Implement strings

MatFileHandler.Tests/MatFileReaderTests.cs

@@ -418,6 +418,22 @@ namespace MatFileHandler.Tests
             Assert.That(diff - datetime[0] < TimeSpan.FromMilliseconds(1));
         }
 
+        /// <summary>
+        /// Test string objects.
+        /// </summary>
+        [Test]
+        public void TestString()
+        {
+            var matFile = GetTests("good")["string"];
+            var s = matFile["s"].Value as IMatObject;
+            var str = new StringAdapter(s);
+            Assert.That(str.Dimensions, Is.EquivalentTo(new[] { 4, 1 }));
+            Assert.That(str[0], Is.EqualTo("abc"));
+            Assert.That(str[1], Is.EqualTo("defgh"));
+            Assert.That(str[2], Is.EqualTo("абвгд"));
+            Assert.That(str[3], Is.EqualTo("æøå"));
+        }
+
         private static AbstractTestDataFactory<IMatFile> GetTests(string factoryName) =>
             new MatTestDataFactory(Path.Combine(TestDirectory, factoryName));
 

MatFileHandler/StringAdapter.cs

@@ -0,0 +1,80 @@
+// Copyright 2017-2018 Alexander Luzgarev
+
+using System;
+using System.Linq;
+using System.Numerics;
+using System.Text;
+
+namespace MatFileHandler
+{
+    /// <summary>
+    /// A better interface for using datetime objects.
+    /// </summary>
+    public class StringAdapter
+    {
+        private readonly int[] dimensions;
+        private readonly string[] strings;
+
+        /// <summary>
+        /// Initializes a new instance of the <see cref="StringAdapter"/> class.
+        /// </summary>
+        /// <param name="array">Source datetime object.</param>
+        public StringAdapter(IArray array)
+        {
+            var matObject = array as IMatObject;
+            if (matObject?.ClassName != "string")
+            {
+                throw new ArgumentException("The object provided is not a string.");
+            }
+
+            var binaryData = matObject["any", 0] as IArrayOf<ulong>;
+
+            (dimensions, strings) = ParseBinaryData(binaryData.Data);
+        }
+
+        /// <summary>
+        /// Gets datetime array dimensions.
+        /// </summary>
+        public int[] Dimensions => dimensions;
+
+        /// <summary>
+        /// Gets string object at given position.
+        /// </summary>
+        /// <param name="list">Indices.</param>
+        /// <returns>Value.</returns>
+        public string this[params int[] list] => strings[Dimensions.DimFlatten(list)];
+
+        private static (int[] dimensions, string[] strings) ParseBinaryData(ulong[] binaryData)
+        {
+            var numberOfDimensions = (int)binaryData[1];
+            var d = new int[numberOfDimensions];
+            for (var i = 0; i < numberOfDimensions; i++)
+            {
+                d[i] = (int)binaryData[i + 2];
+            }
+
+            var numberOfElements = d.NumberOfElements();
+            var start = numberOfDimensions + 2;
+            var lengths = new int[numberOfElements];
+            for (var i = 0; i < numberOfElements; i++)
+            {
+                lengths[i] = (int)binaryData[start + i];
+            }
+
+            var strings = new string[numberOfElements];
+
+            start += numberOfElements;
+            var numberOfUlongsLeft = binaryData.Length - start;
+            var bytes = new byte[numberOfUlongsLeft * sizeof(ulong)];
+            Buffer.BlockCopy(binaryData, start * sizeof(ulong), bytes, 0, bytes.Length);
+            var counter = 0;
+            for (var i = 0; i < numberOfElements; i++)
+            {
+                strings[i] = Encoding.Unicode.GetString(bytes, counter * 2, lengths[i] * 2);
+                counter += lengths[i];
+            }
+
+            return (d, strings);
+        }
+    }
+}

MatFileHandler/SubsystemData.cs

@@ -125,11 +125,9 @@ namespace MatFileHandler
             /// <summary>
             /// Initializes a new instance of the <see cref="ObjectInfo"/> class.
             /// </summary>
-            /// <param name="position">Position of object in the object information table.</param>
             /// <param name="fieldLinks">A dictionary mapping the field indices to "field values" indices.</param>
-            public ObjectInfo(int position, Dictionary<int, int> fieldLinks)
+            public ObjectInfo(Dictionary<int, int> fieldLinks)
             {
-                Position = position;
                 this.fieldLinks = fieldLinks ?? throw new ArgumentNullException(nameof(fieldLinks));
             }
 
@@ -137,11 +135,6 @@ namespace MatFileHandler
             /// Gets mapping between the field indices and "field values" indices.
             /// </summary>
             public IReadOnlyDictionary<int, int> FieldLinks => fieldLinks;
-
-            /// <summary>
-            /// Gets position of object in the object information table.
-            /// </summary>
-            public int Position { get; }
         }
     }
 }

