Merge pull request #889 from planetary-social/bugfix/detect-raw-domai…

…n-urls

Fix URL detection of raw domain names

CHANGELOG.md

@@ -8,6 +8,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+- Fixed URL detection of raw domain names, such as “nos.social” (without the “http” prefix).
 - Fixed the sort order of gallery media to match the order in the note.
 - While composing a note, a space is now automatically inserted after any mention of a user or note to ensure it’s formatted correctly.
 

Nos.xcodeproj/xcshareddata/xcschemes/Nos.xcscheme

@@ -73,6 +73,9 @@
                <Test
                   Identifier = "SocialGraphTests/testFollow()">
                </Test>
+               <Test
+                  Identifier = "SocialGraphTests/testTwoFollows()">
+               </Test>
             </SkippedTests>
          </TestableReference>
          <TestableReference

Nos/Extensions/String+Markdown.swift

@@ -38,22 +38,24 @@ extension String {
     func extractURLs() -> (String, [URL]) {
         var urls: [URL] = []
         let mutableString = NSMutableString(string: self)
-        let regexPattern = "(\\s*)(https?://[^\\s]*)"
-
+        // The following pattern uses rules from the Domain Name System page on Wikipedia:
+        // https://en.wikipedia.org/wiki/Domain_Name_System#Domain_name_syntax,_internationalization
+        let regexPattern = "(\\s*)((https?://)?([a-zA-Z0-9][-a-zA-Z0-9]{0,62}\\.){1,127}[a-z]{2,63}[^\\s]*)"
+
         do {
-            let regex = try NSRegularExpression(pattern: regexPattern, options: [])
+            let regex = try NSRegularExpression(pattern: regexPattern)
             let range = NSRange(location: 0, length: mutableString.length)
 
-            let matches = regex.matches(in: self, options: [], range: range).reversed()
+            let matches = regex.matches(in: self, range: range).reversed()
 
             for match in matches {
                 if let range = Range(match.range(at: 2), in: self), let url = URL(string: String(self[range])) {
-                    urls.insert(url, at: 0) // maintain original order of links (we're looping in reverse)
+                    // maintain original order of links by inserting at index 0 (we're looping in reverse)
+                    urls.insert(url.addingSchemeIfNeeded(), at: 0)
                     let prettyURL = url.truncatedMarkdownLink
                     regex.replaceMatches(
                         in: mutableString, 
-                        options: [], 
-                        range: match.range, 
+                        range: match.range,
                         withTemplate: "$1\(prettyURL)"
                     )
                 }

Nos/Extensions/URL+Extensions.swift

@@ -8,7 +8,18 @@
 import Foundation
 
 extension URL {
-
+
+    /// Returns the URL with the scheme "https" if the URL doesn't have a scheme; othewise, returns self.
+    /// Using a URL with a scheme fixes an issue with opening link previews when there's no scheme.
+    /// - Returns: The URL with the scheme "https" if the URL doesn't have a scheme; otherwise, returns self.
+    func addingSchemeIfNeeded() -> URL {
+        guard scheme == nil else {
+            return self
+        }
+
+        return URL(string: "https://\(absoluteString)") ?? self
+    }
+
     var isImage: Bool {
         let imageExtensions = ["png", "jpg", "jpeg", "gif"]
         return imageExtensions.contains(pathExtension)
@@ -23,18 +34,20 @@ extension URL {
     }
 
     var truncatedMarkdownLink: String {
-        guard var host = host() else {
-            return "[\(absoluteString)](\(absoluteString))"
+        let url = self.addingSchemeIfNeeded()
+
+        guard var host = url.host() else {
+            return "[\(url.absoluteString)](\(url.absoluteString))"
         }
 
         if host.hasPrefix("www.") {
             host = String(host.dropFirst(4))
         }
 
-        if path().isEmpty {
-            return "[\(host)](\(absoluteString))"
+        if url.path().isEmpty {
+            return "[\(host)](\(url.absoluteString))"
         } else {
-            return "[\(host)...](\(absoluteString))"
+            return "[\(host)...](\(url.absoluteString))"
         }
     }
 }

