regexps.Rmd

# Regular expressions

```{r, results = "asis", echo = FALSE}
status("restructuring")
```

## Introduction

You learned the basics of regular expressions in Chapter \@ref(strings), but regular expressions are fairly rich language so it's worth spending some extra time on the details.

The chapter starts by expanding your knowledge of patterns, to cover six important new topics (escaping, anchoring, character classes, shorthand classes, quantifiers, and alternation).
Here we'll focus mostly on the language itself, not the functions that use it.
That means we'll mostly work with toy character vectors, showing the results with `str_view()` and `str_view_all()`.
You'll need to take what you learn here and apply it to data frames with tidyr functions or by combining dplyr and stringr functions.

Next we'll talk about the important concepts of "grouping" and "capturing" which give you new ways to extract variables out of strings using `tidyr::separate_group()`.
Grouping also allows you to use back references which allow you do things like match repeated patterns.

We'll finish by discussing the various "flags" that allow you to tweak the operation of regular expressions and cover a few final details about how regular expressions work.
These aren't particularly important in day-to-day usage, but at little extra understanding of the underlying tools is often helpful.

### Prerequisites

This chapter will use regular expressions as provided by the **stringr** package.

```{r setup, message = FALSE}
library(tidyverse)
```

It's worth noting that the regular expressions used by stringr are very slightly different to those of base R.
That's because stringr is built on top of the [stringi package](https://stringi.gagolewski.com), which is in turn built on top of the [ICU engine](https://unicode-org.github.io/icu/userguide/strings/regexp.html), whereas base R functions (like `gsub()` and `grepl()`) use either the [TRE engine](https://github.com/laurikari/tre) or the [PCRE engine](https://www.pcre.org).
Fortunately, the basics of regular expressions are so well established that you'll encounter few variations when working with the patterns you'll learn in this book (and I'll point them out where important).
You only need to be aware of the difference when you start to rely on advanced features like complex Unicode character ranges or special features that use the `(?…)` syntax.
You can learn more about these advanced features in `vignette("regular-expressions", package = "stringr")`.

Another useful reference is [https://www.regular-expressions.info/](https://www.regular-expressions.info/tutorial.html).
It's not R specific, but it includes a lot more information about how regular expressions actually work.

### Exercises

1.  Explain why each of these strings don't match a `\`: `"\"`, `"\\"`, `"\\\"`.

2.  How would you match the sequence `"'\`?

3.  What patterns will the regular expression `\..\..\..` match?
    How would you represent it as a string?

## Pattern language

You learned the very basics of the regular expression pattern language in Chapter \@ref(strings), and now its time to dig into more of the details.
First, we'll start with **escaping**, which allows you to match characters that the pattern language otherwise treats specially.
Next you'll learn about **anchors**, which allow you to match the start or end of the string.
Then you'll learn about **character classes** and their shortcuts, which allow you to match any character from a set.
We'll finish up with **quantifiers**, which control how many times a pattern can match, and **alternation**, which allows you to match either *this* or *that.*

The terms I use here are the technical names for each component.
They're not always the most evocative of their purpose, but it's very helpful to know the correct terms if you later want to Google for more details.

### Escaping {#regexp-escaping}

In Chapter \@ref(strings), you'll learned how to match a literal `.` by using `fixed(".")`.
But what if you want to match a literal `.` as part of a bigger regular expression?
You'll need to use an **escape**, which tells the regular expression you want it to match exactly, not use its special behavior.
Like strings, regexps use the backslash for escaping, so to match a `.`, you need the regexp `\.`.
Unfortunately this creates a problem.
We use strings to represent regular expressions, and `\` is also used as an escape symbol in strings.
So, as the following example shows, to create the regular expression `\.` we need the string `"\\."`.

```{r}
# To create the regular expression \., we need to use \\.
dot <- "\\."

# But the expression itself only contains one \
str_view(dot)

# And this tells R to look for an explicit .
str_view(c("abc", "a.c", "bef"), "a\\.c")
```

In this book, I'll write regular expression as `\.` and strings that represent the regular expression as `"\\."`.

If `\` is used as an escape character in regular expressions, how do you match a literal `\`?
Well you need to escape it, creating the regular expression `\\`.
To create that regular expression, you need to use a string, which also needs to escape `\`.
That means to match a literal `\` you need to write `"\\\\"` --- you need four backslashes to match one!

```{r}
x <- "a\\b"
str_view(x)
str_view(x, "\\\\")
```

Alternatively, you might find it easier to use the raw strings you learned about in Section \@ref(raw-strings)).
That lets you to avoid one layer of escaping:

```{r}
str_view(x, r"(\\)")
```

The full set of characters with special meanings that need to be escaped is `.^$\|*+?{}[]()`.
In general, look at punctuation characters with suspicion; if your regular expression isn't matching what you think it should, check if you've used any of these characters.

