CSV filetype. In order that the CSV filetype plugin is loaded correctly, vim needs to be enabled to load filetype-plugins. This can be ensured by putting a line like this in your .vimrc:

:filetype plugin on

(see also filetype-plugin-on).

In case this did not work, you need to setup vim like this:

To have Vim automatically detect csv files, you need to do the following.

  1. Create your user runtime directory if you do not have one yet. This directory needs to be in your 'runtime' path. In Unix this would typically the ~/.vim directory, while in Windows this is usually your ~/vimfiles directory. Use :echo expand("~") to find out, what Vim thinks your user directory is. To create this directory, you can do:

    :!mkdir ~/.vim

    for Unix and

    :!mkdir ~/vimfiles

    for Windows.

  2. In that directory you create a file that will detect csv files.

    if exists("did_load_csvfiletype")
    let did_load_csvfiletype=1
    augroup filetypedetect
      au! BufRead,BufNewFile *.csv,*.dat    setfiletype csv
    augroup END

    You save this file as "filetype.vim" in your user runtime diretory:

    :w ~/.vim/filetype.vim
  3. To be able to use your new filetype.vim detection, you need to restart Vim. Vim will then load the csv filetype plugin for all files whose names end with .csv.


The CSV ftplugin provides several Commands. All commands are also provided with the prefix :CSV (e.g. :CSVNrColumns)


If you would like to know, on which column the cursor is, use




Use the bang attribute, if you have a heading in the first line and you want to know the name of the column in which the cursor is:



:NrColumns and :CSVNrColumns outputs the maximum number of columns available. It does this by testing the first 10 lines for the number of columns. This usually should be enough. If you use the '!' attribute, it outputs the number of columns in the current line.


Use :SearchInColumn or :CSVSearchInColumn to search for a pattern within a specific column. The usage is:

:SearchInColumn [<nr>] /{pat}/

So if you would like to search in Column 1 for the word foobar, you enter

:SearchInColumn 1 /foobar/

Instead of / as delimiter, you can use any other delimiter you like. If you don't enter a column, the current column will be used.


:HiColumn or :CSVHiColumn can be used to highlight Column . Currently the plugin uses the WildMenu Highlight Group. If you would like to change this, you need to define the variable g:csv_hiGroup.

If you do not specify a , HiColumn will highlight the column on which the cursor is. Use


to remove any highlighting.

If you want to automatically highlight a column, see csv-hicol


If you would like all columns to be visually arranged, you can use the :ArrangeColumn or :CSVArrangeColumn command:

:[range]ArrangeColumn[!] [<Row>]

Beware, that this will change your file and depending on the size of your file may slow down Vim significantly. This is highly experimental. :ArrangeCommand will try to vertically align all columns by their maximum column size. While the command is run, a progressbar in the statusline 'stl' will be shown.

Use the bang attribute to force recalculating the column width. This is slower, but especially if you have modified the file, this will correctly calculate the width of each column so that they can be correctly aligned. If no column width has been calculated before, the width will be calculated, even if the '!' has not been given.

If is given, will use the Row, to calculate the width, else will calculate the maximum of at least the first 10,000 rows to calculate the width. The limit of 10,000 is set to speed up the processing and can be overriden by setting the "b:csv_arrange_use_all_rows" variable (see below).

If [range] is not given, it defaults to the current line.

By default, the columns will be right-aligned. If you want a different alignment you need to specify this through the b:csv_arrange_align variable. This is a string of flags ('r': right align, 'l': left align, 'c': center alignment, '.': decimal alignment) where each flag defines the alignment for a particular column (starting from left). Missing columns will be right aligned. You can use '*' to repeat the previous value until the end. So this:

:let b:csv_arrange_align = 'lc.'

Will left-align the first column, center align the second column, decimal align the third column and all following columns right align. (Note: decimal aligning might slow down Vim and additionally, if the value is no decimal number it will be right aligned). And this:

:let b:csv_arrange_align = 'l*'

will left align all columns.

If you change the alignment parameter, you need to use the "!" attribute, the next time you run the :ArrangeCol command, otherwise for performance reasons, it won't be considered.

Note, arranging the columns can be very slow on large files or many columns (see csv-slow on how to increase performance for this command). For large files, calculating the column width can take long and take a consierable amount of memory. Therefore, the csv plugin will at most check 10.000 lines for the width. Set the variable b:csv_arrange_use_all_rows to 1 to use all records:

:let b:csv_arrange_use_all_rows = 1

(this could however in the worst case lead to a crash).

To disable the statusline progressbar set the variable g:csv_no_progress:

:let g:csv_no_progress = 1

This will disable the progressbar and slightly improve performance (since no additional redraws are needed).

Note: this command does not work for fixed width columns csv-fixedwidth

See also csv-arrange-autocmd on how to have vim automaticaly arrange a CSV file upon entering it.


If you would like to undo a previous :ArrangeColumn command, you can use this :UnArrangeColumn or :CSVUnArrangeColumn command:


Beware, that is no exact undo of the :ArrangeColumn command, since it strips away all leading blanks for each column. So if previously a column contained only some blanks, this command will strip all blanks.

If [range] is given, it defaults to the current line.


The command :DeleteColumn or :CSVDeleteColumn can be used to delete a specific column.

