I am attempting to use SwiftSoup to scrape some HTML. This example, based on the SwiftSoup github documentation, works fine…
func scrape() throws {
do {
let htmlFromSomeSource = "<html><body><p class="nerp">HerpDerp</p><p class="narf">HoopDoop</p>"
let doc = try! SwiftSoup.parse(htmlFromSomeSource)
let tag = try! doc.select("p").first()!
let tagClass = try! tag.attr("class")
} catch {
print("oh dang")
throw Abort(.notFound)
}
print(tagClass)
}
… Up until I mess with the selector or attribute targets, at which point everything crashes thanks to the implicitly unwrapped optionals (which I assume was just quick-and-dirty code to get smarter people started). That do/catch doesn't seem to help at all.
So what's the Right way? This compiles...
print("is there a doc?")
guard let doc = try? SwiftSoup.parse(response.body.description) else {
print("no doc")
throw Abort(.notFound)
}
print("should halt because there's no img")
guard let tag = try? doc.select("img").first()! else {
print("no paragraph tag")
throw Abort(.notFound)
}
print("should halt because there's no src")
guard let tagClass = try? tag.attr("src") else {
print("no src")
throw Abort(.notFound)
}
... but again if I mess with the selector or attribute it crashes out, "Unexpectedly found nil while unwrapping an Optional value" (after "is there a doc?"). I thought guard would halt the process when it encountered a nil? (If I convert "try?" to "try" the compiler complains that "initializer for conditional binding must have Optional type"…)
If you declare the function as throws you don't need a do - catch block inside the function. Just remove the block and the exclamation marks after try to pass through the errors to the caller function.
func scrape() throws { // add a return type
let htmlFromSomeSource = "<html><body><p class="nerp">HerpDerp</p><p class="narf">HoopDoop</p>"
let doc = try SwiftSoup.parse(htmlFromSomeSource)
guard let tag = try doc.select("p").first() else { throw Abort(.notFound) }
let tagClass = try tag.attr("class")
// return something
}
Say I have a closure with two optional parameters:
(data: MyData?, error: Error?) in
// I want to safely unwrap data & make sure error is nil
if let data = data, let error== nil {
}
The if condition above gives me error: Variable binding in a condition requires an initializer.
I understand I might have used wrong syntax with the let error==nil part. But, what is the correct way to do the condition check ?
For Swift 3 just drop let before error
if let data = data, error == nil {
// do stuff
}
Before Swift 3 the syntax was a little different
if let data = data where error == nil {
}
In your case it can be a good idea to use guard :
(data: MyData?, error: Error?) in
// I want to safely unwrap data & make sure error is nil
guard let data = data, error == nil else {
return
}
// code here
Another option is to use if case with pattern matching:
if case let (data?, nil) = (data, error) {
// ...
}
Here data? is the "optional pattern" and a shortcut for .some(data).
Sometimes, I want to use guard combined with let & where to simplify my code. But I wonder what's the priority of let and where. For example:
class Person {
func check() -> Bool? {
print("checking")
return nil
}
}
func test(dont: Bool) {
let person = Person()
guard let check = person.check() where dont else {
print("should not check")
return
}
print("result: \(check)")
}
test(false)
As you can see the console result, printed output are:
checking
should not check
For the condition of let check = person.check() where dont in guard <condition> else { } syntax, even the expression in where doesn't relate to the results of expression in let, Swift seems to execute let first then check where later. Sometimes in my code, let optional binding takes lots of calculation and where is only a simple condition without relying on let results, should I move the where out of guard? Or I'm wrong about the priority or let & where?
You are right, in your Swift 2 code
guard let check = person.check() where dont else { }
the conditional binding let check = ... is checked first, and only if that succeeds, the boolean condition dont is checked. You can use
guard dont, let check = person.check() else { }
to check the boolean condition first.
This "asymmetry" in the syntax has been removed in Swift 3:
All guard clauses are separated by commas and the
where keyword is not used anymore. So you have
guard let check = person.check(), dont else { }
// or
guard dont, let check = person.check() else { }
The conditions are checked from left to right with short-circuiting,
i.e. if one condition fails then the else-block is executed without
checking the remaining conditions.
I have been reading about Optionals in Swift, and I have seen examples where if let is used to check if an Optional holds a value, and in case it does – do something with the unwrapped value.
However, I have seen that in Swift 2.0 the keyword guard let is used mostly. I wonder whether if let has been removed from Swift 2.0 or if it still possible to be used.
Should I change my programs that contain if let to guard let?
if let and guard let serve similar, but distinct purposes.
The "else" case of guard must exit the current scope. Generally that means it must call return or abort the program. guard is used to provide early return without requiring nesting of the rest of the function.
if let nests its scope, and does not require anything special of it. It can return or not.