### Anchors

By default, regular expressions will match any part of a string.
If you want to match at the start of end you need to **anchor** the regular expression using `^` or `$`.

-   `^` to match the start of the string.
-   `$` to match the end of the string.

```{r}
x <- c("apple", "banana", "pear")
str_view(x, "a")  # match "a" anywhere
str_view(x, "^a") # match "a" at start
str_view(x, "a$") # match "a" at end
```

To remember which is which, try this mnemonic which I learned from [Evan Misshula](https://twitter.com/emisshula/status/323863393167613953): if you begin with power (`^`), you end up with money (`$`).
It's tempting to put `$` at the start, because that's how we write sums of money, but it's not what regular expressions want.

To force a regular expression to only match the full string, anchor it with both `^` and `$`:

```{r}
x <- c("apple pie", "apple", "apple cake")
str_view(x, "apple")
str_view(x, "^apple$")
```

You can also match the boundary between words (i.e. the start or end of a word) with `\b`.
I don't often use this in my R code, but I'll sometimes use it when I'm doing a search in RStudio.
It's useful to find the name of a function that's a component of other functions.
For example, if I want to find all uses of `sum()`, I'll search for `\bsum\b` to avoid matching `summarise`, `summary`, `rowsum` and so on:

```{r}
x <- c("summary(x)", "summarise(df)", "rowsum(x)", "sum(x)")
str_view(x, "sum")
str_view(x, "\\bsum\\b")
```

### Character classes

A **character class**, or character **set**, allows you to match any character in a set.
The basic syntax lists each character you want to match inside of `[]`, so `[abc]` will match a, b, or c.
Inside of `[]` only `-`, `^`, and `\` have special meanings:

-   `-` defines a range. `[a-z]` matches any lower case letter and `[0-9]` matches any number.
-   `^` takes the inverse of the set. `[^abc]`: matches anything except a, b, or c.
-   `\` escapes special characters so `[\^\-\]]`: matches `^`, `-`, or `]`.

```{r}
str_view_all("abcd12345-!@#%.", "[abc]")
str_view_all("abcd12345-!@#%.", "[a-z]")
str_view_all("abcd12345-!@#%.", "[^a-z0-9]")

# You need an escape to match characters that are otherwise
# special inside of []
str_view_all("a-b-c", "[a\\-c]")
```

Remember that regular expressions are case sensitive so if you want to match any lowercase or uppercase letter, you'd need to write `[a-zA-Z0-9]`.

### Shorthand character classes

There are a few character classes that are used so commonly that they get their own shortcut.
You've already seen `.`, which matches any character apart from a newline.
There are three other particularly useful pairs:

-   `\d`: matches any digit; \
    `\D` matches anything that isn't a digit.
-   `\s`: matches any whitespace (e.g. space, tab, newline); \
    `\S` matches anything that isn't whitespace.
-   `\w` matches any "word" character, i.e. letters and numbers; \
    `\W`, matches any non-word character.

Remember, to create a regular expression containing `\d` or `\s`, you'll need to escape the `\` for the string, so you'll type `"\\d"` or `"\\s"`.
The following code demonstrates the different shortcuts with a selection of letters, numbers, and punctuation characters.

```{r}
str_view_all("abcd12345!@#%. ", "\\d+")
str_view_all("abcd12345!@#%. ", "\\D+")
str_view_all("abcd12345!@#%. ", "\\w+")
str_view_all("abcd12345!@#%. ", "\\W+")
str_view_all("abcd12345!@#%. ", "\\s+")
str_view_all("abcd12345!@#%. ", "\\S+")
```

### Quantifiers

The **quantifiers** control how many times a pattern matches.
In Chapter \@ref(strings) you learned about `?` (0 or 1 matches), `+` (1 or more matches), and `*` (0 or more matches).
For example, `colou?r` will match American or British spelling, `\d+` will match one or more digits, and `\s?` will optionally match a single whitespace.

