fix #1

internal/walker/archives/archive.go

@@ -7,6 +7,7 @@ package archives
 
 import (
 	"context"
+	"fmt"
 	"gocatcli/internal/log"
 	"io/fs"
 	"os"
@@ -42,7 +43,7 @@ func GetFiles(path string) ([]*ArchivedFile, error) {
 	}
 	defer fd.Close()
 
-	format, _, err := archiver.Identify(path, fd)
+	format, stream, err := archiver.Identify(path, fd)
 	if err == archiver.ErrNoMatch {
 		log.Debugf("file \"%s\" is not an archive", path)
 		return names, nil
@@ -61,13 +62,19 @@ func GetFiles(path string) ([]*ArchivedFile, error) {
 		return nil
 	}
 
-	ext := format.(archiver.Extractor)
-	ctx := context.Background()
-	err = ext.Extract(ctx, fd, nil, handler)
-	if err != nil {
-		return names, err
+	switch archive := format.(type) {
+	case archiver.Extractor:
+		ctx := context.Background()
+		err = archive.Extract(ctx, stream, nil, handler)
+		if err != nil {
+			return names, err
+		}
+		log.Debugf("got %d file(s) inside %s", len(names), path)
+		return names, nil
+	case archiver.Decompressor:
+		// no children
+		return names, nil
 	}
 
-	log.Debugf("got %d file(s) inside %s", len(names), path)
-	return names, nil
+	return nil, fmt.Errorf("cannot read archive content for %s", path)
 }

internal/walker/walker.go

@@ -115,16 +115,7 @@ func (w *Walker) walk(storageID int, walkPath string, storagePath string, parent
 			// handle archives
 			if w.withArchive && archives.IsArchive(pathUnderRoot) {
 				log.Debugf("%s is archive", pathUnderRoot)
-				child.Type = node.FileTypeArchive
-				archived, _ := archives.GetFiles(pathUnderRoot)
-				for _, arc := range archived {
-					fpath, err := filepath.Rel(storagePath, pathUnderRoot)
-					if err != nil {
-						return err
-					}
-					sub := node.NewArchivedFileNode(storageID, fpath, arc.FileInfo, arc.Path)
-					child.AddChild(sub)
-				}
+				processArchive(pathUnderRoot, storageID, storagePath, child)
 			}
 		}
 		return nil
@@ -133,6 +124,26 @@ func (w *Walker) walk(storageID int, walkPath string, storagePath string, parent
 	return cnt, err
 }
 
+func processArchive(path string, storageID int, storagePath string, child *node.FileNode) {
+	//defer func() {
+	//	r := recover()
+	//	if r != nil {
+	//		log.Errorf("archive indexing failed for %s", path)
+	//	}
+	//}()
+	archived, _ := archives.GetFiles(path)
+	for _, arc := range archived {
+		fpath, err := filepath.Rel(storagePath, path)
+		if err != nil {
+			log.Errorf("archive indexing failed for %s: %v", path, err)
+			return
+		}
+		sub := node.NewArchivedFileNode(storageID, fpath, arc.FileInfo, arc.Path)
+		child.AddChild(sub)
+	}
+	child.Type = node.FileTypeArchive
+}
+
 // Walk walks the filesystem hierarchy
 func (w *Walker) Walk(storageID int, walkPath string, storage *node.StorageNode) (int64, error) {
 	cnt, err := w.walk(storageID, walkPath, walkPath, storage)

tests-ng/test-archive.sh

@@ -25,30 +25,35 @@ out="${tmpd}/output.txt"
 arcdir="${tmpd}/archives"
 mkdir -p "${arcdir}"
 
-# create archive
+# create archives
 tar_arc="${arcdir}/archive1.tar.gz"
 tar -czf "${tar_arc}" "${cur}/../internal"
 
 zip_arc="${arcdir}/archive2.zip"
 zip -r "${zip_arc}" "${cur}/../internal"
 
+echo "test" > "${tmpd}/gzipped-original"
+gzip -c "${tmpd}/gzipped-original" > "${arcdir}/gzipped.gz"
+
 # index
 "${bin}" --debug index -a -C -c "${catalog}" "${arcdir}" arcdir
 [ ! -e "${catalog}" ] && echo "catalog not created" && exit 1
 
 cnt=$(find "${cur}/../internal" | wc -l)
 # +1 storage entry
 # +2 each of the archive file header
-total="$(("${cnt}" + "${cnt}" + 1 + 2))"
+# +1 for gzipped
+total="$(("${cnt}" + "${cnt}" + 1 + 2 + 1))"
 
 echo ">>> test archive ls <<<"
 "${bin}" --debug ls -r -a -c "${catalog}" | sed -e 's/\x1b\[[0-9;]*m//g' > "${out}"
 cat_file "${out}"
 cnt=$(wc -l "${out}" | awk '{print $1}')
 [ "${cnt}" != "${total}" ] && echo "expecting ${total} line (got ${cnt})" && exit 1
 grep '^storage arcdir 0B' "${out}" && (echo "empty storage" && exit 1)
-grep '^archive1.tar.gz' "${out}" || (echo "empty storage" && exit 1)
-grep '^archive2.zip' "${out}" || (echo "empty storage" && exit 1)
+grep '^archive1.tar.gz' "${out}" || (echo "no archive1" && exit 1)
+grep '^archive2.zip' "${out}" || (echo "archive2" && exit 1)
+grep '^gzipped.gz' "${out}" || (echo "no gzipped" && exit 1)
 
 echo "test $(basename "${0}") OK!"
 exit 0