:DeleteColumn 2

will delete column 2.

If you don't specify a column number, it will delete the column on which the cursor is. Alternatively, you can also specify a search string. The plugin will then delete all columns that match the pattern:

:DeleteColumn /foobar

will delete all columns where the pattern "foobar" matches.


Reinitialize the Plugin. Use this, if you have changed the configuration of the plugin (see csv-configuration ). If you use the bang (!) attribute, it will keep the b:delimiter configuration variable.

Header Lines

The :Header or :CSVHeader command splits the csv-buffer and adds a window, that holds a small fraction of the csv file. This is useful, if the first line contains some kind of a heading and you want always to display it. This works similar to fixing a certain line at the top. As optional argument, you can give the number of columns from the top, that shall be displayed. By default, 1 is used (You can define youre own default by setting the b:csv_headerline variable, see csv-header). Use the '!' to close this window. So this

:Header 3

opens at the top a split window, that holds the first 3 lines, is fixed and horizontally 'scrollbind'ed to the csv window and highlighted using the CSVHeaderLine highlighting. To close the header window, use


Note, this won't work with linebreaks in the column.

Note also, that if you already have a horizontal header window (VHeader_CSV), this command will close the horizontal Header window. This is because of a limitation of Vim itself, which doesn't allow to sync the scrolling between two windows horizontally and at the same time have another window only sync its scrolling vertically.

Note: this command does not work for fixed width columns csv-fixedwidth

If you want a vertical header line, use :VHeader or :CSVVHeader. This works similar to the Header_CSV command, except that it will open a vertical split window with the first column always visible. It will always open the first column in the new split window. Use the '!' to close the window. If you specify a count, that many columns will be visible (default: the first). Add the bang to the count, if you only want the specific column to be visible.

:VHeader 2

This will open a vertical split window containing the first 2 columns, while

:VHeader 2!

Opens a new vertical split window containing only the 2 second column.

Note, this won't work with linebreaks in the column. Note also: this command does not work for fixed width columns csv-fixedwidth

Use the :HeaderToggle and :VHeaderToggle command to toggle displaying the horizontal or vertical header line. Alternatively, use :CSVHeaderToggle or :CSVVHeaderToggle


The command :Sort or :CSVSort can be used to sort the csv file on a certain column. If no range is given, is sorts the whole file. Specify the column number to sort on as argument. Use the '!' attribute to reverse the sort order. For example, the following command sorts line 1 til 10 on the 3 column

:1,10Sort 3

While this command

:1,10Sort! 3

reverses the order based on column 3.

The column number can be optionally followed by any of the flags [i], [n], [x] and [o] for [i]gnoring case, sorting by [n]umeric, he[x]adecimal or [o]ctal value.

When no column number is given, it will sort by the column, on which the cursor is currently.

Copy Column

If you need to copy a specific column, you can use the command :CSVColumn or :Column

:[N]Column [a]

Copy column N into register a. This will copy all the values, that are not folded-away (csv-filter) and skip comments.

If you don't specify N, the column of the current cursor position is used. If no register is given, the default register quotequote is used.

Move A Column

You can move one column to the right of another column by using the :CSVMoveColumn or :MoveColumn command

:[range]MoveColumn [source] [dest]

This moves the column number source to the right of column nr destination. If both arguments are not given, move the column on which the cursor is to the right of the current last column. If [range] is not given, MoveColumn moves the entire column, otherwise, it moves the columns only for the lines within the range, e.g. given that your first line is a header line, which you don't want to change

:2,$MoveColumn 1 $

this would move column 1 behind the last column, while keeping the header line as is.

Sum of a Column

You can let Vim output the sum of a column using the :CSVSumCol or :SumCol command

:[range]SumCol [nr] [/format/]

This outputs the result of the column within the range given. If no range is given, this will calculate the sum of the whole column. If is not given, this calculates the sum for the column the cursor is on. Note, that the delimiter will be stripped away from each value and also empty values won't be considered.

By default, Vim uses the a numerical format that uses the '.' as decimal separator while there is no thousands separator. If youre file contains the numbers in a different format, you can use the /format/ option to specify a different thousands separator or a different decimal separator. The format needs to be specified like this:


where 'x' defines the thousands separator and y defines the decimal separator and each one is optional. This means, that

:SumCol 1 /:,/

uses the default thousands separator and ',' as the decimal separator and

:SumCol 2 / :./

uses the Space as thousands separator and the '.' as decimal separator.

Note, if you Vim is compiled without floating point number format (+float), Vim will only aggregate the integer part and therefore won't use the 'y' argument in the /format/ specifier.

See also csv-aggregate-functions

Create new Records

If you want to create one or several records, you can use the :NewRecord or :CSVNewRecord command:

:[range]NewRecord [count]

This will create in each line given by range [count] number of new empty records. If [range] is not specified, creates a new line below the line the cursor is on and if count is not given, it defaults to 1.

Change the delimiter

If you want to change the field delimiter of your file you can use the :CSVNewDelimiter or :NewDelimiter command:

:NewDelimiter char

This changes the field delimiter of your file to the new delimiter "char". Note: Will remove trailing delimiters.