In general, if the if-let block was going to be the rest of the function, or its else clause would have a return or abort in it, then you should be using guard instead. This often means (at least in my experience), when in doubt, guard is usually the better answer. But there are plenty of situations where if let still is appropriate.
Guard can improve clarity
When you use guard you have a much higher expectancy for the guard to succeed and it's somewhat important that if it doesn't succeed, then you just want to exit scope early. Like you guard to see if a file/image exists, if an array isEmpty or not.
func icon() -> UIImage {
guard let image = UIImage(named: "Photo") else {
return UIImage(named: "Default")! //This is your fallback
}
return image //-----------------you're always expecting/hoping this to happen
}
If you write the above code with if-let it conveys to the reading developer that it's more of a 50-50. But if you use guard you add clarity to your code and it implies I expect this to work 95% of the time...if it ever failed, I don't know why it would; it's very unlikely...but then just use this default image instead or perhaps just assert with a meaningful message describing what went wrong!
Avoid guards when they create side effects, guards are to be used as a natural flow. Avoid guards when else clauses introduce side effects.
Guards establish required conditions for code to execute properly,
offering early exit
When you perform significant computation in the positive branch, refactor from if to a guard statement and returns the fallback value
in the else clause
From: Erica Sadun's Swift Style book
Also as a result of the above suggestions and clean code, it's more likely you will want/need to add assertions into failed guard statements, it just improves readability and makes it clear to other developers what you were expecting.
guard let image = UIImage(named: selectedImageName) else { // YESSSSSS
assertionFailure("Missing \(selectedImageName) asset")
return
}
guard let image = UIImage(named: selectedImageName) else { // NOOOOOOO
return
}
From: Erica Sadun's Swift Style book + some modifications
(you won't use asserts/preconditions for if-lets. It just doesn't seem right)
Using guards also help you improve clarity by avoiding pyramid of doom. See Nitin's answer.
Guard keeps code that handles a violated requirement next to the requirement
To be clear, guard isn't always about success vs failure. The more generic way to see it is about handling a violated requirement vs process code that isn't violated.
example:
func getImage(completion: (image: Image)? -> Void) {
guard cache["image1"] == false else {
completion(cache["image1"]!)
}
downloadAndStore("image1") { image in
completion(image)
}
}
In the above the requirement is for the image to not be present in cache. If the image is present then our requirement is violated. We return early. As you can see we also handle the violated code path, right next to its requirement i.e. the structure is not:
if requirement {
.
.
ten lines of code
.
.
} else {
handle requirement
}
The Swift Docs on Control Flow explain the idea behind that:
Using a guard statement for requirements improves the readability of
your code, compared to doing the same check with an if statement.
It lets you write the code that’s typically executed without wrapping it in an else block
it lets you keep the code that handles a violated requirement next to the requirement.
Guard avoids nesting by creating a new variable in the current scope
There is one important difference that I believe no one has explained well.
Both guard let and if let unwrap the variable however
With guard let you are creating a new variable that will exist in the current scope.
With if let you’re only creating a new variable inside the code block.
guard let:
func someFunc(blog: String?) {
guard let blogName = blog else {
print("some ErrorMessage")
print(blogName) // will create an error Because blogName isn't defined yet
return
}
print(blogName) // You can access it here ie AFTER the guard statement!!
//And if I decided to do 'another' guard let with the same name ie 'blogName' then I would create an error!
guard let blogName = blog else { // errorLine: Definition Conflicts with previous value.
print(" Some errorMessage")
return
}
print(blogName)
}
if-let:
func someFunc(blog: String?) {
if let blogName1 = blog {
print(blogName1) // You can only access it inside the code block. Outside code block it doesn't exist!
}
if let blogName1 = blog { // No Error at this line! Because blogName only exists inside the code block ie {}
print(blogName1)
}
}
For more info on if let do see: Why redeclaration of optional binding doesn't create an error
Guard requires scope exiting
(Also mentioned in Rob Napier's answer) :
You MUST have guard defined inside a func. It's major purpose is to abort/return/exit scope, if a condition isn't met:
var str : String?
guard let blogName1 = str else {
print("some error")
return // Error: Return invalid outside of a func
}
print (blogName1)
For if let you don't need to have it inside any func:
var str : String?
if let blogName1 = str {
print(blogName1) // You don't get any errors!
}
guard vs if
It's worth noting that it's more appropriate to see this question as guard let vs if let and guard vs if.
A standalone if doesn't do any unwrapping, neither does a standalone guard. See example below. It doesn't exit early if a value is nil. There are NO optional values. It just exits early if a condition isn't met.
let array = ["a", "b", "c"]
func subscript(at index: Int) -> String?{
guard index > 0, index < array.count else { return nil} // exit early with bad index
return array[index]
}
When to use if-let and when to use guard is often a question of style.
