Follow me on Twitter at @tim_deschryver | Subscribe to the Newsletter | Originally published on timdeschryver.dev.
My first experience with pattern matching was a few years ago when I was fiddling with some functional programming languages like Elixir and F#. The ability to pattern match a value or object was one thing that I was missing when I was writing code in C# (and also in JavaScript).
I was happy to see that C# 7 introduced the first version of pattern matching, which was supporting the bare minimum. Because of my previous encounters with pattern matching, this was a bit of a let-down because it didn't have all the possibilities I was used to. Luckily, the release of C# 8 extended the pattern match syntax by supporting more features, which were further enhanced in C# 9, and it also seems like this trend is continuing in the upcoming versions of C#.
Why pattern matching
So why am I excited about this?
Everything that you can do with pattern matching is also possible without using it, but it won't look as good.
Pattern matching can turn a complex if-else or switch statement into a compact block of code.
Because of this, my developer experience has improved with the addition of pattern matching.
Its strength is that it's expressive, making it easier to read and harder to write bugs.
A bonus is that the compiler acts as a safety net (more on this later), it warns for unhandled cases and prevents cases to conflict with each other.
This isn't something you get with a normal if-else statement, which sadly, has bitten me in the past.
It's good to know that compiler has your back by reporting silly mistakes.
For the rest of this post, let's take a look at some examples of how pattern matching with switch expressions is used.
A simple example
The simplest example is a check for a simple true/false value.
Pattern matching compares an input value (false
in the example below) to a set of defined patterns, think of them as a set of conditions. The patterns inside the switch expression are evaluated from top to bottom, and the first case that causes a match is executed.
var output = false switch
{
true => "true",
false => "false",
};
// output: false
Constant Patterns
We can use pattern matching to check if a variable has a constant value.
The types that are matchable are strings, chars, numbers, and enums.
var output = 4 switch
{
1 => "one",
2 => "two",
3 => "three",
4 => "four",
5 => "five",
};
// output: four
Discard Patterns
Because not every number is included in the switch expression of the previous example, the C# compiler warns that there are uncovered cases.
warning CS8509: The switch expression does not handle all possible values
of its input type (it is not exhaustive).
For example, the pattern '0' is not covered.
To address this warning, we can add a fallback case.
Think of it as a default case from a switch statement, a wildcard that is invoked when no other of the previous patterns are matched. In the example below, the discard operator (represented as an underscore _
) is used to match all other possible values.
var output = 9 switch
{
1 => "one",
2 => "two",
3 => "three",
4 => "four",
5 => "five",
_ => "other"
};
// output: other
Variable Patterns
A matched value can be assigned to a variable.
In the following example, the variable acts as a wildcard that matches any value.
A variable can also be defined when the input value of a Type Pattern matches a Type, the created variable is typed in this case. The variable can then be used in the execution expression (after the =>
) to create a return value.
Furthermore, a when
statement can be added to the pattern to add an extra guard to the pattern.
The when
statement can be used to match a non-constant value against a condition, for example, to invoke a method on the matched variable. Besides matching the object that is being matched, it's also possible to include other objects in the when
statement.
var greetWithName = false;
var output = "Mrs. Kim" switch
{
_ when greetWithName == false => $"Hi",
"Tim" => "Hi Tim!",
var str when str.StartsWith("Mrs.") || str.StartsWith("Mr.") => $"Greetings {str}",
var str => $"Hello ${str}",
};
// output: Mrs. Kim
Relational Patterns
We use relation operators (=
, >
, <
, >=
, <=
) to test if the input is equal, greater, or less than another value.
var output = 4 switch
{
< 3 => "less than 3",
<= 7 => "less than or equal to 7",
< 1 => "less than 10",
_ => "greater than or equal to 10"
};
// output: ?
Did you spot the mistake in the example?
If that isn't the case, you don't have to worry, because this time the compiler throws an error to address my mistake.
error CS8510: The pattern is unreachable. It has already been handled by
a previous arm of the switch expression or it is impossible to match.
Let's fix this mistake by correcting the pattern.
var output = 4 switch
{
< 3 => "less than 3",
<= 7 => "less than or equal to 7",
< 10 => "less than 10",
_ => "greater than or equal to 10"
};
// output: less than or equal to 7
Multiple Patterns
To combine patterns or to negate values, we use the logical operators (and
, or
, not
).
var output = 4 switch
{
1 or 2 or 3 => "1, 2, or 3",
> 3 and <= 6 => "between 3 and 6",
not 7 => "not 7",
_ => "7"
};
// output: between 3 and 6
Tuple Patterns
Matching single values is nice but not very useful in many cases.
To match multiple values, we can pattern match multiple input values by using tuples.
var output = (5, false) switch
{
(< 4, true) => "lower than 4 and true",
(< 4, false) => "lower than 4 and false",
(4, true) => "4 and true",
(5, _) => "five and something",
(_, false) => "something and false",
_ => "something else",
};
// output: five and something
Property Patterns
When the input is an object, we can add a pattern to the properties of an object.
var output = new User("Tim Deschryver", "Developer") switch
{
{ Role: "Admin" } => "the user is an admin",
{ Role: "Administrator" } => "the user is an admin",
{ Name: "Tim", Role: "Developer" } => "the user is Tim and he is a developer",
{ Name: "Tim" } => "the user is Tim and he isn't a developer",
_ => "the user is unknown",
};
// output: the user is Tim and he is a developer
record User(string Name, string Role);
Nested Properties Patterns
It's even possible to match nested properties.
var output = new Member("Tim Deschryver", new MemberDetails(8, false)) switch
{
{ Details: { Blocked: true } } => Array.Empty<string>(),
{ Details: { MonthsSubscribed: < 3 } } => new[] { "comments" },
{ Details: { MonthsSubscribed: < 9 } } => new[] { "comments", "mention" },
_ => new[] { "comments", "mention", "ping" },
};
// output: comments, mention
record Member(string Name, MemberDetails Details);
record MemberDetails(int MonthsSubscribed, bool Blocked);
The above example can be hard to read (especially if you need to drill down into multiple objects).