Check for duplicate records

If you want to check the file for duplicate records, use the command :Duplicate or :CSVDuplicate:

:Duplicate columnlist

Columnlist needs to be a numeric comma-separated list of all columns that you want to check. You can also use a range like '2-5' which means the plugin should check columns 2,3,4 and 5.

If the plugin finds a duplicate records, it outputs its line number (but it only does that at most 10 times).

Normal mode commands

The csv filetype plugin redefines the following keys as:

Key Effect
<C-Right> or L or W Move [count] field forwards
<C-Left> or E or H Move [count] field backwards (but see csv-mapping-H for the movement of H).
<Up> or K Move [count] lines upwards within the same column
<Down> or J Move [count] lines downwards within the same column
<Enter> Dynamically fold all lines away, that don't match the value in the current column. See csv-filter In Replace-mode and Virtual-Replace-mode does not create a new row, but instead moves the cursor to the beginning of the same column, one more line below.
<Space> Dynamically fold all lines away, that match the value in the current column. See csv-filter
<BS> Remove last item from the dynamic filter. See csv-filter

Note how the mapping of 'H' differs from 'E'

H step fields backwards but also stops at where the content of the columns begins.

If you look into this example (with the cursor being '|')

aaa,   bbbb,|ccc

Pressing 'H' moves to

aaa,   |bbbb,ccc

Pressing 'H' again moves to

aaa,|   bbbb,ccc

Pressing 'H' again moves to

|aaa,   bbbb,ccc

While with 'E', the cursor moves to:

 aaa,|  bbbb,ccc

and pressing 'E' again, it would move directly to

|aaa,   bbbb,ccc

Also, the csv plugin defines these text-object:

Field Meaning
if Inner Field (contains everything up to the delimiter)
af Outer Field (contains everything up to and including the delimiter)
iL Inner Line (visually linewise select all lines, that has the same value at the cursor's column)

Note, that the <BS>, <CR>, K and J overlap Vim's default mapping for <CR>, <BS>, J and K respectively. Therefore, this functionality has been mapped to a sane default of <Localleader>J and <LocalLeader>K. If you haven't changed the <Leader> or <LocalLeader> variables, those the <Localleader> is equival to a single backslash '\', e.g. \K would run the lookup function on the word under the cursor and \J would join this line with the previous line.

If you want to prevent the mapping of keys, simply set the global variable g:csv_nomap_ to 1, e.g. to prevent mapping of in csv files, put

let g:csv_nomap_cr = 1

into your .vimrc. Note, the keyname must be lower case.

Converting a CSV File

You can convert your CSV file to a different format with the :ConvertData or :CSVConvertData command


Use the the ! attribute, to convert your data without the delimiter.

This command will interactively ask you for the definition of 3 variables. After which it will convert your csv file into a new format, defined by those 3 variables and open the newly created file in a new window. Those 3 variables define how the text converted.

First, You need to define what has to be done, before converting your column data. That is done with the "pre convert" variable. The content of this variable will be put in front of the new document.

Second, you define, what has to be put after the converted content of your column data. This happens with the "post convert" variable. Basically the contents of this variable will be put after processing the columns.

Last, the columns need to be converted into your format. For this you can specify a printf() format like string, that defines how your data will be converted. You can use '%s' to specify placeholders, which will later be replaced by the content of the actual column.

For example, suppose you want to convert your data into HTML, then you first call the


At this point, Vim will ask you for input. First, you need to specify, what needs to be done before processing the data:

Pre convert text: <html><body><table>

This would specify to put the HTML Header before the actual data can be processed. If the variable g:csv_pre_convert is already defined, Vim will already show you its' content as default value. Simply pressing Enter will use this data. After that, Vim asks, what the end of the converted file needs to look like:

Post convert text: </table></body></html>

So here you are defining how to finish up the HTML file. If the variable g:csv_post_convert is already defined, Vim will already show you its' content as default value which you can confirm by pressing Enter. Last, you define, how your columns need to be converted. Again, Vim asks you for how to do that:

Converted text, use %s for column input:

This time, you can use '%s' expandos. They tell Vim, that they need to be replaced by the actual content of your file. It does by going from the first column in your file and replacing it with the corresponding %s in that order. If there are less '%s' expandos then columns in your file, Vim will skip the columns, that are not used. Again If the variable g:csv_convert is already defined, Vim will already show you its' content as default value which you can confirm by pressing Enter.

After you hit Enter, Vim will convert your data and put it into a new window. It may look like this:


Note, this is only a proof of concept. A better version of converting your data to HTML is bundled with Vim (:TOhtml).

But may be you want your data converted into SQL-insert statements. That could be done like this:

Pre convert text:

(Leave this empty. It won't be used).

Post convert text: Commit;

After inserting the data, commit it into the database.

Converted text, use %s for column input:
Insert into table foobar values ('%s', '%s', %s);

Note, that the last argument is not included within single quotation marks, since in this case the data is assumed to be integer and won't need to be quoted for the database.

After hitting Enter, a new Window will be opened, which might look like this:

Insert into table foobar values('Foobar', '2', 2011);
Insert into table foobar values('Bar', '1', 2011);

Since the command was used with the bang attribute (!), the converted data doesn't include the column delimiters.

Now you can copy it into your database, or further manipulate it.

