cmd/kgz: add extended metadata support to kgz compression

cmd/kgz/README

@@ -3,18 +3,31 @@ kgz
 kgz is like gzip, but supports compressing and decompressing to a different
 directory than the source file is in.
 
-Usage: kgz [-l] source [target]
+Usage: kgz [-l] [-k] [-m] [-x] [--uid N] [--gid N] source [target]
 
-If target is a directory, the basename of the sourcefile will be used
+If target is a directory, the basename of the source file will be used
 as the target filename. Compression and decompression is selected
 based on whether the source filename ends in ".gz".
 
 Flags:
-	-l level	Compression level (0-9). Only meaninful when
+	-l level    Compression level (0-9). Only meaningful when compressing.
-			    compressing a file.
     -u          Do not restrict the size during decompression. As
                 a safeguard against gzip bombs, the maximum size
                 allowed is 32 * the compressed file size.
+    -k          Keep the source file (do not remove it after successful
+                compression or decompression).
+    -m          On decompression, set the file mtime from the gzip header.
+    -x          On compression, include uid/gid/mode/ctime in the gzip Extra
+                field so that decompression can restore them. The Extra payload
+                is an ASN.1 DER-encoded struct.
+    --uid N     When used with -x, set UID in Extra to N (override source).
+    --gid N     When used with -x, set GID in Extra to N (override source).
+
+Metadata notes:
+- mtime is stored in the standard gzip header and restored with -m.
+- uid/gid/mode/ctime are stored in a kgz-specific Extra subfield as an ASN.1
+  DER-encoded struct. Restoring
+  uid/gid may fail without sufficient privileges; such errors are ignored.
 
 
 

cmd/kgz/main.go

@@ -3,23 +3,122 @@ package main
 import (
 	"compress/flate"
 	"compress/gzip"
+	"encoding/asn1"
+	"encoding/binary"
 	"flag"
 	"fmt"
 	"io"
 	"os"
 	"path/filepath"
 	"strings"
+
+	goutilslib "git.wntrmute.dev/kyle/goutils/lib"
+	"golang.org/x/sys/unix"
 )
 
 const gzipExt = ".gz"
 
-func compress(path, target string, level int) error {
+// kgzExtraID is the two-byte subfield identifier used in the gzip Extra field
+// for kgz-specific metadata.
+var kgzExtraID = [2]byte{'K', 'G'}
+
+// buildKGExtra constructs the gzip Extra subfield payload for kgz metadata.
+//
+// The payload is an ASN.1 DER-encoded struct with the following fields:
+//
+//	Version    INTEGER (currently 1)
+//	UID        INTEGER
+//	GID        INTEGER
+//	Mode       INTEGER (permission bits)
+//	CTimeSec   INTEGER (seconds)
+//	CTimeNSec  INTEGER (nanoseconds)
+//
+// The ASN.1 blob is wrapped in a gzip Extra subfield with ID 'K','G'.
+func buildKGExtra(uid, gid, mode uint32, ctimeS int64, ctimeNs int32) []byte {
+	// Define the ASN.1 structure to encode
+	type KGZExtra struct {
+		Version   int
+		UID       int
+		GID       int
+		Mode      int
+		CTimeSec  int64
+		CTimeNSec int32
+	}
+
+	payload, err := asn1.Marshal(KGZExtra{
+		Version:   1,
+		UID:       int(uid),
+		GID:       int(gid),
+		Mode:      int(mode),
+		CTimeSec:  ctimeS,
+		CTimeNSec: ctimeNs,
+	})
+	if err != nil {
+		// On marshal failure, return empty to avoid breaking compression
+		return nil
+	}
+
+	// Wrap in gzip subfield: [ID1 ID2 LEN(lo) LEN(hi) PAYLOAD]
+	extra := make([]byte, 4+len(payload))
+	extra[0] = kgzExtraID[0]
+	extra[1] = kgzExtraID[1]
+	binary.LittleEndian.PutUint16(extra[2:], uint16(len(payload)))
+	copy(extra[4:], payload)
+	return extra
+}
+
+// parseKGExtra scans a gzip Extra blob and returns kgz metadata if present.
+func parseKGExtra(extra []byte) (uid, gid, mode uint32, ctimeS int64, ctimeNs int32, ok bool) {
+	i := 0
+	for i+4 <= len(extra) {
+		id1 := extra[i]
+		id2 := extra[i+1]
+		l := int(binary.LittleEndian.Uint16(extra[i+2 : i+4]))
+		i += 4
+		if i+l > len(extra) {
+			break
+		}
+		if id1 == kgzExtraID[0] && id2 == kgzExtraID[1] {
+			// ASN.1 decode payload
+			payload := extra[i : i+l]
+			var s struct {
+				Version   int
+				UID       int
+				GID       int
+				Mode      int
+				CTimeSec  int64
+				CTimeNSec int32
+			}
+			if _, err := asn1.Unmarshal(payload, &s); err != nil {
+				return 0, 0, 0, 0, 0, false
+			}
+			if s.Version != 1 {
+				return 0, 0, 0, 0, 0, false
+			}
+			return uint32(s.UID), uint32(s.GID), uint32(s.Mode), s.CTimeSec, s.CTimeNSec, true
+		}
+		i += l
+	}
+	return 0, 0, 0, 0, 0, false
+}
+
+func compress(path, target string, level int, includeExtra bool, setUID, setGID int) error {
 	sourceFile, err := os.Open(path)
 	if err != nil {
 		return fmt.Errorf("opening file for read: %w", err)
 	}
 	defer sourceFile.Close()
 
