Today I’ll show some ways to edit text using vim’s command line search and replace plus how to use an external command to help us nicely align columns of text.

NOTE: The asciinema player sometimes positions the player’s bottom bar in such a way that we can’t see the vim commands being typed. Make the player full screen or take the mouse away from the player to make that bottom bar disappear.

Example #1 - php variables to assignment

See it in action bellow (or here) and then take a look at the explanation.

here

So, I was working in some PHP code and I had these variables which I needed to initialize with the values from a POST request.

$id
$nome
$email
$password
$status
$fone1
$fone2
$nome_fantasia
$razao_social
$cpf
$cnpj
$rg
$ie
$cidade
$estado
$rua
$cep
$numero

First we visually selecte those lines, and hit : to start the command line mode. Since it is a visual selection, Vim marks the beginning of the selection with '< and the end with '>, therefore the command line starts as :'<,'>. The comma just separates the range marks.

Then, type down the regexp replace thing. The entire command line looks like this:

:'<,'>s/$\(\w\+\)/$\1 = param('\1');/

Hit <CR> (the ‘Enter’, or ‘Return’ key) and the result is:

<?php
$id = param('id');
$nome = param('nome');
$email = param('email');
$password = param('password');
$status = param('status');
$fone1 = param('fone1');
$fone2 = param('fone2');
$nome_fantasia = param('nome_fantasia');
$razao_social = param('razao_social');
$cpf = param('cpf');
$cnpj = param('cnpj');
$rg = param('rg');
$ie = param('ie');
$cidade = param('cidade');
$estado = param('estado');
$rua = param('rua');
$cep = param('cep');
$numero = param('numero');

That command did a lot, didn’t it‽ Basically, it matches the “word characters” after the $ and saves it in a capturing group: \(\w+\). Then it just adds the $ followed by the previous matched text in the capturing group \1 followed by  = param('\1');, which results in every line with the single variable being turned into $varname = param('varname'); statements.

If the code style of your code base uses assignments in aligned columns, it is possible to use the column command line (available on *nix systems) to nicely align the columns. Visually select the lines again (you may use gv) and then run:

:'<,'>!column -t

And the result is:

<?php
$id             =  param('id');
$nome           =  param('nome');
$email          =  param('email');
$password       =  param('password');
$status         =  param('status');
$fone1          =  param('fone1');
$fone2          =  param('fone2');
$nome_fantasia  =  param('nome_fantasia');
$razao_social   =  param('razao_social');
$cpf            =  param('cpf');
$cnpj           =  param('cnpj');
$rg             =  param('rg');
$ie             =  param('ie');
$cidade         =  param('cidade');
$estado         =  param('estado');
$rua            =  param('rua');
$cep            =  param('cep');
$numero         =  param('numero');

It is important to note the use of the ! character so that the text in the selected lines is sent to the external program, operated on and then placed back (after modification), effectively replacing the old, unaligned columns with the new, aligned ones.

Example #2 - db columns to PHP/PDO bindParam

This one is very similar, actually, but I do think it is worth to take a look at it nonetheless. Here is the raw input lines I had, copied from an SQL statement somewhere else. See asciinema below or here:

nome
email
fone1
fone2
nome_fantasia
razao_social
cpf
cnpj
rg
ie
cidade
estado
rua
cep
numero

Then I run:

:'<,'>s/\w\+/$sth->bindParam(':&', $&, PDO::PARAM_STR);/

And the result is:

<?php
$sth->bindParam(':nome', $nome, PDO::PARAM_STR);
$sth->bindParam(':email', $email, PDO::PARAM_STR);
$sth->bindParam(':fone1', $fone1, PDO::PARAM_STR);
$sth->bindParam(':fone2', $fone2, PDO::PARAM_STR);
$sth->bindParam(':nome_fantasia', $nome_fantasia, PDO::PARAM_STR);
$sth->bindParam(':razao_social', $razao_social, PDO::PARAM_STR);
$sth->bindParam(':cpf', $cpf, PDO::PARAM_STR);
$sth->bindParam(':cnpj', $cnpj, PDO::PARAM_STR);
$sth->bindParam(':rg', $rg, PDO::PARAM_STR);
$sth->bindParam(':ie', $ie, PDO::PARAM_STR);
$sth->bindParam(':cidade', $cidade, PDO::PARAM_STR);
$sth->bindParam(':estado', $estado, PDO::PARAM_STR);
$sth->bindParam(':rua', $rua, PDO::PARAM_STR);
$sth->bindParam(':cep', $cep, PDO::PARAM_STR);
$sth->bindParam(':numero', $numero, PDO::PARAM_STR);

Here, we again match “word characters” with \w\+, but this time without a capturing group. In vim regexp we can use & to refer to the whole match. In this example & is used to place all the matched text in two places in the replacement part of the regexp. That is seen in :& (which produces things like :nome and :email) and $& (which similarly produces things like $nome and $email).

