tar/asm: add FileGetter and concerns in README

tar/asm/README.md

@@ -5,6 +5,21 @@ This library for assembly and disassembly of tar archives, facilitated by
 `github.com/vbatts/tar-split/tar/storage`.
 
 
+Concerns
+--------
+
+For completely safe assembly/disassembly, there will need to be a CAS
+directory, that maps to a checksum in the `storage.Entity` of
+`storage.FileType`.
+
+This is due to the fact that tar archives _can_ allow multiple records for the
+same path, but the last one effectively wins. Even if the prior records had a
+different payload. 
+
+In this way, when assembling an archive from relative paths, if the archive has
+multiple entries for the same path, then all payloads read in from a relative
+path would be identical.
+
 
 Thoughts
 --------

tar/asm/assemble.go

@@ -2,13 +2,11 @@ package asm
 
 import (
 	"io"
-	"os"
-	"path"
 
 	"github.com/vbatts/tar-split/tar/storage"
 )
 
-func NewTarStream(relpath string, up storage.Unpacker) io.ReadCloser {
+func NewTarStream(fg FileGetter, up storage.Unpacker) io.ReadCloser {
 	pr, pw := io.Pipe()
 	go func() {
 		for {
@@ -24,7 +22,16 @@ func NewTarStream(relpath string, up storage.Unpacker) io.ReadCloser {
 					break
 				}
 			case storage.FileType:
-				if err := writeEntryFromRelPath(pw, relpath, entry); err != nil {
+				if entry.Size == 0 {
+					continue
+				}
+				fh, err := fg.Get(entry.Name)
+				if err != nil {
+					pw.CloseWithError(err)
+					break
+				}
+				defer fh.Close()
+				if _, err := io.Copy(pw, fh); err != nil {
 					pw.CloseWithError(err)
 					break
 				}
@@ -33,21 +40,3 @@ func NewTarStream(relpath string, up storage.Unpacker) io.ReadCloser {
 	}()
 	return pr
 }
-
-func writeEntryFromRelPath(w io.Writer, root string, entry *storage.Entry) error {
-	if entry.Size == 0 {
-		return nil
-	}
-
-	// FIXME might should have a check for '../../../../etc/passwd' attempts?
-	fh, err := os.Open(path.Join(root, entry.Name))
-	if err != nil {
-		return err
-	}
-	defer fh.Close()
-	if _, err := io.Copy(w, fh); err != nil {
-		return err
-	}
-
-	return nil
-}

tar/asm/getter.go

@@ -0,0 +1,26 @@
+package asm
+
+import (
+	"io"
+	"os"
+	"path"
+)
+
+type FileGetter interface {
+	// Get returns a stream for the provided file path
+	Get(string) (io.ReadCloser, error)
+}
+
+// NewPathFileGetter returns a FileGetter that is for files relative to path relpath.
+func NewPathFileGetter(relpath string) FileGetter {
+	return &pathFileGetter{root: relpath}
+}
+
+type pathFileGetter struct {
+	root string
+}
+
+func (pfg pathFileGetter) Get(filename string) (io.ReadCloser, error) {
+	// FIXME might should have a check for '../../../../etc/passwd' attempts?
+	return os.Open(path.Join(pfg.root, filename))
+}