Add support for big-endian byte order

SpssLib/Compression/DecompressedDataStream.cs

@@ -3,6 +3,8 @@
 using System.Text;
 using System.IO;
 
+using SpssLib.FileParser;
+
 namespace SpssLib.Compression
 {
     class DecompressedDataStream: Stream
@@ -24,15 +26,15 @@ class DecompressedDataStream: Stream
         private byte[] _systemMissingBytes;
         private byte[] _spacesBytes;
 
-        private BinaryReader _reader;
+        private DualBinaryReader _reader;
 
-        public DecompressedDataStream(Stream compressedDataStream, double bias, double systemMissing)
+        public DecompressedDataStream(Stream compressedDataStream, double bias, double systemMissing, bool isLittleEndian)
         {
             CompressedDataStream = compressedDataStream;
             Bias = bias;
             SystemMissing = systemMissing;
-            _reader = new BinaryReader(compressedDataStream, Encoding.ASCII);
-
+            _reader = new DualBinaryReader(compressedDataStream, Encoding.ASCII);
+            _reader.IsLittleEndian = isLittleEndian;
             _spacesBytes = Encoding.ASCII.GetBytes(SpaceString);
             _systemMissingBytes = BitConverter.GetBytes(SystemMissing);
         }
@@ -163,11 +165,19 @@ private bool ParseNextInstructionSet()
                 {
                 }
 
-                else if (instruction > 0 && instruction < 252) // compressed value
+                else if (instruction > 0 && instruction < 252) // compressed value (small 1-byte integers)
                 {
                     // compute actual value:
                     double value = instruction - Bias;
-                    _elementBuffer[bufferPosition++] = BitConverter.GetBytes(value);
+                    byte[] element = BitConverter.GetBytes(value);
+                    if(_reader.IsLittleEndian)
+                        _elementBuffer[bufferPosition++] = BitConverter.GetBytes(value);
+                    else
+                    {
+                        byte[] revElement = new byte[InstructionSetByteSize];
+                        Helper.ReverseArray(element, revElement);
+                        _elementBuffer[bufferPosition++] = revElement;
+                    }
                 }
                 else if (instruction == 252) // end of file
                 {

SpssLib/DataReader/SpssReader.cs

@@ -17,7 +17,7 @@ public class SpssReader : IDisposable
         // TODO add needed metadata info (use SpssOptions?)
 
         /// <summary>
-        /// A collection of variables read from teh file 
+        /// A collection of variables read from the file 
         /// </summary>
 		public ICollection<Variable> Variables { get; private set; }
         /// <summary>

SpssLib/FileParser/DualBinaryReader.cs

@@ -0,0 +1,113 @@
+using System;
+using System.IO;
+using System.Text;
+
+namespace SpssLib.FileParser
+{
+    public static class Helper
+    {
+        public static void ReverseArray(byte[] source, byte[] target)
+        {
+            int len = source.Length;
+            for (int i = 0; i < len; i++)
+                target[i] = source[len - 1 - i];
+        }
+    }
+
+    // Binary reader that supports both little-endian and big-endian byte orders. It assumes that the current environment is little-indian.
+    public class DualBinaryReader : BinaryReader
+    {
+        public DualBinaryReader(Stream stream, Encoding encoding) : base(stream, encoding)
+        {
+
+        }
+
+        public bool IsLittleEndian { get; set; } = true;
+
+        public override Int16 ReadInt16()
+        {
+            Int16 i;
+
+            if (this.IsLittleEndian)
+                i = base.ReadInt16();
+            else
+            {
+                byte b1 = this.ReadByte();
+                byte b0 = this.ReadByte();
+                i = (short)(b1 << 8 + b0);
+            }
+
+            return i;
+        }
+
+        public override UInt16 ReadUInt16()
+        {
+            UInt16 i;
+
+            if (this.IsLittleEndian)
+                i = base.ReadUInt16();
+            else
+            {
+                byte b1 = this.ReadByte();
+                byte b0 = this.ReadByte();
+                i = (UInt16)(b1 << 8 + b0);
+            }
+
+            return i;
+        }
+
+        public override int ReadInt32()
+        {
+            int i;
+
+            if (this.IsLittleEndian)
+                i = base.ReadInt32();
+            else
+            {
+                i = this.ReadByte();
+
+                for (int k = 0; k < 3; k++)
+                    i = (i << 8) | this.ReadByte();
+            }
+
+            return i;
+        }
+
+        public override uint ReadUInt32()
+        {
+            uint i;
+
+            if (this.IsLittleEndian)
+                i = base.ReadUInt32();
+            else
+            {
+                i = this.ReadByte();
+
+                for (int k = 0; k < 3; k++)
+                    i = (i << 8) | this.ReadByte();
+            }
+
+            return i;
+        }
+
+        public override double ReadDouble()
+        {
+            double f;
+
+            if (this.IsLittleEndian)
+                f = base.ReadDouble();
+            else
+            {
+                UInt64 l = this.ReadByte();
+
+                for(int k = 0; k < 7; k++)
+                    l = (l << 8) | this.ReadByte();
+
+                f = BitConverter.ToDouble( BitConverter.GetBytes(l), 0);
+            }
+
+            return f;
+        }
+
+    }
+}