Say you have func collectionView(collectionView: UICollectionView, numberOfItemsInSection section: Int) -> Int and an optional array of items (var optionalArray: [SomeType]?), and you need to return either 0 if the array is nil (not-set) or the count if the array has a value (is set).
You could implement it like this using if-let:
func collectionView(collectionView: UICollectionView, numberOfItemsInSection section: Int) -> Int
{
if let array = optionalArray {
return array.count
}
return 0
}
or like this using guard:
func collectionView(collectionView: UICollectionView, numberOfItemsInSection section: Int) -> Int
{
guard let array = optionalArray else {
return 0
}
return array.count
}
The examples are functionally identical.
Where guard really shines is when you have a task like validating data, and you want the function to fail early if anything is wrong.
Instead of nesting a bunch of if-lets as you get closer to finishing validation, the "success path" and the now successfully bound optionals are all in the main scope of the method, because the failure paths have all returned already.
I'll try to explain the usefulness of guard statements with some (unoptimized) code.
You have a UI where you are validating text fields for user registration with first name, last name, email, phone and password.
If any textField is not containing valid text, it should make that field firstResponder.
here is the unoptimized code:
//pyramid of doom
func validateFieldsAndContinueRegistration() {
if let firstNameString = firstName.text where firstNameString.characters.count > 0{
if let lastNameString = lastName.text where lastNameString.characters.count > 0{
if let emailString = email.text where emailString.characters.count > 3 && emailString.containsString("#") && emailString.containsString(".") {
if let passwordString = password.text where passwordString.characters.count > 7{
// all text fields have valid text
let accountModel = AccountModel()
accountModel.firstName = firstNameString
accountModel.lastName = lastNameString
accountModel.email = emailString
accountModel.password = passwordString
APIHandler.sharedInstance.registerUser(accountModel)
} else {
password.becomeFirstResponder()
}
} else {
email.becomeFirstResponder()
}
} else {
lastName.becomeFirstResponder()
}
} else {
firstName.becomeFirstResponder()
}
}
You can see above, that all the strings (firstNameString, lastNameString etc) are accessible only within the scope of the if statement. so it creates this "pyramid of doom" and has many issues with it, including readability and ease of moving things around (if the fields' order is altered, you have to rewrite most of this code)
With the guard statement (in the code below), you can see that these strings are available outside the {} and are made use of, if all fields are valid.
// guard let no pyramid of doom
func validateFieldsAndContinueRegistration() {
guard let firstNameString = firstName.text where firstNameString.characters.count > 0 else {
firstName.becomeFirstResponder()
return
}
guard let lastNameString = lastName.text where lastNameString.characters.count > 0 else {
lastName.becomeFirstResponder()
return
}
guard let emailString = email.text where
emailString.characters.count > 3 &&
emailString.containsString("#") &&
emailString.containsString(".") else {
email.becomeFirstResponder()
return
}
guard let passwordString = password.text where passwordString.characters.count > 7 else {
password.becomeFirstResponder()
return
}
// all text fields have valid text
let accountModel = AccountModel()
accountModel.firstName = firstNameString
accountModel.lastName = lastNameString
accountModel.email = emailString
accountModel.password = passwordString
APIHandler.sharedInstance.registerUser(accountModel)
}
If the order of the fields changes, just move respective lines of code up or down, and you are good to go.
This is a very simple explanation and a use case. Hope this helps!
Basic Difference
Guard let
Early exist process from the scope
Require scope existing like return, throw etc.
Create a new variable that can be accessed outside the scope.
if let
Can not access outside the scope.
no need for return statement. But we can write
NOTE: Both are used to unwrap the Optional variable.
guard let vs if let
func anyValue(_ value:String?) -> String {
guard let string = value else {
return ""
}
return string
}
func anyValue(_ value:String?) -> String {
if let string = value {
return string
}
return ""
}
The clearest explanation I saw was in the Github Swift Style Guide:
if adds a level of depth:
if n.isNumber {
// Use n here
} else {
return
}
guard doesn't:
guard n.isNumber else {
return
}
// Use n here
guard
A guard statement is used to transfer program control out of a scope
if one or more conditions aren’t met.
The value of any condition in a guard statement must be of type Bool
or a type bridged to Bool. The condition can also be an optional
binding declaration.
A guard statement has the following form:
guard condition else {
//Generally return
}
if let
Also popular as optional binding.
For accessing optional object we use if let.
if let roomCount = optionalValue {
print("roomCount available")
} else {
print("roomCount is nil")
}
I learnt this from swift with Bob..