+	// Gather file metadata
+	var st unix.Stat_t
+	if err := unix.Stat(path, &st); err != nil {
+		return fmt.Errorf("stat source: %w", err)
+	}
+	fi, err := sourceFile.Stat()
+	if err != nil {
+		return fmt.Errorf("stat source file: %w", err)
+	}
+
 	destFile, err := os.Create(target)
 	if err != nil {
 		return fmt.Errorf("opening file for write: %w", err)
@@ -30,6 +129,27 @@ func compress(path, target string, level int) error {
 	if err != nil {
 		return fmt.Errorf("invalid compression level: %w", err)
 	}
+	// Set header metadata
+	gzipCompressor.ModTime = fi.ModTime()
+	if includeExtra {
+		uid := uint32(st.Uid)
+		gid := uint32(st.Gid)
+		if setUID >= 0 {
+			uid = uint32(setUID)
+		}
+		if setGID >= 0 {
+			gid = uint32(setGID)
+		}
+		mode := uint32(st.Mode & 0o7777)
+		// Use portable helper to gather ctime
+		var cts int64
+		var ctns int32
+		if ft, err := goutilslib.LoadFileTime(path); err == nil {
+			cts = ft.Changed.Unix()
+			ctns = int32(ft.Changed.Nanosecond())
+		}
+		gzipCompressor.Extra = buildKGExtra(uid, gid, mode, cts, ctns)
+	}
 	defer gzipCompressor.Close()
 
 	_, err = io.Copy(gzipCompressor, sourceFile)
@@ -40,7 +160,7 @@ func compress(path, target string, level int) error {
 	return nil
 }
 
-func uncompress(path, target string, unrestrict bool) error {
+func uncompress(path, target string, unrestrict bool, preserveMtime bool) error {
 	sourceFile, err := os.Open(path)
 	if err != nil {
 		return fmt.Errorf("opening file for read: %w", err)
@@ -79,19 +199,40 @@ func uncompress(path, target string, unrestrict bool) error {
 	if err != nil {
 		return fmt.Errorf("uncompressing file: %w", err)
 	}
-
+	// Apply metadata from Extra (uid/gid/mode) if present
+	if gzipUncompressor.Header.Extra != nil {
+		if uid, gid, mode, _, _, ok := parseKGExtra(gzipUncompressor.Header.Extra); ok {
+			// Chmod
+			_ = os.Chmod(target, os.FileMode(mode))
+			// Chown (may fail without privileges)
+			_ = os.Chown(target, int(uid), int(gid))
+		}
+	}
+	// Preserve mtime if requested
+	if preserveMtime {
+		mt := gzipUncompressor.Header.ModTime
+		if !mt.IsZero() {
+			// Set both atime and mtime to mt for simplicity
+			_ = os.Chtimes(target, mt, mt)
+		}
+	}
 	return nil
 }
 
 func usage(w io.Writer) {
-	fmt.Fprintf(w, `Usage: %s [-l] source [target]
+	fmt.Fprintf(w, `Usage: %s [-l] [-k] [-m] [-x] [--uid N] [--gid N] source [target]
 
 kgz is like gzip, but supports compressing and decompressing to a different
 directory than the source file is in.
 
 Flags:
-	-l level	Compression level (0-9). Only meaninful when
+    -l level    Compression level (0-9). Only meaningful when compressing.
-			compressing a file.
+    -u          Do not restrict the size during decompression (gzip bomb guard is 32x).
+    -k          Keep the source file (do not remove it after successful (de)compression).
+    -m          On decompression, set the file mtime from the gzip header.
+    -x          On compression, include uid/gid/mode/ctime in the gzip Extra field.
+    --uid N     When used with -x, set UID in Extra to N (overrides source owner).
+    --gid N     When used with -x, set GID in Extra to N (overrides source group).
 `, os.Args[0])
 }
 
@@ -150,9 +291,19 @@ func main() {
 	var target = "."
 	var err error
 	var unrestrict bool
+	var keep bool
+	var preserveMtime bool
+	var includeExtra bool
+	var setUID int
+	var setGID int
 
 	flag.IntVar(&level, "l", flate.DefaultCompression, "compression level")
 	flag.BoolVar(&unrestrict, "u", false, "do not restrict decompression")
+	flag.BoolVar(&keep, "k", false, "keep the source file (do not remove it)")
+	flag.BoolVar(&preserveMtime, "m", false, "on decompression, set mtime from gzip header")
+	flag.BoolVar(&includeExtra, "x", false, "on compression, include uid/gid/mode/ctime in gzip Extra")
+	flag.IntVar(&setUID, "uid", -1, "when used with -x, set UID in Extra to this value")
+	flag.IntVar(&setGID, "gid", -1, "when used with -x, set GID in Extra to this value")
 	flag.Parse()
 
 	if flag.NArg() < 1 || flag.NArg() > 2 {
@@ -172,12 +323,15 @@ func main() {
 			os.Exit(1)
 		}
 
-		err = uncompress(path, target, unrestrict)
+		err = uncompress(path, target, unrestrict, preserveMtime)
 		if err != nil {
 			os.Remove(target)
 			fmt.Fprintf(os.Stderr, "%s\n", err)
 			os.Exit(1)
 		}
+		if !keep {
+			_ = os.Remove(path)
+		}
 		return
 	}
 
@@ -187,10 +341,13 @@ func main() {
 		os.Exit(1)
 	}
 
-	err = compress(path, target, level)
+	err = compress(path, target, level, includeExtra, setUID, setGID)
 	if err != nil {
 		os.Remove(target)
 		fmt.Fprintf(os.Stderr, "%s\n", err)
 		os.Exit(1)
 	}
+	if !keep {
+		_ = os.Remove(path)
+	}
 }