Quick Guides

for Masterminds

Core Data
with SwiftUI
Learn how to store data in a database with Core Data and SwiftUI

J.D Gauchat
www.jdgauchat.com
Quick Guides for Masterminds
Copyright © 2022 John D Gauchat
All Rights Reserved

No part of this publication may be reproduced, distributed, or transmitted

in any form or by any means, including photocopying, recording, or other
electronic or mechanical methods, without the prior written permission of
the publisher, except in the case of brief quotations embodied in critical
reviews and certain other noncommercial uses permitted by copyright law.

Companies, services, or product names used in this book are for

identification purposes only. All trademarks and registered trademarks are
the property of their respective owners.

Apple™, iPhone™, iPad™, Mac™, among others mentioned in this work, are
trademarks of Apple Inc.

The information in this book is distributed without warranty. Although

every precaution has been taken in the preparation of this work, neither
the author nor the publisher shall have any liability to any person or entity
with respect to any loss or damage caused or alleged to be caused directly
or indirectly by the information contained in this work.

The source code for this book is available at

www.formasterminds.com

Copyright Registration Number: 1140725

1st Edition 2022

Table of Contents

CORE DATA
DATA MODEL
CORE DATA STACK
MANAGED OBJECTS
FETCH REQUEST
ASYNCHRONOUS ACCESS
CORE DATA APPLICATION
PREVIEWS
SORT DESCRIPTORS
PREDICATES
DELETE OBJECTS
MODIFY OBJECTS
CUSTOM FETCH REQUESTS
SECTIONS
TO-MANY RELATIONSHIPS
CORE DATA
Core Data is the system used to store structured data on the device. Core
Data is an Object Graph manager that defines and manages its own objects
and connections and stores them in a database for easy access. We
determine the composition of the objects and the relationships depending
on the information we want to store and the system takes care of encoding
and decoding the objects, preserving consistency and maximizing efficiency
to quickly write and read data.
Data Model

The structure of the Core Data’s Object Graph is defined with a data model.
A Core Data model is a definition of the type of objects the graph is going
to contain (called Entities) and their connections (called Relationships).
A model can be created from code, but Xcode offers a practical editor to
define the structure of the graph. The model is stored in a file and then the
file is compiled and included in the Core Data system created for our
application. Xcode offers a template to create this file.

Figure 1: Option to create a Core Data model in the iOS panel

The file may be created with any name we want but it must have the
extension xcdatamodel. Once created, it is included in our project along
with the rest of the files. Clicking on it reveals the Xcode editor in the
Editor Area.

Figure 2: Model editor


The model contains three main components: Entities, Attributes, and
Relationships. Entities are the objects, Attributes are the objects’
properties, and Relationships are the connections between objects. The
first step is to add Entities to the model. Entities are created from the Add
Entity button at the bottom of the editor (Figure 2, number 1). When we
press this button, Xcode creates an entity with the generic name "Entity".

Figure 3: New Entities

We can change the name of the newly created entity by double-clicking

the item (Figure 3, number 1) or editing the field in the Data Model
Inspector panel (Figure 3, number 2).
An entity defines the objects that are going to be part of the Object Graph,
so the next step is to declare the type of values those objects are going to
manage. For this purpose, entities include Attributes. To add an attribute,
we must select the entity and press the + button under the Attributes area
(Figure 3, number 3) or press the Add Attribute button at the bottom of the
editor (Figure 2, number 2). The attribute is added with the generic name
"attribute" and the data type Undefined. Again, we can change the name
of the attribute by double-clicking on it or from the Data Model Inspector
panel (Figure 3, number 2). For our example, we called the entity Books
and the first attribute title (Figure 4, number 1).

Figure 4: New Attributes



IMPORTANT: The name of entities must start with an upper-case

letter and the names of attributes and relationships with a lower-
case letter.

Every attribute must be associated with a data type for the objects to know
what kind of values they can manage (Figure 4, number 2). Clicking on the
attribute’s type, we can open a menu to select the right data type. The
most frequently used are Integer 16, Integer 32, Integer 64, Double, Float,
String, Boolean, Date, and Binary Data. The Integer 16, 32, or 64 options
are for Int16, Int32, and Int64 values, Double and Float are for Double and Float
values, String is for String values, Boolean is for Bool values, Date is for Date
values, and Binary Data is for Data values.
An entity may contain as many attributes as our objects need. For example,
we may add a few more attributes to complement the information
required for books.

Figure 5: Multiple Attributes


In this example, we have added an attribute called year to store the year in
which the book was published, and two attributes of type Binary Data to
store images (the book's cover and thumbnail). The data types used by
these attributes are analog to Swift data types. The title attribute takes a
String value, the year attribute stores a value of type Int32, and the images
are going to be stored as Data structures.
Most values don't require much consideration, but images are made of big
chunks of data. Storing large amounts of data in a Persistent Store can
affect the system's performance and slow down essential processes like
searching for values or migrating the model. One alternative is to store the
images in separate files, but it can get cumbersome to coordinate
hundreds of files with the data in the database. Fortunately, Core Data can
perform the process for us. All we need to do is to store the image as
Binary Data and select the option Allows External Storage, available in the
Data Model inspector panel inside the Utilities Area, as shown below. After
the option is selected, the images assigned to that attribute are stored in
separate files managed by the system.

Figure 6: Option to store images outside the Persistent Store

We could have also included another attribute for the author’s name, but
here is when we need to think about the structure of the Object Graph and
how the information will be stored. If we include a String type attribute for
the author's name inside the Books entity, every time the user inserts a
new book it will have to type the name of the author. This is error prone,
time consuming, and when several books of the same author are available,
it is impossible to make sure that all share the same exact name (for
example, one book could have the author’s middle name and others just
the first one). Without the certainty of having the exact same name, we
can never incorporate features in our app such as ordering the books by
author or getting the list of books written by a particular author. Things get
worse when, along with the name, we also decide to store other
information about the author, like his or her date of birth or their
nationality. A proper organization of this information demands separate
objects and therefore we must create new entities to represent them.
Additional entities are added to the model in the same way as we did with
the first one. Figure 7, below, shows our model with a new entity called
Authors containing an attribute called name.

Figure 7: Multiple Entities

Entities are blueprints that we use to define the characteristics of the

objects we want to store in the database. For instance, when we want to
store a new book in our example, we create a new object based on the
Books entity. That object will have four properties corresponding to the
values of its title, year, cover, and thumbnail. The same happens when we
want to store information of an author. We create a new object based on
the Authors entity and assign the name of the author to its name property.
At the end, we will have two objects in the database, one for the book and
another for the author. But if we want to retrieve these objects later, we
need a way to know which Books object is related to which Authors object.
To create this connection, the Core Data model includes Relationships.
A relationship is a property in one object containing a reference to another
object. Relationships can have a reference to only one object or a set of
objects. For example, in the Books entity, we can create a relationship that
contains a reference to only one object of the Authors entity, because
there can only be one author per book (for this example we are assuming
that our app is storing books written only by one author). On the contrary,
in the Authors entity, we need to establish a relationship that contains
references to multiple Books objects, because an author may have written
several books, not just one. Core Data calls these relationships according to
the number of objects they may reference. The names are To-One and To-
Many, and they are created pressing the + button in the Relationships area
below the Attributes area. Figure 8, below, shows a relationship called
author we have created for the Books entity of our example.

Figure 8: Relationship for the Books entity

A relationship only needs two values: its name (the name of the property)
and the destination (the type of objects it is referencing), but it requires
some parameters to be set. We must tell the model if the relationship is
going to be optional, define its type (To-One or To-Many), and determine
what should happen to the destination object if the source object is
deleted (the Delete Rule). All these options are available in the Data Model
Inspector panel when the relationship is selected, as shown below.

Figure 9: Relationship settings



By default, the relationship is set as Optional, which means that the source
may be connected to a destination object or not (a book can have an
author or not), the Type of the relationship is set to To-One (a book can
only have one author), and the Delete Rule is set to Nullify. The following
are all the values available for this rule.

Deny: If there is at least one object at the destination, the source

is not deleted (if there is an Authors object assigned to the Books
object, the book is not deleted).
Nullify: The connections between objects are removed, but the
objects at the destination are not deleted (if a Books object is
deleted, the Authors object associated with that book loses the
connection but is not deleted).
Cascade: The objects at the destination are deleted when the
source is deleted (the Authors object is deleted if one of its books is
deleted).
No Action: The objects at the destination are not deleted or
modified (the connections are preserved, even when the source
object does not exist anymore).

To find the right rule for a relationship, we must think in terms of the
information we are dealing with. Is it right to delete the author if one of its
books is deleted? In our case, the answer is simple. An author can have
more than one book, so we cannot delete the author when we delete a
book because there could be other books that are connected to that same
author. Therefore, the Nullify rule set by default is the right one for this
relationship. But this could change when we create the opposite
relationship, connecting the Authors entity to the Books entity. We need
this second relationship to search for books that belong to an author.
Figure 10 shows a relationship called books that we have created for the
Authors entity.

Figure 10: Relationship for the Authors entity

IMPORTANT: Relationships must always be bidirectional. If we set a

relationship from entity A to entity B, we must set the opposite
relationship from entity B to A. Core Data offers another type of
relationship called Fetched Properties to connect entities in only one
direction. You can add a Fetched Property from the area below the
Relationships area in the model’s editor.

The new relationship added in Figure 10 is in the Authors entity, so every

Authors object will have a property called books that we can use to retrieve
the Books objects associated to the author. Because one author can have
many books, the setting of this relationship is going to differ from the
previous one. In this case, we must set the Type of the relationship as To-
Many (to many books) and modify the Delete Rule according to how we
want our application to respond when an author is deleted. If we don't
want to keep books that do not have an author assigned, we should select
the Cascade option, so when an author is deleted all his or her books are
deleted too. But if we don't mind having books with no author around,
then the option should be kept as Nullify.

IMPORTANT: The Delete Rules are a way to ensure that the objects
remaining in the Object Graph are those that our application and the
user need. But we can always set the rule to Nullify and take care of
deleting all the objects ourselves.

There is a third value for the relationship called Inverse. Once we set the
relationships on both sides, it is highly recommended to set this value. It
just tells the model what the name of the opposite relationship is. Core
Data needs this to ensure the consistency of the Object Graph. Figure 11
shows the final setup for both relationships.

Figure 11: Inverse Relationships

Two relationships are simple to follow, but multiple relationships

connecting several entities together can turn the model into an
indecipherable mess. To help us identify every component of the model,
Xcode offers an additional visualization style that displays the entities as
boxes and the relationships as arrows connecting the boxes. The option,
called Editor Style, is at the bottom of the Editor Area (Figure 2, number 3).
Figure 12, below, shows what our model looks like when we switch to this
style. (Notice that the To-Many relationship is represented by double
arrows.)
Figure 12: Graphic Style

Do It Yourself: Create a Multiplatform project. Open the File menu

and select the File option to create a new file. Move to the Core Data
section and select the option Data Model from the iOS panel (Figure
1). Save the file with the name "books". Click on the file to open the
editor (Figure 2). Press the Add Entity button to create two entities
with the names Authors and Books. Create the attributes for these
entities as illustrated in Figures 5 and 6. Create the relationships for
both entities as shown in Figure 11. Set the books relationship as To-
Many and keep the rest of the values by default. Click on the Editor
Style button to see a graphic representation of the model (Figure
12).
Core Data Stack

The creation of the model is just the first step in the definition of the Core
Data system. Once we have all the entities along with their attributes and
relationships set up, we must initialize Core Data. Core Data is created from
a group of objects that are in charge of all the processes required to
manage the data, from the organization of the Object Graph to the storage
of the graph in a database. There is an object that manages the model, an
object that stores the data on file, and an object that intermediates
between this Persistent Store and our own code. The scheme is called
Stack. Figure 13 illustrates a common Core Data stack.