Dynamic filters

If you are on a value and only want to see lines that have the same value in this column, you can dynamically filter the file and fold away all lines not matching the value in the current column. To do so, simply press (Enter). Now Vim will fold away all lines, that don't have the same value in this particular row. Note, that leading blanks and the delimiter is removed and the value is used literally when comparing with other values. If you press on the value, all fields having the same value will be folded away.

The way this is done is, that the value from the column is extracted and a regular expression for that field is generated from it. In the end this regular expression is used for folding the file.

A subsequent or on another value, will add this value to the current applied filter (this is like using the logical AND between the currently active filter and the new value). To remove the last item from the filter, press (backspace). If all items from the filter are removed, folding will be disabled.

If some command messes up the folding, you can use zX to have the folding being reinitialized.

By default, the first line is assumed to be the header and won't be folded away. See also csv-header.

If you have set the g:csv_move_folds variable and the file is modifiable, all folded lines will be moved to the end of the file, so you can view all non-folded lines as one consecutive area (see also csv-move-folds)

To see the active filters, you can use the :Filter or :CSVFilter command. This will show you a small summary, of what filters are active and looks like this:

Nr Match Col Name Value
01 - 07 Price 23.10
02 + 08 Qty 10

This means, there are two filters active. The current active filter is on column 7 (column name is Price) and all values that match 23.10 will be folded away AND all values that don't match a value of 10 in the QTY column will also be folded away. When removing one item from the filter by pressing , it will always remove the last item (highest number in NR column) from the active filter values.

Note, that depending on your csv file and the number of filters you used, applying the filter might actually slow down vim, because a complex regular expression is generated that is applied by the fold expression. Look into the @/ (quote_/) register to see its value.

Use zX to apply the current value of your search register as filter. Use


to reapply all values from the current active filter and fold non-matching items away.

Analyze a Column

If you'd like to know, how the values are distributed among a certain column, you can use the :CSVAnalyze or :Analyze command. So

:Analyze 3

outputs the the distribution of the top 5 values in column 3. This looks like this:

Nr Count % Value
01 20 50% 10
02 10 25% 2
03 10 25% 5

This tells you, that the the value '10' in column 3 occurs 50% of the time (exactly 20 times) and the other 2 values '2' and '5' occur only 10 times, so 25% of the time.

Vertical Folding

Sometimes, you want to hide away certain columns to better view only certain columns without having to horizontally scroll. You can use the :CSVVertFold or :VertFold command to hide certain columns:

:VertFold [<nr>]

This will hide all columns from the first until the number entered. It currently can't hide single columns, because of the way, syntax highlighting is used. This command uses the conceal-feature :syn-conceal to hide away those columns. If no nr is given, hides all columns from the beginning till the current column.



to display all hidden columns again.

Transposing a column

Transposing means to exchange rows and columns. You can transpose the csv file, using the :CSVTranspose or :Transpose:


command. If [range] is not given, it will transpose the complete file, otherwise it will only transpose the lines in the range given. Note, comments will be deleted and transposing does not work with fixed-width columns.

Transforming into a table

You can also transform your csv data into a visual table, using the :CSVTabularize or :CSVTable:


command. This will make a frame around your csv data and substitute all delimiters by '|', so that it will look like a table.

e.g. consider this data:


This will be transformed into:

| First| Second| Third|
|    10|      5|     2|
|     5|      2|    10|
|     2|     10|     5|
|    10|      5|     2|

If your Vim uses an unicode 'encoding', the plugin makes a nice table using special unicode drawing glyphs (but it might be possible, that those chars are not being displayed correctly, if either your terminal or the gui font doesn't have characters for those codepoints). If you use the bang form, each row will be separated by a line. You can also visual select a range of lines and use :Tabularize to have only that range converted into a nice ascii table. Else it try to use the current paragraph and try to transform it.

If you use the '!' bang argument, between each row, a line will be drawn.

In csv files, you can also use the :CSVTabularize command, in different filetypes you can use the :CSVTable command (and is available as plugin so it will be available for non-CSV filetypes).

Set the variable g:csv_table_leftalign=1 if you want the columns to be leftaligned.

Note: Each row must contain exactly as many fields as columns.

Add new empty columns

If you want to add new empty columns to your file you can use the :CSVAddColumn or :AddColumn command:

:[range]AddColumn [column] [count]

By default, this works for the whole file, but you can give a different range to which the AddColumn command applies. If no arguments are given, the new empty column will be added after the column on which the cursor is. You can however add as first argument the column number after which the new column needs to be added.

Additionally, you can also add a count number to add several columns at once after the specified column number. You 0 for the column number, if you want to add several columns after the current column.

Substitute in columns

If you want to substitute only in specific columns, you can use the :CSVSubstitute or :Substitute command:

:[range]Substitute [column/]pattern/string[/flags]

This means in the range and within the given columns replace pattern by string. This works bascially like the :s command, except that you MUST use forward slashes / to delimit the command. The optional part [column/] can take either the form of an address or if you leave it out, substitution will only happen in the current column. Additionally, you can use the 1,5/ form to substitute within the columns 1 till 5 or you can even use 1,$ which means to substitute in each column (so in fact this simplifies to a simple :s command whithin the given range. For the use of [/flags] see :s_flags Here are some examples:

:%Substitute 1,4/foobar/baz/gce

Substitutes in the whole file in columns 1 till 4 the pattern foobar by baz for every match ('g' flag) and asks for confirmation ('c' flag).

