Improve the bzip2 detection by looking for the data or end-of-data

marker that must follow the initial file signature.
Obtained from:	Scott

SVN-Revision: 33

diff --git a/libarchive/archive_read_support_compression_bzip2.c b/libarchive/archive_read_support_compression_bzip2.c
index 372eff0954198bd20a781a945f1083e95d5377b8..e6397df96344104e744de43e8cd54d7c77e17852 100644 (file)
--- a/libarchive/archive_read_support_compression_bzip2.c
+++ b/libarchive/archive_read_support_compression_bzip2.c
@@ -116,17 +116,29 @@ bid(const void *buff, size_t len)
        if (buffer[3] < '1' || buffer[3] > '9')
                return (0);
        bits_checked += 5;
+       if (len < 5)
+               return (bits_checked);
 
-       /*
-        * Research Question: Can we do any more to verify that this
-        * really is BZip2 format??  For 99.9% of the time, the above
-        * test is sufficient, but it would be nice to do a more
-        * thorough check.  It's especially troubling that the BZip2
-        * signature begins with all ASCII characters; a tar archive
-        * whose first filename begins with 'BZh3' would potentially
-        * fool this logic.  (It may also be possible to guard against
-        * such anomalies in archive_read_support_compression_none.)
-        */
+       /* After BZh[1-9], there must be either a data block
+        * which begins with 0x314159265359 or an end-of-data
+        * marker of 0x177245385090. */
+
+       if (buffer[4] == 0x31) {
+               /* Verify the data block signature. */
+               size_t s = len;
+               if (s > 10) s = 10;
+               if (memcmp(buffer + 4, "\x31\x41\x59\x26\x53\x59", s - 4) != 0)
+                       return (0);
+               bits_checked += 8 * (s - 4);
+       } else if (buffer[4] == 0x17) {
+               /* Verify the end-of-data marker. */
+               size_t s = len;
+               if (s > 10) s = 10;
+               if (memcmp(buffer + 4, "\x17\x72\x45\x38\x50\x90", s - 4) != 0)
+                       return (0);
+               bits_checked += 8 * (s - 4);
+       } else
+               return (0);
 
        return (bits_checked);
 }