Use SwiftSoup to parse Open Graph metadata

CHANGELOG.md

@@ -30,6 +30,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Minor refactor of Event+CoreDataClass. [#1443](https://github.com/planetary-social/nos/issues/1443)
 - Refactored feature flag and added a feature flag toggle for “Enable new moderation flow” to Staging builds. [#1496](https://github.com/planetary-social/nos/issues/1496)
 - Refactored list row gradient background.
+- Added SwiftSoup to parse Open Graph metadata. [#1165](https://github.com/planetary-social/nos/issues/1165)
 
 ## [0.1.26] - 2024-09-09Z
 

Nos.xcodeproj/project.xcworkspace/xcshareddata/swiftpm/Package.resolved

@@ -179,6 +179,15 @@
         "revision" : "1228d5a43ca791db0719d42f641c34a00b9c32f3"
       }
     },
+    {
+      "identity" : "swiftsoup",
+      "kind" : "remoteSourceControl",
+      "location" : "https://github.com/scinfu/SwiftSoup",
+      "state" : {
+        "revision" : "3c2c7e1e72b8abd96eafbae80323c5c1e5317437",
+        "version" : "2.7.5"
+      }
+    },
     {
       "identity" : "swiftui-navigation",
       "kind" : "remoteSourceControl",

Nos/Models/OpenGraph/OpenGraphParser.swift

@@ -0,0 +1,15 @@
+import Foundation
+
+/// Parses the Open Graph metadata from an HTML document.
+protocol OpenGraphParser {
+    /// Fetches the Open Graph video metadata from the given HTML document.
+    /// - Parameter html: An HTML document.
+    /// - Returns: The Open Graph video metadata from the HTML.
+    func videoMetadata(html: Data) -> OpenGraphMedia?
+}
+
+/// An Open Graph property in the HTML.
+enum OpenGraphProperty: String {
+    case videoHeight = "og:video:height"
+    case videoWidth = "og:video:width"
+}

Nos/Models/OpenGraph/SoupOpenGraphParser.swift

@@ -0,0 +1,32 @@
+import Foundation
+import SwiftSoup
+
+/// Parses the Open Graph metadata from an HTML document using SwiftSoup.
+struct SoupOpenGraphParser: OpenGraphParser {
+    func videoMetadata(html: Data) -> OpenGraphMedia? {
+        let htmlString = String(decoding: html, as: UTF8.self)
+        guard let document = try? SwiftSoup.parse(htmlString) else { return nil }
+
+        guard let widthString = openGraphProperty(.videoWidth, from: document),
+            let width = Double(widthString) else {
+            return nil
+        }
+        guard let heightString = openGraphProperty(.videoHeight, from: document),
+            let height = Double(heightString) else {
+            return nil
+        }
+
+        return OpenGraphMedia(type: .video, width: width, height: height)
+    }
+}
+
+extension SoupOpenGraphParser {
+    /// Gets the Open Graph property value from the given HTML document.
+    /// - Parameters:
+    ///   - property: The Open Graph property to fetch from the HTML document.
+    ///   - document: The HTML document.
+    /// - Returns: The value of the Open Graph property, or `nil` if none is found.
+    private func openGraphProperty(_ property: OpenGraphProperty, from document: Document) -> String? {
+        try? document.select("meta[property=\(property.rawValue)]").attr("content")
+    }
+}

Nos/Service/Media/MediaService.swift

@@ -47,7 +47,7 @@ struct DefaultMediaService: MediaService {
         }
     }
 
-    /// Loads the content at the given URL and returns its orientation.
+    /// Fetches metadata for the given URL and returns its orientation.
     /// - Parameter url: The URL of the data to download.
     /// - Returns: The orientation of the content.
     /// - Note: For web pages, `landscape` is returned. For videos, we're returning `portrait` until we implement