:%S 3,$/(\d\+)/\1 EUR/g

Substitutes in each column starting from the third each number and appends the EURO suffix to it.

Count Values inside a Column

You can let Vim output the number of values inside a column using the :CSVCountCol command

:[range]CountCol [nr] [distinct]

This outputs the number of [distinct] values visible in the column [nr] If [distinct] is not given, count's all values. Note, header rows and folded rows won't be counted.

See also csv-aggregate-functions

Maximum/Minimum value of a Column

You can let Vim output the 10 maximum/minimum values of a column using the :CSVMaxCol command

:[range]MaxCol [nr][distinct] [/format/]
:[range]MinCol [nr][distinct] [/format/]

This outputs the result of the column within the range given. If no range is given, this will calculate the max value of the whole column. If is not given, this calculates the sum for the column the cursor is on. Note, that the delimiter will be stripped away from each value and also empty values won't be considered.

By default, Vim uses the a numerical format that uses the '.' as decimal separator while there is no thousands separator. If youre file contains the numbers in a different format, you can use the /format/ option to specify a different thousands separator or a different decimal separator. The format needs to be specified like this: /x:y/ where 'x' defines the thousands separator and y defines the decimal separator and each one is optional. This means, that

:MaxCol 1 /:,/

uses the default thousands separator and ',' as the decimal separator and

:MaxCol 2 / :./

uses the Space as thousands separator and the '.' as decimal separator.

If [distinct] is given, only returns the number of distinct values.

Note, if you Vim is compiled without floating point number format (+float), Vim will only aggregate the integer part and therefore won't use the 'y' argument in the /format/ specifier.

See also csv-aggregate-functions

Average value of a Column

You can let Vim output the value of a column using the :CSVAvgCol command

:[range]AvgCol [nr] [/format/]

This outputs the result of the column <nr> within the range given. If no range is given, this will calculate the average value of the whole column. If is not given, this calculates the sum for the column the cursor is on. Note, that the delimiter will be stripped away from each value and also empty values won't be considered.

For the [/format/] part, see MaxCol_CSV.

See also csv-aggregate-functions

Duplicate columns

If you want to add duplicate an existing column you can use the :CSVDupColumn or :DupColumn command:

:[range]DupColumn [column] [count]

By default, this works for the whole file, but you can give a different range to which the command applies. By default it will duplicate the column on which the cursor is, but you can add as first argument <column> which column will be duplicated.

Additionally, you can also provide a <count> to copy several columns at once.

CSV Configuration

The CSV plugin tries to automatically detect the field delimiter for your file, cause although often the file is called CSV (comma separated values), a semicolon is actually used. The column separator is stored in the buffer-local variable b:delimiter. This delimiter is heavily used, because you need it to define a column. Almost all commands use this variable therefore.


To override the automatic detection of the plugin and define the separator manually, use:

:let g:csv_delim=','

to let the comma be the delimiter. This sets the buffer local delimiter variable b:delimiter.

If your file does not consist of delimited columns, but rather is a fixed width csv file, see csv-fixedwidth for configuring the plugin appropriately.

If you changed the delimiter, you should reinitialize the plugin using InitCSV

Note: the delimiter will be used to generate a regular expression that matches a column. Therefore, you need to escape special characters. So instead of '^' use '^'.


The definition, of what a column is, is defined as buffer-local variable b:col. By default this variable is initialized to:

let b:col='\%(\%([^' . b:delimiter . ']*"[^"]*"[^' . b:delimiter . ']*'
    \. b:delimiter . '\)\|\%([^' . b:delimiter . ']*\%(' . b:delimiter
    \. '\|$\)\)\)'

This should take care of quoted delimiters within a column. Those should obviously not count as a delimiter. This regular expression is quite complex and might not always work on some complex cases (e.g. linebreaks within a field, see RFC4180 for some ugly cases that will probably not work with this plugin).

If you changed the b:delimiter variable, you need to redefine the b:col variable, cause otherwise it will not reflect the change. To change the variable from the comma to a semicolon, you could call in your CSV-Buffer this command:

:let b:col=substitute(b:col, ',', ';', 'g')

Check with :echo b:col, if the definition is correct afterwards.

You can also force the plugin to use your own defined regular expression as column. That regular expression should include the delimiter for the columns. To define your own regular expression, set the g:csv_col variable:

let g:csv_col='[^,]*,'

This defines a column as a field delimited by the comma (where no comma can be contained inside a field), similar to how csv-strict works.

You should reinitialize the plugin afterwards InitCSV

Highlighting Group

By default the csv ftplugin uses the WildMenu highlighting Group to define how the HiColumn command highlights columns. If you would like to define a different highlighting group, you need to set this via the g:csv_hiGroup variable. You can e.g. define it in your .vimrc:

:let g:csv_hiGroup = "IncSearch"

You need to restart Vim, if you have changed this variable or use InitCSV

The hl-Title highlighting is used for the Header line that is created by the Header_CSV command. If you prefer a different highlighting, set the g:csv_hiHeader variable to the prefered highlighting:

let g:csv_hiHeader = 'Pmenu'

This would set the header window to the hl-Pmenu highlighting, that is used for the popup menu. To disable the custom highlighting, simply unlet the variable:

unlet g:csv_hiHeader

You should reinitialize the plugin afterwards InitCSV

Strict Columns

The default regular expression to define a column is quite complex (csv-column). This slows down the processing and makes Vim use more memory and it could still not fit to your specific use case.

