This chapter covers two separate but related concepts: search and substitute. Many times, the texts that you are searching for are not straightforward and you must search for a common pattern. By learning how to use meaningful patterns in search and substitute instead of literal strings, you will be able to target any text quickly.
It can be tricky trying to match the case of the search term. If you are searching for the text "Learn Vim", you can easily mistype the case of one letter and get a false search result. Wouldn't it be easier and safer if you can match any case? This is where the option `ignorecase` shines. Just add `set ignorecase` in your vimrc and all your search terms become case insensitive. Now you don't have to do `/Learn Vim` anymore. `/learn vim` will work.
However, there are times when you need to search for a case specific phrase. One way to do that is to turn off `ignorecase` option with `set noignorecase`, but that is a lot of work to turn on and off each time you need to search for a case sensitive phrase.
Is there a setting that allows you to do case insensitive search most of the time, but also know to do case sensitive search when you need it? Turns out there is a way.
Vim has a `smartcase` option to override `ignorecase` if the search pattern *contains at least one uppercase character*. You can combine both `ignorecase` and `smartcase` to perform a case insensitive search when you enter all lowercase characters and a case sensitive search when you enter one or more uppercase characters.
There is one downside. What if you need to search for only a lowercase string? When you do `/hello`, Vim will always match its uppercase variants. What if you don't want to match them? You can use `\C` pattern in front of your search term to tell Vim that the subsequent search term will be case sensitive. If you do `/\Chello`, it will strictly match "hello", not "HELLO" or "Hello".
You can target the first "hello" with `/^hello`. The character that follows `^` must be the first character in a line. To target the last "hello", run `/hello$`. The character before `$` must be the last character in a line.
You can repeat the previous search with `//`. If you have just searched for `/hello`, running `//` is equivalent to running `/hello`. This shortcut can save you some keystrokes especially if you just did a long search term. Also recall that you can also use `n` and `N` to repeat the last search with the same direction and opposite direction, respectively.
What if you want to quickly recall *n* last search term? You can quickly traverse the search history by first pressing `/`, then press `up`/`down` arrow keys (or `Ctrl-n`/`Ctrl-p`) until you find the search term you need. To see all your search history, you can run `:history /`.
When you reach the end of a file while searching, Vim throws an error: `"Search hit the BOTTOM without match for: <your-search>"`. Sometimes this can be a good safeguard from oversearching, but other times you want to cycle the search back to the top again. You can use the `set wrapscan` option to make Vim to search back at the top of the file when you reach the end of the file. To turn this feature off, do `set nowrapscan`.
It is common to search for multiple words at once. If you need to search for *either* "hello vim" or "hola vim", but not "salve vim" or "bonjour vim", you can use the `|` pipe alternative syntax.
To match both "hello" and "hola", you can do `/hello\|hola`. You have to escape (`\`) the pipe (`|`) operator, otherwise Vim will literally search for the string "|".
If you don't want to type `\|` every time, you can use the `magic` syntax (`\v`) at the start of the search: `/\vhello|hola`. I will not cover `magic` in this chapter, but with `\v`, you don't have to escape special characters anymore. To learn more about `\v`, feel free to check out `:h \v`.
Maybe you need to search for a text that is a part of a compound word. If you have these texts:
```
11vim22
vim22
11vim
vim
```
If you need to select "vim" but only when it starts with "11" and ends with "22", you can use `\zs` (starting match) and `\ze` (ending match) operators. Run:
```
/11\zsvim\ze22
```
Vim still has to match the entire pattern "11vim22", but only highlights the pattern sandwiched between `\zs` and `\ze`. Another example:
```
foobar
foobaz
```
If you need to search for the "foo" in "foobaz" but not in "foobar", run:
All your search terms up to this point have been a literal word search. In real life, you may have to use a general pattern to find your text. The most basic pattern is the character range, `[ ]`.
