reduce number of open files while processing nested java archives (#227)

Signed-off-by: Alex Goodman <alex.goodman@anchore.com>

internal/file/opener.go

@@ -0,0 +1,14 @@
+package file
+
+import (
+	"io"
+	"os"
+)
+
+type Opener struct {
+	path string
+}
+
+func (o Opener) Open() (io.ReadCloser, error) {
+	return os.Open(o.path)
+}

internal/file/zip_file_traversal.go

@@ -63,8 +63,8 @@ func TraverseFilesInZip(archivePath string, visitor func(*zip.File) error, paths
 	return nil
 }
 
-func ExtractFromZipToUniqueTempFile(archivePath, dir string, paths ...string) (map[string]io.Reader, error) {
+func ExtractFromZipToUniqueTempFile(archivePath, dir string, paths ...string) (map[string]Opener, error) {
-	results := make(map[string]io.Reader)
+	results := make(map[string]Opener)
 
 	// don't allow for full traversal, only select traversal from given paths
 	if len(paths) == 0 {
@@ -78,11 +78,21 @@ func ExtractFromZipToUniqueTempFile(archivePath, dir string, paths ...string) (m
 		if err != nil {
 			return fmt.Errorf("unable to create temp file: %w", err)
 		}
+		// we shouldn't try and keep the tempfile open as the returned result may have several files, which takes up
+		// resources (leading to "too many open files"). Instead we'll return a file opener to the caller which
+		// provides a ReadCloser. It is up to the caller to handle closing the file explicitly.
+		defer tempFile.Close()
 
 		zippedFile, err := file.Open()
 		if err != nil {
 			return fmt.Errorf("unable to read file=%q from zip=%q: %w", file.Name, archivePath, err)
 		}
+		defer func() {
+			err := zippedFile.Close()
+			if err != nil {
+				log.Errorf("unable to close source file=%q from zip=%q: %+v", file.Name, archivePath, err)
+			}
+		}()
 
 		if file.FileInfo().IsDir() {
 			return fmt.Errorf("unable to extract directories, only files: %s", file.Name)
@@ -103,12 +113,8 @@ func ExtractFromZipToUniqueTempFile(archivePath, dir string, paths ...string) (m
 			return fmt.Errorf("unable to reset file pointer (%s): %w", tempFile.Name(), err)
 		}
 
-		results[file.Name] = tempFile
+		results[file.Name] = Opener{path: tempFile.Name()}
 
-		err = zippedFile.Close()
-		if err != nil {
-			return fmt.Errorf("unable to close source file=%q from zip=%q: %w", file.Name, archivePath, err)
-		}
 		return nil
 	}
 

syft/cataloger/java/archive_parser.go

@@ -5,6 +5,8 @@ import (
 	"io"
 	"strings"
 
+	"github.com/anchore/syft/internal/log"
+
 	"github.com/anchore/syft/internal"
 	"github.com/anchore/syft/internal/file"
 	"github.com/anchore/syft/syft/cataloger/common"
@@ -214,18 +216,28 @@ func (j *archiveParser) discoverPkgsFromNestedArchives(parentPkg *pkg.Package) (
 	}
 
 	// search and parse pom.properties files & fetch the contents
-	readers, err := file.ExtractFromZipToUniqueTempFile(j.archivePath, j.contentPath, j.fileManifest.GlobMatch(archiveFormatGlobs...)...)
+	openers, err := file.ExtractFromZipToUniqueTempFile(j.archivePath, j.contentPath, j.fileManifest.GlobMatch(archiveFormatGlobs...)...)
 	if err != nil {
 		return nil, fmt.Errorf("unable to extract files from zip: %w", err)
 	}
 
 	// discover nested artifacts
-	for archivePath, archiveReader := range readers {
+	for archivePath, archiveOpener := range openers {
-		nestedPath := fmt.Sprintf("%s:%s", j.virtualPath, archivePath)
+		archiveReadCloser, err := archiveOpener.Open()
-		nestedPkgs, err := parseJavaArchive(nestedPath, archiveReader)
 		if err != nil {
+			return nil, fmt.Errorf("unable to open archived file from tempdir: %w", err)
+		}
+		nestedPath := fmt.Sprintf("%s:%s", j.virtualPath, archivePath)
+		nestedPkgs, err := parseJavaArchive(nestedPath, archiveReadCloser)
+		if err != nil {
+			if closeErr := archiveReadCloser.Close(); closeErr != nil {
+				log.Errorf("unable to close archived file from tempdir: %+v", closeErr)
+			}
 			return nil, fmt.Errorf("unable to process nested java archive (%s): %w", archivePath, err)
 		}
+		if err = archiveReadCloser.Close(); err != nil {
+			return nil, fmt.Errorf("unable to close archived file from tempdir: %w", err)
+		}
 
 		// attach the parent package to all discovered packages that are not already associated with a java archive
 		for _, p := range nestedPkgs {

syft/cataloger/java/save_archive_to_tmp.go

@@ -35,6 +35,7 @@ func saveArchiveToTmp(reader io.Reader) (string, string, func(), error) {
 	if err != nil {
 		return contentDir, "", cleanupFn, fmt.Errorf("unable to create archive: %w", err)
 	}
+	defer archiveFile.Close()
 
 	_, err = io.Copy(archiveFile, reader)
 	if err != nil {