As a remedy, C# 10 provides a new syntax called extended property patterns to make this easier on the eyes.
The refactored example looks as follows. Much better, right?
var output = new Member("Tim Deschryver", new MemberDetails(8, false)) switch
{
{ Details.Blocked: true } => Array.Empty<string>(),
{ Details.MonthsSubscribed: < 3 } => new[] { "comments" },
{ Details.MonthsSubscribed: < 9 } => new[] { "comments", "mention" },
_ => new[] { "comments", "mention", "ping" },
};
// output: comments, mention
record Member(string Name, MemberDetails Details);
record MemberDetails(int MonthsSubscribed, bool Blocked);
Type Patterns
Pattern matching can also be used to match a type of an object.
Type patterns are useful when you have a generic handler that acts as a pass-through.
var output = new InventoryItemRemoved(3) as object switch
{
// the variable `added` is of type `InventoryItemAdded`
InventoryItemAdded added => $"Added {added.Amount}",
// the variable `removed` is of type `InventoryItemRemoved`
InventoryItemRemoved removed => $"Removed {removed.Amount}",
InventoryItemDeactivated => "Deactivated",
null => throw new ArgumentNullException()
// the variable `o` is of type `object`
var o => throw new InvalidOperationException($"Unknown {o.GetType().Name}")
};
// output: Removed 3
Examples
The examples we've seen so far are simple and are here to show the different syntaxes to build a pattern.
To unlock the true power of pattern matching, multiple pattern matching strategies can be combined.
// retrieve the rate of an appointment
var holidays = new DateTime[] {...};
var output = new Appointment(DayOfWeek.Friday, new DateTime(2021, 09, 10, 22, 15, 0), false) switch
{
{ SocialRate: true } => 5,
{ Day: DayOfWeek.Sunday } => 25,
Appointment a when holidays.Contains(a.Time) => 25,
{ Day: DayOfWeek.Saturday } => 20,
{ Day: DayOfWeek.Friday, Time.Hour: > 12 } => 20,
{ Time.Hour: < 8 or >= 18 } => 15,
_ => 10,
};
// output: 20
record Appointment(DayOfWeek Day, DateTime Time, bool SocialRate);
// a combination of the variable pattern and the tuple pattern
var output = ("", "Tim") switch
{
var (title, name) when title.Equals("Mrs.") || title.Equals("Mr.") => $"Greetings {title} {name}",
var (_, name) and (_, "Tim") => $"Hi {name}!",
var (_, name) => $"Hello {name}",
};
// output: Hi Tim!
// format a string
var contactInfo = new ContactInfo("Sarah", "Peeters", "0123456789");
var output = contactInfo switch
{
{ TelephoneNumber: not null } or { TelephoneNumber: not "" } => $"{contactInfo.FirstName} {contactInfo.LastName} ({contactInfo.TelephoneNumber})",
_ => $"${contactInfo.FirstName} ${contactInfo.LastName}"
};
// output: Sarah Peeters (0123456789)
// C# Language Highlights: Tuple Pattern Matching https://www.youtube.com/watch?v=v_xKLwTv3AI
IEnumerable<string> sequence = new[] { "foo" };
var output = sequence switch
{
string[] { Length: 0 } => "array with no items",
string[] { Length: 1 } => "array with a single item",
string[] { Length: 2 } => "array with 2 items",
string[] => $"array with more than 2items",
IEnumerable<string> source when !source.Any() => "empty enumerable",
IEnumerable<string> source when source.Count() < 3 => "a small enumerable",
IList<string> list => $"a list with {list.Count} items",
null => "null",
_ => "something else"
};
// output: array with a single item
Official documentation
We've only seen examples of pattern matching expressions in this post.
For more details, please take a look at the official documentation:
- Pattern matching overview
- Use pattern matching to build your class behavior for better code
- Tutorial: Use pattern-matching to build type-driven and data-driven algorithms
Follow me on Twitter at @tim_deschryver | Subscribe to the Newsletter | Originally published on timdeschryver.dev.
Top comments (4)
Hello.
Thanks for this article. I once found a bit of that syntax with
switch
and thewhen
keyword. I was badly wishing there is such thing because this is handy but I didn't know it is so complete. I'm going through the codes and playing them in Visual Studio.I have questions I note as I go so there may be one or more.
Variable Patterns :
There is apparently a small error in the article because the output gives "Hi" alone. Which makes sense to me when I read the code.
_ when greetWithName == false => $"Hi",
applies.Format a string
Modified the code so that phone is
null
.Does not seem to work as expected and goes to the 1st pattern.
Any Idea why?
Regards.
Thanks! I think you found some syntax mistakes in the snippets, and again thanks for pointing them out.
Variable pattern should be:
Format a string should use
and
instead ofor
:Thanks for the fixes.
Tried them and they work.
I must admit that when reading the code this or looked legit.
But when you go back to boolean it makes sense.
TelephoneNumber
isnull
TelephoneNumber not null
β False β 0TelephoneNumber not ""
β True β 10 or 1 is 1 so it executes the line.
0 and 1 is 0 so it filters accordingly.
Regards.
Thanks Tim, I'll look into this for my upcoming 5.0 project. I like what they did after having for years; written more complex code for what amounts to a simple lookup. It looks very similar to plain ole switch case constrcts.