If you need to search for any digit, you probably don't want to type `/0\|1\|2\|3\|4\|5\|6\|7\|8\|9\|0` every single time. Instead, use `/[0-9]` to match for a single digit. The `0-9` expression represents a range of numbers 0-9 that Vim will try to match, so if you are looking for digits between 1 to 5 instead, use `/[1-5]`.
Digits are not the only data types Vim can look up. You can also do `/[a-z]` to search for lowercase alphas and `/[A-Z]` to search for uppercase alphas.
You can combine these ranges together. If you need to search for digits 0-9 and both lowercase and uppercase alphas from "a" to "f" (a hex), you can do `/[0-9a-fA-F]`.
To do a negative search, you can add `^` inside the character range brackets. To search for a non-digit, run `/[^0-9]`. Vim will match any character as long as it is not a digit. Beware that the caret (`^`) inside the range brackets is different from the beginning-of-a-line caret (ex: `/^hello`). If a caret is outside of a pair of brackets and is the first character in the search term, it means "the first character in a line". If a caret is inside a pair of brackets and it is the first character inside the brackets, it means a negative search operator. `/^abc` matches the first "abc" in a line and `/[^abc]` matches any character except for an "a", "b", or "c".
If you need to search for double digits in this text:
```
1aa
11a
111
```
You can use `/[0-9][0-9]` to match a two-digit character, but this method is unscalable. What if you need to match twenty digits? Typing `[0-9]` twenty times is not a fun experience. That's why you need a `count` argument.
You can pass `count` to your search. It has the following syntax:
```
{n,m}
```
By the way, these `count` braces need to be escaped when you use them in Vim. The `count` operator is placed after a single character you want to increment.
Here are the four different variations of the `count` syntax:
-`{n}` is an exact match. `/[0-9]\{2\}` matches the two digit numbers: "11" and the "11" in "111".
-`{n,m}` is a range match. `/[0-9]\{2,3\}` matches between 2 and 3 digit numbers: "11" and "111".
-`{,m}` is an up-to match. `/[0-9]\{,3\}` matches up to 3 digit numbers: "1", "11", and "111".
-`{n,}` is an at-least match. `/[0-9]\{2,\}` matches at least a 2 or more digit numbers: "11" and "111".
The count arguments `\{0,\}` (zero or more) and `\{1,\}` (one or more) are common search patterns and Vim has special operators for them: `*` and `+` (`+` needs to be escaped while `*` works fine without the escape). If you do `/[0-9]*`, it is the same as `/[0-9]\{0,\}`. It searches for zero or more digits. It will match "", "1", "123". By the way, it will also match non-digits like "a", because there is technically zero digit in the letter "a". Think carefully before using `*`. If you do `/[0-9]\+`, it is the same as `/[0-9]\{1,\}`. It searches for one or more digits. It will match "1" and "12".
Vim has predefined ranges for common characters like digits and alphas. I will not go through every single one here, but you can find the full list inside `:h /character-classes`. Here are the useful ones:
```
\d Digit [0-9]
\D Non-digit [^0-9]
\s Whitespace character (space and tab)
\S Non-whitespace character (everything except space and tab)
\w Word character [0-9A-Za-z_]
\l Lowercase alphas [a-z]
\u Uppercase character [A-Z]
```
You can use them like you would use character ranges. To search for any single digit, instead of using `/[0-9]`, you can use `/\d` for a more concise syntax.
When sees the first `"`, it begins the pattern capture. The moment Vim sees the second double quote in a line, it matches the second `"` pattern and stops the pattern capture. Meanwhile, all non-`"` characters between the two `"` are captured by the `[^"]\+` pattern, in this case, the phrase `Vim is awesome!`. This is a common pattern to capture a phrase surrounded by a pair of similar delimiters: to capture a phrase surrounded by a single quote, you can use `/'[^']\+'`.