Nos/Models/NoteParser.swift

@@ -33,7 +33,7 @@ enum NoteParser {
         }
     }
 
-    /// Parses the content and tags stored in a note and returns an attributed text and list of URLs that can be used 
+    /// Parses the content and tags stored in a note and returns an attributed string and list of URLs that can be used 
     /// to display the note in the UI.
     static func parse(content: String, tags: [[String]], context: NSManagedObjectContext) -> (AttributedString, [URL]) {
         var result = replaceTaggedNostrEntities(in: content, tags: tags, context: context)

NosTests/Extensions/String+MarkdownTests.swift

@@ -8,7 +8,21 @@
 import XCTest
 
 class String_MarkdownTests: XCTestCase {
-    func testExtractURLsFromNote() throws {
+    /// Test this function that's not used anwhere.
+    /// Consider removing it after extracting all value from it. (that regex in it looks great)
+    func testFindAndReplaceUnformattedLinksWithNoURLScheme() throws {
+        // Arrange
+        let string = "One: https://nos.social and two: nostr1.com"
+        let expected = "One: [https://nos.social](https://nos.social) and two: [nostr1.com](https://nostr1.com)"
+
+        // Act
+        let result = try string.findAndReplaceUnformattedLinks(in: string)
+
+        // Assert
+        XCTAssertEqual(result, expected)
+    }
+
+    func testExtractURLs() throws {
         // swiftlint:disable line_length
         let string = "Classifieds incoming... 👀\n\nhttps://nostr.build/i/2170fa01a69bca5ad0334430ccb993e41bb47eb15a4b4dbdfbee45585f63d503.jpg"
         let expectedString = "Classifieds incoming... 👀\n\n[nostr.build...](https://nostr.build/i/2170fa01a69bca5ad0334430ccb993e41bb47eb15a4b4dbdfbee45585f63d503.jpg)"
@@ -23,14 +37,26 @@ class String_MarkdownTests: XCTestCase {
         XCTAssertEqual(actualURLs, expectedURLs)
     }
 
-    func testExtractURLsFromNoteWithMultipleURLs() throws {
-        // swiftlint:disable line_length
-        let string = "Classifieds incoming... 👀\n\nhttps://nostr.build/i/2170fa01a69bca5ad0334430ccb993e41bb47eb15a4b4dbdfbee45585f63d503.jpg\n\nhttps://cdn.ymaws.com/nacfm.com/resource/resmgr/images/blog_photos/footprints.jpg"
-        let expectedString = "Classifieds incoming... 👀\n\n[nostr.build...](https://nostr.build/i/2170fa01a69bca5ad0334430ccb993e41bb47eb15a4b4dbdfbee45585f63d503.jpg)\n\n[cdn.ymaws.com...](https://cdn.ymaws.com/nacfm.com/resource/resmgr/images/blog_photos/footprints.jpg)"
-        // swiftlint:enable line_length
+    func testExtractURLsWithMultipleURLs() throws {
+        let string = """
+        A few links...
+
+        https://nostr.build/i/2170fa01a69bca5ad0334430ccb993e41bb47eb15a4b4dbdfbee45585f63d503.jpg
+        nos.social
+        www.nostr.com/get-started
+        """
+        let expectedString = """
+        A few links...
+
+        [nostr.build...](https://nostr.build/i/2170fa01a69bca5ad0334430ccb993e41bb47eb15a4b4dbdfbee45585f63d503.jpg)
+        [nos.social](https://nos.social)
+        [nostr.com...](https://www.nostr.com/get-started)
+        """
+
         let expectedURLs = [
             URL(string: "https://nostr.build/i/2170fa01a69bca5ad0334430ccb993e41bb47eb15a4b4dbdfbee45585f63d503.jpg")!,
-            URL(string: "https://cdn.ymaws.com/nacfm.com/resource/resmgr/images/blog_photos/footprints.jpg")!
+            URL(string: "https://nos.social")!,
+            URL(string: "https://www.nostr.com/get-started")
         ]
 
         // Act
@@ -39,7 +65,19 @@ class String_MarkdownTests: XCTestCase {
         XCTAssertEqual(actualURLs, expectedURLs)
     }
 
