Merge branch 'feature/extra_blank_lines' into develop

Author: Jericho

Source/HttpMultipartParser.UnitTests/ParserScenarios/BlankLinesBeforeFirstBoundary.cs

@@ -0,0 +1,64 @@
+using System.Collections.Generic;
+using System.IO;
+using System.Text;
+using System.Threading.Tasks;
+using Xunit;
+
+namespace HttpMultipartParser.UnitTests.ParserScenarios
+{
+	public class BlankLinesBeforeFirstBoundary
+	{
+		private static readonly string _testData = TestUtil.TrimAllLines(
+			@"--boundary
+            Content-Disposition: form-data; name=""text""
+
+            textdata
+            --boundary--"
+		);
+
+		// This test case has a few blank lines before the first boundary marker
+		// This unusual scenario is described in GH-116
+		// https://github.com/Http-Multipart-Data-Parser/Http-Multipart-Data-Parser/issues/116
+		private static readonly TestData _testCase = new TestData(
+			$"\n\n\n{_testData}", // Intentionally add a few blank lines before the data. These blank lines should be ignored by the parser when attempting to detect the boundary marker
+			new List<ParameterPart> {
+				new ParameterPart("text", "textdata"),
+			},
+			new List<FilePart>()
+		);
+
+		public BlankLinesBeforeFirstBoundary()
+		{
+			foreach (var filePart in _testCase.ExpectedFileData)
+			{
+				filePart.Data.Position = 0;
+			}
+		}
+
+		/// <summary>
+		///     Tests for correct detection of the boundary in the input stream.
+		/// </summary>
+		[Fact]
+		public void CanAutoDetectBoundary()
+		{
+			using (Stream stream = TestUtil.StringToStream(_testCase.Request, Encoding.UTF8))
+			{
+				var parser = MultipartFormDataParser.Parse(stream);
+				Assert.True(_testCase.Validate(parser));
+			}
+		}
+
+		/// <summary>
+		///     Tests for correct detection of the boundary in the input stream.
+		/// </summary>
+		[Fact]
+		public async Task CanAutoDetectBoundaryAsync()
+		{
+			using (Stream stream = TestUtil.StringToStream(_testCase.Request, Encoding.UTF8))
+			{
+				var parser = await MultipartFormDataParser.ParseAsync(stream, Encoding.UTF8).ConfigureAwait(false);
+				Assert.True(_testCase.Validate(parser));
+			}
+		}
+	}
+}

Source/HttpMultipartParser/StreamingBinaryMultipartFormDataParser.cs

@@ -244,6 +244,11 @@ public async Task RunAsync(CancellationToken cancellationToken = default)
 		///     current position of the reader is the start of the file and therefore
 		///     the beginning of the boundary.
 		/// </summary>
+		/// <remarks>
+		/// As of version 8.2.0 (released in June 2023), we ignore blank lines at the
+		/// start of the stream. In previous version, an exception was thrown if any
+		/// blank line was present before the boundary marker.
+		/// </remarks>
 		/// <param name="reader">
 		///     The binary reader to parse.
 		/// </param>
@@ -254,7 +259,16 @@ private static string DetectBoundary(RebufferableBinaryReader reader)
 		{
 			// Presumably the boundary is --|||||||||||||| where -- is the stuff added on to
 			// the front as per the protocol and ||||||||||||| is the part we care about.
-			var line = reader.ReadLine();
+
+			// The following loop ignores blank lines that may be present before the first line of the form.
+			// It's highly unusual to find blank lines at the start of the data but it's a possible scenario described in GH-116.
+			// Please note that we intentionally do NOT check for "string.IsNullOrEmpty(line)" because NULL does
+			// not indicate a blank line. It indicates that we have reached the end of the stream.
+			var line = string.Empty;
+			while (line == string.Empty)
+			{
+				line = reader.ReadLine();
+			}
 
 			// The line must not be empty and must starts with "--".
 			if (string.IsNullOrEmpty(line)) throw new MultipartParseException("Unable to determine boundary: either the stream is empty or we reached the end of the stream");
@@ -283,6 +297,11 @@ private static string DetectBoundary(RebufferableBinaryReader reader)
 		///     current position of the reader is the start of the file and therefore
 		///     the beginning of the boundary.
 		/// </summary>
+		/// <remarks>
+		/// As of version 8.2.0 (released in June 2023), we ignore blank lines at the
+		/// start of the stream. In previous version, an exception was thrown if any
+		/// blank line was present before the boundary marker.
+		/// </remarks>
 		/// <param name="reader">
 		///     The binary reader to parse.
 		/// </param>
@@ -296,7 +315,16 @@ private static async Task<string> DetectBoundaryAsync(RebufferableBinaryReader r
 		{
 			// Presumably the boundary is --|||||||||||||| where -- is the stuff added on to
 			// the front as per the protocol and ||||||||||||| is the part we care about.
-			var line = await reader.ReadLineAsync(cancellationToken).ConfigureAwait(false);
+
+			// The following loop ignores blank lines that may be present before the first line of the form.
+			// It's highly unusual to find blank lines at the start of the data but it's a possible scenario described in GH-116.
+			// Please note that we intentionally do NOT check for "string.IsNullOrEmpty(line)" because NULL does
+			// not indicate a blank line. It indicates that we have reached the end of the stream.
+			var line = string.Empty;
+			while (line == string.Empty)
+			{
+				line = await reader.ReadLineAsync(cancellationToken).ConfigureAwait(false);
+			}
 
 			// The line must not be empty and must starts with "--".
 			if (string.IsNullOrEmpty(line)) throw new MultipartParseException("Unable to determine boundary: either the stream is empty or we reached the end of the stream");