internal/pkg: archive unpack artifact

This unpacks an internal/pkg archive stream used in the upcoming mirror service.

Signed-off-by: Ophestra <cat@gensokyo.uk>

4 files changed, 160 insertions, 0 deletions

diff --git a/internal/pkg/archive.go b/internal/pkg/archive.go
index 8b5398fa..902d2c31 100644
--- a/internal/pkg/archive.go
+++ b/internal/pkg/archive.go
@@ -277,3 +277,117 @@ func SumFS(buf *Checksum, fsys fs.FS, root string) error {
 func SumDir(buf *Checksum, pathname *check.Absolute) error {
 	return SumFS(buf, os.DirFS(pathname.String()), ".")
 }
+
+// archiveArtifact is an [Artifact] unpacking an archive supported by [Reader]
+// backed by a [FileArtifact].
+type archiveArtifact struct {
+	// Caller-supplied backing archive.
+	f Artifact
+}
+
+// NewArchive returns a new [Artifact] backed by the supplied [Artifact]. The
+// source [Artifact] must be a [FileArtifact] and produce a stream compatible
+// with [Reader].
+func NewArchive(a Artifact) Artifact {
+	return archiveArtifact{a}
+}
+
+// Kind returns the hardcoded [Kind] constant.
+func (archiveArtifact) Kind() Kind { return KindArchive }
+
+// Params is a noop.
+func (archiveArtifact) Params(*IContext) {}
+
+func init() {
+	register(KindArchive, func(r *IRReader) Artifact {
+		a := NewArchive(r.Next())
+		if _, ok := r.Finalise(); ok {
+			panic(ErrUnexpectedChecksum)
+		}
+		return a
+	})
+}
+
+// Dependencies returns a slice containing the backing file.
+func (a archiveArtifact) Dependencies() []Artifact {
+	return []Artifact{a.f}
+}
+
+// IsExclusive returns false: [Reader] is fully sequential.
+func (archiveArtifact) IsExclusive() bool { return false }
+
+// Cure cures the [Artifact], producing a directory located at work.
+func (a archiveArtifact) Cure(t *TContext) (err error) {
+	var r io.ReadCloser
+	if r, err = t.Open(a.f); err != nil {
+		return
+	}
+
+	defer func() {
+		closeErr := r.Close()
+		if err == nil {
+			err = closeErr
+		}
+	}()
+
+	if err = os.MkdirAll(t.GetWorkDir().String(), 0700); err != nil {
+		return
+	}
+	var root *os.Root
+	if root, err = os.OpenRoot(t.GetWorkDir().String()); err != nil {
+		return
+	}
+	defer func() {
+		closeErr := root.Close()
+		if err == nil {
+			err = closeErr
+		}
+	}()
+
+	var header *ArchiveHeader
+	ar := NewReader(r)
+	for header, err = ar.Next(); err == nil; header, err = ar.Next() {
+		if header.Mode.IsRegular() {
+			var f *os.File
+			if f, err = root.OpenFile(
+				header.Path,
+				os.O_CREATE|os.O_EXCL|os.O_WRONLY,
+				header.Mode.Perm(),
+			); err != nil {
+				return
+			}
+			if _, err = io.Copy(f, ar); err != nil {
+				_ = f.Close()
+				return
+			} else if err = f.Close(); err != nil {
+				return
+			}
+		} else if header.Mode&fs.ModeSymlink != 0 {
+			var p []byte
+			if p, err = io.ReadAll(ar); err != nil {
+				return
+			}
+
+			if err = root.Symlink(
+				unsafe.String(unsafe.SliceData(p), len(p)),
+				header.Path,
+			); err != nil {
+				return
+			}
+		} else if header.Mode.IsDir() {
+			if header.Path == "." {
+				continue
+			}
+
+			if err = root.Mkdir(header.Path, header.Mode.Perm()); err != nil {
+				return
+			}
+		} else {
+			return InvalidFileModeError(header.Mode)
+		}
+	}
+	if errors.Is(err, io.EOF) {
+		err = nil
+	}
+	return
+}
diff --git a/internal/pkg/archive_test.go b/internal/pkg/archive_test.go
index e04f3a66..29383eda 100644
--- a/internal/pkg/archive_test.go
+++ b/internal/pkg/archive_test.go
@@ -4,11 +4,13 @@ import (
 	"bytes"
 	"io"
 	"io/fs"
+	"maps"
 	"reflect"
 	"testing"
 	"testing/fstest"
 	"unsafe"
 
+	"hakurei.app/check"
 	"hakurei.app/internal/pkg"
 )
 
@@ -168,6 +170,46 @@ var archiveTestdata = fstest.MapFS{
 	"block/uevent": {Mode: 0600},
 }
 
