Home

Awesome

SafeDecoding

Swift Macro enabling safe decoding of structs and classes.

Features

Requirements

Versions

Version 1.4.0

Version 1.3.0

Version 1.2.1

Version 1.2.0

Version 1.1.0

Version 1.0.0

Example Usage

Structs & Classes

A common problem is when an otherwise non-mandatory part of a model contains invalid/missing data, causing an entire payload to fail. In the following example we have a Book model, for which we'll retrieve an array from some backend.

struct Tag: Decodable {
    let name: String
}

struct Book: Decodable {
    let title: String
    let author: String
    let synopsis: String?
    let tags: [Tag]
}

Receiving a corrupted tag would cause the entire payload to fail when fetching a list of Books. In the following JSON, note that the tag name is missing from the book My Sweet Swift Book:

[
    {
        "title": "Dune",
        "author": "Frank Herbert",
        "tags": [
            {
                "name": "Sky-Fi"
            }
        ]
    },
    ...
    {
        "title": "My Sweet Swift Book",
        "author": "Me",
        "tags": [
            {
                "name": "Tech"
            },
            {
            }
        ]
    }
]

With the current declaration of Book and Tag, this would cause the entire payload decoding to fail. In order to allow a class/struct to gain resilience to partial decoding failure, simply add use the @SafeDecoding macro:

@SafeDecoding
struct Book {
    ...
}

This will implement custom decoding for Book, allowing the single invalid tag to fail, while correctly decoding everything else. Safe decoding is achieved by attaching a type extension implementing a custom init(from:) (and declaring conformance to Decodable, if necessary). Within the initializer, safe decoding will be applied to all fitting properties (of type Optional, Array, Set and Dictionary).
@SafeDecoding may be used in any struct or class.

In order to opt-out of safe decoding for a property, simply tag it with @IgnoreSafeDecoding macro. Let's say an invalid synopsis is enough to invalidate the entire Book:

@SafeDecoding
struct Book {
    ...
    @IgnoreSafeDecoding
    let synopsis: String?
    ...
}

This will cause synopsis to not be safely decoded in the initializer.

The @FallbackDecoding macro can be used to grant fallback semantics to properties. @FallbackDecoding must be used with @SafeDecoding, and means decoding will never fail properties it is applied to (even if the type is not otherwise "safe-decodable"):

@SafeDecoding
struct Book {
    @FallbackDecoding(false)
    var isFavourite: Bool
}

The RetryDecoding macro can in turn be used to provide alternative decoding of a property; an alternative decoding type and a "mapper" between types must be provided. An example could be a backend that sometimes returns integers as strings, or booleans as integers:

@SafeDecoding
struct Book {
    @RetryDecoding(String.self, map: { $0.lowercased() == "true" })
    @RetryDecoding(Int.self, map: { $0 != 0 })
    var isFavourite: Bool
}

Retries will be performed in the same order as they are declared in the property. If @FallbackDecoding is used alongside retries, all retries will be attempted before the value specified for fallback is used.

Enums

Decoding of enums uses the same @SafeDecoding macro, but requires the decodingStrategy to be specified. The strategy is itself an enumeration (EnumCaseDecodingStrategy) which holds two cases:

caseByObjectProperty(String)

A property name must be specified; this will be itself decoded as a string matching the cases of the enum. E.g. for the following enum:

@SafeDecoding(decodingStrategy: .caseByObjectProperty("type")
enum MediaAsset {
    case vod(...)
    case series(...)
    case episode(...)
    ...
}

A property type will be decoded, and matched with a case in the MediaAsset type:

{
    "type": "vod",
    "title": "...",
    ...
}

This will result in a .vod being decoded. The specified property may or may not be part of the final model.

caseByNestedObject

The standard strategy used when auto-decoding is used, will attempt to decode a root property for each of the enums cases, then decoding the case's parameters. For the same example specified above, decoding a vod case would require a payload like the one below:

{
    "vod": {
        "title": "...",
        ...
    }
}

The literal name used for an enum case will be the case itself by default, but may be overriden using the @CaseNameDecoding macro:

@SafeDecoding(decodingStrategy: .caseByObjectProperty("type"))
enum MediaAssetKeyed {
    @CaseNameDecoding("ASSET/PROGRAMME")
    case vod(String?, String)
    @CaseNameDecoding("ASSET/SERIES")
    case series(id: String, Set<String>)
    @CaseNameDecoding("ASSET/EPISODE")
    case episode(id: String, title: String, arguments: [String: Double])
}

This will try to match cases by decoding a type property, with the property's type itself matching the specified names:

{
    "type": "ASSET/EPISODE",
    "id": "...",
    ...
}

Error Reporting

Finally, @SafeDecoding can report errors that occur (and are recovered from) during the decoding process. This is done by passing the reporter: parameter to the macro:

@SafeDecoding(reporter: ...)
struct Book {
    ....
}

The same applies to enum decoding:

@SafeDecoding(decodingStrategy: ..., reporter: ...)

enum MediaAsset { ... }

The reporter must conform the SafeDecodingReporter protocol. Upon recovery of decoding errors, the reporter will be called with information about said error. Remember that a reporter is local to its type, i.e. although the same type may be used everywhere each @SafeDecoding usage must be given its reporter expression.

Installation

Swift Package Manager

You can use the Swift Package Manager to install your package by adding it as a dependency to your Package.swift file:

dependencies: [
    .package(url: "https://github.com/renato-iar/SafeDecoding.git", from: "1.0.0")
]