Detect UStar tar archives (#308)

* Detect UStar tar archives

UStar tar archives have a `magic` header field at byte offset 257 in
each entry whose value begins with the string `ustar`. Identify them
with the MIME type `application/x-tar`.

Also add test cases for a number of UStar-compatible formats, created by
GNU tar 1.29 (with `--format=<format-name>`):

* `tar.gnu.tar`
* `tar.oldgnu.tar`
* `tar.posix.tar`
* `tar.ustar.tar`

as well as `tar.star.tar` (created by star 1.6) and, for completeness,
`tar.v7-gnu.tar` (a v7 tar archive created by GNU tar 1.29).

Fixes #307.

internal/magic/archive.go

@@ -74,13 +74,21 @@ func CRX(raw []byte, limit uint32) bool {
 }
 
 // Tar matches a (t)ape (ar)chive file.
-//
-// Signature source: https://www.nationalarchives.gov.uk/PRONOM/Format/proFormatSearch.aspx?status=detailReport&id=385&strPageToDisplay=signatures
 func Tar(raw []byte, _ uint32) bool {
+	// The "magic" header field for files in in UStar (POSIX IEEE P1003.1) archives
+	// has the prefix "ustar". The values of the remaining bytes in this field vary
+	// by archiver implementation.
+	if len(raw) >= 512 && bytes.HasPrefix(raw[257:], []byte{0x75, 0x73, 0x74, 0x61, 0x72}) {
+		return true
+	}
+
 	if len(raw) < 256 {
 		return false
 	}
 
+	// The older v7 format has no "magic" field, and therefore must be identified
+	// with heuristics based on legal ranges of values for other header fields:
+	// https://www.nationalarchives.gov.uk/PRONOM/Format/proFormatSearch.aspx?status=detailReport&id=385&strPageToDisplay=signatures
 	rules := []struct {
 		min, max uint8
 		i        int

mimetype_test.go

@@ -180,12 +180,20 @@ var files = map[string]string{
 	// the timestamps.
 	"not.srt.txt": "text/plain; charset=utf-8",
 	// not.srt.2.txt does not specify milliseconds.
-	"not.srt.2.txt":   "text/plain; charset=utf-8",
-	"svg.1.svg":       "image/svg+xml",
-	"svg.svg":         "image/svg+xml",
-	"swf.swf":         "application/x-shockwave-flash",
-	"tar.tar":         "application/x-tar",
-	"tar.v7.tar":      "application/x-tar",
+	"not.srt.2.txt":  "text/plain; charset=utf-8",
+	"svg.1.svg":      "image/svg+xml",
+	"svg.svg":        "image/svg+xml",
+	"swf.swf":        "application/x-shockwave-flash",
+	"tar.tar":        "application/x-tar",
+	"tar.gnu.tar":    "application/x-tar",
+	"tar.oldgnu.tar": "application/x-tar",
+	"tar.posix.tar":  "application/x-tar",
+	// tar.star.tar was generated with star 1.6.
+	"tar.star.tar":  "application/x-tar",
+	"tar.ustar.tar": "application/x-tar",
+	"tar.v7.tar":    "application/x-tar",
+	// tar.v7-gnu.tar is a v7 tar archive generated with GNU tar 1.29.
+	"tar.v7-gnu.tar":  "application/x-tar",
 	"tcl.tcl":         "text/x-tcl",
 	"tcx.tcx":         "application/vnd.garmin.tcx+xml",
 	"tiff.tiff":       "image/tiff",