From a055be2185447ca7f03be2c18ac8433c0b3b2167 Mon Sep 17 00:00:00 2001
From: Ronny Gunawan <3048897+ronnygunawan@users.noreply.github.com>
Date: Sat, 18 Nov 2023 22:01:08 +0700
Subject: [PATCH] Throw correct exception for incorrect number of columns

---
 CsvSerializer/CsvSerializer.csproj            |  2 +-
 .../Internals/NativeImplGenerator.cs          | 31 +++++++++++--------
 CsvSerializer/SourceGenerator.cs              |  4 ++-
 Tests/SerializerTests.cs                      | 24 ++++++++++++--
 4 files changed, 44 insertions(+), 17 deletions(-)

diff --git a/CsvSerializer/CsvSerializer.csproj b/CsvSerializer/CsvSerializer.csproj
index a9965b1..1077686 100644
--- a/CsvSerializer/CsvSerializer.csproj
+++ b/CsvSerializer/CsvSerializer.csproj
@@ -15,7 +15,7 @@
     <RepositoryUrl>https://github.com/ronnygunawan/csv-serializer</RepositoryUrl>
     <PackageTags>csv serializer deserializer parser core</PackageTags>
     <Description>Fast CSV to object serializer and deserializer.</Description>
-    <Version>2.0.3</Version>
+    <Version>2.0.4</Version>
     <PackageReleaseNotes></PackageReleaseNotes>
     <EnforceExtendedAnalyzerRules>true</EnforceExtendedAnalyzerRules>
     <IsRoslynComponent>true</IsRoslynComponent>
diff --git a/CsvSerializer/Internals/NativeImplGenerator.cs b/CsvSerializer/Internals/NativeImplGenerator.cs
index 7d1b873..10767a4 100644
--- a/CsvSerializer/Internals/NativeImplGenerator.cs
+++ b/CsvSerializer/Internals/NativeImplGenerator.cs
@@ -1,5 +1,4 @@
-using System;
-using System.Collections.Generic;
+using System.Collections.Generic;
 using System.Linq;
 using System.Text;
 using Microsoft.CodeAnalysis;
@@ -92,7 +91,7 @@ private enum ParserState {
 								InTrailingWhiteSpace
 							}
 
-							public static void ReadNextLine(ref ReadOnlyMemory<char> csv, ref Span<ReadOnlyMemory<char>> columns, char separator = ',') {
+							public static int ReadNextLine(ref ReadOnlyMemory<char> csv, ref Span<ReadOnlyMemory<char>> columns, char separator = ',') {
 								ReadOnlySpan<char> span = csv.Span;
 								int startOfLiteral = 0;
 								int endOfLiteral = 0;
@@ -106,13 +105,13 @@ public static void ReadNextLine(ref ReadOnlyMemory<char> csv, ref Span<ReadOnlyM
 											case ParserState.InEscapeSequence:
 												columns[col] = csv[startOfLiteral..i];
 												csv = csv.Slice(csv.Length - 1, 0);
-												return;
+												return col + 1;
 											case ParserState.InQuotedValue:
 												throw new CsvFormatException(csv.ToString(), "End of file in quoted literal.");
 											case ParserState.InTrailingWhiteSpace:
 												columns[col] = csv.Slice(startOfLiteral, endOfLiteral - startOfLiteral + 1);
 												csv = csv.Slice(csv.Length - 1, 0);
-												return;
+												return col + 1;
 										}
 									} else {
 										switch (span[i]) {
@@ -161,11 +160,11 @@ public static void ReadNextLine(ref ReadOnlyMemory<char> csv, ref Span<ReadOnlyM
 													case ParserState.InEscapeSequence:
 														columns[col] = csv[startOfLiteral..i];
 														csv = csv[(i + 1)..];
-														return;
+														return col + 1;
 													case ParserState.InTrailingWhiteSpace:
 														columns[col] = csv.Slice(startOfLiteral, endOfLiteral - startOfLiteral + 1);
 														csv = csv[(i + 1)..];
-														return;
+														return col + 1;
 												}
 												break;
 											case char c:
@@ -1419,18 +1418,24 @@ public List<object> Deserialize(IFormatProvider? provider, char delimiter, bool
 								Span<ReadOnlyMemory<char>> columns = new ReadOnlyMemory<char>[{{propertySymbols.Count}}];
 								bool firstRow = true;
 								while (csv.Length > 0) {
-									NativeStringSplitter.ReadNextLine(ref csv, ref columns, delimiter);
+									try {
+										int columnsRead = NativeStringSplitter.ReadNextLine(ref csv, ref columns, delimiter);
+										if (columnsRead != {{propertySymbols.Count}}) {
+											int endOfLine = csv.Span.IndexOf('\n');
+											string line = endOfLine == -1 ? csv.ToString() : csv.Slice(0, endOfLine).ToString();
+											throw new CsvFormatException(typeof({{fullTypeName}}), line, "Row must consists of {{propertySymbols.Count}} columns.");
+										}
+									} catch (IndexOutOfRangeException) {
+										int endOfLine = csv.Span.IndexOf('\n');
+										string line = endOfLine == -1 ? csv.ToString() : csv.Slice(0, endOfLine).ToString();
+										throw new CsvFormatException(typeof({{fullTypeName}}), line, "Row must consists of {{propertySymbols.Count}} columns.");
+									}
 									if (firstRow) {
 										firstRow = false;
 										if (skipHeader) {
 											continue;
 										}
 									}
-									if (columns.Length != {{propertySymbols.Count}}) {
-										int endOfLine = csv.Span.IndexOf('\n');
-										string line = endOfLine == -1 ? csv.ToString() : csv.Slice(0, endOfLine).ToString();
-										throw new CsvFormatException(typeof({{fullTypeName}}), line, "Row must consists of {{propertySymbols.Count}} columns.");
-									}
 									{{fullTypeName}} item = Activator.CreateInstance<{{fullTypeName}}>();
 									if (provider is null) {
 										{{deserializeWithoutProviderBuilder.ToString().Trim()}}
diff --git a/CsvSerializer/SourceGenerator.cs b/CsvSerializer/SourceGenerator.cs
index aeb51e2..3a1f35e 100644
--- a/CsvSerializer/SourceGenerator.cs
+++ b/CsvSerializer/SourceGenerator.cs
@@ -1,4 +1,6 @@
-using System.Diagnostics;
+#if DEBUG
+using System.Diagnostics;
+#endif
 using Csv.Internals;
 using Microsoft.CodeAnalysis;
 