Again, the use of column from inside vim makes it easy to create columns nicely aligned. Just visually select the target lines again (remember gv) and hit :. Again, the whole command looks like this:

:'<,'>!column -t

And the result is:

<?php
$sth->bindParam(':nome',           $nome,           PDO::PARAM_STR);
$sth->bindParam(':email',          $email,          PDO::PARAM_STR);
$sth->bindParam(':fone1',          $fone1,          PDO::PARAM_STR);
$sth->bindParam(':fone2',          $fone2,          PDO::PARAM_STR);
$sth->bindParam(':nome_fantasia',  $nome_fantasia,  PDO::PARAM_STR);
$sth->bindParam(':razao_social',   $razao_social,   PDO::PARAM_STR);
$sth->bindParam(':cpf',            $cpf,            PDO::PARAM_STR);
$sth->bindParam(':cnpj',           $cnpj,           PDO::PARAM_STR);
$sth->bindParam(':rg',             $rg,             PDO::PARAM_STR);
$sth->bindParam(':ie',             $ie,             PDO::PARAM_STR);
$sth->bindParam(':cidade',         $cidade,         PDO::PARAM_STR);
$sth->bindParam(':estado',         $estado,         PDO::PARAM_STR);
$sth->bindParam(':rua',            $rua,            PDO::PARAM_STR);
$sth->bindParam(':cep',            $cep,            PDO::PARAM_STR);
$sth->bindParam(':numero',         $numero,         PDO::PARAM_STR);

Imagine doing all that replacing and aligning manually, character by character, line by line!

Example #3 - form name attr to php array

In this case, we have an initial list of strings that come from HTML form input name attribute. See asciinema below or here. They look like this (

form-cadastro__nome
form-cadastro__email
form-cadastro__fone1
form-cadastro__fone2
form-cadastro__nome-fantasia
form-cadastro__razao-social
form-cadastro__cpf
form-cadastro__rg
form-cadastro__cnpj
form-cadastro__ie
form-cadastro__estado
form-cadastro__cidade
form-cadastro__bairro
form-cadastro__rua
form-cadastro__numero
form-cadastro__cep
form-cadastro__senha

And we want to retrieve those values from the request and create an associative array with them, but the array key does not contain the substring form-cadastro__. So,

form-cadastro__nome

becomes:

$data['nome'] = param('form-cadastro__nome');

So, if the lines we want to modify are betwee 5 and 21, we would do:

:5,21 s:.\+__\(.\+\):$data['\1'] = param('&');:
:5,21 !column -t

The result is:

$data['nome']           =  param('form-cadastro__nome');
$data['email']          =  param('form-cadastro__email');
$data['fone1']          =  param('form-cadastro__fone1');
$data['fone2']          =  param('form-cadastro__fone2');
$data['nome-fantasia']  =  param('form-cadastro__nome-fantasia');
$data['razao-social']   =  param('form-cadastro__razao-social');
$data['cpf']            =  param('form-cadastro__cpf');
$data['rg']             =  param('form-cadastro__rg');
$data['cnpj']           =  param('form-cadastro__cnpj');
$data['ie']             =  param('form-cadastro__ie');
$data['estado']         =  param('form-cadastro__estado');
$data['cidade']         =  param('form-cadastro__cidade');
$data['bairro']         =  param('form-cadastro__bairro');
$data['rua']            =  param('form-cadastro__rua');
$data['numero']         =  param('form-cadastro__numero');
$data['cep']            =  param('form-cadastro__cep');
$data['senha']          =  param('form-cadastro__senha');

Let’s explain this command:

:5,21 s:.\+__\(.\+\):$data['\1'] = param('&');:
  • 5,21 is the line range, that is, to which lines should vim apply text transformation.
  • s: : : is the substitute command using : as the delimiter.
  • .\+__\(.\+\) matches everything in the line, but also captures the text after __ as group 1 (which will be used on the next steps).
  • $data['\1'] inserts the characters $data[ ] and uses the group \1 so the result is something like $data['foo'], $data['bar'], etc. \1 is whatever was captured in group 1 for each line. The whole match would be something like form-cadastro__foo, but we captured just the foo that comes after __ and therefore the array key gets the shorter name, discarding form-cadastro__ part.
  • param('&') uses & wich is the whole match for that line, not just a specific group. So, if the line was form-cadastro__cnpj, it becomes param('form-cadastro__cnpj').

