internal/pkg: implement caching for files

This change contains primitives for validating and caching single-file artifacts.

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

internal/pkg/dir.go

@@ -0,0 +1,68 @@
+package pkg
+
+import (
+	"crypto/sha512"
+	"encoding/gob"
+	"io"
+	"io/fs"
+	"os"
+
+	"hakurei.app/container/check"
+)
+
+// FlatEntry is the representation of a directory entry via [Flatten].
+type FlatEntry struct {
+	Name string      // base name of the file
+	Mode fs.FileMode // file mode bits
+	Data []byte      // file content or symlink destination
+}
+
+// Flatten writes a deterministic representation of the contents of fsys to w.
+// The resulting data can be hashed to produce a deterministic checksum for the
+// directory.
+func Flatten(fsys fs.FS, root string, w io.Writer) error {
+	e := gob.NewEncoder(w)
+	return fs.WalkDir(fsys, root, func(path string, d fs.DirEntry, err error) error {
+		if err != nil {
+			return err
+		}
+
+		var fi fs.FileInfo
+		fi, err = d.Info()
+		if err != nil {
+			return err
+		}
+
+		ent := FlatEntry{
+			Name: fi.Name(),
+			Mode: fi.Mode(),
+		}
+		if ent.Mode.IsRegular() {
+			if ent.Data, err = fs.ReadFile(fsys, path); err != nil {
+				return err
+			}
+		} else if ent.Mode&fs.ModeSymlink != 0 {
+			var newpath string
+			if newpath, err = fs.ReadLink(fsys, path); err != nil {
+				return err
+			}
+			ent.Data = []byte(newpath)
+		}
+
+		return e.Encode(&ent)
+	})
+}
+
+// HashFS returns a checksum produced by hashing the result of [Flatten].
+func HashFS(fsys fs.FS, root string) (Checksum, error) {
+	h := sha512.New384()
+	if err := Flatten(fsys, root, h); err != nil {
+		return Checksum{}, err
+	}
+	return (Checksum)(h.Sum(nil)), nil
+}
+
+// HashDir returns a checksum produced by hashing the result of [Flatten].
+func HashDir(pathname *check.Absolute) (Checksum, error) {
+	return HashFS(os.DirFS(pathname.String()), ".")
+}