Add comment parsing

SaneTsv/SaneTsv.cs

@@ -32,6 +32,7 @@ public class SaneTsv
   public ColumnType[] ColumnTypes { get; protected set; }
   public Dictionary<string, List<object>> Columns { get; protected set; }
   public List<SaneTsvRecord> Records { get; protected set; }
+  public string FileComment { get; protected set; } = null;
 
   public static SaneTsv ParseSaneTsv(byte[] inputBuffer)
   {
@@ -59,6 +60,8 @@ public class SaneTsv
 
     var fieldBytes = new List<byte>();
     var fields = new List<byte[]>();
+    var currentComment = new StringBuilder();
+
     int numFields = -1;
     int line = 1;
     int currentLineStart = 0;
@@ -201,6 +204,12 @@ public class SaneTsv
             parsed.ColumnTypes[j] = type;
           }
 
+          if (currentComment.Length > 0)
+          {
+            parsed.FileComment = currentComment.ToString();
+            currentComment.Clear();
+          }
+
           fields.Clear();
         }
         else if (numFields != fields.Count)
@@ -209,7 +218,14 @@ public class SaneTsv
         }
         else
         {
-          AddRecord(parsed, fields, line);
+          string comment = null;
+          if (currentComment.Length > 0)
+          {
+            comment = currentComment.ToString();
+            currentComment.Clear();
+          }
+          parsed.Records.Add(new SaneTsvRecord(parsed, ParseCurrentRecord(parsed, fields, line), comment));
+          fields.Clear();
         }
 
         line++;
@@ -217,7 +233,29 @@ public class SaneTsv
       }
       else if (inputBuffer[i] == '#')
       {
-        throw new Exception($"Found unescaped '#' at column {i - currentLineStart}, line {line}");
+        if (i == currentLineStart && format >= FormatType.COMMENTED_TSV)
+        {
+          int j = i;
+          for (; j < inputBuffer.Length && inputBuffer[j] != '\n'; j++) { }
+          if (j < inputBuffer.Length)
+          {
+            var commentBytes = new byte[j - i - 1];
+            Array.Copy(inputBuffer, i + 1, commentBytes, 0, j - i - 1);
+            currentComment.Append(Encoding.UTF8.GetString(commentBytes));
+            currentComment.Append("\n");
+            i = j;
+            currentLineStart = i + 1;
+            line++;
+          }
+          else
+          {
+            throw new Exception("Comments at end of file are not allowed");
+          }
+        }
+        else
+        {
+          throw new Exception($"Found unescaped '#' at line {line}, column {i - currentLineStart}");
+        }
       }
       else
       {
@@ -237,13 +275,20 @@ public class SaneTsv
     }
     else
     {
-      AddRecord(parsed, fields, line);
+      string comment = null;
+      if (currentComment.Length > 0)
+      {
+        comment = currentComment.ToString();
+        currentComment.Clear();
+      }
+      parsed.Records.Add(new SaneTsvRecord(parsed, ParseCurrentRecord(parsed, fields, line), comment));
+      fields.Clear();
     }
 
     return parsed;
   }
 
-  protected static void AddRecord(SaneTsv parsed, List<byte[]> fields, int line)
+  protected static object[] ParseCurrentRecord(SaneTsv parsed, List<byte[]> fields, int line)
   {
     var parsedFields = new object[fields.Count];
     for (int j = 0; j < fields.Count; j++)
@@ -351,8 +396,7 @@ public class SaneTsv
       }
     }
 
-    parsed.Records.Add(new SaneTsvRecord(parsed, parsedFields));
+    return parsedFields;
-    fields.Clear();
   }
 
   public SaneTsvRecord this[int i] => Records[i];
@@ -360,14 +404,16 @@ public class SaneTsv
   public class SaneTsvRecord
   {
     public SaneTsv Parent { get; }
+    public string Comment { get; }
     public object[] Fields { get; }
 
     public object this[string columnName] => Fields[Array.IndexOf(Parent.ColumnNames, columnName)];
 
-    public SaneTsvRecord(SaneTsv parent, object[] fields)
+    public SaneTsvRecord(SaneTsv parent, object[] fields, string comment)
     {
       Parent = parent;
       Fields = fields;
+      Comment = comment;
     }
   }
 }

SaneTsv/readme.md

@@ -51,4 +51,8 @@ Aside from the 'binary' column type, all fields must be UTF-8 encoded text. Each
 
 # Commented TSV
 
+Commented lines start with a '#' character at the beginning of the line. Unescaped '#' characters are not allowed on a line that does not start with a '#'. Any '#' characters in fields must be escaped.
+
+Comments must be UTF-8 encoded text.
+
 Comments after the last record are an error.