You can repeat the last substitute command with either the normal command `&` or by running `:s`. If you have just run `:s/good/awesome/`, running either `&` or `:s` will repeat it.
Also, earlier in this chapter I mentioned that you can use `//` to repeat the previous search pattern. This trick works with the substitution command. If `/good` was done recently and you leave the first substitute pattern argument blank, like in `:s//awesome/`, it is the same as running `:s/good/awesome/`.
Just like many Ex commands, you can pass a range argument into the substitute command. The syntax is:
```
:[range]s/old/new/
```
If you have these expressions:
```
let one = 1;
let two = 2;
let three = 3;
let four = 4;
let five = 5;
```
To substitute the "let" into "const" on lines three to five, you can do:
```
:3,5s/let/const/
```
The substitute command's range syntax is similar to the count syntax in search (`{n,m}`), with minor differences. Here are some variations to pass the range:
-`:,3/let/const/` - if nothing is given before the comma, it represents the current line. Substitute from current line to line 3.
-`:1,s/let/const/` - if nothing is given after the comma, it also represents the current line. Substitute from line 1 to current line.
-`:3s/let/const/` - if only one value is given as range (no comma), it does substitution on that line only.
In Vim, `%` usually means the entire file. If you run `:%s/let/const/`, it will do substitution on all lines.
-`"\0"` the double quotes are literal double quotes. `\0` is a special character representing "the whole matched pattern". The matched pattern here is a single digit number, `\d`. On line one, `\0` has the value of "1". On line two, value of "2". On line three, value of "3", and so on.
Alternatively, `&` also represents "the whole matched pattern" like `\0`. `:s/\d/"&"/` would have also worked.
Let's consider another example. Given these expressions:
-`(\w+) (\w+) ` groups the patterns. `\w` is one of Vim's predefined ranges for a word character (`[0-9A-Za-z_]`). The `( )` surrounding it captures a word character match in a group. Notice the space between the two groupings. `(\w+) (\w+)` captures in two groups. On the first line, the first group captures "one" and the second group captures "two".
-`\2 \1` returns the captured group in a reversed order. `\2` contains the captured string "let" and `\1` the string "one". Having `\2 \1` returns the string "let one".
Recall that `\0` represents the entire matched pattern. You can break the matched string into smaller groups with `( )`. Each group is represented by `\1`, `\2`, `\3`, etc.
Let's do one more example to solidify this matched group concept. If you have these numbers:
```
123
456
789
```
To reverse the order, run:
```
:%s/\v(\d)(\d)(\d)/\3\2\1/
```
The result is:
```
321
654
987
```
Each `(\d)` matches and groups each digit. On the first line, the first `(\d)` has a value of "1", the second `(\d)` "2", and the third `(\d)` "3". They are stored in the variables `\1`, `\2`, and `\3`. In the second half of your substitution, the new pattern `\3\2\1` results in the "321" value on line one.
This is because you now only have two groups. The first group, captured by `(\d\d)`, is stored within `\1` and has the value of "12". The second group, captured by `(\d)`, is stored inside `\2` and has the value of "3". `\2\1` then, returns "312".
There are two ways to solve this. First, you can run the substitute command twice more. Second, you can pass it a global (`g`) flag to substitute all of the matches in a line.
Vim substitutes all pancakes with donuts in one swift command. The global command is one of the several flags the substitute command accepts. You pass flags at the end of the substitute command. Here is a list of useful flags:
```
& Reuse the flags from the previous substitute command. Must be passed as the first flag.
g Replace all matches in the line.
c Ask for substitution confirmation.
e Prevent error message from displaying when substitution fails.
There are more flags that I do not list above. To read about all the flags, check out `:h s_flags`.
By the way, the repeat-substitution commands (`&` and `:s`) do not retain the flags. Running `&` will only repeat `:s/pancake/donut/` without `g`. To quickly repeat the last substitute command with all the flags, run `:&&`.