Figure 13: Core Data stack

The code in our application interacts with the Context to manage the
objects it needs to access and their values, the Context asks the Persistent
Store to read or add new objects to the graph, and the Persistent Store
processes the Object Graph and saves it in a file.
The Core Data framework offers classes to create objects that represent
every part of the stack. The NSManagedObjectModel class manages the model,
the NSPersistentStore class manages a Persistent Store, the
NSPersistentStoreCoordinator class is used to manage all the Persistent Stores
available (a Core Data stack can have multiple Persistent Stores), and the
NSManagedObjectContext creates and manages the context that intermediates
between our app and the store. Although we can instantiate these objects
and create the stack ourselves, the framework offers the NSPersistentContainer
class that takes care of everything for us. The class includes the following
initializer and properties to access each object of the stack.

NSPersistentContainer(name: String)—This initializer creates an

object that defines a Core Data stack. The name
NSPersistentContainer
argument is a string representing the name of the container. This
value must match the name of the Core Data model (the file's name,
without the extension).
managedObjectModel—This property sets or returns an
NSManagedObjectModel object that represents the Core Data model.
persistentStoreCoordinator—This property sets or returns the
NSPersistentStoreCoordinator object that manages all the Persistent Stores
available.
viewContext—This property sets or returns the
object in charge of the stack's context that we
NSManagedObjectContext
use to access and modify the Object Graph.

The NSPersistentContainer object automatically resolves conflicts for us, but

we must configure the context to determine how those conflicts are going
to be solved. The NSManagedObjectContext class includes the following
properties for this purpose.

automaticallyMergesChangesFromParent—This property sets or

returns a Boolean value that determines whether the context
automatically merges the changes in the Persistent Store and the
context.
mergePolicy—This property sets or returns an object that decides
the policy that the context is going to use to merge the changes in the
Persistent Store and the context. The Core Data framework defines
 global variables to set standard policies. The NSErrorMergePolicy variable
returns an error if the objects are different, the
NSMergeByPropertyStoreTrumpMergePolicy variable replaces changes in
memory by the external changes, the
NSMergeByPropertyObjectTrumpMergePolicy replaces the external changes
by the changes in memory, the NSOverwriteMergePolicy variable replaces
the values in the Persistent Store by the current changes, and the
NSRollbackMergePolicy uses the version of the objects in the Persistent
Store.

To create the Core Data stack, we must initialize a new NSPersistentContainer

object and then load the Persistent Stores (one by default). Because the
stores may take time to load, the class offers a specific method for this
purpose.

loadPersistentStores(completionHandler: Closure)—This
method loads the Persistent Stores and executes a closure when the
process is over. The closure receives two arguments, an
NSPersistentStoreDescription object with the configuration of the stack, and
an optional Error value to report errors.

All the communication between our app and the data in the Persistent
Store is done through the context. The context is created by the container
from the NSManagedObjectContext class. This class includes properties and
methods to manage the context and the objects in the Persistent Store.
The following are the most frequently used.

hasChanges—This property returns a Boolean value that indicates if

the context has changes that must be saved to the Persistent Store.
save()—This method saves the changes to the Persistent Store.
 reset()—This method resets the context to a basic state. All the
objects and modifications our app introduced to the context are
ignored.
fetch(NSFetchRequest)—This method returns an array with the
objects requested by the NSFetchRequest object.
delete(NSManagedObject)—This method deletes an object from
the Persistent Store.
count(for: NSFetchRequest)—This method returns the number of
objects found in the Persistent Store by the request. The for argument
specifies the request we want to perform.

When working with Core Data, the Core Data's Persistent Store becomes
our app's data model. Therefore, we can define a specific class to initialize
the Core Data stack, or just do it from our model, as in the following
example.

Listing 1: Initializing the Core Data stack from the model


import SwiftUI
import CoreData

class ApplicationData: ObservableObject {

let container: NSPersistentContainer

init() {
container = NSPersistentContainer(name: "books")
container.viewContext.automaticallyMergesChangesFromParent = true
container.loadPersistentStores(completionHandler: { storeDescription, error in
if let error = error as NSError? {
fatalError("Unresolved error \(error), \(error.userInfo)")
}
})
}
}

The ApplicationData class in this example defines a property called container to
store a reference to the Persistent Store. When the object is initialized, we
create an instance of the NSPersistentContainer class with the name of the
Core Data model (in our example, we called it "books") and configure it to
merge the changes between the context and the Persistent Store (this is
the configuration recommended by Apple). The object creates the stack
but does not load the Persistent Stores, we have to do it ourselves with the
loadPersistentStores() method. After completion, this method executes a
closure with two values: a reference to the Persistent Store just created,
and an Error value to report errors. Errors are infrequent, but if an error
occurs, we should warn the user. For instance, we can modify a state
property to open an Alert View to report the situation. In this example, we
just call the fatalError() function to stop the execution of the app.
Once we have the container, we must get the context from it and share it
with the views, so they can add, fetch, or remove objects from the
Persistent Store. The environment offers the managedObjectContext property
for this purpose. The following are the modifications we must introduced
to the App structure to inject a reference to the context into the
environment.

Listing 2: Injecting the context into the environment


import SwiftUI

@main
struct TestApp: App {
@StateObject var appData = ApplicationData()

var body: some Scene {

WindowGroup {
ContentView()
.environmentObject(appData)
.environment(\.managedObjectContext, appData.container.viewContext)
}
}
}

The model is initialized and injected into the environment as before. To
provide direct access to the Core Data context, we get a reference from the
viewContext property of the NSPersistentContainer object and then assign it to
the environment's managedObjectContext property with the environment()
modifier. From now on, the views can access the Core Data context from
the environment and fetch, add, modify or remove objects in the Persistent
Store.

Do It Yourself: Create a Swift file called ApplicationData.swift for

the model in Listing 1 and update the App structure with the code in
Listing 2. Replace the value of the name argument in the
NSPersistentContainer initializer with the name of your model's file (in
our example, it is called "books"). At this moment, the app doesn't
do anything other than creating the stack. We will see how to read
and store data next.
Managed Objects

Core Data does not store our custom objects; it defines a class called
NSManagedObject for this purpose. Every time we want to store information
in the database, we must create an NSManagedObject object, associate that
object to an Entity, and store the data the entity allows. For example, if we
create an object associated to the Books entity, we are only allowed to
store five values that corresponds to the Entity's attributes and relationship
(title, year, cover, thumbnail, and author). The class includes the following
initializer and methods to create and manage the objects.

NSManagedObject(context: NSManagedObjectContext)—This
initializer creates a new instance of the NSManagedObject class, or a
subclass of it, and adds it to the context specified by the context
argument.
fetchRequest()—This type method generates a fetch request for an
entity. A fetch request is a request we use to fetch objects of a
particular entity from the Persistent Store.
entity()—This type method returns a reference to the entity from
which the managed object was created. It is an object of type
NSEntityDescription with a description of the entity.

To simplify our work, the system allows us to define subclasses of the

NSManagedObject class that correspond to the entities in our model. (Instead
of creating instances of the NSManagedObject class, we create instances of
the Books and Authors classes.) Because this is common practice, Xcode
automatically creates the subclasses for us. All we need to do is to
associate each Entity with a subclass from the Data Model Inspector panel.

Figure 14: Entity's subclass



To ask Xcode to create the subclasses for us, we must select the entities
one by one, click on the Class Definition value for the Codegen option
(Figure 14, number 2), and make sure that the name of the subclass is
specified in the Name field (Figure 14, number 1). Once the options are
set, the classes are automatically created. For example, when we set these
options for the entities in our model, Xcode creates a subclass of
NSManagedObject called Books with the properties title, year, cover, thumbnail, and
author, and a subclass called Authors with the properties name and books. From
now on, all we need to do to store a book in the Persistent Store is to
create an instance of the Books class using the NSManagedObject initializer.

Do It Yourself: Click on the Core Data model file to open the editor.
Select the Books and the Authors entity, open the Data Model
Inspector panel, and make sure that the value of the Codegen option
is set to Class Definition and the name of the class is assigned to the
Name field, as shown in Figure 14.

IMPORTANT: The subclasses of the NSManagedObject class created to

represent each entity in our model are not visible in Xcode. They are
created internally and automatically modified every time entities or
attributes are added or removed from the model. If you decide not
to use these subclasses, you can select the value Manual/None from
the Codegen option and work directly with NSManagedObject objects or
define your own subclasses.
Fetch Request

As previously mentioned, to store an object in the Persistent Store, we

must create an NSManagedObject (or an instance of our subclasses), add it to
the context, and then save the context. The process to get the objects back
from the Persistent Store is the opposite. Instead of moving the changes
from the context to the Persistent Store, we must fetch the objects from
the Persistent Store and move them into the context. Once the objects are
in the context, we can read their properties, modify the values or delete
them, and then save them again. Core Data defines the NSFetchRequest class
to fetch objects from the Persistent Store. The class includes the following
properties for configuration.

predicate—This property sets or returns the predicate used to filter

the objects fetched by the request. It is a value of type NSPredicate; a
Foundation class used to establish logical conditions to filter objects.
sortDescriptors—This property sets or returns an array of sort
descriptors that determine how the objects fetched by the request
should be ordered. It is an array of values of type NSSortDescriptor; a
Foundation class used to sort the objects according to the value of a
property in ascending or descending order.
fetchLimit—This property sets or returns the maximum number of
objects that the request should return. It takes a value of type Int.
propertiesToFetch—This property sets or returns an array of values
that determine the properties we want to get (by default, all the
properties of the NSManagedObject objects are returned). The properties
of an entity (attributes) are represented by objects of the
NSPropertyDescription class, or subclasses of it.

Every time we need to get data from the Persistent Store, we must create
an NSFetchRequest object to determine what type of objects we want. To
simplify our work, SwiftUI includes the @FetchRequest property wrapper.
With this property wrapper, we can create a request or apply our own
NSFetchRequest object. The property wrapper takes care of performing the
request and updating the views with the values returned. It is created from
a structure of type FetchRequest. The following are some of the structure's
initializers.

FetchRequest(sortDescriptors: [SortDescriptor], predicate:

NSPredicate?, animation: Animation?)—This initializer creates a
fetch request with the configuration determined by the arguments
and produces a FetchedResults structure that manages and delivers the
objects to the view. The sortDescriptors argument is an array of
SortDescriptor structures that determine the order of the objects, the
predicate argument is an NSPredicate object that filters the objects, and
the animation argument determines how the changes are going to be
animated.
FetchRequest(fetchRequest: NSFetchRequest, animation:
Animation?)—This initializer creates a fetch request with the
NSFetchRequestobject provided by the fetchRequest argument and
produces a FetchedResults structure that manages and delivers the
objects to the view. The animation argument determines how the
changes are going to be animated.