MatFileHandler/SubsystemDataReader.cs

@@ -50,8 +50,19 @@ namespace MatFileHandler
                 }
             }
 
+            var numberOfEmbeddedObjects = (offsets[4] - offsets[3] - 8) / 16;
+            Dictionary<int, EmbeddedObjectInformation> embeddedObjectPositionsToValues = null;
+            using (var stream = new MemoryStream(info, offsets[3], offsets[4] - offsets[3]))
+            {
+                using (var reader = new BinaryReader(stream))
+                {
+                    embeddedObjectPositionsToValues =
+                        ReadEmbeddedObjectPositionsToValuesMapping(reader, numberOfEmbeddedObjects);
+                }
+            }
+
             var numberOfObjects = ((offsets[5] - offsets[4]) / 24) - 1;
-            Dictionary<int, (int objectPosition, int loadingOrder, int classId)> objectClasses = null;
+            Dictionary<int, ObjectClassInformation> objectClasses = null;
             using (var stream = new MemoryStream(info, offsets[4], offsets[5] - offsets[4]))
             {
                 using (var reader = new BinaryReader(stream))
@@ -60,7 +71,7 @@ namespace MatFileHandler
                 }
             }
 
-            var numberOfObjectPositions = objectClasses.Values.Count(x => x.objectPosition != 0);
+            var numberOfObjectPositions = objectClasses.Values.Count(x => x.ObjectPosition != 0);
 
             Dictionary<int, Dictionary<int, int>> objectPositionsToValues = null;
             using (var stream = new MemoryStream(info, offsets[5], offsets[6] - offsets[5]))
@@ -76,7 +87,8 @@ namespace MatFileHandler
                     classIdToName,
                     fieldNames,
                     objectClasses,
-                    objectPositionsToValues);
+                    objectPositionsToValues,
+                    embeddedObjectPositionsToValues);
 
             var allFields = objectInformation.Values.SelectMany(obj => obj.FieldLinks.Values);
             var data = new Dictionary<int, IArray>();
@@ -84,62 +96,28 @@ namespace MatFileHandler
             {
                 data[i] = TransformOpaqueData(opaqueData[i + 2], subsystemData);
             }
+
             return new SubsystemData(classInformation, objectInformation, data);
         }
 
-        private static IArray TransformOpaqueData(IArray array, SubsystemData subsystemData)
+        private static (Dictionary<int, SubsystemData.ClassInfo>, Dictionary<int, SubsystemData.ObjectInfo>)
-        {
+            GatherClassAndObjectInformation(
-            if (array is MatNumericalArrayOf<uint> uintArray)
+                Dictionary<int, string> classIdToName,
-            {
+                string[] fieldNames,
-                if (uintArray.Data[0] == 3707764736u)
+                Dictionary<int, ObjectClassInformation> objectClasses,
-                {
+                Dictionary<int, Dictionary<int, int>> objectPositionsToValues,
-                    var (dimensions, indexToObjectId, classIndex) = DataElementReader.ParseOpaqueData(uintArray.Data);
+                Dictionary<int, EmbeddedObjectInformation> embeddedObjectPositionsToValues)
-                    return new OpaqueLink(
-                        uintArray.Name,
-                        string.Empty,
-                        string.Empty,
-                        dimensions,
-                        array as DataElement,
-                        indexToObjectId,
-                        classIndex,
-                        subsystemData);
-                }
-            }
-
-            return array;
-        }
-
-        private static Dictionary<int, (int objectPosition, int loadingOrder, int classId)> ReadObjectClasses(BinaryReader reader, int numberOfObjects)
-        {
-            var result = new Dictionary<int, (int, int, int)>();
-            reader.ReadBytes(24);
-            for (var i = 0; i < numberOfObjects; i++)
-            {
-                var classId = reader.ReadInt32();
-                reader.ReadBytes(12);
-                var objectPosition = reader.ReadInt32();
-                var loadingOrder = reader.ReadInt32();
-                result[i + 1] = (objectPosition, loadingOrder, classId);
-            }
-
-            return result;
-        }
-
-        private static (Dictionary<int, SubsystemData.ClassInfo>, Dictionary<int, SubsystemData.ObjectInfo>) GatherClassAndObjectInformation(
-            Dictionary<int, string> classIdToName,
-            string[] fieldNames,
-            Dictionary<int, (int objectPosition, int loadingOrder, int classId)> objectClasses,
-            Dictionary<int, Dictionary<int, int>> objectPositionsToValues)
         {
             var classInfos = new Dictionary<int, SubsystemData.ClassInfo>();
+            var newEmbeddedObjectPositionsToValues = new Dictionary<int, Dictionary<int, int>>();
             foreach (var classId in classIdToName.Keys)
             {
                 var className = classIdToName[classId];
                 var fieldIds = new SortedSet<int>();
                 foreach (var objectPosition in objectPositionsToValues.Keys)
                 {
-                    var keyValuePair = objectClasses.First(pair => pair.Value.objectPosition == objectPosition);
+                    var keyValuePair = objectClasses.First(pair => pair.Value.ObjectPosition == objectPosition);
-                    if (keyValuePair.Value.classId != classId)
+                    if (keyValuePair.Value.ClassId != classId)
                     {
                         continue;
                     }
@@ -149,55 +127,48 @@ namespace MatFileHandler
                         fieldIds.Add(fieldId);
                     }
                 }
+
+                foreach (var objectPosition in embeddedObjectPositionsToValues.Keys)
+                {
+                    var keyValuePair = objectClasses.First(pair => pair.Value.EmbeddedObjectPosition == objectPosition);
+                    if (keyValuePair.Value.ClassId != classId)
+                    {
+                        continue;
+                    }
+
+                    fieldIds.Add(embeddedObjectPositionsToValues[objectPosition].FieldIndex);
+                    var d = new Dictionary<int, int>();
+                    var embeddedInfo = embeddedObjectPositionsToValues[objectPosition];
+                    d[embeddedInfo.FieldIndex] = embeddedInfo.ValueIndex;
+                    newEmbeddedObjectPositionsToValues[objectPosition] = d;
+                }
+
                 var fieldToIndex = new Dictionary<string, int>();
                 foreach (var fieldId in fieldIds)
                 {
                     fieldToIndex[fieldNames[fieldId - 1]] = fieldId;
                 }
+
                 classInfos[classId] = new SubsystemData.ClassInfo(className, fieldToIndex);
             }
 