You can also specify the number of matches precisely:

-   `{n}`: exactly n
-   `{n,}`: n or more
-   `{n,m}`: between n and m

The following code shows how this works for a few simple examples using to `\b` match the or end of a word.

```{r}
x <- " x xx xxx xxxx"
str_view_all(x, "\\bx{2}")
str_view_all(x, "\\bx{2,}")
str_view_all(x, "\\bx{1,3}")
str_view_all(x, "\\bx{2,3}")
```

### Alternation

You can use **alternation** to pick between one or more alternative patterns.
Here are a few examples:

-   Match apple, pear, or banana: `apple|pear|banana`.
-   Match three letters or two digits: `\w{3}|\d{2}`.

### Parentheses and operator precedence

What does `ab+` match.
Does it match "a" followed by one or more "b"s, or does it match "ab" repeated any number of times?
What does `^a|b$` match?
Does it match the complete string a or the complete string b, or does it match a string starting with a or a string starting with "b"?
The answer to these questions is determined by operator precedence, similar to the PEMDAS or BEDMAS rule you might have learned in school.

The question comes down to whether `ab+` is equivalent to `a(b+)` or `(ab)+` and whether `^a|b$` is equivalent to `(^a)|(b$)` or `^(a|b)$`.
Alternation has low precedence which means it affects many characters, whereas quantifiers have high precedence which means it affects few characters.

Quantifiers apply to the preceding pattern, regardless of how many letters define it: `a+` matches one or more "a"s, `\d+` matches one or more digits, and `[aeiou]+` matches one or more vowels.
You can use parentheses to apply a quantifier to a compound pattern.
For example, `([aeiou].)+` matches a vowel followed by any letter, repeated any number of times.

```{r}
str_view(words, "^([aeiou].)+$", match = TRUE)
```

`|` has very low precedence which means that everything to the left or right is included in the group.
For example if you want to match only a complete string, `^apple|pear|banana$` won't work because it will match apple at the start of the string, pear anywhere, and banana at the end.
Instead, you need `^(apple|pear|banana)$`.

Technically the escape, character classes, and parentheses are all operators that have some relative precedence.
But these tend to be less likely to cause confusion, for example you experience with escapes in other situations means it's unlikely that you'd think to write `\(s|d)` to mean `(\s)|(\d)`.

### Exercises

1.  How would you match the literal string `"$^$"`?

2.  Given the corpus of common words in `stringr::words`, create regular expressions that find all words that:

    a.  Start with "y".
    b.  Don't start with "y".
    c.  End with "x".
    d.  Are exactly three letters long. (Don't cheat by using `str_length()`!)
    e.  Have seven letters or more.

    Since `words` is long, you might want to use the `match` argument to `str_view()` to show only the matching or non-matching words.

3.  Create regular expressions that match the British or American spellings of the following words: grey/gray, modelling/modeling, summarize/summarise, aluminium/aluminum, defence/defense, analog/analogue, center/centre, sceptic/skeptic, aeroplane/airplane, arse/ass, doughnut/donut.

4.  What strings will `$a` match?

5.  Create a regular expression that will match telephone numbers as commonly written in your country.

6.  Write the equivalents of `?`, `+`, `*` in `{m,n}` form.

7.  Describe in words what these regular expressions match: (read carefully to see if I'm using a regular expression or a string that defines a regular expression.)

    a.  `^.*$`
    b.  `"\\{.+\\}"`
    c.  `\d{4}-\d{2}-\d{2}`
    d.  `"\\\\{4}"`

8.  Solve the beginner regexp crosswords at <https://regexcrossword.com/challenges/beginner>.

## Flags

The are a number of settings, called **flags**, that you can use to control some of the details of the pattern language.
In stringr, you can supply these by instead of passing a simple string as a pattern, by passing the object created by `regex()`:

```{r, eval = FALSE}
# The regular call:
str_view(fruit, "nana")
# Is shorthand for
str_view(fruit, regex("nana"))
```

This is useful because it allows you to pass additional arguments to control the details of the match the most useful is probably `ignore_case = TRUE` because it allows characters to match either their uppercase or lowercase forms:

```{r}
bananas <- c("banana", "Banana", "BANANA")
str_view(bananas, "banana")
str_view(bananas, regex("banana", ignore_case = TRUE))
```

If you're doing a lot of work with multiline strings (i.e. strings that contain `\n`), `multiline` and `dotall` can also be useful.
`dotall = TRUE` allows `.` to match everything, including `\n`:

```{r}
x <- "Line 1\nLine 2\nLine 3"
str_view_all(x, ".L")
str_view_all(x, regex(".L", dotall = TRUE))
```