A fetch request loads all the objects available in the Persistent Store. This is
not a problem when the number of objects is not significant. But a
Persistent Store can manage thousands of objects, which can consume
resources that the app and the system need to run. Therefore, instead of
the fetch request, the @FetchRequest property wrapper produces a value of
type FetchedResults. This is a structure that takes care of loading into the
context only the objects that are required by the view at any given
moment.
Asynchronous Access

Reading and writing information on a database is a delicate process. The

database may be accessed by different parts of the application and from
different threads, which can cause errors or even data corruption. To make
sure the data is accurate and safe, we must access the database from the
same thread assigned to the Core Data context. For this purpose, the
NSManagedObjectContext class includes the following asynchronous method.

perform(schedule: ScheduledTaskType, Closure)—This

asynchronous method performs a closure in the thread assigned to
the Core Data context. The schedule argument determines how the
closure is going to be executed. It is an enumeration with the values
immediate (the closure runs asynchronously), and enqueued (the closure
runs concurrently). If the argument is ignored, the closure is executed
asynchronously.
Core Data Application

It is time to see how all these tools work together to store and retrieve
data from a Persistent Store in a real application. The purpose of this
application is to show the list of objects stored in the Persistent Store and
create new ones.

Figure 15: Interface to work with Core Data

The initial view lists all the books available in the Persistent Store and
includes a button to open a second view to create new objects with the
values provided by the user.

Listing 3: Fetching objects from the Persistent Store


import SwiftUI
import CoreData

struct ContentView: View {

@FetchRequest(sortDescriptors: [], predicate: nil, animation: .default) private var
listOfBooks: FetchedResults<Books>

var body: some View {

NavigationView {
List {
ForEach(listOfBooks) { book in
RowBook(book: book)
}
 }
.navigationBarTitle("Books")
.toolbar {
ToolbarItem(placement: .navigationBarTrailing) {
NavigationLink(destination: InsertBookView(), label: {
Image(systemName: "plus")
})
}
}
}
}
}
struct RowBook: View {
let book: Books

var imageCover: UIImage {

if let data = book.thumbnail, let image = UIImage(data: data) {
return image
} else {
return UIImage(named: "nopicture")!
}
}
var body: some View {
HStack(alignment: .top) {
Image(uiImage: imageCover)
.resizable()
.scaledToFit()
.frame(width: 80, height: 100)
.cornerRadius(10)
VStack(alignment: .leading, spacing: 2) {
Text(book.title ?? "Undefined")
.bold()
Text(book.author?.name ?? "Undefined")
.foregroundColor(book.author != nil ? .black : .gray)
Text(String(book.year))
.font(.caption)
Spacer()
}.padding(.top, 5)
Spacer()
}
}
}

To have access to Core Data from the view, we must import the framework
with the import keyword, as we did in this example. After this, we are ready
to fetch and store objects from the Persistent Store, and the first step is to
get the list of objects already inserted by the user. To this end, the
ContentView view defines a @FetchRequest property. This property creates a
fetch request to fetch objects of the Books entity in no specific order (the
sortDescriptor argument was declared with an empty array). Notice that the
property wrapper produces a structure of type FetchedResults and this
structure is generic. The data type specified for the structure is the one
that determines the type of objects to fetch. In our case it is Books objects
(FetchedResults<Books>).
Once the @FetchRequest property is ready, we can provide it to a List view to
show the objects on the screen. The ForEach view in our example goes
through this list and creates a RowBook view for each object to show the
values. Reading these values is easy. The attributes of an entity are turned
into properties of the NSManagedObject (the Books object in our example), and
they are available as any other property. The only thing we must consider
is that some of these properties return optional values. For instance, the
title property returns an optional string. To show this value, we use a nil-
coalescing operator (??). If there is a value in the property, we display it,
otherwise, we show the string "Undefined". The author property is a little
bit different. This property doesn't represent an attribute but a
relationship. The value we need is actually the name of the author, which is
returned by the name property inside the Authors object (book.author?.name). If
this property contains a value, we show it, otherwise, we show the text
"Undefined" again. The image is more complicated. The value returned by
the thumbnail property is a Data structure that we must convert to an UIImage
object before creating the Image view with it. There are different ways to
perform this process. In this example, we have decided to do it with a
computed property called imageCover. When the view reads this property,
we check if there is a value in the object's thumbnail property, create an
UIImage object with it, and return it. On the other hand, if the thumbnail
property is empty, or there was a problem converting the data into an
image, we return a UIImage object with the image nopicture from the Asset
Catalog.
Now, it is time to let the user create new Books objects. The ContentView view
includes a NavigationLink in the navigation bar that opens the InsertBookView
view for this purpose.

Listing 4: Adding new objects to the Persistent Store


import SwiftUI
import CoreData

struct InsertBookView: View {

@Environment(\.managedObjectContext) var dbContext
@Environment(\.dismiss) var dismiss
@State private var inputTitle: String = ""
@State private var inputYear: String = ""

var body: some View {

VStack(spacing: 12) {
HStack {
Text("Title:")
TextField("Insert Title", text: $inputTitle)
.textFieldStyle(.roundedBorder)
}
HStack {
Text("Year:")
TextField("Insert Year", text: $inputYear)
.textFieldStyle(.roundedBorder)
}
HStack {
Text("Author:")
Text("Undefined")
.foregroundColor(.gray)
}.frame(minWidth: 0, maxWidth: .infinity, alignment: .leading)
Spacer()
}.padding()
.navigationBarTitle("Add Book")
.toolbar {
ToolbarItem(placement: .navigationBarTrailing) {
Button("Save") {
let newTitle = inputTitle.trimmingCharacters(in: .whitespaces)
let newYear = Int32(inputYear)
if !newTitle.isEmpty && newYear != nil {
 Task(priority: .high) {
await storeBook(title: newTitle, year: newYear!)
}
}
}
}
}
}
func storeBook(title: String, year: Int32) async {
await dbContext.perform {
let newBook = Books(context: dbContext)
newBook.title = title
newBook.year = year
newBook.author = nil
newBook.cover = UIImage(named: "bookcover")?.pngData()
newBook.thumbnail = UIImage(named: "bookthumbnail")?.pngData()
do {
try dbContext.save()
dismiss()
} catch {
print("Error saving record")
}
}
}
}


All the interaction between our code and the Persistent Store is done
through the context. When we want to access the objects already stored,
add new ones, remove them, or modify any of their values, we have to do
it in the context and then move those changes from the context to the
Persistent Store. The @FetchRequest property wrapper automatically gets a
reference of the context from the environment (this is the reference we
injected into the context in the App structure of Listing 2), but when
working directly with Core Data, we must get the reference from the
environment with the @Environment property wrapper and the
managedObjectContext key. In this example, we called this property dbContext.
The view includes two TextField views to let the user insert the title of the
book and the year of publication, and a button in the navigation bar to save
the values. When the button is pressed, the code checks if the values are
valid and then calls an asynchronous method to create and store a new
object in the Persistent Store. We moved the code to an asynchronous
method so that we can implement the perform() method provided by the
context and thus ensure that the process is performed in a safe thread,
which is recommended every time we are adding, removing, or modifying
an object.
The process to add a new book begins with the creation of the new object
with the Books() initializer. This not only creates a new object of type Books
but it also adds it to the context specified by the argument (dbContext). The
next step is to assign the values to the object's properties. We assign the
value of the first input field to the title property, the value of the second
input field to the year property, the images bookcover and bookthumbnail
to the cover and thumbnail properties, respectively (we assign standard
images to every book for now), and the nil value to the author property (we
still don't have an Authors object to associate with this book).
The Books() initializer inserts the new object into the context, but this
change is not permanent. If we close the app after the values are assigned
to the properties, the object is lost. To persist the changes, we must save
the context with the save() method. This should be done every time we
finish a process that modifies the context. The method takes the
information in the context and updates the Persistent Store with it, so
everything is stored permanently on file.

Do It Yourself: This example assumes that you have followed the

previous steps to create a project, prepare a model with the Books
and Authors entities, define the Core Data stack, and inject the
context into the environment in the App structure (Listing 2).
Download the bookcover.png, bookthumbnail.png, and
nopicture.png images from our website and add them to the Asset
Catalog. Update the ContentView.swift file with the code in Listing 3.
Create a Swift file called InsertBookView.swift for the view in Listing
 4. Run the application on the iPhone simulator and press the +
button. Insert a book and press Save.

IMPORTANT: Core Data classes are created when the app is built.
Sometimes, Xcode can't recognize the classes created from the
entities and returns an error. One way to remove the errors is to
build the application by pressing the Command + B keys. If this
doesn't solve the issues, you can try running the application on the
simulator.

There are no view models in Core Data. The objects stored in the Persistent
Store (Books and Authors in our example), represent the application's data
model. They store the values and return them as they are. But the views
need the values to be formatted or casted before showing them on the
screen. For instance, in the RowBook view in Listing 3, we had to process the
value of the thumbnail property with a computed property to turn it into an
UIImage view. We also had to use the nil-coalescing operator to show a
string if there was no value in the title and author properties. And we even
had to cast the value of the year property to a string with the String()
initializer. All this work should not be done by the view, it should be done
by a view model. To create a view model for the Core Data objects, we can
extend the classes defined by the system. For example, we can create an
extension of the Books class to provide computed properties that always
return String values for the views to display, and process the images in the
thumbnail and cover properties, so we don't have to do it inside the view.

Listing 5: Defining a view model for the Core Data objects


import SwiftUI

extension Books {
var showTitle: String {
return title ?? "Undefined"
}
 var showYear: String {
return String(year)
}
var showAuthor: String {
return author?.name ?? "Undefined"
}
var showCover: UIImage {
if let data = cover, let image = UIImage(data: data) {
return image
} else {
return UIImage(named: "nopicture")!
}
}
var showThumbnail: UIImage {
if let data = thumbnail, let image = UIImage(data: data) {
return image
} else {
return UIImage(named: "nopicture")!
}
}
}
extension Authors {
var showName: String {
return name ?? "Undefined"
}
}


Extensions have access to the properties defined by the data type. This
means that we can access those properties, process their values, and
return something else. The example in Listing 5 defines an extension for
the Books class and an extension for the Authors class. The extensions include
computed properties that format the values in the original properties and
return a value for the views to display. In the extension of the Books class,
we also include a property that returns a string with the name of the
author, so the views don't have to read this value from the Authors object
anymore, they can do it directly from the Books object, as shown next.

Listing 6: Using the values from the view model


struct RowBook: View {
 let book: Books

var body: some View {

HStack(alignment: .top) {
Image(uiImage: book.showThumbnail)
.resizable()
.scaledToFit()
.frame(width: 80, height: 100)
.cornerRadius(10)
VStack(alignment: .leading, spacing: 2) {
Text(book.showTitle)
.bold()
Text(book.showAuthor)
Text(book.showYear)
.font(.caption)
Spacer()
}.padding(.top, 5)
Spacer()
}
}
}


The view has been greatly simplified. There is no formatting or processing.

All it has to do now is to read the values from the view model and show
them to the user.

Do It Yourself: Create a Swift file called Extensions.swift for the

code in Listing 5. Update the RowBook view in the ContentView.swift
file with the code in Listing 6. Run the application on the iPhone
simulator. You should see the same interface shown in Figure 15.

The Authors objects are generated and stored the same way as the Books
objects. This demands our application to provide new views where the
user can select and add more objects. For our example, we have decided to
expand our interface with a view that lists the authors already inserted by
the user and another view to insert new ones.

Figure 16: Interface to list and add authors



The view on the left is the InsertBookView view introduced before to insert
new books. This view now shows three input options, an input field to
insert the title of the book, another to insert the year, and the Select
Author button to select the author (Figure 16, left). This button is a
NavigationLink view that opens a view to list all the authors available (Figure
16, center). In turn, this view includes a + button in the navigation bar to
open a view that includes an input field to insert the name of an author
(Figure 16, right). The first step we need to take to create this interface is to
add the Select Author button to the InsertBookView view, as shown below.

