[Tests] Add a test for correct usage of CP437 in ZIP

Author: tsolomko

SWCompression.xcodeproj/project.pbxproj

@@ -263,6 +263,7 @@
 		E6791E1B26FD0094003852A9 /* LZ4.swift in Sources */ = {isa = PBXBuildFile; fileRef = E6791E1A26FD0094003852A9 /* LZ4.swift */; };
 		E6791E3326FD05EC003852A9 /* LZ4Tests.swift in Sources */ = {isa = PBXBuildFile; fileRef = E6791E3226FD05EC003852A9 /* LZ4Tests.swift */; };
 		E6974C5B2701AC2600E06C60 /* test_dict_B5_dictID.lz4 in Resources */ = {isa = PBXBuildFile; fileRef = E6974C5A2701AC2600E06C60 /* test_dict_B5_dictID.lz4 */; };
+		E69FAC922729ACD900D3C406 /* test_dos_latin_us.zip in Resources */ = {isa = PBXBuildFile; fileRef = E69FAC912729ACD900D3C406 /* test_dos_latin_us.zip */; };
 		E6B18E85270884E300F9AB99 /* LZ4CompressionTests.swift in Sources */ = {isa = PBXBuildFile; fileRef = E6B18E84270884E300F9AB99 /* LZ4CompressionTests.swift */; };
 		E6C4150726FE230A00F9D36F /* XxHash32.swift in Sources */ = {isa = PBXBuildFile; fileRef = E6C4150626FE230A00F9D36F /* XxHash32.swift */; };
 		E6D86D2F26FE35C50032CFFA /* XxHash32Tests.swift in Sources */ = {isa = PBXBuildFile; fileRef = E6D86D2E26FE35C50032CFFA /* XxHash32Tests.swift */; };
@@ -541,6 +542,7 @@
 		E6791E1A26FD0094003852A9 /* LZ4.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = LZ4.swift; sourceTree = "<group>"; };
 		E6791E3226FD05EC003852A9 /* LZ4Tests.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = LZ4Tests.swift; sourceTree = "<group>"; };
 		E6974C5A2701AC2600E06C60 /* test_dict_B5_dictID.lz4 */ = {isa = PBXFileReference; lastKnownFileType = file; path = test_dict_B5_dictID.lz4; sourceTree = "<group>"; };
+		E69FAC912729ACD900D3C406 /* test_dos_latin_us.zip */ = {isa = PBXFileReference; lastKnownFileType = archive.zip; path = test_dos_latin_us.zip; sourceTree = "<group>"; };
 		E6B18E84270884E300F9AB99 /* LZ4CompressionTests.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = LZ4CompressionTests.swift; sourceTree = "<group>"; };
 		E6C4150626FE230A00F9D36F /* XxHash32.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = XxHash32.swift; sourceTree = "<group>"; };
 		E6D86D2E26FE35C50032CFFA /* XxHash32Tests.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = XxHash32Tests.swift; sourceTree = "<group>"; };
@@ -910,6 +912,7 @@
 				06F065DD1FFB763300312A82 /* test_zip64.zip */,
 				06D42DA42067B46800C1A98B /* test_custom_extra_field.zip */,
 				064DF452245618A300D285F3 /* bad_cd_ext_ts.zip */,
+				E69FAC912729ACD900D3C406 /* test_dos_latin_us.zip */,
 			);
 			path = ZIP;
 			sourceTree = "<group>";
@@ -1247,6 +1250,7 @@
 				06F066391FFB763400312A82 /* test8.xz in Resources */,
 				06F0662F1FFB763400312A82 /* test9.gz in Resources */,
 				064DF453245618A300D285F3 /* bad_cd_ext_ts.zip in Resources */,
+				E69FAC922729ACD900D3C406 /* test_dos_latin_us.zip in Resources */,
 				E66F36242538726E00076A6E /* test_empty.lzma in Resources */,
 				06F066751FFB763400312A82 /* test1.bz2 in Resources */,
 				064D01A920FD071300CAE058 /* текстовый файл.answer in Resources */,

Sources/ZIP/LittleEndianByteReader+Zip.swift

@@ -28,6 +28,12 @@ extension LittleEndianByteReader {
 fileprivate extension String {
 
     // "Latin-US (DOS)" CP437-2147483120
+    // Note that this encoding is interpreted differently "depending on the context". This concerns codes in 0-31 range
+    // and code 127: sometimes they treated as normal characters (as was defined by IBM) and sometimes as control
+    // characters, matching corresponding ranges of ASCII/UTF-8. The second interpretation, however, doesn't make CP437
+    // compatible with UTF-8, as CP437 contains codes in the 80-FF range which are not valid UTF-8 codes.
+    // In any case, we are constrained to the implementation provided by Foundation, which currently treats them as
+    // control characters.
     static let cp437Encoding = String.Encoding(rawValue: 0x80000400)
     static let cp437Available = String.availableStringEncodings.contains(cp437Encoding)
 

Tests/Test Files

@@ -288,4 +288,32 @@ class ZipTests: XCTestCase {
         XCTAssertEqual(entries[1].data, answerData)
     }
 
+    func testDosLatinUS() throws {
+        // This test checks that CP437 encoding is correctly used when there is no indication the file name is in UTF-8.
+        // We introduced several CP437-specific characters from the 0x80-0xFF range into "test1.answer" to test this.
+        // Note, that we didn't used normal characters from the 0x00-0x7F range that don't match the characters from
+        // UTF-8 with the same codes, since they are interpreted as control characters by Foundation.
+        let testData = try Constants.data(forTest: "test_dos_latin_us", withType: ZipTests.testType)
+        let entries = try ZipContainer.open(container: testData)
+
+        XCTAssertEqual(entries.count, 1)
+        XCTAssertEqual(entries[0].info.name, "teüë1.½n█wΩ±")
+        XCTAssertEqual(entries[0].info.type, .regular)
+        XCTAssertEqual(entries[0].info.fileSystemType, .unix)
+        XCTAssertFalse(entries[0].info.isTextFile)
+        XCTAssertEqual(entries[0].info.compressionMethod, .copy)
+        XCTAssertEqual(entries[0].info.ownerID, 501)
+        XCTAssertEqual(entries[0].info.groupID, 20)
+        XCTAssertEqual(entries[0].info.permissions, Permissions(rawValue: 420))
+        XCTAssertEqual(entries[0].info.dosAttributes, DosAttributes(rawValue: 0))
+        XCTAssertEqual(entries[0].info.comment, "")
+        // Checking times' values is a bit difficult since they are extremely precise.
+        XCTAssertNotNil(entries[0].info.modificationTime)
+        XCTAssertNotNil(entries[0].info.accessTime)
+        XCTAssertNil(entries[0].info.creationTime)
+
+        let answerData = try Constants.data(forAnswer: "test1")
+        XCTAssertEqual(entries[0].data, answerData)
+    }
+
 }