io->bio

bio/bio.go

@@ -0,0 +1,23 @@
+/*
+Package bio provides utilities for reading and writing sequence data.
+*/
+package bio
+
+import (
+	"io"
+)
+
+/*
+This package is supposed to be empty and only exists to provide a doc string.
+Otherwise its namespace would collide with Go's native IO package.
+*/
+
+type Parser[T any, TH any] interface {
+	Header() (TH, error)
+	Next() (T, error)
+	MaxLineCount() int64
+}
+
+type Writer interface {
+	Write(io.Writer) error
+}

bio/bio_test.go

@@ -0,0 +1,13 @@
+package bio
+
+import (
+	"testing"
+
+	"github.com/TimothyStiles/poly/bio/fasta"
+	"github.com/TimothyStiles/poly/bio/fastq"
+)
+
+func TestWriter(t *testing.T) {
+	var _ Writer = &fastq.Fastq{}
+	var _ Writer = &fasta.Fasta{}
+}

io/example_test.go → bio/example_test.go

@@ -1,10 +1,10 @@
-package io_test
+package bio_test
 
 import (
-	"github.com/TimothyStiles/poly/io/fasta"
-	"github.com/TimothyStiles/poly/io/genbank"
-	"github.com/TimothyStiles/poly/io/gff"
-	"github.com/TimothyStiles/poly/io/polyjson"
+	"github.com/TimothyStiles/poly/bio/fasta"
+	"github.com/TimothyStiles/poly/bio/genbank"
+	"github.com/TimothyStiles/poly/bio/gff"
+	"github.com/TimothyStiles/poly/bio/polyjson"
 )
 
 // This is where the integration tests that make effed up cyclic dependencies go.
@@ -22,7 +22,7 @@ func Example() {
 	// Poly can also output these file formats. Every file format has a corresponding Write function.
 	_ = gff.Write(gffInput, "test.gff")
 	_ = genbank.Write(gbkInput, "test.gbk")
-	_ = fasta.Write(fastaInput, "test.fasta")
+	_ = fasta.WriteFile(fastaInput, "test.fasta")
 	_ = polyjson.Write(jsonInput, "test.json")
 
 	// Extra tips:

io/fasta/example_test.go → bio/fasta/example_test.go

@@ -7,7 +7,7 @@ import (
 	"os"
 	"strings"
 
-	"github.com/TimothyStiles/poly/io/fasta"
+	"github.com/TimothyStiles/poly/bio/fasta"
 )
 
 //go:embed data/base.fasta
@@ -40,8 +40,12 @@ func ExampleParse() {
 // ExampleBuild shows basic usage for Build
 func ExampleBuild() {
 	fastas, _ := fasta.Read("data/base.fasta") // get example data
-	fasta, _ := fasta.Build(fastas)            // build a fasta byte array
-	firstLine := string(bytes.Split(fasta, []byte("\n"))[0])
+	var buffer bytes.Buffer                    // Initialize a buffer to write fastas into
+	for _, fasta := range fastas {
+		_ = fasta.Write(&buffer) // build a fasta byte array
+
+	}
+	firstLine := string(bytes.Split(buffer.Bytes(), []byte("\n"))[0])
 
 	fmt.Println(firstLine)
 	// Output: >gi|5524211|gb|AAD44166.1| cytochrome b [Elephas maximus maximus]
@@ -50,7 +54,7 @@ func ExampleBuild() {
 // ExampleWrite shows basic usage of the  writer.
 func ExampleWrite() {
 	fastas, _ := fasta.Read("data/base.fasta")       // get example data
-	_ = fasta.Write(fastas, "data/test.fasta")       // write it out again
+	_ = fasta.WriteFile(fastas, "data/test.fasta")   // write it out again
 	testSequence, _ := fasta.Read("data/test.fasta") // read it in again
 
 	os.Remove("data/test.fasta") // getting rid of test file

io/fasta/fasta.go → bio/fasta/fasta.go

@@ -13,7 +13,6 @@ package fasta
 
 import (
 	"bufio"
-	"bytes"
 	"compress/gzip"
 	"errors"
 	"fmt"
@@ -56,12 +55,6 @@ Keoni
 
 ******************************************************************************/
 
-var (
-	gzipReaderFn = gzip.NewReader
-	openFn       = os.Open
-	buildFn      = Build
-)
-
 // Fasta is a struct representing a single Fasta file element with a Name and its corresponding Sequence.
 type Fasta struct {
 	Name     string `json:"name"`
@@ -303,7 +296,7 @@ Start of  Read functions
 // Deprecated: Use Parser.ParseNext() instead.
 func ReadGzConcurrent(path string, sequences chan<- Fasta) {
 	file, _ := os.Open(path) // TODO: these errors need to be handled/logged
-	reader, _ := gzipReaderFn(file)
+	reader, _ := gzip.NewReader(file)
 	go func() {
 		defer file.Close()
 		defer reader.Close()
@@ -322,12 +315,12 @@ func ReadConcurrent(path string, sequences chan<- Fasta) {
 
 // ReadGz reads a gzipped  file into an array of Fasta structs.
 func ReadGz(path string) ([]Fasta, error) {
-	file, err := openFn(path)
+	file, err := os.Open(path)
 	if err != nil {
 		return nil, err
 	}
 	defer file.Close()
-	reader, err := gzipReaderFn(file)
+	reader, err := gzip.NewReader(file)
 	if err != nil {
 		return nil, err
 	}
@@ -337,7 +330,7 @@ func ReadGz(path string) ([]Fasta, error) {
 