SpssLib/FileParser/MetaData.cs

@@ -69,6 +69,19 @@ internal set
 
         public IList<BaseInfoRecord> InfoRecords { get; private set; }
 
+        public bool IsLittleEndian
+        {
+            get
+            {
+                if (this.MachineIntegerInfo.Endianness == 2)
+                    return true;
+                else if (this.MachineIntegerInfo.Endianness == 1)
+                    return false;
+                else
+                    throw new SpssFileFormatException("Failed to detect endianness.");
+            }
+        }
+
         public double SystemMissingValue { get; private set; }
         // Count number of variables (the number of variable-records with a name,
         // the rest is part of a long string variable), this includes the variables 

SpssLib/FileParser/Records/BaseInfoRecord.cs

@@ -21,7 +21,7 @@ public void WriteRecord(BinaryWriter writer)
             WriteInfo(writer);
         }
 
-        public void FillRecord(BinaryReader reader)
+        public void FillRecord(DualBinaryReader reader)
         {
             ItemSize = reader.ReadInt32();
             ItemCount = reader.ReadInt32();
@@ -49,7 +49,7 @@ protected void CheckInfoHeader(int itemSize = -1, int itemCount = -1)
         }
 
         protected abstract void WriteInfo(BinaryWriter writer);
-        protected abstract void FillInfo(BinaryReader reader);
+        protected abstract void FillInfo(DualBinaryReader reader);
     }
 
     public class UnknownInfoRecord : BaseInfoRecord
@@ -78,7 +78,7 @@ protected override void WriteInfo(BinaryWriter writer)
             writer.Write(Data);
         }
 
-        protected override void FillInfo(BinaryReader reader)
+        protected override void FillInfo(DualBinaryReader reader)
         {
             Data = reader.ReadBytes(ItemCount * ItemSize);
         }

SpssLib/FileParser/Records/CharacterEncodingRecord.cs

@@ -39,7 +39,7 @@ protected override void WriteInfo(BinaryWriter writer)
 			writer.Write(bytes);
 		}
 
-        protected override void FillInfo(BinaryReader reader)
+        protected override void FillInfo(DualBinaryReader reader)
         {
             CheckInfoHeader(1); // items must be of size 1 (byte)
 

SpssLib/FileParser/Records/DictionaryTerminationRecord.cs

@@ -17,7 +17,7 @@ public void WriteRecord(BinaryWriter writer)
 			writer.Write(0); 
 		}
 
-	    public void FillRecord(BinaryReader reader)
+	    public void FillRecord(DualBinaryReader reader)
 	    {
             // skip filler
 	        reader.ReadInt32();

SpssLib/FileParser/Records/DocumentRecord.cs

@@ -20,7 +20,7 @@ public DocumentRecord(IList<string> lines)
             LineCount = lines.Count;
         }
 
-        public void FillRecord(BinaryReader reader)
+        public void FillRecord(DualBinaryReader reader)
         {
             LineCount = reader.ReadInt32();
             LineCollection = new List<string>();

SpssLib/FileParser/Records/EbcdicHeaderRecord.cs

@@ -5,7 +5,7 @@ namespace SpssLib.FileParser.Records
 {
     internal class EbcdicHeaderRecord : IRecord
     {
-        public NotSupportedException Exception => new NotSupportedException("EBCDIC???? Who uses that? Honestly!!");
+        public NotSupportedException Exception => new NotSupportedException("EBCDIC records not supported.");
 
         public RecordType RecordType => RecordType.EbcdicHeaderRecord;
 
@@ -14,7 +14,7 @@ public void WriteRecord(BinaryWriter writer)
             throw Exception;
         }
 
-        public void FillRecord(BinaryReader reader)
+        public void FillRecord(DualBinaryReader reader)
         {
             throw Exception;
         }

SpssLib/FileParser/Records/HeaderRecord.cs

@@ -58,10 +58,24 @@ public void WriteRecord(BinaryWriter writer)
 			writer.Write(new byte[3]);
 	    }
 