Nos/Service/Media/OpenGraphService.swift

@@ -0,0 +1,40 @@
+import Foundation
+
+/// A service that fetches metadata for a URL.
+protocol OpenGraphService {
+    /// Fetches metadata for the given URL.
+    /// - Parameter url: The URL to fetch.
+    /// - Returns: The Open Graph metadata for the URL.
+    func fetchMetadata(for url: URL) async throws -> OpenGraphMetadata
+}
+
+/// A default implementation for `OpenGraphService`.
+struct DefaultOpenGraphService: OpenGraphService {
+    let session: URLSessionProtocol
+    let parser: OpenGraphParser
+
+    func fetchMetadata(for url: URL) async throws -> OpenGraphMetadata {
+        let request = URLRequest(url: url)
+        let (data, _) = try await session.data(for: request)
+        let videoMetadata = parser.videoMetadata(html: data)
+        return OpenGraphMetadata(media: videoMetadata)
+    }
+}
+
+/// Open Graph metadata for a URL.
+struct OpenGraphMetadata: Equatable {
+    let media: OpenGraphMedia?
+}
+
+/// Open Graph metadata for media, such as an image or video.
+struct OpenGraphMedia: Equatable {
+    let type: OpenGraphMediaType?
+    let width: Double?
+    let height: Double?
+}
+
+/// The type of Open Graph media.
+enum OpenGraphMediaType {
+    case image
+    case video
+}

Nos/Views/Components/Media/LinkView.swift

@@ -15,7 +15,7 @@ struct LinkView: View {
 }
 
 #Preview("Video") {
-    LinkView(url: URL(string: "https://youtu.be/5qvdbyRH9wA?si=y_KTgLR22nH0-cs8")!)
+    LinkView(url: URL(string: "https://www.youtube.com/watch?v=sB6HY8r983c")!)
 }
 
 #Preview("Image") {

NosTests/Models/OpenGraph/MockOpenGraphParser.swift

@@ -0,0 +1,8 @@
+import Foundation
+
+/// A mock Open Graph parser that can be used for testing.
+struct MockOpenGraphParser: OpenGraphParser {
+    func videoMetadata(html: Data) -> OpenGraphMedia? {
+        nil
+    }
+}

NosTests/Models/OpenGraph/SoupOpenGraphParserTests.swift

@@ -0,0 +1,50 @@
+import Foundation
+import XCTest
+
+class SoupOpenGraphParserTests: XCTestCase {
+    let sampleHTML = """
+    <!DOCTYPE html>
+    <html>
+    <head>
+        <meta property="og:video:width" content="2560">
+        <meta property="og:video:height" content="1440">
+    </head>
+    <body>
+        <h1>Sample HTML</h1>
+    </body>
+    </html>
+    """
+
+    // swiftlint:disable:next implicitly_unwrapped_optional
+    var youTubeHTML: Data!
+
+    override func setUpWithError() throws {
+        youTubeHTML = try htmlData(filename: "youtube_video_toxic")
+    }
+
+    func test_parse() throws {
+        // Arrange
+        let parser = SoupOpenGraphParser()
+        let data = try XCTUnwrap(sampleHTML.data(using: .utf8))
+
+        // Act
+        let videoMetadata = try XCTUnwrap(parser.videoMetadata(html: data))
+
+        // Assert
+        XCTAssertEqual(videoMetadata.width, 2560)
+        XCTAssertEqual(videoMetadata.height, 1440)
+    }
+
+    func test_parse_youTube() throws {
+        // Arrange
+        let parser = SoupOpenGraphParser()
+        let data = try XCTUnwrap(youTubeHTML)
+
+        // Act
+        let videoMetadata = try XCTUnwrap(parser.videoMetadata(html: data))
+
+        // Assert
+        XCTAssertEqual(videoMetadata.width, 1280)
+        XCTAssertEqual(videoMetadata.height, 720)
+    }
+}

NosTests/Service/Media/DefaultOpenGraphServiceTests.swift

@@ -0,0 +1,16 @@
+import XCTest
+
+class DefaultOpenGraphServiceTests: XCTestCase {
+    func test_fetchMetadata() async throws {
+        // Arrange
+        let url = try XCTUnwrap(URL(string: "https://youtu.be/5qvdbyRH9wA?si=y_KTgLR22nH0-cs8"))
+        let subject = DefaultOpenGraphService(session: MockURLSession(), parser: MockOpenGraphParser())
+        let expected = OpenGraphMetadata(media: nil)
+
+        // Act
+        let metadata = try await subject.fetchMetadata(for: url)
+
+        // Assert
+        XCTAssertEqual(metadata, expected)
+    }
+}

NosTests/TestHelpers/XCTestCase+JSONData.swift → NosTests/TestHelpers/XCTestCase+FileData.swift

@@ -5,4 +5,9 @@ extension XCTestCase {
         let url = try XCTUnwrap(Bundle.current.url(forResource: filename, withExtension: "json"))
         return try Data(contentsOf: url)
     }
+
+    func htmlData(filename: String) throws -> Data {
+        let url = try XCTUnwrap(Bundle.current.url(forResource: filename, withExtension: "html"))
+        return try Data(contentsOf: url)
+    }
 }