Merge pull request #10 from iamcarbon/cq11

Improve Code Quality (7 of x)

Src/Protsyk.PMS.FullText.Core.UnitTests/TextPositionTests.cs

@@ -22,31 +22,31 @@ private void TextIndex(IFullTextIndex index)
     {
         Assert.Equal(new TextPosition[]
                         {
-                            TextPosition.P(0, 5),
-                            TextPosition.P(6, 5)
+                            new(0, 5),
+                            new(6, 5)
                         },
                         index.GetPositions(1, 1));
 
         Assert.Equal(new TextPosition[]
                         {
-                            TextPosition.P(0, 5),
-                            TextPosition.P(6, 9),
-                            TextPosition.P(16, 6)
+                            new(0, 5),
+                            new(6, 9),
+                            new(16, 6)
                         },
                         index.GetPositions(2, 1));
 
         Assert.Equal(new TextPosition[]
                         {
-                            TextPosition.P(0, 11),
-                            TextPosition.P(12, 2),
-                            TextPosition.P(15, 4),
-                            TextPosition.P(20, 8),
-                            TextPosition.P(30, 8),
-                            TextPosition.P(39, 3),
-                            TextPosition.P(43, 4),
-                            TextPosition.P(49, 4),
-                            TextPosition.P(54, 2),
-                            TextPosition.P(57, 9)
+                            new(0, 11),
+                            new(12, 2),
+                            new(15, 4),
+                            new(20, 8),
+                            new(30, 8),
+                            new(39, 3),
+                            new(43, 4),
+                            new(49, 4),
+                            new(54, 2),
+                            new(57, 9)
                         },
                     index.GetPositions(6, 1));
 
@@ -71,4 +71,15 @@ private void TextIndex(IFullTextIndex index)
         Assert.Equal("very", doc.TokenAt(15));
         Assert.Equal("fantastic", doc.TokenAt(60));
     }
+
+    [Theory]
+    [InlineData("[0,0]", 0, 0)]
+    [InlineData("[1,5]", 1, 5)]
+    public void CanParse(string text, int offset, int length)
+    {
+        var result = TextPosition.Parse(text);
+
+        Assert.Equal(offset, result.Offset);
+        Assert.Equal(length, result.Length);
+    }
 }

Src/Protsyk.PMS.FullText.Core/Automata/DFA.cs

@@ -7,7 +7,7 @@ namespace Protsyk.PMS.FullText.Core.Automata;
 /// </summary>
 public class DFA
 {
-    public static readonly int NoState = -1;
+    public const int NoState = -1;
 
     private readonly List<List<(CharRange, int)>> transitions = new();
     private readonly HashSet<int> final = new ();
@@ -33,7 +33,7 @@ public void AddTransition(int from, int to, CharRange c)
 
     public int Next(int s, char c)
     {
-        if (s == NoState)
+        if (s is NoState)
         {
             return NoState;
         }

Src/Protsyk.PMS.FullText.Core/Automata/NFA.cs

@@ -126,7 +126,7 @@ private void EpsilonClosure(HashSet<int> setState)
         return transitions.Where(t => t.Item1 == from_state);
     }
 
-    class SetStateComparer : IEqualityComparer<HashSet<int>>
+    private sealed class SetStateComparer : IEqualityComparer<HashSet<int>>
     {
         public bool Equals(HashSet<int> x, HashSet<int> y)
         {

Src/Protsyk.PMS.FullText.Core/Collections/BtreePersistent.cs

@@ -1405,8 +1405,8 @@ public static DataLink FromBytes(ReadOnlySpan<byte> buffer)
 
     private readonly struct NodeData
     {
-        public static readonly int HeaderLength = 3 * sizeof(int);
-        public static readonly int DataHeaderLength = HeaderLength + sizeof(int);
+        public const int HeaderLength = 3 * sizeof(int);
+        public const int DataHeaderLength = HeaderLength + sizeof(int);
 
         private readonly byte[] data;
 

Src/Protsyk.PMS.FullText.Core/Collections/PersistentHashTable.cs

@@ -12,7 +12,7 @@ public class PersistentHashTable<TKey, TValue> : IEnumerable<KeyValuePair<TKey,
     private const int MinCapacity = 8;
     private static ReadOnlySpan<byte> Header => "PMS-HASH"u8;
     private static readonly int HeaderSize = Header.Length;
-    private static readonly int IndexRecordSize = sizeof(long) + sizeof(int);
+    private const int IndexRecordSize = sizeof(long) + sizeof(int);
 
     private readonly IDataSerializer<TKey> keySerializer;
     private readonly IDataSerializer<TValue> valueSerializer;
@@ -54,7 +54,7 @@ public PersistentHashTable(IPersistentStorage dataStorage, int capacity, IEquali
             this.dataStorage.WriteInt32LittleEndian(HeaderSize + sizeof(int), count);
 