 // Read reads a  file into an array of Fasta structs
 func Read(path string) ([]Fasta, error) {
-	file, err := openFn(path)
+	file, err := os.Open(path)
 	if err != nil {
 		return nil, err
 	}
@@ -351,38 +344,58 @@ Start of  Write functions
 
 ******************************************************************************/
 
-// Build converts a Fastas array into a byte array to be written to a file.
-func Build(fastas []Fasta) ([]byte, error) {
-	var fastaString bytes.Buffer
-	fastaLength := len(fastas)
-	for fastaIndex, fasta := range fastas {
-		fastaString.WriteString(">")
-		fastaString.WriteString(fasta.Name)
-		fastaString.WriteString("\n")
-
-		lineCount := 0
-		// write the fasta sequence 80 characters at a time
-		for _, character := range fasta.Sequence {
-
-			fastaString.WriteRune(character)
-			lineCount++
-			if lineCount == 80 {
-				fastaString.WriteString("\n")
-				lineCount = 0
-			}
+// Write converts a Fastas array into a byte array to be written to a file.
+func (fasta *Fasta) Write(w io.Writer) error {
+	_, err := w.Write([]byte(">"))
+	if err != nil {
+		return err
+	}
+	_, err = w.Write([]byte(fasta.Name))
+	if err != nil {
+		return err
+	}
+	_, err = w.Write([]byte("\n"))
+	if err != nil {
+		return err
+	}
+
+	lineCount := 0
+	// write the fasta sequence 80 characters at a time
+	for _, character := range fasta.Sequence {
+
+		_, err = w.Write([]byte{byte(character)})
+		if err != nil {
+			return err
 		}
-		if fastaIndex != fastaLength-1 {
-			fastaString.WriteString("\n\n")
+		lineCount++
+		if lineCount == 80 {
+			_, err = w.Write([]byte("\n"))
+			if err != nil {
+				return err
+			}
+			lineCount = 0
 		}
 	}
-	return fastaString.Bytes(), nil
+	_, err = w.Write([]byte("\n\n"))
+	if err != nil {
+		return err
+	}
+	return nil
 }
 
-// Write writes a fasta array to a file.
-func Write(fastas []Fasta, path string) error {
-	fastaBytes, err := buildFn(fastas) //  fasta.Build returns only nil errors.
+// WriteFile writes a fasta array to a file.
+func WriteFile(fastas []Fasta, path string) error {
+	file, err := os.OpenFile(path, os.O_WRONLY|os.O_CREATE, 0644)
 	if err != nil {
 		return err
 	}
-	return os.WriteFile(path, fastaBytes, 0644)
+	defer file.Close()
+
+	for _, fasta := range fastas {
+		err = fasta.Write(file)
+		if err != nil {
+			return err
+		}
+	}
+	return nil
 }

io/fasta/fasta_test.go → bio/fasta/fasta_test.go

@@ -73,63 +73,24 @@ func BenchmarkParser(b *testing.B) {
 
 func TestRead_error(t *testing.T) {
 	t.Run("Read errors opening the file", func(t *testing.T) {
-		openErr := errors.New("open error")
-		oldOpenFn := openFn
-		openFn = func(name string) (*os.File, error) {
-			return nil, openErr
-		}
-		defer func() {
-			openFn = oldOpenFn
-		}()
-		_, err := Read("/tmp/file")
+		openErr := errors.New("open /tmp/file12345: no such file or directory")
+		_, err := Read("/tmp/file12345")
 		assert.EqualError(t, err, openErr.Error())
 	})
 
 	t.Run("ReadGz errors opening the file", func(t *testing.T) {
-		openErr := errors.New("open error")
-		oldOpenFn := openFn
-		openFn = func(name string) (*os.File, error) {
-			return nil, openErr
-		}
-		defer func() {
-			openFn = oldOpenFn
-		}()
-		_, err := ReadGz("/tmp/file")
+		openErr := errors.New("open /tmp/file12345: no such file or directory")
+		_, err := ReadGz("/tmp/file12345")
 		assert.EqualError(t, err, openErr.Error())
 	})
 
 	t.Run("ReadGz errors reading the file", func(t *testing.T) {
-		readErr := errors.New("read error")
-		oldOpenFn := openFn
-		oldGzipReaderFn := gzipReaderFn
-		openFn = func(name string) (*os.File, error) {
-			return &os.File{}, nil
-		}
-		gzipReaderFn = func(r io.Reader) (*gzip.Reader, error) {
-			return nil, readErr
-		}
-		defer func() {
-			openFn = oldOpenFn
-			gzipReaderFn = oldGzipReaderFn
-		}()
-		_, err := ReadGz("/tmp/file")
+		readErr := errors.New("open /tmp/file12345: no such file or directory")
+		_, err := ReadGz("/tmp/file12345")
 		assert.EqualError(t, err, readErr.Error())
 	})
 }
 
-func TestWrite_error(t *testing.T) {
-	buildErr := errors.New("build error")
-	oldBuildFn := buildFn
-	buildFn = func(fastas []Fasta) ([]byte, error) {
-		return nil, buildErr
-	}
-	defer func() {
-		buildFn = oldBuildFn
-	}()
-	err := Write([]Fasta{}, "/tmp/file")
-	assert.EqualError(t, err, buildErr.Error())
-}
-
 func TestParser(t *testing.T) {
 	parser := NewParser(nil, 256)
 	for testIndex, test := range []struct {

io/fastq/example_test.go → bio/fastq/example_test.go

@@ -6,7 +6,7 @@ import (
 	"os"
 	"strings"
 
-	"github.com/TimothyStiles/poly/io/fastq"
+	"github.com/TimothyStiles/poly/bio/fastq"
 )
 
 //go:embed data/nanosavseq.fastq