#region Copyright notice and license // Protocol Buffers - Google's data interchange format // Copyright 2008 Google Inc. All rights reserved. // http://github.com/jskeet/dotnet-protobufs/ // Original C++/Java/Python code: // http://code.google.com/p/protobuf/ // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. #endregion using System; using System.IO; using Google.ProtocolBuffers.TestProtos; using NUnit.Framework; using System.Diagnostics; namespace Google.ProtocolBuffers { [TestFixture] public class CodedInputStreamTest { /// /// Helper to construct a byte array from a bunch of bytes. The inputs are /// actually ints so that I can use hex notation and not get stupid errors /// about precision. /// private static byte[] Bytes(params int[] bytesAsInts) { byte[] bytes = new byte[bytesAsInts.Length]; for (int i = 0; i < bytesAsInts.Length; i++) { bytes[i] = (byte) bytesAsInts[i]; } return bytes; } /// /// Parses the given bytes using ReadRawVarint32() and ReadRawVarint64() and /// private static void AssertReadVarint(byte[] data, ulong value) { CodedInputStream input = CodedInputStream.CreateInstance(data); Assert.AreEqual((uint) value, input.ReadRawVarint32()); input = CodedInputStream.CreateInstance(data); Assert.AreEqual(value, input.ReadRawVarint64()); Assert.IsTrue(input.IsAtEnd); // Try different block sizes. for (int bufferSize = 1; bufferSize <= 16; bufferSize *= 2) { input = CodedInputStream.CreateInstance(new SmallBlockInputStream(data, bufferSize)); Assert.AreEqual((uint) value, input.ReadRawVarint32()); input = CodedInputStream.CreateInstance(new SmallBlockInputStream(data, bufferSize)); Assert.AreEqual(value, input.ReadRawVarint64()); Assert.IsTrue(input.IsAtEnd); } // Try reading directly from a MemoryStream. We want to verify that it // doesn't read past the end of the input, so write an extra byte - this // lets us test the position at the end. MemoryStream memoryStream = new MemoryStream(); memoryStream.Write(data, 0, data.Length); memoryStream.WriteByte(0); memoryStream.Position = 0; Assert.AreEqual((uint) value, CodedInputStream.ReadRawVarint32(memoryStream)); Assert.AreEqual(data.Length, memoryStream.Position); } /// /// Parses the given bytes using ReadRawVarint32() and ReadRawVarint64() and /// expects them to fail with an InvalidProtocolBufferException whose /// description matches the given one. /// private static void AssertReadVarintFailure(InvalidProtocolBufferException expected, byte[] data) { CodedInputStream input = CodedInputStream.CreateInstance(data); try { input.ReadRawVarint32(); Assert.Fail("Should have thrown an exception."); } catch (InvalidProtocolBufferException e) { Assert.AreEqual(expected.Message, e.Message); } input = CodedInputStream.CreateInstance(data); try { input.ReadRawVarint64(); Assert.Fail("Should have thrown an exception."); } catch (InvalidProtocolBufferException e) { Assert.AreEqual(expected.Message, e.Message); } // Make sure we get the same error when reading directly from a Stream. try { CodedInputStream.ReadRawVarint32(new MemoryStream(data)); Assert.Fail("Should have thrown an exception."); } catch (InvalidProtocolBufferException e) { Assert.AreEqual(expected.Message, e.Message); } } [Test] public void ReadVarint() { AssertReadVarint(Bytes(0x00), 0); AssertReadVarint(Bytes(0x01), 1); AssertReadVarint(Bytes(0x7f), 127); // 14882 AssertReadVarint(Bytes(0xa2, 0x74), (0x22 << 0) | (0x74 << 7)); // 2961488830 AssertReadVarint(Bytes(0xbe, 0xf7, 0x92, 0x84, 0x0b), (0x3e << 0) | (0x77 << 7) | (0x12 << 14) | (0x04 << 21) | (0x0bL << 28)); // 64-bit // 7256456126 AssertReadVarint(Bytes(0xbe, 0xf7, 0x92, 0x84, 0x1b), (0x3e << 0) | (0x77 << 7) | (0x12 << 14) | (0x04 << 21) | (0x1bL << 28)); // 41256202580718336 AssertReadVarint(Bytes(0x80, 0xe6, 0xeb, 0x9c, 0xc3, 0xc9, 0xa4, 0x49), (0x00 << 0) | (0x66 << 7) | (0x6b << 14) | (0x1c << 21) | (0x43L << 28) | (0x49L << 35) | (0x24L << 42) | (0x49L << 49)); // 11964378330978735131 AssertReadVarint(Bytes(0x9b, 0xa8, 0xf9, 0xc2, 0xbb, 0xd6, 0x80, 0x85, 0xa6, 0x01), (0x1b << 0) | (0x28 << 7) | (0x79 << 14) | (0x42 << 21) | (0x3bUL << 28) | (0x56UL << 35) | (0x00UL << 42) | (0x05UL << 49) | (0x26UL << 56) | (0x01UL << 63)); // Failures AssertReadVarintFailure( InvalidProtocolBufferException.MalformedVarint(), Bytes(0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x00)); AssertReadVarintFailure( InvalidProtocolBufferException.TruncatedMessage(), Bytes(0x80)); } /// /// Parses the given bytes using ReadRawLittleEndian32() and checks /// that the result matches the given value. /// private static void AssertReadLittleEndian32(byte[] data, uint value) { CodedInputStream input = CodedInputStream.CreateInstance(data); Assert.AreEqual(value, input.ReadRawLittleEndian32()); Assert.IsTrue(input.IsAtEnd); // Try different block sizes. for (int blockSize = 1; blockSize <= 16; blockSize *= 2) { input = CodedInputStream.CreateInstance( new SmallBlockInputStream(data, blockSize)); Assert.AreEqual(value, input.ReadRawLittleEndian32()); Assert.IsTrue(input.IsAtEnd); } } /// /// Parses the given bytes using ReadRawLittleEndian64() and checks /// that the result matches the given value. /// private static void AssertReadLittleEndian64(byte[] data, ulong value) { CodedInputStream input = CodedInputStream.CreateInstance(data); Assert.AreEqual(value, input.ReadRawLittleEndian64()); Assert.IsTrue(input.IsAtEnd); // Try different block sizes. for (int blockSize = 1; blockSize <= 16; blockSize *= 2) { input = CodedInputStream.CreateInstance( new SmallBlockInputStream(data, blockSize)); Assert.AreEqual(value, input.ReadRawLittleEndian64()); Assert.IsTrue(input.IsAtEnd); } } [Test] public void ReadLittleEndian() { AssertReadLittleEndian32(Bytes(0x78, 0x56, 0x34, 0x12), 0x12345678); AssertReadLittleEndian32(Bytes(0xf0, 0xde, 0xbc, 0x9a), 0x9abcdef0); AssertReadLittleEndian64(Bytes(0xf0, 0xde, 0xbc, 0x9a, 0x78, 0x56, 0x34, 0x12), 0x123456789abcdef0L); AssertReadLittleEndian64( Bytes(0x78, 0x56, 0x34, 0x12, 0xf0, 0xde, 0xbc, 0x9a), 0x9abcdef012345678UL); } [Test] public void DecodeZigZag32() { Assert.AreEqual(0, CodedInputStream.DecodeZigZag32(0)); Assert.AreEqual(-1, CodedInputStream.DecodeZigZag32(1)); Assert.AreEqual(1, CodedInputStream.DecodeZigZag32(2)); Assert.AreEqual(-2, CodedInputStream.DecodeZigZag32(3)); Assert.AreEqual(0x3FFFFFFF, CodedInputStream.DecodeZigZag32(0x7FFFFFFE)); Assert.AreEqual(unchecked((int) 0xC0000000), CodedInputStream.DecodeZigZag32(0x7FFFFFFF)); Assert.AreEqual(0x7FFFFFFF, CodedInputStream.DecodeZigZag32(0xFFFFFFFE)); Assert.AreEqual(unchecked((int) 0x80000000), CodedInputStream.DecodeZigZag32(0xFFFFFFFF)); } [Test] public void DecodeZigZag64() { Assert.AreEqual(0, CodedInputStream.DecodeZigZag64(0)); Assert.AreEqual(-1, CodedInputStream.DecodeZigZag64(1)); Assert.AreEqual(1, CodedInputStream.DecodeZigZag64(2)); Assert.AreEqual(-2, CodedInputStream.DecodeZigZag64(3)); Assert.AreEqual(0x000000003FFFFFFFL, CodedInputStream.DecodeZigZag64(0x000000007FFFFFFEL)); Assert.AreEqual(unchecked((long) 