Listing 7: Selecting the author


struct InsertBookView: View {
@Environment(\.managedObjectContext) var dbContext
@Environment(\.dismiss) var dismiss

@State private var selectedAuthor: Authors? = nil

@State private var inputTitle: String = ""
@State private var inputYear: String = ""

var body: some View {

VStack(spacing: 12) {
HStack {
Text("Title:")
TextField("Insert Title", text: $inputTitle)
.textFieldStyle(.roundedBorder)
}
HStack {
 Text("Year:")
TextField("Insert Year", text: $inputYear)
.textFieldStyle(.roundedBorder)
}
HStack(alignment: .top) {
Text("Author:")
VStack(alignment: .leading, spacing: 8) {
Text(selectedAuthor?.name ?? "Undefined")
.foregroundColor(selectedAuthor != nil ? Color.black : Color.gray)
NavigationLink(destination: AuthorsView(selected: $selectedAuthor), label: {
Text("Select Author")
})
}
}.frame(minWidth: 0, maxWidth: .infinity, alignment: .leading)
Spacer()
}.padding()
.navigationBarTitle("Add Book")
.toolbar {
ToolbarItem(placement: .navigationBarTrailing) {
Button("Save") {
let newTitle = inputTitle.trimmingCharacters(in: .whitespaces)
let newYear = Int32(inputYear)
if !newTitle.isEmpty && newYear != nil {
Task(priority: .high) {
await storeBook(title: newTitle, year: newYear!)
}
}
}
}
}
}
func storeBook(title: String, year: Int32) async {
await dbContext.perform {
let newBook = Books(context: dbContext)
newBook.title = title
newBook.year = year
newBook.author = selectedAuthor
newBook.cover = UIImage(named: "bookcover")?.pngData()
newBook.thumbnail = UIImage(named: "bookthumbnail")?.pngData()
do {
try dbContext.save()
dismiss()
} catch {
print("Error saving record")
}
}
}
}


Every time an author is selected or created, we must get its Authors object
and send it back to the InsertBookView view to assign it to the book. To store
this value, the view includes a @State property called selectedAuthor. If the
property contains an Authors object, we show the value of its name property
to the user, otherwise, we show the text "Undefined". (Notice that in this
case we don't use the properties from the view models created before
because the selectedAuthor property is an optional, so we have to check its
value anyway.) Below the name, we include a NavigationLink button to open
a view called AuthorsView to let the user select an author.

Listing 8: Listing all the authors available


import SwiftUI

struct AuthorsView: View {

@FetchRequest(sortDescriptors: [], predicate: nil, animation: .default) private var
listOfAuthors: FetchedResults<Authors>
@Environment(\.dismiss) var dismiss
@Binding var selected: Authors?

var body: some View {

List {
ForEach(listOfAuthors) { author in
HStack {
Text(author.showName)
}
.frame(minWidth: 0, maxWidth: .infinity, minHeight: 0, maxHeight: .infinity, alignment:
.leading)
.contentShape(Rectangle())
.onTapGesture {
selected = author
dismiss()
}
}
}
.navigationBarTitle("Authors")
.toolbar {
ToolbarItem(placement: .navigationBarTrailing) {
 NavigationLink(destination: InsertAuthorView(), label: {
Image(systemName: "plus")
})
}
}
}
}
struct AuthorsView_Previews: PreviewProvider {
static var previews: some View {
AuthorsView(selected: .constant(nil))
}
}


This view lists the authors already inserted by the user. The @FetchRequest
property is called listOfAuthors and it is defined to work with Authors objects,
but other than that, the rest of the code is the same we used to list books.
The only significant difference is that the rows now include the
onTapGesture() modifier to let the user select an author. When the user taps
on the name of an author, we assign the Authors object to a @Binding
property called selected and close the view. Because the selected property is
connected to the selectedAuthor property defined in the InsertBookView view,
the name of the author selected by the user will be shown on the screen.
If there are no authors in the Persistent Store yet, or the author the user is
looking for is not on the list, the user can press a button in the navigation
bar that opens the InsertAuthorView view to insert a new author.

Listing 9: Inserting new authors


import SwiftUI
import CoreData

struct InsertAuthorView: View {

@Environment(\.managedObjectContext) var dbContext
@Environment(\.dismiss) var dismiss
@State private var inputName: String = ""

var body: some View {

VStack {
HStack {
 Text("Name:")
TextField("Insert Name", text: $inputName)
.textFieldStyle(.roundedBorder)
}
HStack {
Spacer()
Button("Save") {
let newName = inputName.trimmingCharacters(in: .whitespaces)
if !newName.isEmpty {
Task(priority: .high) {
await storeAuthor(name: newName)
}
}
}
}
Spacer()
}.padding()
.navigationBarTitle("Add Author")
}
func storeAuthor(name: String) async {
await dbContext.perform {
let newAuthor = Authors(context: dbContext)
newAuthor.name = name
do {
try dbContext.save()
dismiss()
} catch {
print("Error saving record")
}
}
}
}


This is a simple view. It includes one TextField view to insert the name of the
author and a button to save it. When the button is pressed, we follow the
same procedure as before. The Authors object is initialized and stored in the
context, the name inserted by the user is assigned to the object's name
property, and the context is saved to make the changes permanent.
With these additions, our basic app is complete. Now authors can be
assigned to new books. When we pressed the Select Author button, the
app opens a view with all the authors available (Figure 16, center). If there
are no authors yet or the author we want is not on the list, we can press
the + button to insert a new one (Figure 16, right). Every time we select an
author from the list, the app goes back to the view with the book’s
information and shows the name of the selected author on the screen
(Figure 16, left). The Authors object that represents the author is assigned to
the book’s author property and therefore the name of the author is now
shown on the list of books.

Do It Yourself: Update the InsertBookView view with the code in

Listing 7. Create a SwiftUI View file called AuthorsView.swift for the
view in Listing 8 and another called InsertAuthorView.swift for the
code in Listing 9. Run the application on the iPhone simulator. Press
the + button to insert a new book. Press the Select Author button.
Press the + button to add an author. After adding the author, click on
it to select it. You should see the name of the author on the screen.
Previews

So far, we have run the application on the iPhone simulator or a device.

This is because Apple recommends to create the Persistent Store for
previews in memory, so all the objects added for testing are removed as
soon as the preview is over. Therefore, if we want to develop the interface
on the canvas, we need to define two Persistent Stores, one for the
application and another for the previews. In the following example, we
achieve this with a type property. The model works as before, but if we
read this property, we get an ApplicationData class that creates a Persistent
Store in memory. To know what kind of Persistent Store to create, we add a
Boolean argument to the ApplicationData initializer called preview. If the value
of this argument is false, the Persistent Store is created as always,
otherwise, it is created in memory.

Listing 10: Creating a Persistent Store for previews


import SwiftUI
import CoreData

class ApplicationData: ObservableObject {

let container: NSPersistentContainer

static var preview: ApplicationData = {

let model = ApplicationData(preview: true)
return model
}()

init(preview: Bool = false) {

container = NSPersistentContainer(name: "books")
if preview {
container.persistentStoreDescriptions.first!.url = URL(fileURLWithPath: "/dev/null")
}
container.loadPersistentStores(completionHandler: { (storeDescription, error) in
if let error = error as NSError? {
fatalError("Unresolved error \(error), \(error.userInfo)")
}
})
}
}


This model creates the Persistent Store as before, but it also includes a
type property called preview to create a Persistent Store in memory for the
previews. The closure assigned to this property initializes the ApplicationData
object with the value true. The initializer checks this value and assigns a URL
structure with a null URL to the url property of the NSPersistentStoreDescription
object returned by the persistentStoreDescriptions property of the
NSPersistentContainer class. The NSPersistentStoreDescription object is used by Core
Data to create and load the Persistent Store, so when we assign a URL with
the "/dev/null" path to this object, the Persistent Store is created in
memory.
Now that we have a type property to create the Persistent Store for
previews, we can use it in the PreviewProvider structures. All we need to do is
to inject it into the environment as we did in the App structure, but this
time we access the viewContext property of the NSPersistentContainer object
returned by the ApplicationData object created by the preview property, as
shown next.

Listing 11: Accessing the Core Data context from the preview

struct ContentView_Previews: PreviewProvider {
static var previews: some View {
ContentView()
.environment(\.managedObjectContext, ApplicationData.preview.container.viewContext)
}
}


Do It Yourself: Update the ApplicationData class with the code in

Listing 10. Update PreviewProvider structure in the ContentView.swift
file with the code in Listing 11. Use the same environment() modifier to
inject the context in the preview of every view we have created so
far. Select the ContentView view and activate the live preview. You
should be able to add new books and see the list on the canvas.
Sort Descriptors

Objects returned from a request are usually in the order they were
created, but this is not guaranteed. Foundation defines two data types to
sort objects, the NSSortDescriptor class to define NSFetchRequest objects, and
the SortDescriptor structure, designed to work with the @FetchRequest property
wrapper. The most useful in SwiftUI applications is the SortDescriptor
structure. From these structure, we can specify an order according to the
values of a property. The structure includes the following initializers.

SortDescriptor(KeyPath, order: SortOrder)—This initializer

creates a SortDescriptor structure that sorts the objects according to the
value of the property specified by the first argument. This argument is
the key path of the property we want to use to sort the objects, and
the order argument determines if the objects will be sorted in
ascending or descending order. It is an enumeration with the values
forward and reverse.

SortDescriptor(KeyPath, comparator: StandardComparator,

order: SortOrder)—This initializer creates a SortDescriptor structure
that sorts the objects according to the value of the property specified
by the first argument. This argument is the key path of the property
we want to use to sort the objects. The comparator argument
determines how to compare the values. It is a structure with the type
properties lexical, localized, and localizedStandard (default). And the order
argument determines if the objects will be sorted in ascending or
descending order. It is an enumeration with the values forward and
reverse.

The @FetchRequest property wrapper includes the sortDescriptors argument to

define the order of the objects in the request. All we have to do to define
the order is to create at least one SortDescriptor structure and assign it to this
argument. The following example sorts the Books objects by title in
ascending order.

Listing 12: Sorting the books by title


struct ContentView: View {
@FetchRequest(sortDescriptors: [SortDescriptor(\Books.title, order: .forward)], predicate: nil,
animation: .default) var listOfBooks: FetchedResults<Books>

var body: some View {

NavigationView {
List {
ForEach(listOfBooks) { book in
RowBook(book: book)
}
}
.navigationBarTitle("Books")
.toolbar {
ToolbarItem(placement: .navigationBarTrailing) {
NavigationLink(destination: InsertBookView(), label: {
Image(systemName: "plus")
})
}
}
}
}
}


The sortDescriptors argument takes an array of SortDescriptor structures, so

we can specify multiple conditions to sort the list. The final order is
established according to the location of the SortDescriptor structures in the
array. For example, we can sort the books by author first and then by year.

Listing 13: Sorting books by author and year


@FetchRequest(sortDescriptors: [SortDescriptor(\Books.author?.name, order: .forward),
SortDescriptor(\.year, order: .forward)], predicate: nil, animation: .default) var listOfBooks:
FetchedResults<Books>

The values are sorted according to a comparator. The String structure
defines the StandardComparator structure to create them. The structure
includes type properties to create instances for common comparators. The
currently available are lexical, localizedStandard (default), and localized. The
structure returned by the lexical property sorts the values alphabetically. For
instance, if we have an array with the values 1, 2, and 10, the order will be
1, 10, 2 (the value 10 begins with 1, which comes before 2). This also
applies to letters—uppercase letters come before lowercase letters, so
words beginning with a lowercase letter will be moved to the end of the
list. The structure returned by the localizedStandard property sorts the values
numerically (1, 2, 10), which also applies to letters—words beginning with
uppercase letters will be sorted along with lowercase letters. And finally,
the structure returned by the localized property sorts the values
alphabetically as the lexical structure, but uppercase and lowercase letters
are sorted together, as the localizedStandard structure.
By default, the values are sorted with the localizedStandard structure. The
following example shows how to apply a lexical structure instead.