+func TestArchiveArtifact(t *testing.T) {
+	t.Parallel()
+
+	want := maps.Clone(archiveTestdata)
+	want["."].Mode = fs.ModeDir | 0500
+
+	checkWithCache(t, []cacheTestCase{
+		{"unpack", 0, nil, func(t *testing.T, base *check.Absolute, c *pkg.Cache) {
+			var buf bytes.Buffer
+			if err := pkg.Write(archiveTestdata, ".", &buf); err != nil {
+				t.Fatal(err)
+			}
+
+			cureMany(t, c, []cureStep{
+				{"sample", pkg.NewArchive(
+					pkg.NewFile("", buf.Bytes()),
+				), ignorePathname, expectsFS(want), nil},
+			})
+		}, expectsFS{
+			".": {Mode: fs.ModeDir | 0700},
+
+			"checksum": {Mode: fs.ModeDir | 0700},
+			"checksum/CBPcoVHuVUTVRCMbRl8J30RSSzm_tyfuXaZ-HlZsanY1sY50meOVmgaWDrGKbx9F":                    {Mode: fs.ModeDir | 0500},
+			"checksum/CBPcoVHuVUTVRCMbRl8J30RSSzm_tyfuXaZ-HlZsanY1sY50meOVmgaWDrGKbx9F/block":              {Mode: fs.ModeDir | 0700},
+			"checksum/CBPcoVHuVUTVRCMbRl8J30RSSzm_tyfuXaZ-HlZsanY1sY50meOVmgaWDrGKbx9F/block/uevent":       {Mode: 0600},
+			"checksum/CBPcoVHuVUTVRCMbRl8J30RSSzm_tyfuXaZ-HlZsanY1sY50meOVmgaWDrGKbx9F/devices":            {Mode: fs.ModeDir | 0700},
+			"checksum/CBPcoVHuVUTVRCMbRl8J30RSSzm_tyfuXaZ-HlZsanY1sY50meOVmgaWDrGKbx9F/devices/empty":      {Mode: fs.ModeDir | 0700},
+			"checksum/CBPcoVHuVUTVRCMbRl8J30RSSzm_tyfuXaZ-HlZsanY1sY50meOVmgaWDrGKbx9F/devices/sub":        {Mode: fs.ModeDir | 0700},
+			"checksum/CBPcoVHuVUTVRCMbRl8J30RSSzm_tyfuXaZ-HlZsanY1sY50meOVmgaWDrGKbx9F/devices/sub/uevent": {Mode: 0600, Data: []byte("add")},
+			"checksum/CBPcoVHuVUTVRCMbRl8J30RSSzm_tyfuXaZ-HlZsanY1sY50meOVmgaWDrGKbx9F/devices/uevent":     {Mode: 0600, Data: []byte("add")},
+
+			"identifier": {Mode: fs.ModeDir | 0700},
+			"identifier/3oYyAbRJ_we7AgWo1BRcRcnxXFk3mAQ0Qui2nGQMi8GIJNJQtvUC6P2IeoA5mbjD": {Mode: fs.ModeSymlink | 0777, Data: []byte("../checksum/CBPcoVHuVUTVRCMbRl8J30RSSzm_tyfuXaZ-HlZsanY1sY50meOVmgaWDrGKbx9F")},
+
+			"substitute": {Mode: fs.ModeDir | 0700},
+			"work":       {Mode: fs.ModeDir | 0700},
+		}},
+	})
+}
+
 func BenchmarkArchiveRead(b *testing.B) {
 	var buf bytes.Buffer
 	if err := pkg.Write(archiveTestdata, ".", &buf); err != nil {
diff --git a/internal/pkg/ir_test.go b/internal/pkg/ir_test.go
index 0cb4a6ef..ad6baa76 100644
--- a/internal/pkg/ir_test.go
+++ b/internal/pkg/ir_test.go
@@ -98,6 +98,8 @@ func TestIRRoundtrip(t *testing.T) {
 		{"file", pkg.NewFile("stub", []byte("stub"))},
 
 		{"decompress", pkg.NewDecompress(pkg.NewFile("", []byte{0}), pkg.Bzip2)},
+
+		{"archive", pkg.NewArchive(pkg.NewFile("", []byte{0}))},
 	}
 	testCasesCache := make([]cacheTestCase, len(testCases))
 	for i, tc := range testCases {
diff --git a/internal/pkg/pkg.go b/internal/pkg/pkg.go
index 09d91f73..b8a997d8 100644
--- a/internal/pkg/pkg.go
+++ b/internal/pkg/pkg.go
@@ -510,6 +510,8 @@ const (
 	KindFile
 	// KindDecompress is the kind of [Artifact] returned by [NewDecompress].
 	KindDecompress
+	// KindArchive is the kind of [Artifact] returned by [NewArchive].
+	KindArchive
 
 	// _kindEnd is the total number of kinds and does not denote a kind.
 	_kindEnd