-        public void FillRecord(BinaryReader reader)
+        public void FillRecord(DualBinaryReader reader)
         {
             ProductName = new string(reader.ReadChars(60));
             LayoutCode = reader.ReadInt32();
+            switch(LayoutCode)
+            {
+                case 2:
+                case 3:
+                    reader.IsLittleEndian = true;
+                    break;
+                case 0x02000000:
+                case 0x03000000:
+                    reader.IsLittleEndian = false;
+                    break;
+                default:
+                    throw new SpssFileFormatException("Failed to detect endianness.");
+            }
+
             NominalCaseSize = reader.ReadInt32();
             Compressed = reader.ReadInt32() == 1;
             WeightIndex = reader.ReadInt32();

SpssLib/FileParser/Records/IRecord.cs

@@ -6,7 +6,7 @@ internal interface IRecord
     {
         RecordType RecordType { get; }
         void WriteRecord(BinaryWriter writer);  // TODO: split to internal interface
-        void FillRecord(BinaryReader reader);
+        void FillRecord(DualBinaryReader reader);
         void RegisterMetadata(MetaData metaData);
     }
 

SpssLib/FileParser/Records/IRecordParser.cs

@@ -6,7 +6,7 @@ namespace SpssLib.FileParser.Records
     internal interface IRecordParser
     {
         RecordType Accepts { get; }
-        IRecord ParseRecord(BinaryReader reader);
+        IRecord ParseRecord(DualBinaryReader reader);
     }
 
     internal class GeneralRecordParser<TRecord> : IRecordParser where TRecord : IRecord
@@ -18,7 +18,7 @@ public GeneralRecordParser(RecordType accepts)
             Accepts = accepts;
         }
 
-        public IRecord ParseRecord(BinaryReader reader)
+        public IRecord ParseRecord(DualBinaryReader reader)
         {
             TRecord record = CreateRecord();
             record.FillRecord(reader);

SpssLib/FileParser/Records/InfoRecordParser.cs

@@ -15,7 +15,7 @@ public InfoRecordParser(IDictionary<int, Type> infoRecordsTypes)
             _infoRecordsTypes = infoRecordsTypes;
         }
 
-        public IRecord ParseRecord(BinaryReader reader)
+        public IRecord ParseRecord(DualBinaryReader reader)
         {
             IRecord record = CreateRecord(reader);
             record.FillRecord(reader);

SpssLib/FileParser/Records/MachineFloatingPointInfoRecord.cs

@@ -35,7 +35,7 @@ protected override void WriteInfo(BinaryWriter writer)
             writer.Write(MissingLowestValue);
         }
 
-        protected override void FillInfo(BinaryReader reader)
+        protected override void FillInfo(DualBinaryReader reader)
         {
             CheckInfoHeader(8, 3);
 

SpssLib/FileParser/Records/MachineIntegerInfoRecord.cs

@@ -91,7 +91,7 @@ protected override void WriteInfo(BinaryWriter writer)
 			writer.Write(CharacterCode);
 		}
 
-        protected override void FillInfo(BinaryReader reader)
+        protected override void FillInfo(DualBinaryReader reader)
         {
             // Must have 8 int of 32 bits
             CheckInfoHeader(itemSize:4, itemCount:8);

SpssLib/FileParser/Records/ValueLabelRecord.cs

@@ -59,7 +59,7 @@ public void WriteRecord(BinaryWriter writer)
 			}
 		}
 
-        public void FillRecord(BinaryReader reader)
+        public void FillRecord(DualBinaryReader reader)
         {
             LabelCount = reader.ReadInt32();
             _labelsRaw = new Dictionary<byte[], KeyValuePair<byte, byte[]>>();

SpssLib/FileParser/Records/VariableDataInfoRecord.cs

@@ -96,7 +96,7 @@ protected void BuildDataArray()
             Data = buffer;
         }
 
-        protected override void FillInfo(BinaryReader reader)
+        protected override void FillInfo(DualBinaryReader reader)
         {
             CheckInfoHeader(1);
             Data = reader.ReadBytes(ItemCount);

SpssLib/FileParser/Records/VariableDisplayParameterRecord.cs

@@ -101,7 +101,7 @@ protected override void WriteInfo(BinaryWriter writer)
             }
         }
 
-        protected override void FillInfo(BinaryReader reader)
+        protected override void FillInfo(DualBinaryReader reader)
         {
             CheckInfoHeader(4);
 

SpssLib/FileParser/Records/VariableRecord.cs

@@ -472,7 +472,7 @@ public void WriteRecord(BinaryWriter writer)
 
 	    }
 
-        public void FillRecord(BinaryReader reader)
+        public void FillRecord(DualBinaryReader reader)
         {
             Type = reader.ReadInt32();
             HasVariableLabel = reader.ReadInt32() == 1;