diff --git a/Tests/SerializerTests.cs b/Tests/SerializerTests.cs
index 5f49d45..02d2536 100644
--- a/Tests/SerializerTests.cs
+++ b/Tests/SerializerTests.cs
@@ -183,8 +183,8 @@ public void CsvCanBeDeserializedToPrivateType() {
 		[Fact]
 		public void HeaderNameCanBeRenamedUsingAttribute() {
 			ModelWithColumnNames[] items = {
-				new ModelWithColumnNames { ItemName = "Coconut", WeightInGrams = 900 },
-				new ModelWithColumnNames { ItemName = "Jackfruit", WeightInGrams = 1200 }
+				new() { ItemName = "Coconut", WeightInGrams = 900 },
+				new() { ItemName = "Jackfruit", WeightInGrams = 1200 }
 			};
 			string csv = CsvSerializer.Serialize(items, withHeaders: true);
 			string[] lines = csv.Split("\r\n");
@@ -233,6 +233,26 @@ string Serialize<T>(T item) {
 				True,102,-100,-200,200,-3000,3000,-40000,40000,1E+14,1.7837193718273812E+19,989898989898,"CSV Serializer","8/23/2019 12:00:00 AM","http://localhost:5000/",OK
 				""");
 		}
+
+		[Fact]
+		public void CannotDeserializeRowWithExtraColumns() {
+			string csv = """
+				"Bool","Byte","SByte","Short","UShort","Int","UInt","Long","ULong","Float","Double","Decimal","String","DateTime","Uri","StatusCode"
+				True,102,-100,-200,200,-3000,3000,-40000,40000,1E+14,1.7837193718273812E+19,989898989898,"CSV Serializer","08/23/2019 00:00:00","http://localhost:5000/",OK,foo,bar
+				""";
+			new Action(() => CsvSerializer.Deserialize<Model>(csv, hasHeaders: true, provider: CultureInfo.InvariantCulture))
+				.Should().Throw<CsvFormatException>("Row must consists of 16 columns.");
+		}
+
+		[Fact]
+		public void CannotDeserializeRowWithMissingColumns() {
+			string csv = """
+				"Bool","Byte","SByte","Short","UShort","Int","UInt","Long","ULong","Float","Double","Decimal","String","DateTime","Uri","StatusCode"
+				True,102,-100,-200,200,-3000,3000,-40000,40000,1E+14,1.7837193718273812E+19,989898989898,"CSV Serializer","08/23/2019 00:00:00","http://localhost:5000/"
+				""";
+			new Action(() => CsvSerializer.Deserialize<Model>(csv, hasHeaders: true, provider: CultureInfo.InvariantCulture))
+				.Should().Throw<CsvFormatException>("Row must consists of 16 columns.");
+		}
 	}
 
 	public class ExcelModel {