From 6a6a6f6f2cbbc24690f79e99b329adfd2e10956f Mon Sep 17 00:00:00 2001
From: Eric Coissac <eric.coissac@metabarcoding.org>
Date: Mon, 16 Oct 2023 15:34:06 +0200
Subject: [PATCH] Correctly handle empty files

Former-commit-id: d166aa352ce4bf32739ddc2f7d1c9967918822fd
---
 pkg/obiformats/embl_read.go      | 19 +++++++------------
 pkg/obiformats/fastaseq_read.go  | 11 -----------
 pkg/obiformats/genbank_read.go   | 19 +++++++------------
 pkg/obiformats/universal_read.go |  5 +++++
 4 files changed, 19 insertions(+), 35 deletions(-)

diff --git a/pkg/obiformats/embl_read.go b/pkg/obiformats/embl_read.go
index 947a3fe..a29136b 100644
--- a/pkg/obiformats/embl_read.go
+++ b/pkg/obiformats/embl_read.go
@@ -4,13 +4,10 @@ import (
 	"bufio"
 	"bytes"
 	"io"
-	"os"
 	"path"
 	"strconv"
 	"strings"
 
-	gzip "github.com/klauspost/pgzip"
-
 	log "github.com/sirupsen/logrus"
 
 	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiiter"
@@ -276,23 +273,21 @@ func ReadEMBL(reader io.Reader, options ...WithOption) obiiter.IBioSequence {
 
 func ReadEMBLFromFile(filename string, options ...WithOption) (obiiter.IBioSequence, error) {
 	var reader io.Reader
-	var greader io.Reader
 	var err error
 
 	options = append(options, OptionsSource(obiutils.RemoveAllExt((path.Base(filename)))))
 
-	reader, err = os.Open(filename)
+	reader, err = Ropen(filename)
+
+	if err == ErrNoContent {
+		log.Infof("file %s is empty", filename)
+		return ReadEmptyFile(options...)
+	}
+
 	if err != nil {
 		log.Printf("open file error: %+v", err)
 		return obiiter.NilIBioSequence, err
 	}
 
-	// Test if the flux is compressed by gzip
-	//greader, err = gzip.NewReader(reader)
-	greader, err = gzip.NewReaderN(reader, 1<<24, 2)
-	if err == nil {
-		reader = greader
-	}
-
 	return ReadEMBL(reader, options...), nil
 }
diff --git a/pkg/obiformats/fastaseq_read.go b/pkg/obiformats/fastaseq_read.go
index 34b439e..9774970 100644
--- a/pkg/obiformats/fastaseq_read.go
+++ b/pkg/obiformats/fastaseq_read.go
@@ -114,19 +114,8 @@ func FastaChunkReader(r io.Reader, size int, cutHead bool) (chan FastxChunk, err
 			end := []byte{}
 
 			for err == nil && n > 0 {
-				// fmt.Println("============end=========================")
-				// fmt.Println(string(end))
-				// fmt.Println("------------buff------------------------")
-				// fmt.Println(string(buff))
 				buff = Concatenate(end, buff)
-				// fmt.Println("------------buff--pasted----------------")
-				// fmt.Println(string(buff))
 				buff, end = lastFastaCut(buff)
-				// fmt.Println("----------------buff--cutted------------")
-				// fmt.Println(string(buff))
-				// fmt.Println("------------------end-------------------")
-				// fmt.Println(string(end))
-				// fmt.Println("========================================")
 				if len(buff) > 0 {
 					out <- FastxChunk{
 						Bytes: bytes.Clone(buff),
diff --git a/pkg/obiformats/genbank_read.go b/pkg/obiformats/genbank_read.go
index 40b3b1c..7e54874 100644
--- a/pkg/obiformats/genbank_read.go
+++ b/pkg/obiformats/genbank_read.go
@@ -4,14 +4,11 @@ import (
 	"bufio"
 	"bytes"
 	"io"
-	"os"
 	"path"
 	"regexp"
 	"strconv"
 	"strings"
 
-	gzip "github.com/klauspost/pgzip"
-
 	log "github.com/sirupsen/logrus"
 
 	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiiter"
@@ -173,23 +170,21 @@ func ReadGenbank(reader io.Reader, options ...WithOption) obiiter.IBioSequence {
 
 func ReadGenbankFromFile(filename string, options ...WithOption) (obiiter.IBioSequence, error) {
 	var reader io.Reader
-	var greader io.Reader
 	var err error
 
 	options = append(options, OptionsSource(obiutils.RemoveAllExt((path.Base(filename)))))
 
-	reader, err = os.Open(filename)
+	reader, err = Ropen(filename)
+
+	if err == ErrNoContent {
+		log.Infof("file %s is empty", filename)
+		return ReadEmptyFile(options...)
+	}
+
 	if err != nil {
 		log.Printf("open file error: %+v", err)
 		return obiiter.NilIBioSequence, err
 	}
 
-	// Test if the flux is compressed by gzip
-	//greader, err = gzip.NewReader(reader)
-	greader, err = gzip.NewReaderN(reader, 1<<24, 2)
-	if err == nil {
-		reader = greader
-	}
-
 	return ReadGenbank(reader, options...), nil
 }
diff --git a/pkg/obiformats/universal_read.go b/pkg/obiformats/universal_read.go
index e2d2f5f..d0f75bf 100644
--- a/pkg/obiformats/universal_read.go
+++ b/pkg/obiformats/universal_read.go
@@ -148,6 +148,11 @@ func ReadSequencesFromFile(filename string,
 
 	file, err = Ropen(filename)
 
+	if err == ErrNoContent {
+		log.Infof("file %s is empty", filename)
+		return ReadEmptyFile(options...)
+	}
+
 	if err != nil {
 		log.Fatalf("open file error: %v", err)
 		return obiiter.NilIBioSequence, err