Refactor and add a names subcommand

Author: bjeight

Diff for: cli.go

@@ -11,7 +11,7 @@ var (
 		Use:               "fastats {command}",
 		Short:             "Very simple statistics from fasta files",
 		Long:              ``,
-		Version:           "0.5.0",
+		Version:           "0.6.0",
 		CompletionOptions: cobra.CompletionOptions{DisableDefaultCmd: true},
 	}
 )
@@ -26,13 +26,16 @@ func main() {
 var f bool
 var c bool
 var p string
+var d bool
 
 func init() {
 	rootCmd.PersistentFlags().BoolVarP(&f, "file", "f", false, "calculate statistics per file (default is per record)")
 	rootCmd.PersistentFlags().BoolVarP(&c, "count", "c", false, "print counts (default is proportions)")
+	rootCmd.PersistentFlags().BoolVarP(&d, "description", "d", false, "write record descriptions (default is IDs)")
 
 	rootCmd.PersistentFlags().Lookup("file").NoOptDefVal = "true"
 	rootCmd.PersistentFlags().Lookup("count").NoOptDefVal = "true"
+	rootCmd.PersistentFlags().Lookup("description").NoOptDefVal = "true"
 
 	rootCmd.AddCommand(atCmd)
 	rootCmd.AddCommand(gcCmd)
@@ -43,6 +46,7 @@ func init() {
 	rootCmd.AddCommand(softCmd)
 	rootCmd.AddCommand(patternCmd)
 	rootCmd.AddCommand(numCmd)
+	rootCmd.AddCommand(nameCmd)
 
 	patternCmd.Flags().StringVarP(&p, "pattern", "p", "", "arbitrary pattern to parse")
 }
@@ -52,7 +56,7 @@ var atCmd = &cobra.Command{
 	Short:                 "AT content",
 	DisableFlagsInUseLine: true,
 	RunE: func(cmd *cobra.Command, args []string) (err error) {
-		err = pattern(os.Stdout, args, "ATat", f, c)
+		err = pattern(os.Stdout, args, "ATat", f, c, d)
 		return err
 	},
 }
@@ -62,7 +66,7 @@ var gcCmd = &cobra.Command{
 	Short:                 "GC content",
 	DisableFlagsInUseLine: true,
 	RunE: func(cmd *cobra.Command, args []string) (err error) {
-		err = pattern(os.Stdout, args, "GCgc", f, c)
+		err = pattern(os.Stdout, args, "GCgc", f, c, d)
 		return err
 	},
 }
@@ -72,7 +76,7 @@ var atgcCmd = &cobra.Command{
 	Short:                 "ATGC content",
 	DisableFlagsInUseLine: true,
 	RunE: func(cmd *cobra.Command, args []string) (err error) {
-		err = pattern(os.Stdout, args, "ATGCatgc", f, c)
+		err = pattern(os.Stdout, args, "ATGCatgc", f, c, d)
 		return err
 	},
 }
@@ -82,7 +86,7 @@ var nCmd = &cobra.Command{
 	Short:                 "N content",
 	DisableFlagsInUseLine: true,
 	RunE: func(cmd *cobra.Command, args []string) (err error) {
-		err = pattern(os.Stdout, args, "Nn", f, c)
+		err = pattern(os.Stdout, args, "Nn", f, c, d)
 		return err
 	},
 }
@@ -92,7 +96,7 @@ var gapCmd = &cobra.Command{
 	Short:                 "Gap content",
 	DisableFlagsInUseLine: true,
 	RunE: func(cmd *cobra.Command, args []string) (err error) {
-		err = pattern(os.Stdout, args, "-", f, c)
+		err = pattern(os.Stdout, args, "-", f, c, d)
 		return err
 	},
 }
@@ -102,7 +106,7 @@ var softCmd = &cobra.Command{
 	Short:                 "Softmasked content",
 	DisableFlagsInUseLine: true,
 	RunE: func(cmd *cobra.Command, args []string) (err error) {
-		err = pattern(os.Stdout, args, "atgcn", f, c)
+		err = pattern(os.Stdout, args, "atgcn", f, c, d)
 		return err
 	},
 }
@@ -114,7 +118,7 @@ var patternCmd = &cobra.Command{
 	Short:                 "Arbitrary PATTERN content",
 	DisableFlagsInUseLine: true,
 	RunE: func(cmd *cobra.Command, args []string) (err error) {
-		err = pattern(os.Stdout, args, p, f, c)
+		err = pattern(os.Stdout, args, p, f, c, d)
 		return err
 	},
 }
@@ -124,7 +128,7 @@ var lenCmd = &cobra.Command{
 	Short:                 "Sequence length",
 	DisableFlagsInUseLine: true,
 	RunE: func(cmd *cobra.Command, args []string) (err error) {
-		err = length(os.Stdout, args, p, f, c)
+		err = length(os.Stdout, args, p, f, c, d)
 		return err
 	},
 }
@@ -134,7 +138,17 @@ var numCmd = &cobra.Command{
 	Short:                 "Number of records",
 	DisableFlagsInUseLine: true,
 	RunE: func(cmd *cobra.Command, args []string) (err error) {
-		err = num(os.Stdout, args, p, f, c)
+		err = num(os.Stdout, args, p, f, c, d)
+		return err
+	},
+}
+
+var nameCmd = &cobra.Command{
+	Use:                   "names <infile[s]>",
+	Short:                 "Record names",
+	DisableFlagsInUseLine: true,
+	RunE: func(cmd *cobra.Command, args []string) (err error) {
+		err = names(os.Stdout, args, p, f, c, d)
 		return err
 	},
 }