Example #4 - html option to php array

This time, we have several lines with HTML option tags like this:

<option value='SN'>State Name</option>

And we want to turn it into PHP’s associative 'key' ⇒ 'value' array entries:

'sn' => 'State Name',

The array key should be lowercased. The array value remains untouched. This is how it is done:

'<,'>s:[^"]\+"\(\u\u\)">\([^<]\+\).*$:'\l\1' => '\2',

The matching part goes like this:

  • match everything up to the first double quote: [^"]\+";
  • then match and store two uppercase chars in a capturing group : \(\u\u\);
  • match the closing > for opening option tag;
  • match and store the next characters up to but not including the next <: \([^<]\+\);
  • match the rest of the line.

And then the replacment part:

  • lowercase the text of the first capturing group: \l\1 (problem here) and place it inside single quotes to make the PHP array key.
  • add the chars (required by php array syntax);
  • then place the contents of the second capturing groups in single quotes as well;
  • the array trailing comma;

So, something like

<option value='SN'>State Name</option>

becomes

'sN' => 'State Name',

My first attempt was this:

'<,'>s:[^"]\+"\(\u\u\)">\([^<]\+\).*$:'\l\1' => '\2',

Note that from the matching part we intentionally do not capture some text. Some characters need only to help us match and delimit stuff we need, but are not used in the replacement part later; only the captured characters are used in the replacement part of the regexp.

The problem is that the 'sn' array key should be all lowercase, but our first attempt results in something like 'xX', that is, the second character is still uppercase.

Using \L\1 lowercases the entire rest of the regexp replace thing, so, even the array values become all lower case, that is, instead of 'sn' => 'State Name' we end up with 'sn' => 'state name'. Not what we want.

Of course, I asked for help in #vim, and surely enough, @LeoNerd told me about \E. Bingo! Running

'<,'>s:[^"]\+"\(\u\u\)">\([^<]\+\).*$:'\L\1\E' => '\2',
#                                          ^^

on the input:

<option value="AC">Acre</option>
<option value="AL">Alagoas</option>
<option value="AP">Amapá</option>
<option value="AM">Amazonas</option>
<option value="BA">Bahia</option>
<option value="CE">Ceará</option>
<option value="DF">Distrito Federal</option>
<option value="ES">Espírito Santo</option>
<option value="GO">Goiás</option>
<option value="MA">Maranhão</option>
<option value="MT">Mato Grosso</option>
<option value="MS">Mato Grosso do Sul</option>
<option value="MG">Minas Gerais</option>
<option value="PA">Pará</option>
<option value="PB">Paraíba</option>
<option value="PR">Paraná</option>
<option value="PE">Pernambuco</option>
<option value="PI">Piauí</option>
<option value="RJ">Rio de Janeiro</option>
<option value="RN">Rio Grande do Norte</option>
<option value="RS">Rio Grande do Sul</option>
<option value="RO">Rondônia</option>
<option value="RR">Roraima</option>
<option value="SC">Santa Catarina</option>
<option value="SP">São Paulo</option>
<option value="SE">Sergipe</option>
<option value="TO">Tocantins</option>

produces exactly the output we want:

'ac' => 'Acre',
'al' => 'Alagoas',
'ap' => 'Amapá',
'am' => 'Amazonas',
'ba' => 'Bahia',
'ce' => 'Ceará',
'df' => 'Distrito Federal',
'es' => 'Espírito Santo',
'go' => 'Goiás',
'ma' => 'Maranhão',
'mt' => 'Mato Grosso',
'ms' => 'Mato Grosso do Sul',
'mg' => 'Minas Gerais',
'pa' => 'Pará',
'pb' => 'Paraíba',
'pr' => 'Paraná',
'pe' => 'Pernambuco',
'pi' => 'Piauí',
'rj' => 'Rio de Janeiro',
'rn' => 'Rio Grande do Norte',
'rs' => 'Rio Grande do Sul',
'ro' => 'Rondônia',
'rr' => 'Roraima',
'sc' => 'Santa Catarina',
'sp' => 'São Paulo',
'se' => 'Sergipe',
'to' => 'Tocantins',

Final Tip

If you don’t like to use visual selections, you can still operate on a renge of lines using their line numbers. For instance, if the lines you want to handle are in the range 5 to 13, you can just do:

:5, 13 s/this/that/

or

:5, 13 !column -t

There is more about this range of lines thing. Take a look at the help sections described below.

Help and man pages

Vim:

:help gv
:help visual.txt
:help :range
:help !
:help pattern.txt # prepare to be blown away!
:help s/\&
:help s/\l
:help s/\L
:help s/\E

Note

Most of these help sections will be very hard to understand. It requires familiarity with lots of concepts. If you like, you can start reading :help usr_01.txt.

You can also take a look at:

:help usr_toc.txt

Column command:

man 1 column