Return underlying Seeker from Identify if possible (mholt#327)

* [feat] don't require io.Seeker for identify * tidy up * Refactor and simplify with some bug fixes * Clarify returned Reader in godoc comment * if underlying reader supports seek use that * update comment * Update formats.go Co-authored-by: Matt Holt <[email protected]> Co-authored-by: Matthew Holt <[email protected]>
drewstinnett · Mar 21, 2022 · d3e979a · d3e979a
1 parent dda6eaa
commit d3e979a
Show file tree

Hide file tree

Showing 2 changed files with 30 additions and 0 deletions.
diff --git a/formats.go b/formats.go
@@ -319,7 +319,14 @@ func (rr *rewindReader) rewind() {
 // bytes, then from the underlying stream. After calling this,
 // no more rewinding is allowed since reads from the stream are
 // not recorded, so rewinding properly is impossible.
+// If the underlying reader implements io.Seeker, then the
+// underlying reader will be used directly.
 func (rr *rewindReader) reader() io.Reader {
+	if ras, ok := rr.Reader.(io.Seeker); ok {
+		if _, err := ras.Seek(0, io.SeekStart); err == nil {
+			return rr.Reader
+		}
+	}
 	return io.MultiReader(bytes.NewReader(rr.buf.Bytes()), rr.Reader)
 }
 

diff --git a/formats_test.go b/formats_test.go
@@ -387,3 +387,26 @@ func TestIdentifyFindFormatByStreamContent(t *testing.T) {
 		})
 	}
 }
+
+func TestIdentifyAndOpenZip(t *testing.T) {
+	f, err := os.Open("testdata/test.zip")
+	checkErr(t, err, "opening zip")
+	defer f.Close()
+
+	format, reader, err := Identify("test.zip", f)
+	checkErr(t, err, "identifying zip")
+	if format.Name() != ".zip" {
+		t.Fatalf("unexpected format found: expected=.zip actual:%s", format.Name())
+	}
+
+	err = format.(Extractor).Extract(context.Background(), reader, nil, func(ctx context.Context, f File) error {
+		rc, err := f.Open()
+		if err != nil {
+			return err
+		}
+		defer rc.Close()
+		_, err = io.ReadAll(rc)
+		return err
+	})
+	checkErr(t, err, "extracting zip")
+}