CsvFieldIndexer.Search: Naive search implementation.

CsvLib.Tests/CsvFieldIndexerTests.cs

@@ -1,3 +1,4 @@
+using System.Text;
 using CsvLib;
 
 namespace CvsLib;
@@ -172,4 +173,77 @@ public class CsvFieldIndexerTests
 
     #endregion GenerateIndex
 
+    #region Search
+    
+    [Fact]
+    public void Search__TwoLinesWithTwoQuotedColumns__OneIndexFirstRow()
+    {
+        // --- Arrange
+        string strText = """
+                         "Hello","test"
+                         "Hello","World"
+                         """;
+        StringReader sr = new(strText);
+        CsvFieldIndexer indexer = new();
+        indexer.GenerateIndex(sr);
+
+        // --- Act
+        byte[] bText = Encoding.UTF8.GetBytes(strText);
+        MemoryStream ms = new(bText);
+        List<long> indexes = indexer.Search(ms, "test");
+
+        // --- Assert
+
+        Assert.Single(indexes);
+        Assert.Equal(0, indexes[0]);
+    }
+
+    [Fact]
+    public void Search__TwoLinesWithTwoQuotedColumns__OneIndexSecondRow()
+    {
+        // --- Arrange
+        string strText = """
+                         "Hello","World"
+                         "Hello","test"
+                         """;
+        StringReader sr = new(strText);
+        CsvFieldIndexer indexer = new();
+        indexer.GenerateIndex(sr);
+
+        // --- Act
+        byte[] bText = Encoding.UTF8.GetBytes(strText);
+        MemoryStream ms = new(bText);
+        List<long> indexes = indexer.Search(ms, "test");
+
+        // --- Assert
+
+        Assert.Single(indexes);
+        Assert.Equal(16, indexes[0]);
+    }
+
+    [Fact]
+    public void Search__TwoLinesWithTwoQuotedColumnsTwoMatches__OneIndexSecondRow()
+    {
+        // --- Arrange
+        string strText = """
+                         "Hello","World"
+                         "test","test"
+                         """;
+        StringReader sr = new(strText);
+        CsvFieldIndexer indexer = new();
+        indexer.GenerateIndex(sr);
+
+        // --- Act
+        byte[] bText = Encoding.UTF8.GetBytes(strText);
+        MemoryStream ms = new(bText);
+        List<long> indexes = indexer.Search(ms, "test");
+
+        // --- Assert
+
+        Assert.Single(indexes);
+        Assert.Equal(16, indexes[0]);
+    }
+    
+    #endregion Search
+    
 }

CsvLib/BufferedTextReader.cs

@@ -39,11 +39,6 @@ namespace CsvLib
             return read;
         }
 
-        public override int Peek()
-        {
-            return _baseReader.Peek();
-        }
-
         public int Position
         {
             get { return _position; }

CsvLib/ByteArraySearcher.cs

@@ -32,8 +32,7 @@ public class ByteArraySearcher
                 }
             }
 
-            if (found)
-                return true;
+            if (found) { return true; }
         }
 
         return false;

CsvLib/CsvFieldIndexer.cs

@@ -119,6 +119,7 @@ namespace CsvLib
             using FileStream stream = new(file, FileMode.Open);
             using StreamReader streamReader = new(stream, Encoding.Default, true, 4096);
             GenerateIndex(streamReader);
+            stream.Close();
         }
 
         public void GenerateIndex(TextReader textReader)
@@ -263,5 +264,61 @@ namespace CsvLib
                 SaveFile(indexFile);
             }
         }
+
+        public List<long> Search(string fileName, string textToSearch, Action<float> notifyProgress = null)
+        {
+            List<long> index;
+            using FileStream streamIn = new(fileName, FileMode.Open);
+            try
+            {
+                index = Search(streamIn, textToSearch, notifyProgress);
+            }
+            finally
+            {
+                streamIn.Close();
+            }
+            return index ?? new List<long>();
+        }
+        
+        public List<long> Search(Stream streamIn, string textToSearch, Action<float> notifyProgress = null)
+        {
+            // TODO: Use MemoryMappedFile for better IO performance
+            DateTime datePrevious = DateTime.UtcNow;
+            List<long> newIndexes = new();
+            byte[] bText = Encoding.UTF8.GetBytes(textToSearch);
+            ByteArraySearcher searcher = new(bText);
+            byte[] buffer = new byte[1024];
+            for (int j = 0; j < _fieldIndex.Count; j++)
+            {
+                for (int i = 0; i < _fieldIndex[j].Count; i += 2)
+                {
+                    TimeSpan tsElapsed = DateTime.UtcNow - datePrevious;
+                    if (tsElapsed.TotalMilliseconds > 200)
+                    {
+                        datePrevious = DateTime.UtcNow;
+                        notifyProgress?.Invoke(j/(float)_fieldIndex.Count);
+                    }
+                    
+                    long offset = _fieldIndex[j][i];
+                    int length = (int)(_fieldIndex[j][i + 1] - offset) + 1;
+                    
+                    if (buffer.Length < length)
+                    {
+                        buffer = new byte[length];
+                    }
+                    streamIn.Seek(offset, SeekOrigin.Begin);
+                    int read = streamIn.Read(buffer, 0, length);
+                    if (read != length) { throw new Exception($"Search: Expected {length} bytes, but read {read}"); }
+
+                    bool matches = searcher.Contains(buffer, length);
+                    if (matches == false) { continue; }
+                    
+                    newIndexes.Add(_index[j]);
+                    break;
+                }
+            }
+
+            return newIndexes;
+        }
     }
 }