Typical Else-If
func checkDrinkingAge() {
let canDrink = true
if canDrink {
print("You may enter")
// More Code
// More Code
// More Code
} else {
// More Code
// More Code
// More Code
print("Let me take you to the jail")
}
}
Issues with Else-If
Nested brackets
Have to read every line to spot the error message
Guard Statement
A guard block only runs if the condition is false, and it will exit out of the function through return. If the condition is true, Swift ignores the guard block. It provides an early exit and fewer brackets.+
func checkDrinkProgram() {
let iCanDrink = true
guard iCanDrink else {
// if iCanDrink == false, run this block
print("Let's me take you to the jail")
return
}
print("You may drink")
// You may move on
// Come on.
// You may leave
// You don't need to read this.
// Only one bracket on the bottom: feeling zen.
}
Unwrap Optionals with Else-If
A guard statement is not only useful for replacing a typical conditional block with an else-if statement, but also great for unwrapping optionals by minimizing the number of brackets. To compare, let's first begin how to unwrap multiple optionals with else-if.
First, let us create three optionals that will be unwrapped.
var publicName: String? = "Bob Lee"
var publicPhoto: String? = "Bob's Face"
var publicAge: Int? = nil
The Worst Nightmare
func unwrapOneByOne() {
if let name = publicName {
if let photo = publicPhoto {
if let age = publicAge {
print("Bob: \(name), \(photo), \(age)")
} else {
print("age is mising")
}
} else {
print("photo is missing")
}
} else {
print("name is missing")
}
}
The code above certainly works but violates the DRY principle. It's atrocious. Let us break it down.+
Slightly Better
The code below is more readable than above.+
func unwrapBetter() {
if let name = publicName {
print("Yes name")
} else {
print("No name")
return
}
if let photo = publicPhoto {
print("Yes photo")
} else {
print("No photo")
return
}
if let age = publicAge {
print("Yes age")
} else {
print("No age")
return
}
}
Unwrap with Guard
The else-if statements can be replaced with guard.+
func unwrapOneByOneWithGuard() {
guard let name = publicName else {
print("Name missing")
return
}
guard let photo = publicPhoto else {
print("Photo missing")
return
}
guard let age = publicAge else {
print("Age missing")
return
}
print(name)
print(photo)
print(age)
}
Unwrap Multiple Optionals with Else-If
So far, you've been unwrapping optionals one by one. Swift allows us to unwrap multiple optionals at once. If one of them contains nil, it will execute the else block.
func unwrap() {
if let name = publicName, let photo = publicPhoto, let age = publicAge {
print("Your name is \(name). I see your face right here, \(photo), you are \(age)")
} else {
// if any one of those is missing
print("Something is missing")
}
}
Be aware that when you unwrap multiple optionals at once, you can't identify which contains nil
Unwrap Multiple Optionals with Guard
Of course, we should use guard over else-if.+
func unwrapWithGuard() {
guard let name = publicName, let photo = publicPhoto, let age = publicAge else {
// if one or two of the variables contain "nil"
print("Something is missing")
return
}
print("Your name is \(name). I see your, \(photo). You are \(age).")
// Animation Logic
// Networking
// More Code, but still zen
}
The main difference between guard and if statements in swift are:
The if statement is used to run code when a condition is met.
The guard statement is used to run code when a condition is not met.
I'm a bit new to swift optionals and am looking for a graceful way of dealing with transforming optionals to non-optionals in a map function.
I'd like my map transformation to not include nulls and also to have the most succint optionals syntax.
let result = JSON as? [Dictionary<String,String>]
//how can I avoid the result != nil check here with a more graceful optionals syntax
if (result != nil && response!.statusCode == 200 && error == nil){
let strHighlights = result!.map { json in
//is there a more 'optionals' syntax for this so that it doesn't return nulls in the map?
return json["uname"]! //i would like to avoid the bang! here as that could cause an error
}
callback(success: true ,errorMsg: nil, highlights:strHighlights)
}
else{
callback(success: false,errorMsg:error?.localizedDescription, highlights:[])
}
Can I do this more gracefully than checking to see if result != nil in the 3rd line? Also, in the map function I'd prefer not to do the hard ! cast in the json["uname"] in case it doesn't exist. Can I use optional syntax to gracefully ignore a nil value in that optional so that it doesn't get returned in the map? I can do a json["uname"] != nil check but that seems a bit verbose and makes me wonder if there is a more succint optionals syntax in swift for that.
I think this is what you want:
if let result = result, response = response
where response.statusCode == 200 && error == nil {
let strHighlights = result.reduce([String]()) { array, json in
if let name = json["uname"] {
return array + [name]
} else {
return array
}
}
callback(success: true ,errorMsg: nil, highlights:strHighlights)
} else {
callback(success: false,errorMsg:error?.localizedDescription, highlights:[])
}
(No exclamation marks required)
Oh and btw: Every kind of those functions (map, filter, reversed, ... Can be expressed in a single reduce function. Also combinations of them can often be represented with a single reduce function.