issue-3: rationalize fastq readers

michaeljon · May 27, 2022 · 1b94999 · 1b94999
1 parent 7274b61
commit 1b94999
Show file tree

Hide file tree

Showing 7 changed files with 114 additions and 285 deletions.
diff --git a/Readers/AbstractReader.cs b/Readers/AbstractReader.cs
@@ -0,0 +1,70 @@
+using System;
+using System.IO;
+using System.IO.Compression;
+
+namespace Ovation.FasterQC.Net
+{
+    public abstract class AbstractReader : ISequenceReader
+    {
+        private readonly FileStream inputStream;
+
+        private readonly GZipStream? gzipStream;
+
+        protected readonly BufferedStream bufferedStream;
+
+        protected readonly int bufferSize = 128 * 1024;
+
+        private bool disposedValue;
+
+        protected ulong sequencesRead = 0;
+
+        public ulong SequencesRead => sequencesRead;
+
+        public double ApproximateCompletion =>
+            100.0 * inputStream.Position / inputStream.Length;
+
+        public AbstractReader(string filename, bool gzipped = true)
+        {
+            var fileStreamOptions = new FileStreamOptions()
+            {
+                Mode = FileMode.Open,
+                BufferSize = bufferSize,
+            };
+
+            if (gzipped == true)
+            {
+                inputStream = File.Open(filename, fileStreamOptions);
+                gzipStream = new GZipStream(inputStream, CompressionMode.Decompress);
+                bufferedStream = new BufferedStream(gzipStream, bufferSize);
+            }
+            else
+            {
+                inputStream = File.Open(filename, fileStreamOptions);
+                bufferedStream = new BufferedStream(inputStream, bufferSize);
+            }
+        }
+
+        public abstract bool ReadSequence(out Sequence? sequence);
+
+        protected virtual void Dispose(bool disposing)
+        {
+            if (!disposedValue)
+            {
+                if (disposing)
+                {
+                    bufferedStream?.Dispose();
+                    gzipStream?.Dispose();
+                    inputStream?.Dispose();
+                }
+
+                disposedValue = true;
+            }
+        }
+
+        public void Dispose()
+        {
+            Dispose(true);
+            GC.SuppressFinalize(this);
+        }
+    }
+}
diff --git a/Readers/BamReader.cs b/Readers/BamReader.cs
@@ -1,46 +1,25 @@
 using System;
 using System.IO;
-using System.IO.Compression;
 using System.Text;
 using static Ovation.FasterQC.Net.Utils.CliOptions;
 
 namespace Ovation.FasterQC.Net
 {
-    public class BamReader : ISequenceReader
+    public class BamReader : AbstractReader
     {
-        private readonly FileStream inputStream;
-
-        private readonly GZipStream gzipStream;
-
-        private readonly BufferedStream bufferedStream;
-
         private readonly BinaryReader binaryReader;
 
         private bool disposedValue;
 
-        private ulong sequencesRead = 0;
-
-        public ulong SequencesRead => sequencesRead;
-
-        public BamReader(string bam)
+        public BamReader(string bam) :
+            base(bam, false)
         {
-            var bufferSize = 128 * 1024;
-
-            var fileStreamOptions = new FileStreamOptions()
-            {
-                Mode = FileMode.Open,
-                BufferSize = bufferSize,
-            };
-
-            inputStream = File.Open(bam, fileStreamOptions);
-            gzipStream = new GZipStream(inputStream, CompressionMode.Decompress);
-            bufferedStream = new BufferedStream(gzipStream, bufferSize);
             binaryReader = new BinaryReader(bufferedStream);
 
             ConsumeHeader();
         }
 
-        public bool ReadSequence(out Sequence? sequence)
+        public override bool ReadSequence(out Sequence? sequence)
         {
             try
             {
@@ -57,9 +36,6 @@ public bool ReadSequence(out Sequence? sequence)
             }
         }
 
-        public double ApproximateCompletion =>
-            100.0 * inputStream.Position / inputStream.Length;
-
         private void ConsumeHeader()
         {
             var magic = binaryReader.ReadBytes(4);
@@ -174,26 +150,17 @@ private BamAlignment ReadSequence()
             return bamAlignment;
         }
 