However, it is hard to tell which forward slashes (`/`) are part of the substitution pattern and which ones are the delimiters. You can change the delimiter with any single-byte characters (except for alphabets, numbers, or `"`, `|`, and `\`). Let's replace them with `+`. The substitution command above then can be rewritten as:
-`\<.` is comprised of two parts: `\<` to match the start of a word and `.` to match any character. `\<` operator makes the following character to be the first character of a word. Since `.` is the next character, it will match the first character of any word.
-`{-}` performs non-greedy match of 0 or more of the preceding atom.
-`\zsMississippi` makes "Mississippi" the start of the match.
-`(...){3}` looks for the third match.
You have seen the `{3}` syntax before. It is a type of `{n,m}`. In this case, you have `{3}` which will match exactly the third match. The new trick here is `{-}`. It is a non-greedy match. It finds the shortest match of the given pattern. In this case, `(.{-}Mississippi)` matches the least amount of "Mississippi" preceded by any character. Contrast this with `(.*Mississippi)` where it finds the longest match of the given pattern.
If you use `(.{-}Mississippi)`, you get five matches: "One Mississippi", "Two Mississippi", etc. If you use `(.*Mississippi)`, you get one match: the last "Mississippi". To learn more check out `:h /\{-` and `:h non-greedy`.
Let's do a simpler example. If you have the string:
```
abc1de1
```
You can match "abc1de1" (greedy) with:
```
/a.*1
```
You can match "abc1" (non-greedy) with:
```
/a.\{-}1
```
So if you need to uppercase the longest match (greedy), run:
```
:s/a.*1/\U&/g
```
To get:
```
ABC1DEFG1
```
If you need to uppercase the shortest match (non-greedy), run:
First, capture both `food.txt` and `animal.txt` inside `:args`. Recall from earlier chapters that `:args` can be used to create a list of file names. There are several ways to do this from inside Vim:
You can also run the commands above from outside Vim, passing the files as *arguments* for Vim (hence it is called the "args" command). From the terminal, run
When Vim starts, you will find `food.txt` and `animal.txt` inside `:args`.
Either way, when you run `:args`, you should see:
```
[food.txt] animal.txt
```
To go to the next or previous argument on the list, type `:next` or `:previous`. Now that all the relevant files are stored inside the argument list, you can perform a multi-file substitution with the `:argdo` command. Run:
```
:argdo %s/dog/chicken/
```
This performs substitution against the all files inside the `:args` list. Finally, save the changed files with:
```
:argdo update
```
`:args` and `:argdo` are useful tools to apply command line commands across multiple files. Try it with other commands!
Alternatively, you can also run the substitute command across multiple files with macros. Let's start by getting the relevant files into the args list. Run:
```
:args animal.txt food.txt
qq
:%s/dog/chicken/g
:wnext
q
99@q
```
Here is the breakdown of the steps:
-`:args animal.txt food.txt` lists the relevant files into the `:args` list.
-`:wnext` writes (saves) the file then go to the next file on the `args` list. It's like running `:w` and `:next` at the same time.
-`q` stops the macro recording.
-`99@q` executes the macro ninety-nine times. Vim will stop the macro execution after it encounters the first error, so Vim won't actually execute the macro ninety-nine times.
The ability to do search well is a necessary skill in editing. Mastering the search lets you to utilize the flexibility of regular expressions to search for any pattern in a file. Take your time to learn these. Actually do the searches and substitutions in this chapter yourself. I once read a book about regular expression without actually doing it and I forgot almost everything I read afterwards. Active coding is the best way to master any skill.
A good way to improve your pattern matching skill is whenever you need to search for a pattern (like "hello 123"), instead of querying for the literal search term (`/hello 123`), try to come up with a pattern for it (`/\v(\l+) (\d+)`). Many of these regular expression concepts are also applicable in general programming, not only when using Vim.
Now that you learned about advanced search and substitution in Vim, let's learn one of the most versatile commands, the global command.