             var objectInfos = new Dictionary<int, SubsystemData.ObjectInfo>();
             foreach (var objectPosition in objectPositionsToValues.Keys)
             {
-                var keyValuePair = objectClasses.First(pair => pair.Value.objectPosition == objectPosition);
+                var keyValuePair = objectClasses.First(pair => pair.Value.ObjectPosition == objectPosition);
-                objectInfos[keyValuePair.Key] = new SubsystemData.ObjectInfo(objectPosition, objectPositionsToValues[objectPosition]);
+                objectInfos[keyValuePair.Key] = new SubsystemData.ObjectInfo(objectPositionsToValues[objectPosition]);
+            }
+
+            foreach (var objectPosition in embeddedObjectPositionsToValues.Keys)
+            {
+                var keyValuePair = objectClasses.First(pair => pair.Value.EmbeddedObjectPosition == objectPosition);
+                objectInfos[keyValuePair.Key] =
+                    new SubsystemData.ObjectInfo(newEmbeddedObjectPositionsToValues[objectPosition]);
             }
 
             return (classInfos, objectInfos);
         }
 
-        private static Dictionary<int, int> ReadFieldToFieldDataMapping(BinaryReader reader)
-        {
-            var length = reader.ReadInt32();
-            var result = new Dictionary<int, int>();
-            for (var i = 0; i < length; i++)
-            {
-                var x = reader.ReadInt32();
-                var y = reader.ReadInt32();
-                var index = x * y;
-                var link = reader.ReadInt32();
-                result[index] = link;
-            }
-            return result;
-        }
-
-        private static Dictionary<int, Dictionary<int, int>> ReadObjectPositionsToValuesMapping(BinaryReader reader, int numberOfValues)
-        {
-            var result = new Dictionary<int, Dictionary<int, int>>();
-            reader.ReadBytes(8);
-            for (var objectPosition = 1; objectPosition <= numberOfValues; objectPosition++)
-            {
-                result[objectPosition] = ReadFieldToFieldDataMapping(reader);
-                var position = reader.BaseStream.Position;
-                if (position % 8 != 0)
-                {
-                    reader.ReadBytes(8 - (int)(position % 8));
-                }
-            }
-            return result;
-        }
-
         private static Dictionary<int, string> ReadClassNames(
             BinaryReader reader,
             string[] fieldNames,
@@ -221,6 +192,98 @@ namespace MatFileHandler
             return result;
         }
 