             // Index
-            var zero = new byte[IndexRecordSize*1024];
+            var zero = new byte[IndexRecordSize * 1_024];
             var i = 0;
             while (i < capacity)
             {
@@ -63,7 +63,7 @@ public PersistentHashTable(IPersistentStorage dataStorage, int capacity, IEquali
                 {
                     toWrite = (capacity-i) % 1024;
                 }
-                this.dataStorage.WriteAll(headerSize + i*IndexRecordSize, zero.AsSpan(0, IndexRecordSize * toWrite));
+                this.dataStorage.WriteAll(headerSize + i* IndexRecordSize, zero.AsSpan(0, IndexRecordSize * toWrite));
                 i += toWrite;
             }
         }

Src/Protsyk.PMS.FullText.Core/Common/Compression/CombinationList.cs

@@ -5,11 +5,11 @@ namespace Protsyk.PMS.FullText.Core.Common.Compression;
 
 public partial class HuTuckerBuilder
 {
-    class CombinationList
+    private sealed class CombinationList
     {
         ListItem First { get; set; }
         ListItem Last { get; set; }
-        public List<TreeLeaf> list = new List<TreeLeaf>();
+        public List<TreeLeaf> list = new();
 
         public void Add(TreeLeaf item)
         {
@@ -34,10 +34,12 @@ public void Add(TreeLeaf item)
 
         public TreeNode Combine(ListItem left, ListItem right)
         {
-            var node = new TreeNode();
-            node.Left = left;
-            node.Right = right;
-            node.W = left.W + right.W;
+            var node = new TreeNode
+            {
+                Left = left,
+                Right = right,
+                W = left.W + right.W
+            };
 
             if (left.Next != right)
             {

Src/Protsyk.PMS.FullText.Core/Common/Compression/HuffmanEncodingBuilder.cs

@@ -66,15 +66,15 @@ public string GetCode()
         }
     }
 
-    class HuffmanEncoding : VarLenCharEncoding
+    private sealed class HuffmanEncoding : VarLenCharEncoding
     {
         public HuffmanEncoding(IEncodingNode root)
             : base(root)
         {
         }
     }
 
-    class TreeNode : IEncodingNode
+    private class TreeNode : IEncodingNode
     {
         private static int nextId = 0;
 
@@ -88,7 +88,7 @@ public TreeNode(int weight)
         }
     }
 
-    class CharNode : TreeNode, IEncodingLeafNode
+    private sealed class CharNode : TreeNode, IEncodingLeafNode
     {
         public readonly char c;
 
@@ -101,7 +101,7 @@ public CharNode(char c, int weight)
         char IEncodingLeafNode.V => c;
     }
 
-    class CombineNode : TreeNode, IEncodingTreeNode
+    private sealed class CombineNode : TreeNode, IEncodingTreeNode
     {
         public readonly TreeNode left;
         public readonly TreeNode right;

Src/Protsyk.PMS.FullText.Core/Common/Compression/VarLenCharEncoding.cs

@@ -450,7 +450,7 @@ IEnumerator IEnumerable.GetEnumerator()
             return GetEnumerator();
         }
 
-        class ByteToBitEn : IEnumerator<byte>
+        private sealed class ByteToBitEn : IEnumerator<byte>
         {
             private readonly IEnumerator<byte> data;
             private int c;

Src/Protsyk.PMS.FullText.Core/Common/Compression/VarLenEncoding.cs

@@ -30,5 +30,5 @@ public int GetMaxEncodedLength(int maxTokenLength)
         => 4 * maxTokenLength;
 
     public IDfaMatcher<byte> CreateMatcher(IDfaMatcher<char> charMatcher, int maxLength)
-        => (IDfaMatcher<byte>)new DecodingMatcherForVarLenCharEncoding(charMatcher, encoding);
+        => new DecodingMatcherForVarLenCharEncoding(charMatcher, encoding);
 }

Src/Protsyk.PMS.FullText.Core/Helpers/StringSplitter.cs

@@ -0,0 +1,45 @@
+namespace Protsyk.PMS;
+
+internal ref struct StringSplitter
+{
+    private readonly ReadOnlySpan<char> _text;
+    private readonly char _separator;
+
+    private int _position;
+
+    public StringSplitter(ReadOnlySpan<char> text, char separator)
+    {
+        _text = text;
+        _separator = separator;
+        _position = 0;
+    }
+
+    public bool TryRead(out ReadOnlySpan<char> result)
+    {
+        if (_position == _text.Length)
+        {
+            result = default;
+
+            return false;
+        }
+
+        int start = _position;
+
+        int separatorIndex = _text.Slice(_position).IndexOf(_separator);
+
+        if (separatorIndex > -1)
+        {
+            _position += separatorIndex + 1;
+
+            result = _text.Slice(start, separatorIndex);
+        }
+        else
+        {
+            _position = _text.Length;
+
+            result = _text.Slice(start);
+        }
+
+        return true;
+    }
+}

