Allow semicolon in file path for Swift URL v2 (swiftlang#1882)

* (174175082) Allow semicolon in file path for Swift URL v2

* Clarify 'compatibility' -> 'encodeSemicolons'

Author: jrflat

Sources/FoundationEssentials/URL/URL_C+File.swift

@@ -122,7 +122,7 @@ private func parseFileSystemRepresentation(
             isDirectory: isDirectory
         )
         let path = pathBuffer.span.extracting(first: finalLength)
-        return URL.parseFinalFileSystemRepresentation(path: path, flags: &flags)
+        return URL.parseFinalFileSystemRepresentation(path: path, flags: &flags, encodeSemicolons: true)
     }
 }
 
@@ -159,7 +159,7 @@ private func parsePOSIX(_ path: CFString, flags: inout _URLFlags, isDirectory: B
             isDirectory: isDirectory
         )
         let path = pathBuffer.span.extracting(first: finalLength)
-        return URL.parseFinalFileSystemRepresentation(path: path, flags: &flags)
+        return URL.parseFinalFileSystemRepresentation(path: path, flags: &flags, encodeSemicolons: true)
     }
 }
 
@@ -187,7 +187,7 @@ private func parseHFS(_ path: String, flags: inout _URLFlags, isDirectory: Bool)
     guard !path.isEmpty else {
         return ""
     }
-    return URL.parseUTF8Path(path, flags: &flags, isDirectory: isDirectory)
+    return URL.parseUTF8Path(path, flags: &flags, isDirectory: isDirectory, encodeSemicolons: true)
 }
 
 // Note this does not percent-encode the path

Sources/FoundationEssentials/URL/URL_File.swift

@@ -76,7 +76,8 @@ extension URL {
     ///   see `finalPathLength(updating:currentLength:flags:isDirectory)`.
     static func parseFinalFileSystemRepresentation(
         path: borrowing Span<UInt8>,
-        flags: inout _URLFlags
+        flags: inout _URLFlags,
+        encodeSemicolons: Bool
     ) -> String {
         guard path.count > 0 else {
             return ""
@@ -108,8 +109,8 @@ extension URL {
                 URLEncoder.percentEncodeUnchecked(
                     input: pathBuffer,
                     output: .init(rebasing: encodedBuffer[pathStart...]),
-                    // Encode ";" for compatibility
-                    component: .pathNoSemicolon,
+                    // Encode ";" for CF/NSURL (encodeSemicolons: true)
+                    component: encodeSemicolons ? .pathNoSemicolon : .path,
                     skipAlreadyEncoded: false
                 )
             }
@@ -121,7 +122,7 @@ extension URL {
         }
     }
 
-    static func parseUTF8Path(_ path: String, flags: inout _URLFlags, isDirectory: Bool) -> String {
+    static func parseUTF8Path(_ path: String, flags: inout _URLFlags, isDirectory: Bool, encodeSemicolons: Bool) -> String {
         var path = path
         return path.withUTF8 { pathBuffer in
             // Allocate an extra byte in case we need to append a directory slash.
@@ -133,7 +134,7 @@ extension URL {
                     isDirectory: isDirectory
                 )
                 let path = $0.span.extracting(first: pathLength)
-                return parseFinalFileSystemRepresentation(path: path, flags: &flags)
+                return parseFinalFileSystemRepresentation(path: path, flags: &flags, encodeSemicolons: encodeSemicolons)
             }
         }
     }
@@ -144,7 +145,7 @@ extension URL {
         #if FOUNDATION_FRAMEWORK
         #if !os(watchOS)
         if path.utf8Span.isKnownASCII {
-            return parseUTF8Path(path, flags: &flags, isDirectory: isDirectory)
+            return parseUTF8Path(path, flags: &flags, isDirectory: isDirectory, encodeSemicolons: false)
         }
         #endif
         // Convert path to its decomposed file system representation
@@ -163,7 +164,7 @@ extension URL {
                     isDirectory: isDirectory
                 )
                 let path = pathBuffer.span.extracting(first: finalLength)
-                return parseFinalFileSystemRepresentation(path: path, flags: &flags)
+                return parseFinalFileSystemRepresentation(path: path, flags: &flags, encodeSemicolons: false)
             }
 
             // Decomposition failed or there was an embedded null byte.
@@ -176,10 +177,10 @@ extension URL {
             // to a "file not found" error, this is more practical and debuggable
             // than returning an empty URL or crashing with fatalError().
 
-            return parseUTF8Path(path, flags: &flags, isDirectory: isDirectory)
+            return parseUTF8Path(path, flags: &flags, isDirectory: isDirectory, encodeSemicolons: false)
         }
         #else
-        return parseUTF8Path(path, flags: &flags, isDirectory: isDirectory)
+        return parseUTF8Path(path, flags: &flags, isDirectory: isDirectory, encodeSemicolons: false)
         #endif
     }
 
@@ -190,6 +191,6 @@ extension URL {
         guard !path.isEmpty else {
             return ""
         }
-        return parseUTF8Path(path, flags: &flags, isDirectory: isDirectory)
+        return parseUTF8Path(path, flags: &flags, isDirectory: isDirectory, encodeSemicolons: false)
     }
 }

Sources/FoundationEssentials/URL/URL_Info.swift

@@ -282,7 +282,7 @@ extension _URLInfo {
                 isDirectory: isDirectory
             )
             let path = buffer.span.extracting(first: finalLength)
-            return URL.parseFinalFileSystemRepresentation(path: path, flags: &flags)
+            return URL.parseFinalFileSystemRepresentation(path: path, flags: &flags, encodeSemicolons: false)
         }
 
         let base = resolveBaseURL(base, updating: &flags)

Tests/FoundationEssentialsTests/URLTests.swift

@@ -561,6 +561,22 @@ private struct URLTests {
         try FileManager.default.removeItem(at: URL(filePath: "\(tempDirectory.path)/tmp-dir"))
     }
 
+    @Test func filePathAPIsWithSemicolon() throws {
+        // The NSURL and CFURL file path APIs encode ";" in file paths
+        // for compatibility. URL and other modern parsers do not.
+        var url = URL(filePath: "/path;to/file")
+        #expect(url.path == "/path;to/file")
+        #expect(url.relativeString == "file:///path;to/file")
+
+        url.append(path: "hello;world")
+        #expect(url.path == "/path;to/file/hello;world")
+        #expect(url.relativeString == "file:///path;to/file/hello;world")
+
+        url.appendPathExtension("some;ext")
+        #expect(url.path == "/path;to/file/hello;world.some;ext")
+        #expect(url.relativeString == "file:///path;to/file/hello;world.some;ext")
+    }
+
     #if FOUNDATION_FRAMEWORK
     @Test func fileSystemRepresentations() throws {
         let base = "/base/"