Get commented typing working

SaneTsv/SaneTsv.cs

@@ -3,9 +3,14 @@ using System.Text;
 
 namespace NathanMcRae;
 
-public class Tsv<T> where T : SaneTsv.SaneTsvRecord
+public class Tsv<T> where T : SaneTsv.TsvRecord
 {
-  public List<T> Records { get; set; }
+  public virtual List<T> Records { get; set; }
+}
+
+public class CommentedTsv<T>: Tsv<T> where T : SaneTsv.TsvRecord
+{
+  public override List<T> Records { get; set; }
   public string FileComment { get; set; } = null;
 }
 
@@ -54,25 +59,34 @@ public class SaneTsv
     }
   }
 
-  public static Tsv<T> ParseSimpleTsv<T>(byte[] inputBuffer) where T : SaneTsvRecord, new()
+  public static Tsv<T> ParseSimpleTsv<T>(byte[] inputBuffer) where T : TsvRecord, new()
   {
     return Parse<T>(inputBuffer, FormatType.SIMPLE_TSV);
   }
 
-  public static Tsv<T> ParseTypedTsv<T>(byte[] inputBuffer) where T : SaneTsvRecord, new()
+  public static Tsv<T> ParseTypedTsv<T>(byte[] inputBuffer) where T : TsvRecord, new()
   {
     return Parse<T>(inputBuffer, FormatType.TYPED_TSV);
   }
 
-  public static Tsv<T> ParseCommentedTsv<T>(byte[] inputBuffer) where T : SaneTsvRecord, new()
+  public static CommentedTsv<T> ParseCommentedTsv<T>(byte[] inputBuffer) where T : CommentedTsvRecord, new()
   {
-    return Parse<T>(inputBuffer, FormatType.COMMENTED_TSV);
+    // TODO: add the file comment?
+    return  (CommentedTsv<T>)Parse<T>(inputBuffer, FormatType.COMMENTED_TSV);
   }
 
   // TODO: Have parsing errors include line / column #
-  protected static Tsv<T> Parse<T>(byte[] inputBuffer, FormatType format) where T : SaneTsvRecord, new()
+  protected static Tsv<T> Parse<T>(byte[] inputBuffer, FormatType format) where T : TsvRecord, new()
   {
-    var parsed = new Tsv<T>();
+    Tsv<T> parsed;
+    if (format == FormatType.COMMENTED_TSV)
+    {
+      parsed = new CommentedTsv<T>();
+    }
+    else
+    {
+      parsed = new Tsv<T>();
+    }
     parsed.Records = new List<T>();
 
     var headerTypes = new List<Type>();
@@ -82,7 +96,7 @@ public class SaneTsv
 
     foreach (PropertyInfo property in typeof(T).GetProperties())
     {
-      SaneTsvColumnAttribute attribute = (SaneTsvColumnAttribute)Attribute.GetCustomAttribute(property, typeof(SaneTsvColumnAttribute));
+      TypedTsvColumnAttribute attribute = (TypedTsvColumnAttribute)Attribute.GetCustomAttribute(property, typeof(TypedTsvColumnAttribute));
       if (attribute == null)
       {
         continue;
@@ -248,9 +262,17 @@ public class SaneTsv
 
           if (currentComment.Length > 0)
           {
-            parsed.FileComment = currentComment.ToString();
+            if (parsed is CommentedTsv<T> commentedParsed)
+            {
+              commentedParsed.FileComment = currentComment.ToString();
               currentComment.Clear();
             }
+            else
+            {
+              throw new Exception("Found a file comment, but parser wasn't expecting a comment");
+            }
+          }
+          
 
           fields.Clear();
         }
@@ -333,11 +355,24 @@ public class SaneTsv
     return parsed;
   }
 
-  protected static T ParseCurrentRecord<T>(Type[] columnTypes, PropertyInfo[] properties, List<byte[]> fields, string comment, int line) where T : SaneTsvRecord, new()
+  protected static T ParseCurrentCommentedRecord<T> (Type[] columnTypes, PropertyInfo[] properties, List<byte[]> fields, string comment, int line) where T : CommentedTsvRecord, new()
+  {
+    return (T)ParseCurrentRecord<T>(columnTypes, properties, fields, comment, line);
+  }
+
+  protected static T ParseCurrentRecord<T>(Type[] columnTypes, PropertyInfo[] properties, List<byte[]> fields, string comment, int line) where T : TsvRecord, new()
   {
     T record = new T();
 