If you know, that in your data file, the delimiter cannot be contained inside the fields quoted or escaped, you can speed up processing (this is quite noticeable when using the ArrangeColumn_CSV command) by setting the g:csv_strict_columns variable:

let g:csv_strict_columns = 1

This would define a column as this regex:

let b:col = '\%([^' . b:delimiter . ']*' . b:delimiter . '\`$\)'

Much simpler then the default column definition, isn't it? See also csv-column and csv-delimiter

You can disable the effect if you unlet the variable:

unlet g:csv_strict_columns

You should reinitialize the plugin afterwards InitCSV

For example when opening a CSV file you get the Error E363: pattern uses more memory than 'maxmempattern'. In this case, either increase the 'maxmempattern' or set the g:csv_strict_columns variable.


The CSV plugin comes with a function to syntax highlight csv files. Basically allt it does is highlight the columns and the header line.

By default, the delimiter will not be displayed, if Vim supports conceal of syntax items and instead draws a vertical line. If you don't want that, simply set the g:csv_noconceal variable in your .vimrc

let g:csv_no_conceal = 1

and to disable it, simply unlet the variable

unlet g:csv_no_conceal

You should reinitialize the plugin afterwards InitCSV Note: You can also set the 'conceallevel' option to control how the concealed chars will be displayed.

If you want to customize the syntax colors, you can define your own groups. The CSV plugin will use already defined highlighting groups, if they are already defined, otherwise it will define its own defaults which should be visible with 8, 16, 88 and 256 color terminals. For that it uses the CSVColumnHeaderOdd and CSVColumnHeaderEven highlight groups for syntax coloring the first line. All other lines get either the CSVColumnOdd or CSVColumnEven highlighting.

In case you want to define your own highlighting groups, you can define your own syntax highlighting like this in your .vimrc

hi CSVColumnEven term=bold ctermbg=4 guibg=DarkBlue
hi CSVColumnOdd  term=bold ctermbg=5 guibg=DarkMagenta
hi CSVColumnHeaderEven ...
hi CSVColumnHeaderOdd ...

Alternatively, you can simply link those highlighting groups to some other ones, you really like:

hi link CSVColumnOdd MoreMsg
hi link CSVColumnEven Question

If you do not want column highlighting, set the variable g:csv_no_column_highlight to 1

:let g:csv_no_column_highlight = 1

Note, these changes won't take effect, until you restart Vim.


RFC4180 allows newlines in double quoted strings. By default, the csv-plugin won't recognize newlines inside fields. It is however possible to make the plugin aware of newlines within quoted strings. To enable this, set

let g:csv_nl = 1

and to disable it again, simply unset the variable

unlet g:csv_nl

It is a good idea to reinitialize the plugin afterwards InitCSV

Note, this might not work correctly in all cases. The syntax highlighting seems to change on cursor movements. This could possibly be a bug in the syntax highlighting engine of Vim. Also, WhatColumn_CSV can't handle newlines inside fields and will most certainly be wrong.

Highlight column automatically

You can let vim automatically highlight the column on which the cursor is. This works by defining an CursorMoved autocommand to always highlight the column, when the cursor is moved in normal mode. Note, this does not update the highlighting, if the Cursor is moved in Insert mode. To enable this, define the g:csv_highlight_column variable like this

let g:csv_highlight_column = 'y'

and to disable it again, simply unset the variable

unlet g:csv_highlight_column

It is a good idea to reinitialize the plugin afterwards InitCSV

Fixed width columns

Sometimes there are no real columns, but rather the file is fixed width with no distinct delimiters between each column. The CSV plugin allows you to handle such virtual columns like csv columns, if you define where each column starts.

Note: Except for ArrangeColumn_CSV and the Header_CSV commands, all commands work in either mode. Those two commands won't do anything in the case of fixedwidth columns, since they don't really make sense here.

Manual setup

You can do this, by setting the buffer-local variable b:csv_fixed_width like this

let b:csv_fixed_width="1,5,9,13,17,21"

This defines that each column starts at multiples of 4. Be sure, to issue this command in the buffer, that contains your file, otherwise, it won't have an effect, since this is a buffer-local option (local-option)

After setting this variable, you should reinitialize the plugins using InitCSV

Setup using a Wizard

Alternatively, you can setup the fixed width columns using the :CSVFixed command. This provides a simple wizard to select each column. If you enter the command:


The first column will be highlighted and Vim outputs: <Cursor>, <Space>, <ESC>, <BS>, <CR> This means, you can now use those 5 keys to configure the fixed-width columns:

Key Effect
<Cursor> Use Cursor Left () and Cursor Right () to move the highlighting bar.
<Space> If you press , this column will be fixed and remain highlighted and there will be another bar, you can move using the Cursor keys. This means this column will be considered to be the border between 2 fixed with columns.
<ESC> Abort
<BS> Press the backspace key, to remove the last column you fixed with the key.
<CR> Use Enter to finish the wizard. This will use all fixed columns to define the fixed width columns of your csv file. The plugin will be initialized and syntax highlighting should appear.