Diff for: fasta.go renamed to fasta/fasta.go

@@ -1,4 +1,4 @@
-package main
+package fasta
 
 import (
 	"bufio"
@@ -124,3 +124,13 @@ func (r *Reader) Read() (FastaRecord, error) {
 
 	return FR, err
 }
+
+func (r *Reader) Seek(offset int) error {
+	for i := 0; i < offset; i++ {
+		_, err := r.r.ReadByte()
+		if err != nil {
+			return err
+		}
+	}
+	return nil
+}

Diff for: fasta_test.go renamed to fasta/fasta_test.go

@@ -1,4 +1,4 @@
-package main
+package fasta
 
 import (
 	"bytes"

Diff for: len.go

@@ -3,13 +3,15 @@ package main
 import (
 	"fmt"
 	"io"
+
+	"github.com/bjeight/fastats/fasta"
 )
 
 // length() is fastats len in the cli. It writes the appropriate header (which depends on the cli
 // arguments), then passes lengthRecords() + the cli arguments + the writer to collectCommandLine,
 // which processes the fasta file(s) from the command line or stdin, depending on what is provided
 // by the user.
-func length(w io.Writer, filepaths []string, pattern string, file bool, counts bool) error {
+func length(w io.Writer, filepaths []string, pattern string, file bool, counts bool, description bool) error {
 
 	// print the correct header to stdout, depending on whether the statistics are
 	// to be calculated per file or per record
@@ -26,7 +28,7 @@ func length(w io.Writer, filepaths []string, pattern string, file bool, counts b
 	}
 
 	// pass lengthRecords + the cli arguments to template() for processesing the fasta file(s)
-	err := collectCommandLine(w, lengthRecords, filepaths, pattern, file, counts)
+	err := collectCommandLine(w, lengthRecords, filepaths, pattern, file, counts, description)
 	if err != nil {
 		return err
 	}
@@ -35,7 +37,7 @@ func length(w io.Writer, filepaths []string, pattern string, file bool, counts b
 }
 
 // lengthRecords does the work of fastats len for one fasta file at a time.
-func lengthRecords(r *Reader, args arguments, w io.Writer) error {
+func lengthRecords(r *fasta.Reader, args arguments, w io.Writer) error {
 
 	// get the file name in case we need to print it to stdout
 	filename := filenameFromFullPath(args.filepath)
@@ -57,7 +59,7 @@ func lengthRecords(r *Reader, args arguments, w io.Writer) error {
 		if args.file {
 			l_total += len(record.Seq)
 		} else {
-			s := fmt.Sprintf("%s\t%d\n", record.ID, len(record.Seq))
+			s := fmt.Sprintf("%s\t%d\n", return_record_name(record, args.description), len(record.Seq))
 			_, err := w.Write([]byte(s))
 			if err != nil {
 				return err

Diff for: len_test.go

@@ -4,11 +4,13 @@ import (
 	"bytes"
 	"fmt"
 	"testing"
+
+	"github.com/bjeight/fastats/fasta"
 )
 
 func Test_length(t *testing.T) {
 	out := new(bytes.Buffer)
-	err := length(out, []string{}, "", false, false)
+	err := length(out, []string{}, "", false, false, false)
 	if err != nil {
 		t.Error(err)
 	}
@@ -21,7 +23,7 @@ func Test_length(t *testing.T) {
 
 func Test_lengthFile(t *testing.T) {
 	out := new(bytes.Buffer)
-	err := length(out, []string{}, "", true, false)
+	err := length(out, []string{}, "", true, false, false)
 	if err != nil {
 		t.Error(err)
 	}
@@ -35,7 +37,7 @@ stdin	0
 
 func Test_lengthCounts(t *testing.T) {
 	out := new(bytes.Buffer)
-	err := length(out, []string{}, "", false, true)
+	err := length(out, []string{}, "", false, true, false)
 	if err != nil {
 		t.Error(err)
 	}
@@ -48,7 +50,7 @@ func Test_lengthCounts(t *testing.T) {
 