-    record.Comment = comment;
+    if (record is CommentedTsvRecord commentedRecord)
+    {
+      commentedRecord.Comment = comment;
+    }
+    else if (comment != null)
+    {
+      throw new Exception($"Found comment for line {line}, but format does not support comments");
+    }
+    
     record.Line = line;
 
     for (int j = 0; j < fields.Count; j++)
@@ -705,7 +740,22 @@ public class SaneTsv
     }
   }
 
-  public static byte[] SerializeTypedTsv<T>(IList<T> data) where T : SaneTsvRecord
+  public static byte[] SerializeSimpleTsv<T>(IList<T> data) where T : TsvRecord
+  {
+    return SerializeTsv<T>(data, FormatType.TYPED_TSV);
+  }
+
+  public static byte[] SerializeTypedTsv<T>(IList<T> data) where T : TsvRecord
+  {
+    return SerializeTsv<T>(data, FormatType.TYPED_TSV);
+  }
+
+  public static byte[] SerializeCommentedTsv<T>(IList<T> data, string fileComment) where T : CommentedTsvRecord
+  {
+    return SerializeTsv<T>(data, FormatType.COMMENTED_TSV);
+  }
+
+  protected static byte[] SerializeTsv<T>(IList<T> data, FormatType tsvFormat) where T : TsvRecord
   {
     var bytes = new List<byte>();
 
@@ -716,14 +766,20 @@ public class SaneTsv
 
     foreach (PropertyInfo property in typeof(T).GetProperties())
     {
-      SaneTsvColumnAttribute attribute = (SaneTsvColumnAttribute)Attribute.GetCustomAttribute(property, typeof(SaneTsvColumnAttribute));
+      TsvColumnAttribute attribute = (TsvColumnAttribute)Attribute.GetCustomAttribute(property, typeof(TsvColumnAttribute));
       if (attribute == null)
       {
         continue;
       }
 
-      headerNames.Add(attribute.ColumnName ?? property.Name);
+      string headerName = attribute.ColumnName ?? property.Name;
-      headerTypes.Add(attribute.ColumnType ?? GetColumnFromType(property.PropertyType));
+      headerNames.Add(headerName);
+      Type headerType = attribute.ColumnType ?? GetColumnFromType(property.PropertyType);
+      if (tsvFormat == FormatType.SIMPLE_TSV && headerType != typeof(StringType))
+      {
+        throw new Exception($"Serializing Simple TSV requires all columns be of type string, but column '{headerName}' has type '{headerType}'");
+      }
+      headerTypes.Add(headerType);
       headerPropertyInfos.Add(property);
       // TODO: Check that the property type and given column type are compatible
       columnCount++;
@@ -990,42 +1046,64 @@ public class SaneTsv
     }
   }
 
-  public class SaneTsvRecord
+  public class TsvRecord
+  {
+    public int? Line { get; set; }
+    public TsvRecord(int? line)
+    {
+      Line = line;
+    }
+
+    public TsvRecord() { }
+  }
+
+  public class CommentedTsvRecord : TsvRecord
   {
     public string Comment { get; set; }
-    public int? Line { get; set; }
 
-    public SaneTsvRecord(string comment, int? line)
+    public CommentedTsvRecord(string comment, int? line)
     {
       Comment = comment;
       Line = line;
     }
 
-    public SaneTsvRecord() { }
+    public CommentedTsvRecord() { }
   }
 
-  public class TestRecord : SaneTsvRecord
+  public class TestRecord : CommentedTsvRecord
   {
-    [SaneTsvColumn("my-column")]
+    [TypedTsvColumn("my-column")]
     public string MyColumn { get; set; }
   }
 
   // TODO: Add column ordering
-  public class SaneTsvColumnAttribute : Attribute
+  public class TsvColumnAttribute : Attribute
   {
     public string ColumnName { get; }
-    public Type ColumnType { get; }
+    public virtual Type ColumnType { get; }
 
-    public SaneTsvColumnAttribute() { }
+    public TsvColumnAttribute() 
-
-    public SaneTsvColumnAttribute(string columnName)
     {
+      ColumnType = typeof(StringType);
+    }
+    public TsvColumnAttribute(string columnName) 
+    {
+      ColumnType = typeof(StringType);
       ColumnName = columnName;
     }
+  }
 
-    public SaneTsvColumnAttribute(string columnName, Type columnType)
+  // TODO: Add column ordering
+  public class TypedTsvColumnAttribute : TsvColumnAttribute
+  {
+    public override Type ColumnType { get; }
+
+    public TypedTsvColumnAttribute() { }
+
+    public TypedTsvColumnAttribute(string columnName) : base(columnName) { }
+
+    public TypedTsvColumnAttribute(string columnName, Type columnType) : base(columnName)
     {
-      ColumnName = columnName;
       if (columnType.BaseType != typeof(ColumnType))
       {
         throw new Exception("Column type must inherit from SaneTsv.ColumnType");
@@ -1033,7 +1111,7 @@ public class SaneTsv
       ColumnType = columnType;
     }
 
