Add zlib support to extractGzip (#713)

* Add zlib support to extractGzip

Signed-off-by: egibs <[email protected]>

* Appease the linter

Signed-off-by: egibs <[email protected]>

* Use Z instead of zlib

Signed-off-by: egibs <[email protected]>

---------

Signed-off-by: egibs <[email protected]>

pkg/action/archive.go

@@ -5,6 +5,7 @@ import (
 	"archive/zip"
 	"compress/bzip2"
 	"compress/gzip"
+	"compress/zlib"
 	"context"
 	"errors"
 	"fmt"
@@ -19,6 +20,7 @@ import (
 	"github.com/cavaliergopher/cpio"
 	"github.com/cavaliergopher/rpm"
 	"github.com/chainguard-dev/clog"
+	"github.com/chainguard-dev/malcontent/pkg/programkind"
 
 	"github.com/egibs/go-debian/deb"
 
@@ -225,7 +227,6 @@ func extractTar(ctx context.Context, d string, f string) error {
 // extractGzip extracts .gz archives.
 func extractGzip(ctx context.Context, d string, f string) error {
 	logger := clog.FromContext(ctx).With("dir", d, "file", f)
-	logger.Debug("extracting gzip")
 
 	// Check if the file is valid
 	_, err := os.Stat(f)
@@ -239,23 +240,50 @@ func extractGzip(ctx context.Context, d string, f string) error {
 	}
 	defer gf.Close()
 
-	gr, err := gzip.NewReader(gf)
+	// Determine if we're extracting a gzip- or zlib-compressed file
+	ft, err := programkind.File(f)
 	if err != nil {
-		return fmt.Errorf("failed to create gzip reader: %w", err)
+		return fmt.Errorf("failed to determine file type: %w", err)
 	}
-	defer gr.Close()
+
+	logger.Debugf("extracting %s", ft.Ext)
 
 	base := filepath.Base(f)
 	target := filepath.Join(d, base[:len(base)-len(filepath.Ext(base))])
 
-	ef, err := os.Create(target)
-	if err != nil {
-		return fmt.Errorf("failed to create extracted file: %w", err)
-	}
-	defer ef.Close()
+	switch ft.Ext {
+	case "gzip":
+		gr, err := gzip.NewReader(gf)
+		if err != nil {
+			return fmt.Errorf("failed to create gzip reader: %w", err)
+		}
+		defer gr.Close()
 
-	if _, err := io.Copy(ef, io.LimitReader(gr, maxBytes)); err != nil {
-		return fmt.Errorf("failed to copy file: %w", err)
+		ef, err := os.Create(target)
+		if err != nil {
+			return fmt.Errorf("failed to create extracted file: %w", err)
+		}
+		defer ef.Close()
+
+		if _, err := io.Copy(ef, io.LimitReader(gr, maxBytes)); err != nil {
+			return fmt.Errorf("failed to copy file: %w", err)
+		}
+	case "Z":
+		zr, err := zlib.NewReader(gf)
+		if err != nil {
+			return fmt.Errorf("failed to create zlib reader: %w", err)
+		}
+		defer zr.Close()
+
+		ef, err := os.Create(target)
+		if err != nil {
+			return fmt.Errorf("failed to create extracted file: %w", err)
+		}
+		defer ef.Close()
+
+		if _, err := io.Copy(ef, io.LimitReader(zr, maxBytes)); err != nil {
+			return fmt.Errorf("failed to copy file: %w", err)
+		}
 	}
 
 	return nil

pkg/programkind/programkind.go

@@ -18,6 +18,7 @@ import (
 // file extension to MIME type, if it's a good scanning target.
 var supportedKind = map[string]string{
 	"7z":      "",
+	"Z":       "application/zlib",
 	"asm":     "",
 	"bash":    "application/x-bsh",
 	"bat":     "application/bat",
@@ -39,6 +40,7 @@ var supportedKind = map[string]string{
 	"expect":  "text/x-expect",
 	"fish":    "text/x-fish",
 	"go":      "text/x-go",
+	"gzip":    "application/gzip",
 	"h":       "text/x-h",
 	"hh":      "text/x-h",
 	"html":    "",
@@ -57,9 +59,9 @@ var supportedKind = map[string]string{
 	"pyc":     "application/x-python-code",
 	"rb":      "text/x-ruby",
 	"rs":      "text/x-rust",
-	"script":  "text/x-generic-script",
 	"scpt":    "application/x-applescript",
 	"scptd":   "application/x-applescript",
+	"script":  "text/x-generic-script",
 	"service": "text/x-systemd",
 	"sh":      "application/x-sh",
 	"so":      "application/x-sharedlib",
@@ -101,6 +103,8 @@ func makeFileType(path string, ext string, mime string) *FileType {
 }
 
 // File detects what kind of program this file might be.
+//
+//nolint:cyclop // ignore complexity of 38
 func File(path string) (*FileType, error) {
 	// Follow symlinks and return cleanly if the target does not exist
 	_, err := filepath.EvalSymlinks(path)
@@ -176,6 +180,10 @@ func File(path string) (*FileType, error) {
 		return Path(".c"), nil
 	case strings.Contains(s, "BEAMAtU8"):
 		return Path(".beam"), nil
+	case hdr[0] == '\x1f' && hdr[1] == '\x8b':
+		return Path(".gzip"), nil
+	case hdr[0] == '\x78' && hdr[1] == '\x5E':
+		return Path(".Z"), nil
 	}
 	return nil, nil
 }