0xFFFFFFFFC0000000L), CodedInputStream.DecodeZigZag64(0x000000007FFFFFFFL)); Assert.AreEqual(0x000000007FFFFFFFL, CodedInputStream.DecodeZigZag64(0x00000000FFFFFFFEL)); Assert.AreEqual(unchecked((long) 0xFFFFFFFF80000000L), CodedInputStream.DecodeZigZag64(0x00000000FFFFFFFFL)); Assert.AreEqual(0x7FFFFFFFFFFFFFFFL, CodedInputStream.DecodeZigZag64(0xFFFFFFFFFFFFFFFEL)); Assert.AreEqual(unchecked((long) 0x8000000000000000L), CodedInputStream.DecodeZigZag64(0xFFFFFFFFFFFFFFFFL)); } [Test] public void ReadWholeMessage() { TestAllTypes message = TestUtil.GetAllSet(); byte[] rawBytes = message.ToByteArray(); Assert.AreEqual(rawBytes.Length, message.SerializedSize); TestAllTypes message2 = TestAllTypes.ParseFrom(rawBytes); TestUtil.AssertAllFieldsSet(message2); // Try different block sizes. for (int blockSize = 1; blockSize < 256; blockSize *= 2) { message2 = TestAllTypes.ParseFrom(new SmallBlockInputStream(rawBytes, blockSize)); TestUtil.AssertAllFieldsSet(message2); } } [Test] public void SkipWholeMessage() { TestAllTypes message = TestUtil.GetAllSet(); byte[] rawBytes = message.ToByteArray(); // Create two parallel inputs. Parse one as unknown fields while using // skipField() to skip each field on the other. Expect the same tags. CodedInputStream input1 = CodedInputStream.CreateInstance(rawBytes); CodedInputStream input2 = CodedInputStream.CreateInstance(rawBytes); UnknownFieldSet.Builder unknownFields = UnknownFieldSet.CreateBuilder(); uint tag; string name; while (input1.ReadTag(out tag, out name)) { uint tag2; Assert.IsTrue(input2.ReadTag(out tag2, out name)); Assert.AreEqual(tag, tag2); unknownFields.MergeFieldFrom(tag, input1); input2.SkipField(); } } /// /// Test that a bug in SkipRawBytes has been fixed: if the skip /// skips exactly up to a limit, this should bnot break things /// [Test] public void SkipRawBytesBug() { byte[] rawBytes = new byte[] {1, 2}; CodedInputStream input = CodedInputStream.CreateInstance(rawBytes); int limit = input.PushLimit(1); input.SkipRawBytes(1); input.PopLimit(limit); Assert.AreEqual(2, input.ReadRawByte()); } public void ReadHugeBlob() { // Allocate and initialize a 1MB blob. byte[] blob = new byte[1 << 20]; for (int i = 0; i < blob.Length; i++) { blob[i] = (byte) i; } // Make a message containing it. TestAllTypes.Builder builder = TestAllTypes.CreateBuilder(); TestUtil.SetAllFields(builder); builder.SetOptionalBytes(ByteString.CopyFrom(blob)); TestAllTypes message = builder.Build(); // Serialize and parse it. Make sure to parse from an InputStream, not // directly from a ByteString, so that CodedInputStream uses buffered // reading. TestAllTypes message2 = TestAllTypes.ParseFrom(message.ToByteString().CreateCodedInput()); Assert.AreEqual(message.OptionalBytes, message2.OptionalBytes); // Make sure all the other fields were parsed correctly. TestAllTypes message3 = TestAllTypes.CreateBuilder(message2) .SetOptionalBytes(TestUtil.GetAllSet().OptionalBytes) .Build(); TestUtil.AssertAllFieldsSet(message3); } [Test] public void ReadMaliciouslyLargeBlob() { MemoryStream ms = new MemoryStream(); CodedOutputStream output = CodedOutputStream.CreateInstance(ms); uint tag = WireFormat.MakeTag(1, WireFormat.WireType.LengthDelimited); output.WriteRawVarint32(tag); output.WriteRawVarint32(0x7FFFFFFF); output.WriteRawBytes(new byte[32]); // Pad with a few random bytes. output.Flush(); ms.Position = 0; CodedInputStream input = CodedInputStream.CreateInstance(ms); uint testtag; string ignore; Assert.IsTrue(input.ReadTag(out testtag, out ignore)); Assert.AreEqual(tag, testtag); try { ByteString bytes = null; input.ReadBytes(ref