read gzip archives

This commit is contained in:
Motiejus Jakštys 2021-05-24 00:11:58 +03:00
parent a45e512697
commit fd244bfc23
4 changed files with 84 additions and 5 deletions

View File

@ -3,6 +3,8 @@ package tartest
import ( import (
"archive/tar" "archive/tar"
"bytes" "bytes"
"compress/gzip"
"fmt"
"io" "io"
"testing" "testing"
@ -53,6 +55,20 @@ func (tb Tarball) Buffer() *bytes.Buffer {
return &buf return &buf
} }
// Gzip returns a gzipped buffer
func (tb Tarball) Gzip() *bytes.Buffer {
var buf bytes.Buffer
w := gzip.NewWriter(&buf)
_, err := io.Copy(w, tb.Buffer())
if err != nil {
panic(fmt.Errorf("Gzip(): %w", err))
}
if err := w.Close(); err != nil {
panic(fmt.Errorf("gzip.Close(): %w", err))
}
return &buf
}
// Tar tars the Dir // Tar tars the Dir
func (d Dir) Tar(tw *tar.Writer) error { func (d Dir) Tar(tw *tar.Writer) error {
hdr := &tar.Header{ hdr := &tar.Header{

View File

@ -2,9 +2,12 @@ package tartest
import ( import (
"bytes" "bytes"
"compress/gzip"
"io"
"testing" "testing"
"github.com/stretchr/testify/assert" "github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
) )
func TestTarball(t *testing.T) { func TestTarball(t *testing.T) {
@ -23,6 +26,18 @@ func TestTarball(t *testing.T) {
Dir{Name: "bin"}, Dir{Name: "bin"},
Hardlink{Name: "entrypoint2"}, Hardlink{Name: "entrypoint2"},
} }
assert.Equal(t, want, got) assert.Equal(t, want, got)
} }
func TestGzip(t *testing.T) {
tb := Tarball{File{Name: "entrypoint.sh", Contents: bytes.NewBufferString("hello")}}
tbuf := tb.Buffer()
tgz, err := gzip.NewReader(tb.Gzip())
require.NoError(t, err)
var uncompressed bytes.Buffer
_, err = io.Copy(&uncompressed, tgz)
require.NoError(t, err)
assert.Equal(t, tbuf.Bytes(), uncompressed.Bytes())
}

View File

@ -2,6 +2,8 @@ package rootfs
import ( import (
"archive/tar" "archive/tar"
"bytes"
"compress/gzip"
"encoding/json" "encoding/json"
"errors" "errors"
"fmt" "fmt"
@ -9,7 +11,7 @@ import (
"path/filepath" "path/filepath"
"strings" "strings"
"github.com/motiejus/code/undocker/internal/bytecounter" "github.com/motiejus/code/undocker/internal/bytecounter"
"go.uber.org/multierr" "go.uber.org/multierr"
) )
@ -43,6 +45,7 @@ func (r *RootFS) WriteTo(w io.Writer) (n int64, err error) {
wr := bytecounter.New(w) wr := bytecounter.New(w)
tr := tar.NewReader(r.rd) tr := tar.NewReader(r.rd)
tw := tar.NewWriter(wr) tw := tar.NewWriter(wr)
var closer func() error
defer func() { defer func() {
err = multierr.Append(err, tw.Close()) err = multierr.Append(err, tw.Close())
n = wr.N n = wr.N
@ -110,7 +113,7 @@ func (r *RootFS) WriteTo(w io.Writer) (n int64, err error) {
if _, err := r.rd.Seek(no.offset, io.SeekStart); err != nil { if _, err := r.rd.Seek(no.offset, io.SeekStart); err != nil {
return n, err return n, err
} }
tr = tar.NewReader(r.rd) tr, closer = readTar(r.rd)
for { for {
hdr, err := tr.Next() hdr, err := tr.Next()
if err == io.EOF { if err == io.EOF {
@ -138,9 +141,11 @@ func (r *RootFS) WriteTo(w io.Writer) (n int64, err error) {
continue continue
} }
} }
file2layer[hdr.Name] = i file2layer[hdr.Name] = i
} }
if err := closer(); err != nil {
return n, err
}
} }
// construct directories to whiteout, for each layer. // construct directories to whiteout, for each layer.
@ -151,7 +156,7 @@ func (r *RootFS) WriteTo(w io.Writer) (n int64, err error) {
if _, err := r.rd.Seek(no.offset, io.SeekStart); err != nil { if _, err := r.rd.Seek(no.offset, io.SeekStart); err != nil {
return n, err return n, err
} }
tr = tar.NewReader(r.rd) tr, closer = readTar(r.rd)
for { for {
hdr, err := tr.Next() hdr, err := tr.Next()
if err == io.EOF { if err == io.EOF {
@ -173,6 +178,9 @@ func (r *RootFS) WriteTo(w io.Writer) (n int64, err error) {
return n, err return n, err
} }
} }
if err := closer(); err != nil {
return n, err
}
} }
return n, nil return n, nil
} }
@ -223,3 +231,31 @@ func whiteoutDirs(whreaddir map[string]int, nlayers int) []*tree {
} }
return ret return ret
} }
// readTar creates a tar reader from a targzip or tar
func readTar(r io.Reader) (*tar.Reader, func() error) {
var buf bytes.Buffer
w := &discarder{w: &buf}
r2 := io.TeeReader(r, w)
gz, err := gzip.NewReader(r2)
if err == nil {
w.discard = true
buf.Reset()
return tar.NewReader(gz), gz.Close
}
return tar.NewReader(io.MultiReader(&buf, r)), func() error { return nil }
}
// discarder is a pass-through writer until asked to 'discard' its writes.
// useful for proxying writes from a TeeReader until a certain point.
type discarder struct {
w io.Writer
discard bool
}
func (d *discarder) Write(p []byte) (int, error) {
if d.discard {
return len(p), nil
}
return d.w.Write(p)
}

View File

@ -167,6 +167,18 @@ func TestRootFS(t *testing.T) {
file{Name: "a/fileb"}, file{Name: "a/fileb"},
}, },
}, },
{
name: "archived layer",
image: tarball{
file{Name: "layer1/layer.tar", Contents: layer1.Gzip()},
file{Name: "layer0/layer.tar", Contents: layer0.Gzip()},
manifest{"layer0/layer.tar", "layer1/layer.tar"},
},
want: []extractable{
dir{Name: "/", UID: 0},
file{Name: "/file", UID: 1, Contents: bytes.NewBufferString("from 1")},
},
},
} }
for _, tt := range tests { for _, tt := range tests {