Listing 14: Sorting with a lexical comparator


@FetchRequest(sortDescriptors: [SortDescriptor(\Books.title, comparator: .lexical, order: .forward)],
predicate: nil, animation: .default) var listOfBooks: FetchedResults<Books>


Do It Yourself: Update the ContentView view with the code in Listing

12 and try any of the @FetchRequest properties defined above to see
how sort descriptors work. Change the value of the order argument
to see how the order changes in each case (forward and reverse).

If later we need to change the order of the objects, we can assign new
SortDescriptor structures to the FetchedResults structure created by the
@FetchRequest property wrapper. For this purpose, the structure includes the
 property. In the following example, we add a button to the
sortDescriptors
navigation bar that opens a menu with three options to sort the values.

Listing 15: Modifying the sorting criteria


struct ContentView: View {
@FetchRequest(sortDescriptors: [SortDescriptor(\Books.title, order: .forward)], predicate: nil,
animation: .default) var listOfBooks: FetchedResults<Books>

var body: some View {

NavigationView {
List {
ForEach(listOfBooks) { book in
RowBook(book: book)
}
}
.navigationBarTitle("Books")
.toolbar {
ToolbarItem(placement: .navigationBarLeading) {
Menu("Sort") {
Button("Sort by Title", action: {
let sort = SortDescriptor(\Books.title, order: .forward)
listOfBooks.sortDescriptors = [sort]
})
Button("Sort by Author", action: {
let sort = SortDescriptor(\Books.author?.name, order: .forward)
listOfBooks.sortDescriptors = [sort]
})
Button("Sort by Year", action: {
let sort = SortDescriptor(\Books.year, order: .forward)
listOfBooks.sortDescriptors = [sort]
})
}
}
ToolbarItem(placement: .navigationBarTrailing) {
NavigationLink(destination: InsertBookView(), label: {
Image(systemName: "plus")
})
}
}
}
}
}

The button is created by a Menu view, and the menu includes three Button
views for each option. When a button is pressed, we create a SortDescriptor
structure with the configuration we want and then assign it to the
sortDescriptors property of the FetchedResults structure. The first button sorts
the books by title (the configuration by default), the second button sorts
the books by author, and the third button by year.

Figure 17: Different sorting criteria


Predicates

The requests performed in previous examples are getting all the objects
associated to a particular entity and the values of all their properties. The
Core Data framework defines a class called NSPredicate to fetch only the
objects that pass a filter. For example, we could get only the books that
were published in the year 1983 or the authors which names start with
"Stephen". The class defines the following initializer to create a predicate
with all the conditions we need.

NSPredicate(format: String, argumentArray: [Any]?)—This

initializer creates an NSPredicate object with the conditions set by the
format argument. The argumentArray argument is an optional array
of values that replace placeholders in the string assigned to the
format argument. The argumentArray argument may be ignored or
replaced by a list of values separated by commas.

To filter values in a request, we must create the NSPredicate object and

assign it to the predicate argument of the @FetchRequest property wrapper.
The following example assigns an NSPredicate object to the request in the
ContentView view to search for the books that has the value 1983 assigned to
their year property.

Listing 16: Filtering books by year


struct ContentView: View {
@FetchRequest(sortDescriptors: [], predicate: NSPredicate(format: "year = 1983")) var
listOfBooks: FetchedResults<Books>

var body: some View {

NavigationView {
List {
ForEach(listOfBooks) { book in
RowBook(book: book)
}
 }
.navigationBarTitle("Books")
.toolbar {
ToolbarItem(placement: .navigationBarTrailing) {
NavigationLink(destination: InsertBookView(), label: {
Image(systemName: "plus")
})
}
}
}
}
}


If we are trying to search for a value in a relationship, we can concatenate

the properties with dot notation, as we did before to read the name
property in the Authors object. For example, the following request searches
for books written by Stephen King.

Listing 17: Filtering books by author


@FetchRequest(sortDescriptors: [], predicate: NSPredicate(format: "author.name = 'Stephen
King'")) var listOfBooks: FetchedResults<Books>


Do It Yourself: Update the ContentView view with the code in Listing

15 and try the @FetchRequest property defined in Listing 17 to see how
predicates work. Add books with different authors and years to test
the filters.

Of course, these are predefined conditions, but we can allow the user to
insert a value and use new predicates to perform a search. For this, we
need to replace the current predicate with a new one configured with the
new values we want to filter. To assign a predicate to the request, the
FetchedResults structure includes the nspredicate property. The following
example shows how to implement this property and define new predicates
to search for books by year.
Listing 18: Creating new predicates to search values in the Persistent Store

struct ContentView: View {
@FetchRequest(sortDescriptors: [], predicate: nil, animation: .default) var listOfBooks:
FetchedResults<Books>
@State private var search: String = ""

var body: some View {

NavigationView {
List {
ForEach(listOfBooks) { book in
RowBook(book: book)
}
}
.navigationBarTitle("Books")
.toolbar {
ToolbarItem(placement: .navigationBarTrailing) {
NavigationLink(destination: InsertBookView(), label: {
Image(systemName: "plus")
})
}
}
.searchable(text: $search, prompt: Text("Insert year"))
.onChange(of: search) { value in
if value.count == 4 {
if let year = Int32(value) {
listOfBooks.nsPredicate = NSPredicate(format: "year = %@", NSNumber(value:
year))
}
} else {
listOfBooks.nsPredicate = nil
}
}
}
}
}


To add a search bar, we implement the searchable() modifier, and to perform

the search, we implement the onChange() modifier. This modifier checks the
value of a state property called search. If the value of this property changes,
the modifier executes a closure where we replace the current predicate by
a new one. When the value inserted by the user is a number, we filter
books by year, otherwise, we assign a nil value to the nsPredicate property
and show all the books available. (Notice that the NSPredicate initializer only
takes Property List values, so we must cast the Int32 value into an
NSNumber.)

Figure 18: Searching

Do It Yourself: Update the ContentView view with the code in Listing

18. Run the application on the iPhone simulator. Insert some books
and then insert a year in the input field. Only the books published
that year should be shown on the list, as illustrated in Figure 18.

IMPORTANT: The placeholder %@ is replaced by the value

specified in the arguments between quotes. If you need to add the
value to the predicate without the quotes, you must use the
placeholder %K instead (called Dynamic Key). This is also useful to
insert key paths into the format string.

Predicates use comparison and logical operators like those offered by Swift.
For example, we can compare values with the operators =, !=, >, <, >= and
<=, and also concatenate conditions with the characters && (or the word
AND), || (or the word OR) and ! (or the word NOT). Predicates also include
keywords for a more meticulous search. The following are the most
frequently used.

BEGINSWITH—The condition determined by this keyword is true

when the expression on the left begins with the expression on the
right.
CONTAINS—The condition determined by this keyword is true when
the expression on the left contains the expression on the right.
ENDSWITH—The condition determined by this keyword is true when
the expression on the left ends with the expression on the right.
LIKE—The condition determined by this keyword is true when the
expression on the left is equal to the expression on the right.
IN—The condition determined by this keyword is true when the
expression on the left is equal to any of the values included in the
expression on the right. The values are provided as an array between
parentheses.

These keywords may be accompanied by the characters c or d between

square brackets to specify a case and diacritic insensitive search. For
example, we may search for books which authors' names begin with the
characters inserted by the user, but without considering uppercase or
lowercase letters ([c]).

Listing 19: Filtering values with predicate keywords


.searchable(text: $search, prompt: Text("Insert Name"))
.onChange(of: search) { value in
if !value.isEmpty {
listOfBooks.nsPredicate = NSPredicate(format: "author.name BEGINSWITH[c] %@",
value)
} else {
listOfBooks.nsPredicate = nil
}
}


Another useful keyword is CONTAINS, used to search for a value within a

string. For instance, we can check if the characters inserted by the user can
be found in the title of a book.

Listing 20: Searching for characters in a string


.searchable(text: $search, prompt: Text("Insert Title"))
.onChange(of: search) { value in
if !value.isEmpty {
listOfBooks.nsPredicate = NSPredicate(format: "title CONTAINS[dc] %@", value)
} else {
listOfBooks.nsPredicate = nil
}
}


The predicate in Listing 20 implements the dc modifiers to make the search

case insensitive and ignore diacritic characters (the small marks used in
some languages to change the pronunciation of a letter). For example, we
can search for the term "éry" (with a stress in the letter e) and the request
will return books which title contains the characters "ery" (e.g., "Misery").

Do It Yourself: Update the searchable() and onChange() modifiers in

your project with the example you want to try. Run the application
on the iPhone simulator and perform a search to see how predicate
keywords work.
Delete Objects

Deleting objects in the Persistent Store is no different than any other

values in the model. We must apply the onDelete() modifier to the ForEach
view and then remove from the model the objects in the indexes received
by the closure.

Listing 21: Deleting objects


struct ContentView: View {
@Environment(\.managedObjectContext) var dbContext
@FetchRequest(sortDescriptors: [], predicate: nil, animation: .default) var listOfBooks:
FetchedResults<Books>
@State private var search: String = ""

var body: some View {

NavigationView {
List {
ForEach(listOfBooks) { book in
RowBook(book: book)
}
.onDelete(perform: { indexes in
Task(priority: .high) {
await deleteBook(indexes: indexes)
}
})
}
.navigationBarTitle("Books")
.toolbar {
ToolbarItem(placement: .navigationBarTrailing) {
NavigationLink(destination: InsertBookView(), label: {
Image(systemName: "plus")
})
}
}
.searchable(text: $search, prompt: Text("Insert Title"))
.onChange(of: search) { value in
if !value.isEmpty {
listOfBooks.nsPredicate = NSPredicate(format: "title CONTAINS[dc] %@", value)
} else {
listOfBooks.nsPredicate = nil
}
 }
}
}
func deleteBook(indexes: IndexSet) async {
await dbContext.perform {
for index in indexes {
dbContext.delete(listOfBooks[index])
}
do {
try dbContext.save()
} catch {
print("Error deleting objects")
}
}
}
}


In our example, the objects retrieved from the Persistent Store are stored
in the listOfBooks property. The indexes received by the closure in the
onDelete() modifier are the indexes of the objects the user wants to delete
from this collection. Therefore, to remove the objects selected by the user,
we iterate through the indexes of all the objects to be removed with a for in
loop, get the object from the listOfBooks collection using each index, and
remove it from the context with the delete() method. Once the objects are
removed from the context, the listOfBooks property is automatically updated
and the changes are reflected on the screen. The last step is to save the
context with the save() method to modify the Persistent Store with the
changes in the context, as we did before. (Notice that in this example we
had to get a direct reference to the context from the environment with the
@Environment property wrapper.)

Do It Yourself: Update the ContentView view with the code in Listing

21. Run the application on the iPhone simulator. Drag a book to the
left. You should see the Delete button. Press the button to remove
the book. In this example, we have implemented the code to remove
the object inside the view, but you can move it to a method or an
observable object, and also implement the EditButton() view or any
other tools to provide more options to the user.
Modify Objects

Modifying an object in the Persistent Store is easy. When the user selects a
row, we pass the object to a view that provides input fields to modify its
values, assign the values inserted by the user to the object, and save the
context to make the changes permanent. First, we must add the navigation
link in the ContentView view to allow the user to select a book.