-    public SaneTsvColumnAttribute(Type columnType)
+    public TypedTsvColumnAttribute(Type columnType)
     {
       if (columnType.BaseType != typeof(ColumnType))
       {

SaneTsv/SaneTsvTest/Program.cs

@@ -3,39 +3,39 @@ using System.Text;
 
 internal class Program
 {
-  public class TestRecord : SaneTsv.SaneTsvRecord
+  public class TestRecord : SaneTsv.TsvRecord
   {
-    [SaneTsv.SaneTsvColumn("string-test")]
+    [SaneTsv.TypedTsvColumn("string-test")]
     public string StringTest { get; set; }
 
-    [SaneTsv.SaneTsvColumn("bool-test")]
+    [SaneTsv.TypedTsvColumn("bool-test")]
     public bool BoolTest { get; set; }
 
-    [SaneTsv.SaneTsvColumn("float32-test")]
+    [SaneTsv.TypedTsvColumn("float32-test")]
     public float Float32Test { get; set; }
 
-    [SaneTsv.SaneTsvColumn("float32-le-test", typeof(SaneTsv.Float32LEType))]
+    [SaneTsv.TypedTsvColumn("float32-le-test", typeof(SaneTsv.Float32LEType))]
     public float Float32LETest { get; set; }
 
-    [SaneTsv.SaneTsvColumn("float64-test")]
+    [SaneTsv.TypedTsvColumn("float64-test")]
     public double Float64Test { get; set; }
 
-    [SaneTsv.SaneTsvColumn("float64-le-test", typeof(SaneTsv.Float64LEType))]
+    [SaneTsv.TypedTsvColumn("float64-le-test", typeof(SaneTsv.Float64LEType))]
     public double Float64LETest { get; set; }
 
-    [SaneTsv.SaneTsvColumn("uint32-test")]
+    [SaneTsv.TypedTsvColumn("uint32-test")]
     public UInt32 UInt32Test { get; set; }
 
-    [SaneTsv.SaneTsvColumn("uint64-test")]
+    [SaneTsv.TypedTsvColumn("uint64-test")]
     public UInt64 UInt64Test { get; set; }
 
-    [SaneTsv.SaneTsvColumn("int32-test")]
+    [SaneTsv.TypedTsvColumn("int32-test")]
     public Int32 Int32Test { get; set; }
 
-    [SaneTsv.SaneTsvColumn("int64-test")]
+    [SaneTsv.TypedTsvColumn("int64-test")]
     public Int64 Int64Test { get; set; }
 
-    [SaneTsv.SaneTsvColumn("binary-test")]
+    [SaneTsv.TypedTsvColumn("binary-test")]
     public byte[] BinaryTest { get; set; }
 
     public TestRecord(string stringTest, bool boolTest, float float32Test, float float32LETest, double float64Test, double float64LETest, UInt32 uInt32Test, UInt64 uInt64Test, Int32 int32Test, Int64 int64Test, byte[] binaryTest)
@@ -56,48 +56,48 @@ internal class Program
     public TestRecord() { }
   }
 
-  public class BoolTestRecord : SaneTsv.SaneTsvRecord
+  public class BoolTestRecord : SaneTsv.CommentedTsvRecord
   {
-    [SaneTsv.SaneTsvColumn("column1:ty#pe")]
+    [SaneTsv.TypedTsvColumn("column1:ty#pe")]
     public bool Column1 { get; set; }
 
-    [SaneTsv.SaneTsvColumn]
+    [SaneTsv.TypedTsvColumn]
     public byte[] column2 { get; set; }
 
-    [SaneTsv.SaneTsvColumn("columnthree\nyep")]
+    [SaneTsv.TypedTsvColumn("columnthree\nyep")]
     public string Column3 { get; set; }
   }
 
-  public class BoolTestRecord2 : SaneTsv.SaneTsvRecord
+  public class BoolTestRecord2 : SaneTsv.CommentedTsvRecord
   {
-    [SaneTsv.SaneTsvColumn("column1:type")]
+    [SaneTsv.TypedTsvColumn("column1:type")]
     public bool Column1 { get; set; }
 
-    [SaneTsv.SaneTsvColumn]
+    [SaneTsv.TypedTsvColumn]
     public byte[] column2 { get; set; }
 
-    [SaneTsv.SaneTsvColumn("columnthree\nyep")]
+    [SaneTsv.TypedTsvColumn("columnthree\nyep")]
     public string Column3 { get; set; }
   }
 