And `multiline = TRUE` allows `^` and `$` to match the start and end of each line rather than the start and end of the complete string:

```{r}
x <- "Line 1\nLine 2\nLine 3"
str_view_all(x, "^Line")
str_view_all(x, regex("^Line", multiline = TRUE))
```

If you're writing a complicated regular expression and you're worried you might not understand it in the future, `comments = TRUE` can be super useful.
It allows you to use comments and white space to make complex regular expressions more understandable.
Spaces and new lines are ignored, as is everything after `#`.
(Note that I'm using a raw string here to minimise the number of escapes needed)

```{r}
phone <- regex(r"(
  \(?     # optional opening parens
  (\d{3}) # area code
  [) -]?  # optional closing parens, space, or dash
  (\d{3}) # another three numbers
  [ -]?   # optional space or dash
  (\d{3}) # three more numbers
  )", comments = TRUE)

str_match("514-791-8141", phone)
```

If you're using comments and want to match a space, newline, or `#`, you'll need to escape it:

```{r}
str_view("x x #", regex("x #", comments = TRUE))
str_view("x x #", regex(r"(x\ \#)", comments = TRUE))
```

## Practice

To put these ideas in practice we'll solve a few semi-authentic problems using the `words` and `sentences` datasets built into stringr.
`words` is a list of common English words and `sentences` is a set of simple sentences originally used for testing voice transmission.

```{r}
str_view(head(words))
str_view(head(sentences))
```

Let's find all sentences that start with the:

```{r}
str_view(sentences, "^The", match = TRUE)
str_view(sentences, "^The\\b", match = TRUE)
```

All sentences that use a pronoun:

Modify to create simple set of positive and negative examples (if you later get more into programming and learn about unit tests, I highly recommend unit testing your regular expressions. This doesn't guarantee you won't get it wrong, but it ensures that you'll never make the same mistake twice.)

```{r}
str_view_all(sentences, "\\b(he|she|it)\\b", match = TRUE)
str_view_all(head(sentences), "\\b(he|she|it)\\b", match = FALSE)

str_view_all(sentences, regex("\\b(he|she|it)\\b", ignore_case = TRUE), match = TRUE)
```

All words that only contain consonants:

```{r}
str_view(words, "[^aeiou]+", match = TRUE)
str_view(words, "^[^aeiou]+$", match = TRUE)
```

This is a case where flipping the problem around can make it easier to solve.
Instead of looking for words that containing only consonant, we could look for words that don't contain any vowels:

```{r}
words[!str_detect(words, "[aeiou]")]
```

Can we find evidence for or against the rule "i before e except after c"?
To look for words that support this rule we want i follows e following any letter that isn't c, i.e. `[^c]ie`.
The opposite branch is `cei`:

```{r}
str_view(words, "[^c]ie|cei", match = TRUE)
```

To look for words that don't follow this rule, we just switch the i and the e:

```{r}
str_view(words, "[^c]ei|cie", match = TRUE)
```

Consist only of vowel-consonant or consonant-vowel pairs?

```{r}
str_view(words, "^([aeiou][^aeiou])+$", match = TRUE)
str_view(words, "^([^aeiou][aeiou])+$", match = TRUE)
```

Could combine in two ways: by making one complex regular expression or using `str_detect()` with Boolean operators:

```{r}
str_view(words, "^((([aeiou][^aeiou])+)|([^aeiou][aeiou]+))$", match = TRUE)

vc <- str_detect(words, "^([aeiou][^aeiou])+$")
cv <- str_detect(words, "^([^aeiou][aeiou])+$")
words[cv | vc]
```

This only handles words with even number of letters?
What if we also wanted to allow odd numbers?
i.e. cvc or vcv.

```{r}
vc <- str_detect(words, "^([aeiou][^aeiou])+[aeiou]?$")
cv <- str_detect(words, "^([^aeiou][aeiou])+[^aeiou]?$")
words[cv | vc]
```

If we wanted to require the words to be at least four characters long we could modify the regular expressions switching `+` for `{2,}` or we could combine the results with `str_length()`:

```{r}
words[(cv | vc) & str_length(words) >= 4]
```

Do any words contain all vowels?

```{r}
str_view(words, "a.*e.*i.*o.*u", match = TRUE)
str_view(words, "e.*a.*u.*o.*i", match = TRUE)
```

```{r}
words[
  str_detect(words, "a") &
  str_detect(words, "e") &
  str_detect(words, "i") &
  str_detect(words, "o") &
  str_detect(words, "u")
]
```