bytes); Assert.Fail("Should have thrown an exception!"); } catch (InvalidProtocolBufferException) { // success. } } private static TestRecursiveMessage MakeRecursiveMessage(int depth) { if (depth == 0) { return TestRecursiveMessage.CreateBuilder().SetI(5).Build(); } else { return TestRecursiveMessage.CreateBuilder() .SetA(MakeRecursiveMessage(depth - 1)).Build(); } } private static void AssertMessageDepth(TestRecursiveMessage message, int depth) { if (depth == 0) { Assert.IsFalse(message.HasA); Assert.AreEqual(5, message.I); } else { Assert.IsTrue(message.HasA); AssertMessageDepth(message.A, depth - 1); } } [Test] public void MaliciousRecursion() { ByteString data64 = MakeRecursiveMessage(64).ToByteString(); ByteString data65 = MakeRecursiveMessage(65).ToByteString(); AssertMessageDepth(TestRecursiveMessage.ParseFrom(data64), 64); try { TestRecursiveMessage.ParseFrom(data65); Assert.Fail("Should have thrown an exception!"); } catch (InvalidProtocolBufferException) { // success. } CodedInputStream input = data64.CreateCodedInput(); input.SetRecursionLimit(8); try { TestRecursiveMessage.ParseFrom(input); Assert.Fail("Should have thrown an exception!"); } catch (InvalidProtocolBufferException) { // success. } } [Test] public void SizeLimit() { // Have to use a Stream rather than ByteString.CreateCodedInput as SizeLimit doesn't // apply to the latter case. MemoryStream ms = new MemoryStream(TestUtil.GetAllSet().ToByteString().ToByteArray()); CodedInputStream input = CodedInputStream.CreateInstance(ms); input.SetSizeLimit(16); try { TestAllTypes.ParseFrom(input); Assert.Fail("Should have thrown an exception!"); } catch (InvalidProtocolBufferException) { // success. } } [Test] public void ResetSizeCounter() { CodedInputStream input = CodedInputStream.CreateInstance( new SmallBlockInputStream(new byte[256], 8)); input.SetSizeLimit(16); input.ReadRawBytes(16); try { input.ReadRawByte(); Assert.Fail("Should have thrown an exception!"); } catch (InvalidProtocolBufferException) { // Success. } input.ResetSizeCounter(); input.ReadRawByte(); // No exception thrown. try { input.ReadRawBytes(16); // Hits limit again. Assert.Fail("Should have thrown an exception!"); } catch (InvalidProtocolBufferException) { // Success. } } /// /// Tests that if we read an string that contains invalid UTF-8, no exception /// is thrown. Instead, the invalid bytes are replaced with the Unicode /// "replacement character" U+FFFD. /// [Test] public void ReadInvalidUtf8() { MemoryStream ms = new MemoryStream(); CodedOutputStream output = CodedOutputStream.CreateInstance(ms); uint tag = WireFormat.MakeTag(1, WireFormat.WireType.LengthDelimited); output.WriteRawVarint32(tag); output.WriteRawVarint32(1); output.WriteRawBytes(new byte[] {0x80}); output.Flush(); ms.Position = 0; CodedInputStream input = CodedInputStream.CreateInstance(ms); uint testtag; string ignored; Assert.IsTrue(input.ReadTag(out testtag, out ignored)); Assert.AreEqual(tag, testtag); string text = null; input.ReadString(ref text); Assert.AreEqual('\ufffd', text[0]); } /// /// A stream which limits the number of bytes it reads at a time. /// We use this to make sure that CodedInputStream doesn't screw up when /// reading in small blocks. /// private sealed class SmallBlockInputStream : MemoryStream { private readonly int blockSize; public SmallBlockInputStream(byte[] data, int blockSize) : base(data) { this.blockSize = blockSize; } public override int Read(byte[] buffer, int offset, int count) { return base.Read(buffer, offset, Math.Min(count, blockSize)); } } } }