-        protected virtual void Dispose(bool disposing)
+        protected override void Dispose(bool disposing)
         {
             if (!disposedValue)
             {
                 if (disposing)
                 {
                     binaryReader?.Dispose();
-                    bufferedStream?.Dispose();
-                    gzipStream?.Dispose();
-                    inputStream?.Dispose();
                 }
 
                 disposedValue = true;
             }
         }
-
-        public void Dispose()
-        {
-            Dispose(true);
-            GC.SuppressFinalize(this);
-        }
     }
 }
diff --git a/Readers/FastqLineReader.cs b/Readers/FastqLineReader.cs
diff --git a/Readers/FastqReader.cs b/Readers/FastqReader.cs
@@ -1,86 +1,39 @@
 using System;
 using System.IO;
-using System.IO.Compression;
+using System.Text;
 using static Ovation.FasterQC.Net.Utils.CliOptions;
 
 namespace Ovation.FasterQC.Net
 {
-    public class FastqReader : ISequenceReader
+    public class FastqReader : AbstractReader
     {
-        private readonly FileStream inputStream;
-
-        private readonly GZipStream? gzipStream;
-
-        private readonly BufferedStream bufferedStream;
-
-        private readonly BinaryReader binaryReader;
+        protected readonly StreamReader streamReader;
 
         private bool disposedValue;
 
-        private ulong sequencesRead = 0;
-
-        public ulong SequencesRead => sequencesRead;
-
-        public double ApproximateCompletion =>
-            100.0 * inputStream.Position / inputStream.Length;
-
-        public FastqReader(string fastq, bool gzipped = true)
+        public FastqReader(string fastq, bool gzipped)
+            : base(fastq, gzipped)
         {
-            var bufferSize = 128 * 1024;
-
-            var fileStreamOptions = new FileStreamOptions()
-            {
-                Mode = FileMode.Open,
-                BufferSize = bufferSize,
-            };
-
-            if (gzipped == true)
-            {
-                inputStream = File.Open(fastq, fileStreamOptions);
-                gzipStream = new GZipStream(inputStream, CompressionMode.Decompress);
-                bufferedStream = new BufferedStream(gzipStream, bufferSize);
-                binaryReader = new BinaryReader(bufferedStream);
-            }
-            else
-            {
-                inputStream = File.Open(fastq, fileStreamOptions);
-                bufferedStream = new BufferedStream(inputStream, bufferSize);
-                binaryReader = new BinaryReader(bufferedStream);
-            }
+            streamReader = new StreamReader(bufferedStream, Encoding.ASCII, false, bufferSize);
         }
 
-        public bool ReadSequence(out Sequence? sequence)
+        public override bool ReadSequence(out Sequence? sequence)
         {
-            // this is clearly dangerous, instead read a large chunk of the file
-            // and then walk through it returning only the consumed portion while
-            // keeping track of the last byte consumed on the stream
-            byte[] bytes = new byte[1024];
-
-            int offset = 0;
-            int line = 0;
-            int[] endOfLines = new int[4];
-
             try
             {
-                while (line < 4)
+                if (streamReader.EndOfStream == true)
                 {
-                    var b = binaryReader.ReadByte();
-                    if (b == (byte)'\n')
-                    {
-                        endOfLines[line++] = offset;
-                    }
-                    else
-                    {
-                        bytes[offset++] = b;
-                    }
+                    On(Settings.Verbose, () => Console.Error.WriteLine("End of stream"));
+                    sequence = null;
+                    return false;
                 }
 
-                for (var read = endOfLines[1]; read < endOfLines[2]; read++)
-                {
-                    bytes[read] &= 0xdf;
-                }
+                var identifier = Encoding.ASCII.GetBytes(streamReader.ReadLine() ?? "");
+                var read = Encoding.ASCII.GetBytes(streamReader.ReadLine() ?? "");
+                var blank = Encoding.ASCII.GetBytes(streamReader.ReadLine() ?? "");
+                var quality = Encoding.ASCII.GetBytes(streamReader.ReadLine() ?? "");
 
-                sequence = new Sequence(bytes, endOfLines);
+                sequence = new Sequence(0, identifier, read, blank, quality);
                 sequencesRead++;
                 return true;
             }
@@ -92,26 +45,17 @@ public bool ReadSequence(out Sequence? sequence)
             }
         }
 
-        protected virtual void Dispose(bool disposing)
+        protected override void Dispose(bool disposing)
         {
             if (!disposedValue)
             {
                 if (disposing)
                 {
-                    binaryReader?.Dispose();
-                    bufferedStream?.Dispose();
-                    gzipStream?.Dispose();
-                    inputStream?.Dispose();
+                    streamReader?.Dispose();
                 }
 
                 disposedValue = true;
             }
         }
-
-        public void Dispose()
-        {
-            Dispose(true);
-            GC.SuppressFinalize(this);
-        }
     }
 }