Listing 22: Selecting the book to modify


ForEach(listOfBooks) { book in
NavigationLink(destination: ModifyBookView(book: book), label: {
RowBook(book: book)
.id(UUID())
})
}


In this example, we apply the id() modifier to the RowBook view. We have
introduced this modifier before. It assigns a unique identifier to each view.
In this case, we use a new UUID value, which means that every time the
view is recreated, the identifier will be different. This makes the system
believe that this is a different view and therefore it updates the content,
showing on the screen the new values inserted by the user.
The NavigationLink view opens a view called ModifyBookView to allow the user
to modify the values of the selected book. The following is our
implementation of this view.

Listing 23: Modifying the values of a managed object


import SwiftUI
import CoreData

struct ModifyBookView: View {

@Environment(\.managedObjectContext) var dbContext
@Environment(\.dismiss) var dismiss
@State private var selectedAuthor: Authors? = nil
@State private var inputTitle: String = ""
@State private var inputYear: String = ""
@State private var valuesLoaded: Bool = false

let book: Books?

var body: some View {

VStack(spacing: 12) {
HStack {
Text("Title:")
TextField("Insert Title", text: $inputTitle)
.textFieldStyle(.roundedBorder)
}
HStack {
Text("Year:")
TextField("Insert Year", text: $inputYear)
.textFieldStyle(.roundedBorder)
}
HStack(alignment: .top) {
Text("Author:")
VStack(alignment: .leading, spacing: 8) {
Text(selectedAuthor?.name ?? "Undefined")
.foregroundColor(selectedAuthor != nil ? Color.black : Color.gray)
NavigationLink(destination: AuthorsView(selected: $selectedAuthor), label: {
Text("Select Author")
})
}
}.frame(minWidth: 0, maxWidth: .infinity, alignment: .leading)
Spacer()
}.padding()
.navigationBarTitle("Modify Book")
.toolbar {
ToolbarItem(placement: .navigationBarTrailing) {
Button("Save") {
let newTitle = inputTitle.trimmingCharacters(in: .whitespaces)
let newYear = Int32(inputYear)
if !newTitle.isEmpty && newYear != nil {
Task(priority: .high) {
await saveBook(title: newTitle, year: newYear!)
}
}
}
}
}
.onAppear {
if !valuesLoaded {
 selectedAuthor = book?.author
inputTitle = book?.title ?? ""
inputYear = book?.showYear ?? ""
valuesLoaded = true
}
}
}
func saveBook(title: String, year: Int32) async {
await dbContext.perform {
book?.title = title
book?.year = year
book?.author = selectedAuthor
do {
try dbContext.save()
dismiss()
} catch {
print("Error saving record")
}
}
}
}
struct ModifyBookView_Previews: PreviewProvider {
static var previews: some View {
ModifyBookView(book: nil)
.environment(\.managedObjectContext, ApplicationData.preview.container.viewContext)
}
}


This view defines a constant called book to receive the Books object that
represents the book selected by the user. Because the user needs to see
the book to be able to modify it, we implement the onAppear() modifier to
initialize state properties with the values in the book property. Notice that
there is an additional property called valuesLoaded that we check to only load
the values when the property is false. This is to make sure that the values
are only loaded from the Persistent Store when the user selected a book
from the ContentView view, but not when a new author has been selected
from the AuthorsView view.
The view includes two TextField views for the user to be able to modify the
title and the year, and the same Select Author button included before to
open the AuthorsView view for the user to select an author. When the Save
button is pressed, the values inserted by the user are assigned to the
object's properties and the context is saved.
In this opportunity, we do not create a new Books object, we just modify the
properties of the Books object received by the view. The result is shown
below.

Figure 19: Interface to modify managed objects

Do It Yourself: Update the ForEach view in the ContentView view with

the code in Listing 22. Create a SwiftUI View file called
ModifyBookView.swift for the view in Listing 23. Run the application
on the iPhone simulator. Tap on a book to select it. Change the title
and press Save. You should see the changes on the list.
Custom Fetch Requests

So far, we have let the @FetchRequest property wrapper create the request
for us, but there are situations in which we must create our own requests
to process the values in the Persistent Store. Because a request has to be
associated to an entity, subclasses of the NSManagedObject class, like Books
and Authors, include the fetchRequest() method. This method returns an
NSFetchRequest object with a fetch request associated to the entity. To
perform this request, the context includes the fetch() method. The following
example creates a request when the view appears on the screen and when
an object is removed to count the number of books in the Persistent Store
and show the value at the top of the list.

Listing 24: Counting the books available


struct ContentView: View {
@Environment(\.managedObjectContext) var dbContext
@FetchRequest(sortDescriptors: [SortDescriptor(\Books.title, order: .forward)], predicate: nil,
animation: .default) var listOfBooks: FetchedResults<Books>
@State private var totalBooks: Int = 0

var body: some View {

NavigationView {
List {
HStack {
Text("Total Books")
Spacer()
Text("\(totalBooks)")
.bold()
}.foregroundColor(Color.green)

ForEach(listOfBooks) { book in
NavigationLink(destination: ModifyBookView(book: book), label: {
RowBook(book: book)
.id(UUID())
})
}
.onDelete(perform: { indexes in
for index in indexes {
dbContext.delete(listOfBooks[index])
 countBooks()
}
do {
try dbContext.save()
} catch {
print("Error deleting objects")
}
})
}
.navigationBarTitle("Books")
.toolbar {
ToolbarItem(placement: .navigationBarTrailing) {
NavigationLink(destination: InsertBookView(), label: {
Image(systemName: "plus")
})
}
}
.onAppear {
countBooks()
}
}
}
func countBooks() {
let request: NSFetchRequest<Books> = Books.fetchRequest()
if let list = try? self.dbContext.fetch(request) {
totalBooks = list.count
}
}
}


To store the number of books, this view defines a @State property called
totalBooks, and to show the value, it includes an HStack with two Text views on
top of the list. The request is created by a method called countBooks(). The
method is called when the view appears and when an object is removed. It
creates a request for the Books entity, and then executes the request in the
context with the fetch() method. If there are no errors, this method returns
an array with all the objects that match the request. In this case, we didn't
define any predicate, so the array contains all the Books objects in the
Persistent Store. Finally, we count the objects with the count property and
assign the value to the totalBooks property to show it to the user.
Figure 20: Total number of books in the Persistent Store

Do It Yourself: Update the ContentView view with the code in Listing

24. Run the application on the iPhone simulator. You should see the
total number of books on the screen. Slide a row to the left and
press the Delete button. You should see the number on the screen
go down by one unit.

Although this is a legitimate way to count objects in the Persistent Store, it

loads all the objects into memory and consumes too many resources. To
avoid any issues, the NSManagedObjectContext class includes the count()
method. This method returns an integer with the number of objects we
would get if we call the fetch() method with the same request. The method
does not fetch the objects, so we can call it without being afraid of
consuming too much memory. The following example improves the
countBooks() method using the count() method.

Listing 25: Counting objects with the count() method


func countBooks() {
let request: NSFetchRequest<Books> = Books.fetchRequest()
if let count = try? self.dbContext.count(for: request) {
totalBooks = count
}
}

 Do It Yourself: Update the countBooks() method from the previous
example with the code in Listing 25. The method counts the Books
objects as before, but without consuming resources.

If what we want is to get the number of objects associated to a To-Many

relationship, we just have to count the number of items returned by the
property that represents the relationship. For example, we can count the
number of books of every author and show it along with the name.

Listing 26: Counting the books of each author


struct AuthorsView: View {
@FetchRequest(sortDescriptors: [], predicate: nil, animation: .default) private var listOfAuthors:
FetchedResults<Authors>
@Environment(\.dismiss) var dismiss
@Binding var selected: Authors?

var body: some View {

List {
ForEach(listOfAuthors) { author in
HStack {
Text(author.showName)
Spacer()
Text(String(author.books?.count ?? 0))
}
.frame(minWidth: 0, maxWidth: .infinity, minHeight: 0, maxHeight: .infinity, alignment:
.leading)
.contentShape(Rectangle())
.onTapGesture {
selected = author
dismiss()
}
}
}
.navigationBarTitle("Authors")
.toolbar {
ToolbarItem(placement: .navigationBarTrailing) {
NavigationLink(destination: InsertAuthorView(), label: {
Image(systemName: "plus")
})
 }
}
}
}


This view includes a Text view that shows the total number of books next to
the author's name. To get this value, we count the number of items in the
books property or show the value 0 if the property is equal to nil (no books
have been assigned to the author).

Do It Yourself: Update the AuthorsView view with the code in Listing

26. Run the application on the iPhone simulator. Select a book and
press the Select Author button. You should see the list of authors
available and the number of books assigned to each author on the
right.

Another handy application of custom requests and the count() method is to

check for duplicates. Storing duplicate values is something every
application should avoid. For example, if we insert an author that already
exists, two Authors objects with the same name will be stored in the
Persistent Store. To avoid this situation, we can use a request with a
predicate that looks for authors of the same name before creating a new
object. The following example modifies the InsertAuthorView view to check if
the author inserted by the user already exists in the Persistent Store.

Listing 27: Checking for duplicates


import SwiftUI
import CoreData

struct InsertAuthorView: View {

@Environment(\.managedObjectContext) var dbContext
@Environment(\.dismiss) var dismiss
@State private var inputName: String = ""
 var body: some View {
VStack {
HStack {
Text("Name:")
TextField("Insert Name", text: $inputName)
.textFieldStyle(.roundedBorder)
}
HStack {
Spacer()
Button("Save") {
let newName = inputName.trimmingCharacters(in: .whitespaces)
if !newName.isEmpty {
Task(priority: .high) {
await storeAuthor(name: newName)
}
}
dismiss()
}
}
Spacer()
}.padding()
.navigationBarTitle("Add Author")
}
func storeAuthor(name: String) async {
await dbContext.perform {
let request: NSFetchRequest<Authors> = Authors.fetchRequest()
request.predicate = NSPredicate(format: "name = %@", name)
if let total = try? self.dbContext.count(for: request), total == 0 {
let newAuthor = Authors(context: dbContext)
newAuthor.name = name
do {
try dbContext.save()
} catch {
print("Error saving record")
}
}
}
}
}


The code in Listing 27 creates a request for the Authors entity and uses the
value inserted in the TextField view to create a predicate. The predicate
looks for objects with the name equal to the value of the newName constant.
Using this request, we call the count() method in the context to get the total
amount of objects that match the conditions in the predicate. If the value
returned is 0, we know that there are no authors in the Persistent Store
with that name and we can proceed.

Do It Yourself: Update the InsertAuthorView view in your project with

the code in Listing 27. Run the application on the iPhone simulator.
Select a book, press the Select Author button, and press the Add
Author button. You should only be able to insert a new author if
there is no other author in the Persistent Store with the same name.

Another situation where it is necessary to create a custom fetch request is

when the request needs to be generated dynamically. For instance, our
application may include an option to list books by author. In this case, we
don't know which author the user have selected until the view is opened,
so we must create the predicate dynamically.
For our example, we are going to replace the NavigationLink view in the
ContentView view, so instead of opening a view to modify the book, it opens
a new view to list the books that belong to the selected author.

Listing 28: Selecting an author


NavigationLink(destination: AuthorBooksView(selectedAuthor: book.author), label: {
RowBook(book: book)
})


Now, when a book is selected, the app opens the AuthorBooksView view with
a reference to the Authors object that represents the author selected by the
user. In this view, we must use the reference to create a dynamic fetch
request, as shown next.