-    func testExtractURLsFromImageNote() throws {
+    func testExtractURLsDoesNotInterpretAllDotsAsURLs() throws {
+        // Arrange
+        let string = "No links...and just some tricks for the extractor..to try to trip it up. ...Ready for It?"
+
+        // Act
+        let (actualString, actualURLs) = string.extractURLs()
+
+        // Assert
+        XCTAssertEqual(actualString, string)
+        XCTAssertTrue(actualURLs.isEmpty)
+    }
+
+    func testExtractURLsWithImage() throws {
         let string = "Hello, world!https://cdn.ymaws.com/footprints.jpg"
         let expectedString = "Hello, world![cdn.ymaws.com...](https://cdn.ymaws.com/footprints.jpg)"
         let expectedURLs = [
@@ -52,7 +90,7 @@ class String_MarkdownTests: XCTestCase {
         XCTAssertEqual(actualURLs, expectedURLs)
     }
 
-    func testExtractURLsFromImageNoteWithExtraNewlines() throws {
+    func testExtractURLsWithImageWithExtraNewlines() throws {
         let string = "https://cdn.ymaws.com/footprints.jpg\n\nHello, world!"
         let expectedString = "[cdn.ymaws.com...](https://cdn.ymaws.com/footprints.jpg)\n\nHello, world!"
         let expectedURLs = [

NosTests/Fixtures/URLExtensionTests.swift

@@ -9,6 +9,39 @@ import XCTest
 
 final class URLExtensionTests: XCTestCase {
 
+    func test_addingSchemeIfNeeded_adds_scheme_when_there_is_no_scheme() {
+        let subject = URL(string: "nos.social")!
+        let expected = URL(string: "https://nos.social")!
+        let result = subject.addingSchemeIfNeeded()
+        XCTAssertEqual(result, expected)
+    }
+
+    func test_addingSchemeIfNeeded_adds_scheme_when_url_has_path() {
+        let subject = URL(string: "nos.social/about")!
+        let expected = URL(string: "https://nos.social/about")!
+        let result = subject.addingSchemeIfNeeded()
+        XCTAssertEqual(result, expected)
+    }
+
+    func test_addingSchemeIfNeeded_adds_scheme_when_url_has_subdomain() {
+        let subject = URL(string: "www.nos.social")!
+        let expected = URL(string: "https://www.nos.social")!
+        let result = subject.addingSchemeIfNeeded()
+        XCTAssertEqual(result, expected)
+    }
+
+    func test_addingSchemeIfNeeded_does_not_add_scheme_when_http_scheme_already_exists() {
+        let subject = URL(string: "http://nos.social")!
+        let result = subject.addingSchemeIfNeeded()
+        XCTAssertEqual(result, subject)
+    }
+
+    func test_addingSchemeIfNeeded_does_not_add_scheme_when_nostr_scheme_already_exists() {
+        let subject = URL(string: "nostr:npub1234")!
+        let result = subject.addingSchemeIfNeeded()
+        XCTAssertEqual(result, subject)
+    }
+
     func testTruncatedMarkdownLink_withEmptyPathExtension() {
         let url = URL(string: "https://subdomain.example.com")!
         XCTAssertEqual(url.truncatedMarkdownLink, "[subdomain.example.com](https://subdomain.example.com)")
@@ -28,6 +61,16 @@ final class URLExtensionTests: XCTestCase {
         let url = URL(string: "https://www.example.com")!
         XCTAssertEqual(url.truncatedMarkdownLink, "[example.com](https://www.example.com)")
     }
+
+    func testTruncatedMarkdownLink_noScheme_withWWW_removesWWW() {
+        let url = URL(string: "www.nostr.com/get-started")!
+        XCTAssertEqual(url.truncatedMarkdownLink, "[nostr.com...](https://www.nostr.com/get-started)")
+    }
+
+    func testTruncatedMarkdownLink_noScheme_withWWW_noPath_doesNotIncludeEllipsis() {
+        let url = URL(string: "www.nos.social")!
+        XCTAssertEqual(url.truncatedMarkdownLink, "[nos.social](https://www.nos.social)")
+    }
 
     func testTruncatedMarkdownLink_withShortPath() {
         let url = URL(string: "https://nips.be/1")!