Note: This only works, if your Vim has the 'colorcolumn' option available (This won't work with Vim < 7.3 and also not with a Vim without +syntax feature).

CSV Header lines

By default, dynamic filtering csv-filter will not fold away the first line. If you don't like that, you can define your header line using the variable b:csv_fold_headerline, e.g.

let b:csv_headerline = 0

to disable, that a header line won't be folded away. If your header line instead is on line 5, simply set this variable to 5. This also applies to the Header_CSV command.

Number format

When using the SumCol_CSV command, you can specify a certain number format using the /x:y/ argument. You can however also configure the plugin to detect a different number format than the default number format (which does not support a thousands separator and uses the '.' as decimal separator).

To specify a different thousands separator by default, use

let b:csv_thousands_sep = ' '

to have the space use as thousands separator and

let b:csv_decimal_sep = ','

to use the comma as decimal separator.

Move folded lines

If you use dynamic filters (see csv-filter), you can configure the plugin to move all folded lines to the end of the file. This only happens if you set the variable

let g:csv_move_folds = 1

and the file is modifiable. This let's you see all non-folded records as a consecutive area without being disrupted by folded lines.

Using comments

Strictly speaking, in csv files there can't be any comments. You might however still wish to comment or annotate certain sections in your file, so the CSV plugin supports Comments.

Be default, the CSV plugin will use the 'commentstring' setting to identify comments. If this option includes the '%s' it will consider the part before the '%s' as leading comment marker and the part behind it as comment delimiter.

You can however define your own comment marker, using the variable g:csv_comment. Like with the 'commentstring' setting, you can use '%s' expandos, that will denote where the actual comment text belongs. To define your own comment string, put this in your .vimrc

:let g:csv_comment = '#'

Which will use the '#' sign as comment leader like in many scripting languages.

After setting this variable, you should reinitialize the plugins using InitCSV

By default, the csv plugin sets the 'foldtext' option. If you don't want this, set the variable g:csv_disable_fdt in your .vimrc

:let g:csv_disable_fdt = 1


The csv plugins also defines some functions, that can be used for scripting when a csv file is open


CSVPat({column}[, {pattern}])

This function returns the pattern for the selected column. If only columns is given, returns the regular expression used to search for the pattern .* in that column (which means the content of that column). Alternatively, an optional pattern can be given, so the return string can be directly feeded to the / or :s command, e.g. type:

:s/<C-R>=CSVPat(3, 'foobar')<cr>/baz

where the means pressing Control followed by R followed by = (see c_CTRL-R_=). A prompt will apear, with the '=' as the first character on which you can enter expressions.

In this case enter CSVPat(3, 'foobar') which returns the pattern to search for the string 'foobar' in the third column. After you press enter, the returned pattern will be put after the :s command so you can directly enter / and the substitute string.

CSVField(x,y[, orig])

This function returns the field at index (x,y) (starting from 1). If the parameter orig is given, returns the column "as is" (e.g. including delimiter and leading and trailing whitespace, otherwise that will be stripped.)


If the name parameter is given, returns the name of the column, else returns the index of the current column, starting at 1.

CSVSum(col, fmt, startline, endline)

Returns the sum for column col. Uses fmt to parse number format (see :CSVSumCol) startline and endline specify the lines to consider, if empty, will be first and last line.

CSVCount(col, fmt, startline, endline[, distinct])

Returns the count of values for column col. If the optional parameter [distinct] is given, only returns the distinct number of values.

CSVMax(col, fmt, startline, endline)

Returns the 10 largest values for column col.

CSVMin(col, fmt, startline, endline)

Returns the 10 smallest values for column col.

CSVAvg(col, fmt, startline, endline)

Returns the average value for column col.

CSV Tips and Tricks

Here, there you'll find some small tips and tricks that might help when working with CSV files.


Suppose you want to include the column, on which the cursor is, into your statusline. You can do this, by defining in your .vimrc the 'statusline' like this:

function MySTL()
    if has("statusline")
        hi User1 term=standout ctermfg=0 ctermbg=11 guifg=Black guibg=Yellow
        let stl = ...
        if exists("*CSV_WCol")
                let csv = '%1*%{&ft=~"csv" ? CSV_WCol() : ""}%*'
                let csv = ''
        return stl.csv
set stl=%!MySTL()

This will draw in your statusline right aligned the current column and max column (like 1/10), if you are inside a CSV file. The column info will be drawn using the User1 highlighting (hl-User1), that has been defined in the second line of the function. In the third line of your function, put your desired 'statusline' settings as expression. Note the section starting with 'if exists(..)' guards against not having loaded the filetype plugin.

Note: vim-airline by default supports the csv plugin and enables a nice little csv statusline which helps for navigating within a csv file. For details, see the Vim-Airline documentation.