Listing 29: Listing books by author


import SwiftUI

struct AuthorBooksView: View {

@FetchRequest(sortDescriptors: []) var listOfBooks: FetchedResults<Books>

var selectedAuthor: Authors?

init(selectedAuthor: Authors?) {
self.selectedAuthor = selectedAuthor
if selectedAuthor != nil {
_listOfBooks = FetchRequest(sortDescriptors: [SortDescriptor(\Books.title, order:
.forward)], predicate: NSPredicate(format: "author = %@", selectedAuthor!), animation:
.default)
}
}
var body: some View {
List {
ForEach(listOfBooks) { book in
Text(book.title ?? "Undefined")
}
}
.navigationBarTitle(selectedAuthor?.name ?? "Undefined")
}
}
struct AuthorBooksView_Previews: PreviewProvider {
static var previews: some View {
AuthorBooksView(selectedAuthor: nil)
}
}


The view receives a reference to the Authors object that represent the
author selected by the user, but we can't use this value in the @FetchRequest
property wrapper because the property is not available until the whole
structure is initialized. We must define a basic request and then assign a
new FetchRequest structure to the property wrapper from the initializer with
the predicate we need to only get the books published by the selected
author.
To access the underlining structure of the property wrapper, we must
precede the property name with an underscore (_listOfBooks). In this
example, we initialize the selectedAuthor property, and then assign a new
FetchRequest structure to the property wrapper with a predicate that filters
the books by the author assigned to that property. The result is shown
below.

Figure 21: Dynamic fetch request

Do It Yourself: Update the NavigationLink view in the ContentView view

with the code in Listing 28. Create a SwiftUI View file called
AuthorBooksView.swift for the view in Listing 29. Run the application
on the iPhone simulator. Select a book. You should see the titles of
all the books published by the selected author on the screen.
Sections

The information in the Persistent Store can be presented in sections. For

instance, we can create a section for every author, each of which contains
the author's books. To create these sections, SwiftUI includes the
@SectionedFetchRequest property wrapper. The main difference between this
property wrapper and the @FetchRequest property wrapper is the addition of
an argument to specify the section identifier.

SectionedFetchRequest(sectionIdentifier: KeyPath,
sortDescriptors: [SortDescriptor], predicate: NSPredicate?,
animation: Animation?)—This initializer creates a request with the
configuration determined by the arguments and produces a
SectionedFetchResults structure that manages and delivers the objects to
the view. The sectionIdentifier argument is a key path to the property
that the request is going to use to create the sections, the
sortDescriptors argument is an array of SortDescriptor structures that
determine the order of the objects, the predicate argument is an
NSPredicate object that filters the objects, and the animation argument
determines how the changes are going to be animated.
SectionedFetchRequest(fetchRequest: NSFetchRequest,
sectionIdentifier: KeyPath, animation: Animation?)—This
initializer creates a request with the NSFetchRequest object provided by
the fetchRequest argument and produces a SectionedFetchResults
structure that manages and delivers the objects to the view. The
sectionIdentifier argument is a key path to the property that the
request is going to use to create the sections, and the animation
argument determines how the changes are going to be animated.

The @SectionedFetchRequest property wrapper produces a value of type

SectionedFetchResults. This is a generic structure that works with two data
types: the data type of the property used to identify the sections and the
data type of the objects we are fetching.
The sections are created from the Section structure, defined by the
SectionedFetchResults structure. The Section structure includes the following
properties to describe the section.

id—This property returns the value used to identify the section.

startIndex—This property returns an integer value that represents
the index of the first item in the section.
endIndex—This property returns an integer value that represents
the index that comes after the index of the last item in the section.

The following example shows how to define the @SectionedFetchRequest

property wrapper to create a section for each author.

Listing 30: Fetching objects for sections


struct ContentView: View {
@SectionedFetchRequest(sectionIdentifier: \Books.author?.name, sortDescriptors:
[SortDescriptor(\Books.author?.name, order: .forward)], predicate: nil, animation: .default)
private var sectionBooks: SectionedFetchResults<String?, Books>

var body: some View {

NavigationView {
List {
ForEach(sectionBooks) { section in
Section(header: Text(section.id ?? "Undefined")) {
ForEach(section) { book in
NavigationLink(destination: ModifyBookView(book: book), label: {
RowBook(book: book)
.id(UUID())
})
}
}
}
}
.navigationBarTitle("Books")
.toolbar {
 ToolbarItem(placement: .navigationBarTrailing) {
NavigationLink(destination: InsertBookView(), label: {
Image(systemName: "plus")
})
}
}
}
}
}


In this example, we use the name property of the Authors entity to identify
the sections, so one section is created per author. Because a book may not
have an associated author, the author property of the Books object may
return nil, so the value used to identify the section is an optional String.
That's the reason why we declare the SectionedFetchResults data types as
<String?, Books>.
To present the data in sections, we need to define a ForEach loop for the
sections, and another for the objects in each section. The sections are
identified with a Section view, and the section's label is created with a Text
view from the value returned by the id property (in our example, this is the
author's name).

Figure 22: Sections

Do It Yourself: Update the ContentView view with the code in Listing

30. Run the application on the iPhone simulator. You should see the
 books organized by author into sections, as shown in Figure 22.

If we want to create the sections based on a different value, we must add it

to the Core Data model. For instance, to organize the books alphabetically
into sections, one section per letter, we need to add an attribute to the
Books entity to store the title's first letter.

Figure 23: Attribute to create alphabetical sections

In this example, we have added an attribute called firstLetter to the Books

entity. When a book is added or modified, we must get the first letter of
the book's title and store it in this property. The following are the changes
we need to introduce to the storeBook() method in the InsertBookView view.

Listing 31: Storing the first letter of the book's title


func storeBook(title: String, year: Int32) async {
await dbContext.perform {
let newBook = Books(context: dbContext)
newBook.title = title
newBook.year = year
newBook.author = selectedAuthor
newBook.cover = UIImage(named: "bookcover")?.pngData()
newBook.thumbnail = UIImage(named: "bookthumbnail")?.pngData()

var letter = String(title.first!).uppercased()

if Int(letter) != nil {
letter = "#"
}
newBook.firstLetter = letter
 do {
try dbContext.save()
dismiss()
} catch {
print("Error saving record")
}
}
}


When the user presses the Save button to save the book, we read the first
property to get the letter in the string and make sure that is an uppercase
letter. If the letter is actually a number, we replace it with a # character. (All
books which titles begin with a number will be listed in a unique section
identified with the # character.) Finally, the letter is stored in the firstLetter
property and the book is saved.
We need to perform the same process when a book is modified. The
following are the changes we need to introduce to the saveBook() method in
the ModifyBookView view.

Listing 32: Modifying the book's first letter


func saveBook(title: String, year: Int32) async {
await dbContext.perform {
book?.title = title
book?.year = year
book?.author = selectedAuthor

var letter = String(title.first!).uppercased()

if Int(letter) != nil {
letter = "#"
}
book?.firstLetter = letter

do {
try dbContext.save()
dismiss()
} catch {
print("Error saving record")
}
}
}


Now that every book knows the letter it belongs to, we can list them in
alphabetical sections. The following is the new @SectionedFetchRequest
property wrapper we need for the ContentView view.

Listing 33: Modifying the book's first letter


@SectionedFetchRequest(sectionIdentifier: \Books.firstLetter, sortDescriptors:
[SortDescriptor(\Books.title, order: .forward)], predicate: nil, animation: .default) private var
sectionBooks: SectionedFetchResults<String?, Books>


Now the section identifier is the value of the firstLetter property and the
books are sorted by title. The result is shown below.

Figure 24: Books in alphabetical sections

Do It Yourself: Update the toolbar() modifier in the InsertBookView

view with the code in Listing 31, the toolbar() modifier in the
ModifyBookView view with the code in Listing 32, and the
@SectionedFetchRequest property wrapper in the ContentView view with
the code in Listing 33. Uninstall the app to remove the books in the
Persistent Store. Run the application again on the iPhone simulator.
Add a few books. You should see the books organized in alphabetical
sections, as shown in Figure 24.
To-Many Relationships

The previous examples assumed that there was only one author per book,
but sometimes multiple authors collaborate to write a book. To assign
multiple Authors objects to a book, we must turn the author relationship of
the Books entity into a To-Many relationship, as shown below.

Figure 25: Many-To-Many relationship

Now, both relationships are of type To-Many, which means we can assign
multiple books to an author and multiple authors to a book. This
introduces a problem. Before, every time we wanted to assign an author to
a book, we just had to create a new Authors object and assign it to the
book's author property. Core Data took care of adding the book to the books
property of the Authors object, along with the rest of the books associated
to that author. But we can't do that when both relationships are To-Many.
In that case, we must read and write the values ourselves.
The values of a To-Many relationship are stored in an NSSet object. This is a
class defined by the Foundation framework to store sets of values. To read
the values in an NSSet, we can cast it as a Swift set, but to turn a Swift set or
an array into an NSSet object, we must implement the following initializers.

NSSet(set: Set)—This initializer creates an NSSet object with the set

provided by the attribute.
 NSSet(array: Array)—This initializer creates an NSSet object with the
array provided by the attribute.

In our application, the first place we need to read these values is in the
view model (the Extensions.swift file). We must create a property that
returns a string with the list of authors separated by comma.

Listing 34: Creating a string with the name of the authors from the view
model

import SwiftUI

extension Books {
var showTitle: String {
return title ?? "Undefined"
}
var showYear: String {
return String(year)
}
var showAuthors: String {
var authors: String!
if let list = author as? Set<Authors> {
let listNames = list.map({ $0.name ?? "Undefined" })
if !listNames.isEmpty {
authors = listNames.joined(separator: ", ")
}
}
return authors ?? "Undefined"
}
var showCover: UIImage {
if let data = cover, let image = UIImage(data: data) {
return image
} else {
return UIImage(named: "nopicture")!
}
}
var showThumbnail: UIImage {
if let data = thumbnail, let image = UIImage(data: data) {
return image
} else {
return UIImage(named: "nopicture")!
}
 }
}
extension Authors {
var showName: String {
return name ?? "Undefined"
}
}


The showAuthors property in this view model replaces the showAuthor property
implemented before. To create the list of names, we cast the value of the
author property to a Set<Authors> value. This creates a Swift set with Authors
objects representing all the authors assigned to the book, so we can map
the values into an array of strings and call the joined() method to create a
single string with the names separated by comma.
Now, we can show all the authors of a book on the list by reading the
showAuthors property, as shown next.

Listing 35: Showing the list of authors


import SwiftUI
import CoreData

struct ContentView: View {

@FetchRequest(sortDescriptors: [SortDescriptor(\Books.title, order: .forward)], predicate: nil,
animation: .default) var listOfBooks: FetchedResults<Books>

var body: some View {

NavigationView {
List {
ForEach(listOfBooks) { book in
NavigationLink(destination: ModifyBookView(book: book), label: {
RowBook(book: book)
.id(UUID())
})
}
}
.navigationBarTitle("Books")
.toolbar {
ToolbarItem(placement: .navigationBarTrailing) {
NavigationLink(destination: InsertBookView(), label: {
 Image(systemName: "plus")
})
}
}
}
}
}
struct RowBook: View {
let book: Books

var body: some View {

HStack(alignment: .top) {
Image(uiImage: book.showThumbnail)
.resizable()
.scaledToFit()
.frame(width: 80, height: 100)
.cornerRadius(10)
VStack(alignment: .leading, spacing: 2) {
Text(book.showTitle)
.bold()
Text(book.showAuthors)
.lineLimit(2)
Text(book.showYear)
.font(.caption)
Spacer()
}.padding(.top, 5)
Spacer()
}
}
}


The next step is to allow the users to associate multiple authors with a
book. The following are the changes we must introduce to the
InsertBookView view to allow the user to select multiple authors when a new
book is added.

