Merge pull request #3340 from SwiftPackageIndex/fix-3339

Replace GitHub protected cached images in README files with their original uncached versions

Author: daveverwer

Sources/App/Views/PackageController/PackageReadme+Model.swift

@@ -56,6 +56,7 @@ extension PackageReadme {
             readmeElement.rewriteRelativeImages(to: repoTriple)
             readmeElement.rewriteRelativeLinks(to: repoTriple)
             readmeElement.fixInlineAnchors()
+            readmeElement.fixProtectedCachedImages()
             readmeElement.disableTurboOnLinks()
             return readmeElement
         }
@@ -124,6 +125,21 @@ extension Element {
         }
     }
 
+    func fixProtectedCachedImages() {
+        do {
+            let imageElements = try select("img[data-canonical-src]")
+            for imageElement in imageElements {
+                let originalUrl = try imageElement.attr("data-canonical-src")
+                if originalUrl.hasPrefix("http") {
+                    try imageElement.attr("src", originalUrl)
+                }
+            }
+        } catch {
+            // Errors are being intentionally eaten here. The worst that can happen if the
+            // HTML selection/parsing fails is that cached images don't get corrected.
+        }
+    }
+
     func disableTurboOnLinks() {
         do {
             let linkElements = try select("a")

Tests/AppTests/PackageReadmeModelTests.swift

@@ -130,6 +130,30 @@ class PackageReadmeModelTests: SnapshotTestCase {
         assertSnapshot(of: html, as: .lines)
     }
 
+    func test_Element_fixProtectedCachedImages() throws {
+        // setup
+        let element = Element.extractReadme("""
+            <div id="readme">
+                <article>
+                    <p>README content.</p>
+                    <img src="https://example.com/standard.png" />
+                    <img src="https://camo.githubusercontent.com/0123456789" data-canonical-src="https://example.com/cached.png" />
+                    <img src="https://camo.githubusercontent.com/0123456789" data-canonical-src="" />
+                    <img src="https://camo.githubusercontent.com/0123456789" />
+                    <img src="https://example.com/0123456789" data-canonical-src="https://example.com/cached.png" />
+                    <img />
+                </article>
+            </div>
+            """)
+
+        // MUT
+        element?.fixProtectedCachedImages()
+
+        // validate
+        let html = try XCTUnwrap(try element?.html())
+        assertSnapshot(of: html, as: .lines)
+    }
+
     func test_Element_disableTurboOnLinks() throws {
         // setup
         let element = Element.extractReadme("""

Tests/AppTests/__Snapshots__/PackageReadmeModelTests/test_Element_fixProtectedCachedImages.1.txt

@@ -0,0 +1,7 @@
+<p>README content.</p> 
+<img src="https://example.com/standard.png" /> 
+<img src="https://example.com/cached.png" data-canonical-src="https://example.com/cached.png" /> 
+<img src="https://camo.githubusercontent.com/0123456789" data-canonical-src="" /> 
+<img src="https://camo.githubusercontent.com/0123456789" /> 
+<img src="https://example.com/cached.png" data-canonical-src="https://example.com/cached.png" /> 
+<img />