Src/Protsyk.PMS.FullText.Core/IndexModels/Occurrence.cs

@@ -1,72 +1,67 @@
 using System.Globalization;
-using System.Text.RegularExpressions;
 
 namespace Protsyk.PMS.FullText.Core;
 
 public readonly struct Occurrence : IEquatable<Occurrence>, IComparable<Occurrence>
 {
-    #region Fields
     /// <summary>
     /// Not valid Id
     /// </summary>
-    public static readonly ulong NoId = 0;
+    public const ulong NoId = 0;
 
     /// <summary>
     /// Empty occurrence
     /// </summary>
     public static readonly Occurrence Empty = new(NoId, NoId, NoId);
 
-    private static readonly Regex regParse = new Regex("^\\[(?<docId>\\d+),(?<fieldId>\\d+),(?<tokenId>\\d+)\\]$",
-                                                       RegexOptions.Compiled | RegexOptions.Singleline);
+    public Occurrence(ulong documentId, ulong fieldId, ulong tokenId)
+    {
+        DocumentId = documentId;
+        FieldId = fieldId;
+        TokenId = tokenId;
+    }
 
     /// <summary>
     /// Document Id
     /// </summary>
-    public readonly ulong DocumentId;
+    public ulong DocumentId { get; }
 
     /// <summary>
     /// Fields Id
     /// </summary>
-    public readonly ulong FieldId;
+    public ulong FieldId { get; }
 
     /// <summary>
     /// Token Id
     /// </summary>
-    public readonly ulong TokenId;
-    #endregion
+    public ulong TokenId { get; }
 
     #region Static Methods
 
-    public static Occurrence Parse(string text)
+    public static Occurrence Parse(ReadOnlySpan<char> text)
     {
-        var match = regParse.Match(text);
-        if (!match.Success)
+        if (text is [ '[', .. var inner, ']' ])
         {
-            throw new InvalidOperationException($"Occurrence text has invalid format: {text}");
+            var splitter = new StringSplitter(inner, ',');
+
+            if (splitter.TryRead(out var t0) && ulong.TryParse(t0, NumberStyles.None, CultureInfo.InvariantCulture, out ulong docId) &&
+                splitter.TryRead(out var t1) && ulong.TryParse(t1, NumberStyles.None, CultureInfo.InvariantCulture, out ulong fieldId) &&
+                splitter.TryRead(out var t2) && ulong.TryParse(t2, NumberStyles.None, CultureInfo.InvariantCulture, out ulong tokenId))
+            {
+                return new Occurrence(docId, fieldId, tokenId);
+            }           
         }
 
-        return new Occurrence(
-            ulong.Parse(match.Groups["docId"].ValueSpan, provider: CultureInfo.InvariantCulture),
-            ulong.Parse(match.Groups["fieldId"].ValueSpan, provider: CultureInfo.InvariantCulture),
-            ulong.Parse(match.Groups["tokenId"].ValueSpan, provider: CultureInfo.InvariantCulture));
+        throw new InvalidOperationException($"Occurrence text has invalid format. Was {text}");
     }
-    #endregion
 
-    #region Methods
-    public Occurrence(ulong documentId, ulong fieldId, ulong tokenId)
-    {
-        this.DocumentId = documentId;
-        this.FieldId = fieldId;
-        this.TokenId = tokenId;
-    }
+    #endregion   
 
     public override string ToString()
     {
         return string.Create(CultureInfo.InvariantCulture, $"[{DocumentId},{FieldId},{TokenId}]");
     }
-    #endregion
-
-    #region IEquatable<Occurrence>
+
     public bool Equals(Occurrence other)
     {
         return DocumentId == other.DocumentId &&
@@ -81,8 +76,6 @@ public override bool Equals(object obj)
 
     public override int GetHashCode() => HashCode.Combine(DocumentId, FieldId, TokenId);
 
-    #endregion
-
     #region IComparable
     public int CompareTo(Occurrence other)
     {

Src/Protsyk.PMS.FullText.Core/IndexModels/PostingListAddress.cs

@@ -2,7 +2,7 @@
 
 public readonly struct PostingListAddress : IEquatable<PostingListAddress>
 {
-    public static PostingListAddress Null = new(-1);
+    public static readonly PostingListAddress Null = new(-1);
 
     public readonly long Offset;