+        private static Dictionary<int, EmbeddedObjectInformation> ReadEmbeddedObjectPositionsToValuesMapping(
+            BinaryReader reader, int numberOfObjects)
+        {
+            var result = new Dictionary<int, EmbeddedObjectInformation>();
+            reader.ReadBytes(8);
+            for (var objectPosition = 1; objectPosition <= numberOfObjects; objectPosition++)
+            {
+                var a = reader.ReadInt32();
+                var fieldIndex = reader.ReadInt32();
+                var c = reader.ReadInt32();
+                var valueIndex = reader.ReadInt32();
+                result[objectPosition] = new EmbeddedObjectInformation(fieldIndex, valueIndex);
+            }
+
+            return result;
+        }
+
+        private static string[] ReadFieldNames(byte[] bytes, int startPosition, int numberOfFields)
+        {
+            var result = new string[numberOfFields];
+            var position = startPosition;
+            for (var i = 0; i < numberOfFields; i++)
+            {
+                var list = new List<byte>();
+                while (bytes[position] != 0)
+                {
+                    list.Add(bytes[position]);
+                    position++;
+                }
+
+                result[i] = Encoding.ASCII.GetString(list.ToArray());
+                position++;
+            }
+
+            return result;
+        }
+
+        private static Dictionary<int, int> ReadFieldToFieldDataMapping(BinaryReader reader)
+        {
+            var length = reader.ReadInt32();
+            var result = new Dictionary<int, int>();
+            for (var i = 0; i < length; i++)
+            {
+                var x = reader.ReadInt32();
+                var y = reader.ReadInt32();
+                var index = x * y;
+                var link = reader.ReadInt32();
+                result[index] = link;
+            }
+
+            return result;
+        }
+
+        private static Dictionary<int, ObjectClassInformation> ReadObjectClasses(
+            BinaryReader reader,
+            int numberOfObjects)
+        {
+            var result = new Dictionary<int, ObjectClassInformation>();
+            reader.ReadBytes(24);
+            for (var i = 0; i < numberOfObjects; i++)
+            {
+                var classId = reader.ReadInt32();
+                reader.ReadBytes(8);
+                var embeddedObjectPosition = reader.ReadInt32();
+                var objectPosition = reader.ReadInt32();
+                var loadingOrder = reader.ReadInt32();
+                result[i + 1] =
+                    new ObjectClassInformation(embeddedObjectPosition, objectPosition, loadingOrder, classId);
+            }
+
+            return result;
+        }
+
+        private static Dictionary<int, Dictionary<int, int>> ReadObjectPositionsToValuesMapping(
+            BinaryReader reader,
+            int numberOfObjects)
+        {
+            var result = new Dictionary<int, Dictionary<int, int>>();
+            reader.ReadBytes(8);
+            for (var objectPosition = 1; objectPosition <= numberOfObjects; objectPosition++)
+            {
+                result[objectPosition] = ReadFieldToFieldDataMapping(reader);
+                var position = reader.BaseStream.Position;
+                if (position % 8 != 0)
+                {
+                    reader.ReadBytes(8 - (int)(position % 8));
+                }
+            }
+
+            return result;
+        }
+
         private static (int[] offsets, int newPosition) ReadOffsets(byte[] bytes, int startPosition)
         {
             var position = startPosition;
@@ -238,29 +301,65 @@ namespace MatFileHandler
 
                     break;
                 }
+
                 offsets.Add(next);
             }
 
             return (offsets.ToArray(), position);
         }
 
-        private static string[] ReadFieldNames(byte[] bytes, int startPosition, int numberOfFields)
+        private static IArray TransformOpaqueData(IArray array, SubsystemData subsystemData)
         {
-            var result = new string[numberOfFields];
+            if (array is MatNumericalArrayOf<uint> uintArray)
-            var position = startPosition;
-            for (var i = 0; i < numberOfFields; i++)
             {
-                var list = new List<byte>();
+                if (uintArray.Data[0] == 3707764736u)
-                while (bytes[position] != 0)
                 {
-                    list.Add(bytes[position]);
+                    var (dimensions, indexToObjectId, classIndex) = DataElementReader.ParseOpaqueData(uintArray.Data);
-                    position++;
+                    return new OpaqueLink(
+                        uintArray.Name,
+                        string.Empty,
+                        string.Empty,
+                        dimensions,
+                        array as DataElement,
+                        indexToObjectId,
+                        classIndex,
+                        subsystemData);
                 }
-                result[i] = Encoding.ASCII.GetString(list.ToArray());
-                position++;
             }
 
-            return result;
+            return array;
+        }
+
+        private struct EmbeddedObjectInformation
+        {
+            public EmbeddedObjectInformation(int fieldIndex, int valueIndex)
+            {
+                FieldIndex = fieldIndex;
+                ValueIndex = valueIndex;
+            }
+
+            public int FieldIndex { get; }
+
+            public int ValueIndex { get; }
+        }
+
+        private struct ObjectClassInformation
+        {
+            public ObjectClassInformation(int embeddedObjectPosition, int objectPosition, int loadingOrder, int classId)
+            {
+                EmbeddedObjectPosition = embeddedObjectPosition;
+                ObjectPosition = objectPosition;
+                LoadingOrder = loadingOrder;
+                ClassId = classId;
+            }
+
+            public int ClassId { get; }
+
+            public int EmbeddedObjectPosition { get; }
+
+            public int LoadingOrder { get; }
+
+            public int ObjectPosition { get; }
         }
     }
-}
+}