 func Test_lengthFileCounts(t *testing.T) {
 	out := new(bytes.Buffer)
-	err := length(out, []string{}, "", true, true)
+	err := length(out, []string{}, "", true, true, false)
 	if err != nil {
 		t.Error(err)
 	}
@@ -72,16 +74,17 @@ ATG
 ATGN
 `)
 	fastaR := bytes.NewReader(fastaData)
-	r := NewReader(fastaR)
+	r := fasta.NewReader(fastaR)
 	out := new(bytes.Buffer)
 
 	lengthRecords(
 		r,
 		arguments{
-			filepath: "/path/to/myfile.fasta",
-			file:     false,
-			counts:   false,
-			pattern:  "",
+			filepath:    "/path/to/myfile.fasta",
+			file:        false,
+			counts:      false,
+			description: false,
+			pattern:     "",
 		},
 		out,
 	)
@@ -109,16 +112,17 @@ ATG
 ATGN
 `)
 	fastaR := bytes.NewReader(fastaData)
-	r := NewReader(fastaR)
+	r := fasta.NewReader(fastaR)
 	out := new(bytes.Buffer)
 
 	lengthRecords(
 		r,
 		arguments{
-			filepath: "/path/to/myfile.fasta",
-			file:     true,
-			counts:   false,
-			pattern:  "",
+			filepath:    "/path/to/myfile.fasta",
+			file:        true,
+			counts:      false,
+			description: false,
+			pattern:     "",
 		},
 		out,
 	)

Diff for: lib.go

@@ -5,29 +5,33 @@ import (
 	"os"
 	"path/filepath"
 	"strings"
+
+	"github.com/bjeight/fastats/fasta"
 )
 
 type arguments struct {
-	filepath string
-	file     bool
-	counts   bool
-	pattern  string
+	filepath    string
+	file        bool
+	counts      bool
+	description bool
+	pattern     string
 }
 
-type fastatsFunction func(*Reader, arguments, io.Writer) error
+type fastatsFunction func(*fasta.Reader, arguments, io.Writer) error
 
 // For every file provided on the command line, template applies the correct functionality based on the cli arguments.
 // If no files are provided, it signals that we should try to read an uncompressed fasta file from stdin.
-func collectCommandLine(w io.Writer, fn fastatsFunction, filepaths []string, pattern string, file bool, count bool) error {
+func collectCommandLine(w io.Writer, fn fastatsFunction, filepaths []string, pattern string, file bool, count bool, description bool) error {
 
 	// for every file provided on the command line...
 	for _, fp := range filepaths {
 		// wrap the arguments up in s struct
 		a := arguments{
-			filepath: fp,
-			file:     file,
-			counts:   count,
-			pattern:  pattern,
+			filepath:    fp,
+			file:        file,
+			counts:      count,
+			description: description,
+			pattern:     pattern,
 		}
 		// and pass them to the correct function (defined when template is called)
 		err := applyFastatsFunction(fn, a, w)
@@ -39,10 +43,11 @@ func collectCommandLine(w io.Writer, fn fastatsFunction, filepaths []string, pat
 	// if there were no files provided on the command line, attemp to read from stdin
 	if len(filepaths) == 0 {
 		a := arguments{
-			filepath: "stdin",
-			file:     file,
-			counts:   count,
-			pattern:  pattern,
+			filepath:    "stdin",
+			file:        file,
+			counts:      count,
+			pattern:     pattern,
+			description: description,
 		}
 		err := applyFastatsFunction(fn, a, w)
 		if err != nil {
@@ -58,9 +63,9 @@ func collectCommandLine(w io.Writer, fn fastatsFunction, filepaths []string, pat
 // it the command line arguments and the writer from the scope above)
 func applyFastatsFunction(fn fastatsFunction, args arguments, w io.Writer) error {
 	// open stdin or a file
-	var r *Reader
+	var r *fasta.Reader
 	if args.filepath == "stdin" {
-		r = NewReader(os.Stdin)
+		r = fasta.NewReader(os.Stdin)
 	} else {
 		f, err := os.Open(args.filepath)
 		if err != nil {
@@ -70,9 +75,9 @@ func applyFastatsFunction(fn fastatsFunction, args arguments, w io.Writer) error
 		// depending on whwether the fasta file is compressed or not, provide the correct reader
 		switch filepath.Ext(args.filepath) {
 		case ".gz", ".bgz":
-			r = NewZReader(f)
+			r = fasta.NewZReader(f)
 		default:
-			r = NewReader(f)
+			r = fasta.NewReader(f)
 		}
 	}
 
@@ -89,3 +94,11 @@ func filenameFromFullPath(filepath string) string {
 	sa := strings.Split(filepath, "/")
 	return sa[len(sa)-1]
 }
+
+func return_record_name(fr fasta.FastaRecord, description bool) string {
+	if description {
+		return fr.Description
+	} else {
+		return fr.ID
+	}
+}