Listing 36: Assigning multiple authors to a book


struct InsertBookView: View {
@Environment(\.managedObjectContext) var dbContext
@Environment(\.dismiss) var dismiss
@State private var selectedAuthors: [Authors] = []
@State private var inputTitle: String = ""
@State private var inputYear: String = ""

var showAuthors: String {

var authors = "Undefined"
if !selectedAuthors.isEmpty {
let listNames = selectedAuthors.map({ $0.name ?? "Undefined" })
if !listNames.isEmpty {
authors = listNames.joined(separator: ", ")
}
}
return authors
}
var body: some View {
VStack(spacing: 12) {
HStack {
Text("Title:")
TextField("Insert Title", text: $inputTitle)
.textFieldStyle(.roundedBorder)
}
HStack {
Text("Year:")
TextField("Insert Year", text: $inputYear)
.textFieldStyle(.roundedBorder)
}
HStack(alignment: .top) {
Text("Authors:")
VStack(alignment: .leading, spacing: 8) {
Text(showAuthors)
.foregroundColor(selectedAuthors.count > 0 ? Color.black : Color.gray)
NavigationLink(destination: AuthorsView(selected: $selectedAuthors), label: {
Text("Select Authors")
})
}
}.frame(minWidth: 0, maxWidth: .infinity, alignment: .leading)
Spacer()
}.padding()
.navigationBarTitle("Add Book")
.toolbar {
ToolbarItem(placement: .navigationBarTrailing) {
Button("Save") {
let newTitle = inputTitle.trimmingCharacters(in: .whitespaces)
let newYear = Int32(inputYear)
if !newTitle.isEmpty && newYear != nil {
Task(priority: .high) {
await storeBook(title: newTitle, year: newYear!)
}
 }
}
}
}
}
func storeBook(title: String, year: Int32) async {
await dbContext.perform {
let newBook = Books(context: dbContext)
newBook.title = title
newBook.year = year
newBook.author = NSSet(array: selectedAuthors)
newBook.cover = UIImage(named: "bookcover")?.pngData()
newBook.thumbnail = UIImage(named: "bookthumbnail")?.pngData()

var letter = String(title.first!).uppercased()

if Int(letter) != nil {
letter = "#"
}
newBook.firstLetter = letter

do {
try dbContext.save()
dismiss()
} catch {
print("Error saving record")
}
}
}
}


The values are now stored in an array, so we always know which ones were
selected by the user. To show the list of authors, we define a computed
property called showAuthors that follows the same process performed by the
view model; it maps the Authors objects and returns a string with the names
separated by comma. The view includes a Text view that reads this property
and shows the names on the screen.
When the user decides to save the book, we perform the inverse
procedure. The values in the selectedAuthors array are stored in an NSSet
object and assigned to the author property.
To allow the user to select the authors when a book is modified, we must
also apply these changes to the ModifyBookView view, as shown next.
Listing 37: Modifying the authors of a book

struct ModifyBookView: View {
@Environment(\.managedObjectContext) var dbContext
@Environment(\.dismiss) var dismiss
@State private var selectedAuthors: [Authors] = []
@State private var inputTitle: String = ""
@State private var inputYear: String = ""
@State private var valuesLoaded: Bool = false

let book: Books?

var showAuthors: String {

var authors = "Undefined"
if !selectedAuthors.isEmpty {
let listNames = selectedAuthors.map({ $0.name ?? "Undefined" })
if !listNames.isEmpty {
authors = listNames.joined(separator: ", ")
}
}
return authors
}
var body: some View {
VStack(spacing: 12) {
HStack {
Text("Title:")
TextField("Insert Title", text: $inputTitle)
.textFieldStyle(.roundedBorder)
}
HStack {
Text("Year:")
TextField("Insert Year", text: $inputYear)
.textFieldStyle(.roundedBorder)
}
HStack(alignment: .top) {
Text("Author:")
VStack(alignment: .leading, spacing: 8) {
Text(showAuthors)
.foregroundColor(selectedAuthors.count > 0 ? Color.black : Color.gray)
NavigationLink(destination: AuthorsView(selected: $selectedAuthors), label: {
Text("Select Authors")
})
}
}.frame(minWidth: 0, maxWidth: .infinity, alignment: .leading)
 Spacer()
}.padding()
.navigationBarTitle("Modify Book")
.toolbar {
ToolbarItem(placement: .navigationBarTrailing) {
Button("Save") {
let newTitle = inputTitle.trimmingCharacters(in: .whitespaces)
let newYear = Int32(inputYear)
if !newTitle.isEmpty && newYear != nil {
Task(priority: .high) {
await saveBook(title: newTitle, year: newYear!)
}
}
}
}
}
.onAppear {
if let list = book?.author as? Set<Authors>, !valuesLoaded {
selectedAuthors = Array(list)
inputTitle = book?.title ?? ""
inputYear = book?.showYear ?? ""
valuesLoaded = true
}
}
}
func saveBook(title: String, year: Int32) async {
await dbContext.perform {
book?.title = title
book?.year = year
book?.author = NSSet(array: selectedAuthors)

var letter = String(title.first!).uppercased()

if Int(letter) != nil {
letter = "#"
}
book?.firstLetter = letter

do {
try dbContext.save()
dismiss()
} catch {
print("Error saving record")
}
}
}
}

This view is very similar to the InsertBookView view, the only difference is a
new Boolean @State property called changesAdded that we use to know if the
user is coming from the ContentView view or the AuthorsView view. If the user
opened this view from the ContentView view, we need to load the list of
authors from the Persistent Store and assign them to the selectedAuthors
property when the view appears. Otherwise, the content of the
selectedAuthors property is determined by the authors selected by the user in
the AuthorsView view. Next are the changes we need to introduce to this
view.

Listing 38: Selecting multiple authors


import SwiftUI

struct AuthorsView: View {

@FetchRequest(sortDescriptors: [], predicate: nil, animation: .default) private var listOfAuthors:
FetchedResults<Authors>
@Environment(\.dismiss) var dismiss
@Binding var selected: [Authors]

var body: some View {

List {
ForEach(listOfAuthors) { author in
HStack {
Text(author.showName)
if selected.contains(where: { $0.name == author.name }) {
Image(systemName: "checkmark")
.foregroundColor(Color.blue)
.frame(width: 25, height: 25)
}
}
.frame(minWidth: 0, maxWidth: .infinity, minHeight: 0, maxHeight: .infinity, alignment:
.leading)
.contentShape(Rectangle())
.onTapGesture {
if selected.contains(where: { $0.name == author.name }) {
if let index = selected.firstIndex(of: author) {
selected.remove(at: index)
}
} else {
 selected.append(author)
}
dismiss()
}
}
}
.navigationBarTitle("Authors")
.toolbar {
ToolbarItem(placement: .navigationBarTrailing) {
NavigationLink(destination: InsertAuthorView(), label: {
Image(systemName: "plus")
})
}
}
}
}
struct AuthorsView_Previews: PreviewProvider {
static var previews: some View {
AuthorsView(selected: .constant([]))
}
}


As always, this view displays all available authors. To show which one has
been previously selected, we add an Image view with a checkmark to the
row of the authors that are already in the selectedAuthors array. Then, when a
row is tapped by the user, we check whether the author was previously
selected or not. If it was selected, we remove it from the selectedAuthors
array, otherwise, we add it to the array. This makes sure that the array only
contains the authors currently selected by the user.

Do It Yourself: Open the Core Data model. Select the Books entity
and change the Type of the author relationship to To-Many (Figure
25). Update the Extensions.swift file with the code in Listing 34, the
ContentView view with the code in Listing 35, the InsertBookView with the
code in Listing 36, the ModifyBookView view with the code in Listing 37,
and the AuthorsView view with the code in Listing 38. Run the
application on the iPhone simulator. Press the + button to add a new
 book. Press the Select Authors button to select an author. You should
be able to add as many authors as you want.

The To-Many to To-Many relationships also change the way we search for
values. For instance, we cannot search for a book by author as we did
before because now a book may be associated to many authors. Instead,
we must tell the predicate to search for the value inside the set of authors.
For this purpose, predicates can include the following keywords.

ANY—This keyword returns true when the condition is true for some
of the values in the set.
ALL—This keyword returns true when the condition is true for all the
values in the set.
NONE—This keyword returns true when the condition is false for all
the values in the set.

We have introduced predicate keywords earlier. They are included in the

format string to determine the way the predicate filters the data. For our
example, we can add the ANY keyword in front of the comparison to get
the books associated with at least one author with a specific name, as
shown next.

Listing 39: Fetching books by author


.onAppear {
let request: NSFetchRequest<Books> = Books.fetchRequest()
request.predicate = NSPredicate(format: "ANY author.name == %@", "Stephen King")
if let list = try? dbContext.fetch(request) {
for book in list {
print(book.title!)
}
}
}

This example defines an onAppear() modifier that we can add to a view in the
ContentView view. It creates a request that finds all the books associated with
an author named "Stephen King". The predicate reads all the Authors
objects in the relationship and returns the book when one of the names
matches the string.

Do It Yourself: Add the modifier of Listing 39 to the List view in the

ContentViewview. You also need to get access to the Core Data context
with the @Environment property wrapper, as in
@Environment(\.managedObjectContext) var dbContext. Run the application on
the iPhone simulator and insert a few books with the author
Stephen King. You should see the names of the books associated
with that author printed on the console.

The example we have been working on so far turns the NSSet object
returned by the author relationship into an array of Authors objects and
then adds or removes authors from this array, but if we need to add or
remove values directly from the NSSet object, we must turn it into an
NSMutableSet object. This class creates a mutable set and therefore it allows
us to add or remove values from it. To create an NSMutableSet object from an
NSSet object, the NSManagedObject class includes the following method.

mutableSetValue(forKey: String)—This method reads the NSSet

object of the relationship indicated by the forKey attribute and
returns an NSMutableSet with the values.

The NSMutableSet class includes the following methods to add and remove
items in the set.

add(Any)—This method adds the object specified by the argument

to the set.
 remove(Any)—This method removes the object specified by the
argument from the set.

The following example shows a possible implementation of these methods.

We get the object representing the author with the name "Stephen King"
and then remove that author from every book.

Listing 40: Systematically removing authors from books


.onAppear {
let request: NSFetchRequest<Authors> = Authors.fetchRequest()
request.predicate = NSPredicate(format: "name == %@", "Stephen King")
if let list = try? dbContext.fetch(request), list.count > 0 {
let author = list[0]

Task(priority: .high) {
await dbContext.perform {
for book in listOfBooks {
let authorSet = book.mutableSetValue(forKey: "author")
authorSet.remove(author)
book.author = authorSet
}
try? dbContext.save()
}
}
}
}


This example updates the onAppear() modifier defined before to modify the
Books objects in the Persistent Store as soon as the view is loaded. First, we
perform a request to get the Authors object with the name "Stephen King".
Then, we use a for in loop to modify the books loaded by the @FetchRequest
property wrapper. In the loop, we turn the NSSet object returned by the
author relationship into an NSMutableSet object, remove from the set the
Authors object fetched before with the remove() method, and assign the
result back to the author relationship, effectively removing that author
from every book.
Do It Yourself: Update the onAppear() modifier with the code in
Listing 40. Run the application on the iPhone simulator. The author
Stephen King should be removed from every book.
Find more books at
www.formasterminds.com

J.D Gauchat
www.jdgauchat.com