Merge pull request #1836 from allevato/module-alias-raw-identifiers

Allow module aliases to be raw identifiers.

Author: allevato

Sources/SwiftDriver/Driver/Driver.swift

@@ -2977,16 +2977,18 @@ extension Driver {
                                      with moduleName: String,
                                      onError diagnosticsEngine: DiagnosticsEngine) -> [String: String]? {
     var moduleAliases: [String: String]? = nil
-    let validate = { (_ arg: String, allowModuleName: Bool) -> Bool in
-      if !arg.sd_isSwiftIdentifier {
+    // validatingModuleName should be true when validating the alias target (an actual module
+    // name), or false when validating the alias name (which can be a raw identifier).
+    let validate = { (_ arg: String, validatingModuleName: Bool) -> Bool in
+      if (validatingModuleName && !arg.sd_isSwiftIdentifier) || !arg.sd_isValidAsRawIdentifier {
         diagnosticsEngine.emit(.error_bad_module_name(moduleName: arg, explicitModuleName: true))
         return false
       }
       if arg == "Swift" {
         diagnosticsEngine.emit(.error_stdlib_module_name(moduleName: arg, explicitModuleName: true))
         return false
       }
-      if !allowModuleName, arg == moduleName {
+      if !validatingModuleName, arg == moduleName {
         diagnosticsEngine.emit(.error_bad_module_alias(arg, moduleName: moduleName))
         return false
       }

Sources/SwiftDriver/Utilities/StringAdditions.swift

@@ -21,6 +21,15 @@ extension String {
     return start.isValidSwiftIdentifierStart &&
            continuation.allSatisfy { $0.isValidSwiftIdentifierContinuation }
   }
+
+  /// Whether this string is a valid Swift raw identifier.
+  public var sd_isValidAsRawIdentifier: Bool {
+    guard !isEmpty else {
+      return false
+    }
+    return unicodeScalars.allSatisfy { $0.isValidRawIdentifierScalar }
+           && !unicodeScalars.allSatisfy { $0.isPermittedRawIdentifierWhitespace }
+  }
 }
 
 extension DefaultStringInterpolation {
@@ -120,6 +129,43 @@ extension Unicode.Scalar {
   /// `true` if this character is an ASCII digit: [0-9]
   var isASCIIDigit: Bool { (0x30...0x39).contains(value) }
 
+  /// `true` if this scalar is valid in a Swift raw identifier.
+  var isValidRawIdentifierScalar: Bool {
+    // A raw identifier is terminated by a backtick, and the backslash is reserved for possible
+    // future escaping.
+    if self == "`" || self == "\\" {
+      return false
+    }
+    // Unprintable ASCII control characters are forbidden.
+    if (0x0000...0x001F).contains(value) || value == 0x007F {
+      return false;
+    }
+    return !isForbiddenRawIdentifierWhitespace
+  }
+
+  var isForbiddenRawIdentifierWhitespace: Bool {
+    // This is the set of code points satisfying the `White_Space` property, excluding the set
+    // satisfying the `Pattern_White_Space` property, and excluding any other ASCII non-printables
+    // and Unicode separators. In other words, the only whitespace code points allowed in a raw
+    // identifier are U+0020, and U+200E/200F (LTR/RTL marks) (see
+    // `isPermittedRawIdentifierWhitespace` below).
+    return (0x0009...0x000D).contains(value) ||
+           value == 0x0085 ||
+           value == 0x00A0 ||
+           value == 0x1680 ||
+           (0x2000...0x200A).contains(value) ||
+           (0x2028...0x2029).contains(value) ||
+           value == 0x202F ||
+           value == 0x205F ||
+           value == 0x3000
+  }
+
+  var isPermittedRawIdentifierWhitespace: Bool {
+    return value == 0x0020 ||
+           value == 0x200E ||
+           value == 0x200F
+  }
+
   /// `true` if this is a body character of a C identifier,
   /// which is [a-zA-Z0-9_].
   func isCIdentifierBody(allowDollar: Bool = false) -> Bool {

Tests/SwiftDriverTests/StringAdditionsTests.swift

@@ -76,4 +76,16 @@ final class StringAdditionsTests: XCTestCase {
         XCTAssertFalse("".sd_isSwiftIdentifier,
                        "Private-use characters aren't valid Swift identifiers")
     }
+
+    func testRawIdentifiers() {
+        XCTAssertTrue("plain".sd_isValidAsRawIdentifier)
+        XCTAssertTrue("has spaces".sd_isValidAsRawIdentifier)
+        XCTAssertTrue("$^has/other!characters@#".sd_isValidAsRawIdentifier)
+
+        XCTAssertFalse("has`backtick".sd_isValidAsRawIdentifier)
+        XCTAssertFalse("has\\backslash".sd_isValidAsRawIdentifier)
+        XCTAssertFalse("has\u{0000}control\u{007F}characters".sd_isValidAsRawIdentifier)
+        XCTAssertFalse("has\u{00A0}forbidden\u{2028}whitespace".sd_isValidAsRawIdentifier)
+        XCTAssertFalse(" ".sd_isValidAsRawIdentifier)
+    }
 }

Tests/SwiftDriverTests/SwiftDriverTests.swift

@@ -3360,11 +3360,22 @@ final class SwiftDriverTests: XCTestCase {
       $1.expect(.error("module alias \"Foo\" should be different from the module name \"Foo\""))
     }
 
+    // A module alias is allowed to be a valid raw identifier, not just a regular Swift identifier.
+    try assertNoDriverDiagnostics(
+      args: "swiftc", "foo.swift", "-module-name", "Foo", "-module-alias", "//car/far:par=Bar", "-emit-module", "-emit-module-path", "/tmp/dir/Foo.swiftmodule"
+    )
+    // The alias target (an actual module name), however, may not be a raw identifier.
     try assertDriverDiagnostics(
-      args: ["swiftc", "foo.swift", "-module-name", "Foo", "-module-alias", "C-ar=Bar", "-emit-module", "-emit-module-path", "/tmp/dir/Foo.swiftmodule"]
+      args: ["swiftc", "foo.swift", "-module-name", "Foo", "-module-alias", "Bar=C-ar", "-emit-module", "-emit-module-path", "/tmp/dir/Foo.swiftmodule"]
     ) {
       $1.expect(.error("module name \"C-ar\" is not a valid identifier"))
     }
+    // We should still diagnose names that are not valid raw identifiers.
+    try assertDriverDiagnostics(
+      args: ["swiftc", "foo.swift", "-module-name", "Foo", "-module-alias", "C`ar=Bar", "-emit-module", "-emit-module-path", "/tmp/dir/Foo.swiftmodule"]
+    ) {
+      $1.expect(.error("module name \"C`ar\" is not a valid identifier"))
+    }
 
     try assertDriverDiagnostics(
       args: ["swiftc", "foo.swift", "-module-name", "Foo", "-module-alias", "Car=Bar", "-module-alias", "Train=Car", "-emit-module", "-emit-module-path", "/tmp/dir/Foo.swiftmodule"]