-  public class SerdeTestRecord : SaneTsv.SaneTsvRecord
+  public class SerdeTestRecord : SaneTsv.CommentedTsvRecord
   {
-    [SaneTsv.SaneTsvColumn("column1")]
+    [SaneTsv.TypedTsvColumn("column1")]
     public bool Column1 { get; set; }
 
-    [SaneTsv.SaneTsvColumn]
+    [SaneTsv.TypedTsvColumn]
     public byte[] column2 { get; set; }
 
-    [SaneTsv.SaneTsvColumn("columnthree\nyep")]
+    [SaneTsv.TypedTsvColumn("columnthree\nyep")]
     public string Column3 { get; set; }
   }
 
-  public class FloatTestRecord : SaneTsv.SaneTsvRecord
+  public class FloatTestRecord : SaneTsv.CommentedTsvRecord
   {
-    [SaneTsv.SaneTsvColumn("somefloat")]
+    [SaneTsv.TypedTsvColumn("somefloat")]
     public double SomeFloat { get; set; }
 
-    [SaneTsv.SaneTsvColumn("binfloat", typeof(SaneTsv.Float64LEType))]
+    [SaneTsv.TypedTsvColumn("binfloat", typeof(SaneTsv.Float64LEType))]
     public double BinFloat { get; set; }
   }
 
@@ -148,7 +148,7 @@ internal class Program
         "\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" +
         "\nFALSE\tnother\tno\\ther";
 
-      Tsv<BoolTestRecord2> parsed = SaneTsv.ParseCommentedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
+      CommentedTsv<BoolTestRecord2> parsed = SaneTsv.ParseCommentedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
     }
 
     //{
@@ -188,8 +188,6 @@ internal class Program
       }
     }
 
-    Console.WriteLine("Done with tests");
-
     {
       string testName = "Serde test";
 
@@ -215,5 +213,87 @@ internal class Program
         Console.WriteLine($"Failed {testName}");
       }
     }
+
+    {
+      string testName = "Trying to parse a not commented record as a Commented TSV test";
+
+      // These should not compile:
+
+      //byte[] serialized = SaneTsv.SerializeCommentedTsv(data);
+      // Gives this error: error CS7036: There is no argument given that corresponds to the required parameter 'fileComment' of 'SaneTsv.SerializeCommentedTsv<T>(IList<T>, string)'
+
+      //Tsv<TestRecord> parsed = SaneTsv.ParseCommentedTsv<TestRecord>(serialized);
+      // Gives this error:  error CS0311: The type 'Program.TestRecord' cannot be used as type parameter 'T' in the generic type or method 'SaneTsv.ParseCommentedTsv<T>(byte[])'. There is no implicit reference conversion from 'Program.TestRecord' to 'NathanMcRae.SaneTsv.CommentedTsvRecord'.
+    }
+
+    {
+      string testName = "Try to parsed a Commented TSV as a Simple TSV";
+
+      string testString1 = "#This is a file comment\n" +
+        "#One more file comment line\n" +
+        "column1:type:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" +
+        "\n#This is a comment" +
+        "\n#Another comment line" +
+        "\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" +
+        "\nFALSE\tnother\tno\\ther";
+
+      try
+      {
+        Tsv<BoolTestRecord2> parsed = SaneTsv.ParseSimpleTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
+
+        Console.WriteLine($"Failed {testName}");
+      }
+      catch (Exception e)
+      {
+        Console.WriteLine($"Passed {testName}");
+      }
+    }
+
+    {
+      string testName = "Try to parsed a Commented TSV as a Typed TSV";
+
+      string testString1 = "#This is a file comment\n" +
+        "#One more file comment line\n" +
+        "column1:type:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" +
+        "\n#This is a comment" +
+        "\n#Another comment line" +
+        "\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" +
+        "\nFALSE\tnother\tno\\ther";
+
+      try
+      {
+        Tsv<BoolTestRecord2> parsed = SaneTsv.ParseTypedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
+
+        Console.WriteLine($"Failed {testName}");
+      }
+      catch (Exception e)
+      {
+        Console.WriteLine($"Passed {testName}");
+      }
+    }
+
+    {
+      string testName = "Try to parsed a Typed TSV as a Simple TSV";
+
+      string testString1 = 
+        "column1:type:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" +
+        "\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" +
+        "\nFALSE\tnother\tno\\ther";
+
+      try
+      {
+        Tsv<BoolTestRecord2> parsed = SaneTsv.ParseSimpleTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
+
+        Console.WriteLine($"Failed {testName}");
+      }
+      catch (Exception e)
+      {
+        Console.WriteLine($"Passed {testName}");
+      }
+    }
+
+    Console.WriteLine("Done with tests");
   }
 }
+
+