The CSV_WCol() function controls, what will be output. In the simplest case, when no argument is given, it simply returns on which column the cursor is. This would look like '1/10' which means the cursor is on the first of 10 columns. If you rather like to know the name of the column, simply give as parameter to the function the string "Name". This will return the column name as it is printed on the first line of that column. This can be adjusted, to have the column name printed into the statusline (see csv-stl above) by replacing the line

let csv = '%1*%{&ft=~"csv" ? CSV_WCol() : ""}%*'

by e.g.

let csv = '%1*%{&ft=~"csv" ? CSV_WCol("Name") . " " . CSV_WCol() : ""}%*'

which will output "Name 2/10" if the cursor is in the second column which is named "Name".

Slow CSV plugin

Processing a csv file using ArrangeColumn_CSV can be quite slow, because Vim needs to calculate the width for each column and then replace each column by itself widened by spaces to the optimal length. Unfortunately, csv files tend to be quite big. Remember, for a file with 10,000 lines and 50 columns Vim needs to process each cell, which accumulates to 500,000 substitutions. It might take some time, until Vim is finished.

You can speed up things a little bit, if you omit the '!' attribute to the ArrangeColumn (but this will only work, if the width has been calculated before, e.g. by issuing a :1ArrangeColumn command to arrange only the first line. Additionally you can also configure how this command behaves by setting some configuration variables.

Also note, using dynamic filters (csv-filter), can slow down Vim considerably, since they internally work with complex regular expressions, and if you have a large file, containing many columns, you might hit a performance penalty (especially, if you want to filter many columns). It's best to avoid those functions if you are using a large csv file (so using strict columns csv-strict might help a little and also setting 're' to 1 might also alleviate it a little).

Defining custom aggregate functions

The CSV plugin already defines the SumCol_CSV command, to let you calculate the sum of all values of a certain column within a given range. This will consider all values within the range, that are not folded away (csv-filter), and also skip comments and the header lines. The delimiter will be deleted from each field.

But it may be, that you don't need the sum, but would rather want to have the average of all values within a certain column. You can define your own function and let the plugin call it for a column like this:

  1. You define your own custom function in the after directory of your vim runtime path after-directory (see also #2 below)

    fun! My_CSV_Average(col)
        let sum=0
        for item in a:col
        let sum+=item
        return sum/len(a:col)

    This function takes a list as argument, and calculates the average for all items in the list. You could also make use of Vim's eval() function and write your own Product function like this

    fun! My_CSV_Product(col)
        return eval(join(a:col, '*'))
  2. Now define your own custom command, that calls your custom function for a certain column

    command! -buffer -nargs=? -range=% AvgCol
    \ :echo csv#EvalColumn(<q-args>,
    \ "My_CSV_Average", <line1>,<line2>)

    This command should best be put into a file called csv.vim and save it into your ~/.vim/after/ftplugin/ directory. Create directories that don't exist yet. For Windows, this would be the $VIMRUNTIME/vimfiles/after/ftplugin directory.

  3. Make sure, your .vimrc includes a filetype plugin setting like this

    filetype plugin on

    This should make sure, that all the necessary scripts are loaded by Vim.

After restarting Vim, you can now use your custom command definition :AvgCol. Use a range, for the number of lines you want to evaluate and optionally use an argument to specify which column you want to be evaluated

:2,$AvgCol 7

This will evaluate the average of column seven (assuming, line 1 is the header line, which should not be taken into account).

Note: this plugin already defines an average function.

Autocommand on opening/closing files

If you want your CSV files to always be displayed like a table, you can achieve this using the ArrangeColumn_CSV command and some autocommands. Define these autocommands in your .vimrc

aug CSV_Editing
        au BufRead,BufWritePost *.csv :%ArrangeColumn
        au BufWritePre *.csv :%UnArrangeColumn
aug end

Upon Entering a csv file, Vim will visually arrange all columns and before writing, those columns will be collapsed again. The BufWritePost autocommand makes sure, that after the file has been written successfully, the csv file will again be visually arranged.

You can also simply set the variable

let g:csv_autocmd_arrange = 1

in your vimrc and an autocmd will be installed, that visually arranges your csv file whenever you open them for editing. Alternatively, you can restrict this setting to files below a certain size. For example, if you only want to enable this feature for files smaller than 1 MB, put this into your .vimrc

let g:csv_autocmd_arrange      = 1
let g:csv_autocmd_arrange_size = 1024*1024

Note, this is highly experimental and especially on big files, this might slow down Vim considerably.

Syntax error when opening a CSV file

If you see this error:

CSV Syntax:Invalid column pattern, using default pattern \%([^,]*,\|$\)

This happens usually, when the syntax script is read before the filetype plugin, so the plugin did not have a chance to setup the column delimiter correctly.

The easy way to fix it, is to make sure the :syntax on (:syn-on) statement comes after the :filetype plugin (:filetype-plugin-on) statement in your .vimrc

Alternatively, you can simply call InitCSV and ignore the error.

Calculate new columns

Suppose you have a table like this:

Index Value Value2
1 100 3
2 20 4

And you need one more column, that is the calculated product of column 2 and 3, you can make use of the provided CSVField() function using a sub-replace-expression of an :s command. In this case, you would do this:

:2,3s/$/\=printf("%s%.2f", b:delimiter,

Note: Enter as single line. The result will be this:

Index Value Value2
1 100 3
2 20 4