All sentences that contain a color:

```{r}
str_view(sentences, "\\b(red|green|blue)\\b", match = TRUE)
```

```{r}
colors <- colors()
head(colors)
colors %>% str_view("\\d", match = TRUE)
colors <- colors[!str_detect(colors, "\\d")]

pattern <- str_c("\\b(", str_flatten(colors, "|"), ")\\b")
str_view(sentences, pattern, match = TRUE)
```

Get rid of the modifiers.

```{r}
pattern <- str_c(".(", str_flatten(colors, "|"), ")$")
str_view(colors, pattern, match = TRUE)
colors[!str_detect(colors, pattern)]

prefix <- c("dark", "light", "medium", "pale")
pattern <- str_c("^(", str_flatten(prefix, "|"), ")")
colors[!str_detect(colors, pattern)]
```

## Grouping and capturing

Parentheses are an important tool to control operator precedence in regular expressions.
But they also have an important additional effect: they create **capturing groups** that allow you to use to sub-components of the match.
There are three main ways you can use them:

-   To match a repeated pattern
-   To include a matched pattern in the replacement
-   To extract individual components of the match

### Backreferences

You can refer to the same text as previously matched by a capturing group with **backreferences**, like `\1`, `\2` etc.
For example, the following regular expression finds all fruits that have a repeated pair of letters:

```{r}
str_view(fruit, "(..)\\1", match = TRUE)
```

And this regexp finds all words that start and end with the same pair of letters:

```{r}
str_view(words, "^(..).*\\1$", match = TRUE)
```

You can also use backreferences with `str_replace()` and `str_replace_all()`.
The following code will switch the order of the second and third words:

```{r}
sentences %>% 
  str_replace("(\\w+) (\\w+) (\\w+)", "\\1 \\3 \\2") %>% 
  head(5)
```

### Extracting groups

You can also make use of groups with tidyr's `separate_groups()` which puts each `()` group into its own column.
This provides a natural complement to the other separate functions that you learned about in ...

stringr also provides a lower-level function for extract matches called `str_match()`.
It returns a matrix, so isn't as easy to work with, but it's useful to know about for the connection.

```{r}
sentences %>% 
  str_match("the (\\w+) (\\w+)") %>% 
  head()
```

### Named groups

If you have many groups, referring to them by position can get confusing.
It's possible to give them a name with `(?<name>…)`.
You can refer to it with `\k<name>`.

```{r}
str_view(words, "^(?<first>.).*\\k<first>$", match = TRUE)
```

You can also use named groups as an alternative to the `col_names` argument to `tidyr::separate_groups()`.

### Non-capturing groups

Occasionally, you'll want to use parentheses without creating matching groups.
You can create a non-capturing group with `(?:)`.
Typically, however, you'll find it easier to just ignore that result in the output of `str_match()`.

```{r}
x <- c("a gray cat", "a grey dog")
str_match(x, "(gr(e|a)y)")
str_match(x, "(gr(?:e|a)y)")
```

### Exercises

1.  Describe, in words, what these expressions will match:

    a.  `(.)\1\1`
    b.  `"(.)(.)\\2\\1"`
    c.  `(..)\1`
    d.  `"(.).\\1.\\1"`
    e.  `"(.)(.)(.).*\\3\\2\\1"`

2.  Construct regular expressions to match words that:

    a.  Who's first letter is the same as the last letter, and the second letter is the same as the second to last letter.
    b.  Contain one letter repeated in at least three places (e.g. "eleven" contains three "e"s.)

## Regular expression engine

Regular expressions work by stepping through a string letter by letter.

<https://www.regular-expressions.info/engine.html>

Backtracking: if the regular expression doesn't match, it'll back up.

### Overlapping

Matches never overlap, and the regular expression engine only starts looking for a new match after the end of the last match.
For example, in `"abababa"`, how many times will the pattern `"aba"` match?
Regular expressions say two, not three:

```{r}
str_count("abababa", "aba")
str_view_all("abababa", "aba")
```

### Zero width matches

It's possible for a regular expression to match no character, i.e. the space between too characters.
This typically happens when you use a quantifier that allows zero matches:

```{r}
str_view_all("abcdef", "c?")
```

But anchors also create zero-width matches:

```{r}
str_view_all("this is a sentence", "\\b")
str_view_all("this is a sentence", "^")
```

And `str_replace()` can insert characters there:

```{r}
str_replace_all("this is a sentence", "\\b", "-")
str_replace_all("this is a sentence", "^", "-")
```