utarfile: Fix read/write handling of nulls in tar header.

For reading, the size is always terminated by a null, so just ignore it by
using 11 for the uctypes entry (this fixes a regression introduced in
7128d42).

For writing, the size must always be terminated by a null.

Signed-off-by: Damien George <[email protected]>

Author: dpgeorge

micropython/utarfile-write/manifest.py

@@ -1,4 +1,4 @@
-metadata(description="Adds write (create/append) support to utarfile.", version="0.1")
+metadata(description="Adds write (create/append) support to utarfile.", version="0.1.1")
 
 require("utarfile")
 package("utarfile")

micropython/utarfile-write/utarfile/write.py

@@ -11,9 +11,9 @@
 # http://www.gnu.org/software/tar/manual/html_node/Standard.html
 _TAR_HEADER = {
     "name": (uctypes.ARRAY | 0, uctypes.UINT8 | 100),
-    "mode": (uctypes.ARRAY | 100, uctypes.UINT8 | 7),
-    "uid": (uctypes.ARRAY | 108, uctypes.UINT8 | 7),
-    "gid": (uctypes.ARRAY | 116, uctypes.UINT8 | 7),
+    "mode": (uctypes.ARRAY | 100, uctypes.UINT8 | 8),
+    "uid": (uctypes.ARRAY | 108, uctypes.UINT8 | 8),
+    "gid": (uctypes.ARRAY | 116, uctypes.UINT8 | 8),
     "size": (uctypes.ARRAY | 124, uctypes.UINT8 | 12),
     "mtime": (uctypes.ARRAY | 136, uctypes.UINT8 | 12),
     "chksum": (uctypes.ARRAY | 148, uctypes.UINT8 | 8),
@@ -26,12 +26,6 @@
 _RECORDSIZE = const(_BLOCKSIZE * 20)  # length of records
 
 
-# Write a string into a bytearray by copying each byte.
-def _setstring(b, s, maxlen):
-    for i, c in enumerate(s.encode("utf-8")[:maxlen]):
-        b[i] = c
-
-
 def _open_write(self, name, mode, fileobj):
     if mode == "w":
         if not fileobj:
@@ -72,18 +66,18 @@ def addfile(self, tarinfo, fileobj=None):
         if not name.endswith("/"):
             name += "/"
     hdr = uctypes.struct(uctypes.addressof(buf), _TAR_HEADER, uctypes.LITTLE_ENDIAN)
-    _setstring(hdr.name, name, 100)
-    _setstring(hdr.mode, "%06o " % (tarinfo.mode & 0o7777), 7)
-    _setstring(hdr.uid, "%06o " % tarinfo.uid, 7)
-    _setstring(hdr.gid, "%06o " % tarinfo.gid, 7)
-    _setstring(hdr.size, "%011o " % size, 12)
-    _setstring(hdr.mtime, "%011o " % tarinfo.mtime, 12)
-    _setstring(hdr.typeflag, "5" if tarinfo.isdir() else "0", 1)
+    hdr.name[:] = name.encode("utf-8")[:100]
+    hdr.mode[:] = b"%07o\0" % (tarinfo.mode & 0o7777)
+    hdr.uid[:] = b"%07o\0" % tarinfo.uid
+    hdr.gid[:] = b"%07o\0" % tarinfo.gid
+    hdr.size[:] = b"%011o\0" % size
+    hdr.mtime[:] = b"%011o\0" % tarinfo.mtime
+    hdr.typeflag[:] = b"5" if tarinfo.isdir() else b"0"
     # Checksum is calculated with checksum field all blanks.
-    _setstring(hdr.chksum, " " * 8, 8)
+    hdr.chksum[:] = b"        "
     # Calculate and insert the actual checksum.
     chksum = sum(buf)
-    _setstring(hdr.chksum, "%06o\0" % chksum, 7)
+    hdr.chksum[:] = b"%06o\0 " % chksum
     # Emit the header.
     self.f.write(buf)
     self.offset += len(buf)

micropython/utarfile/manifest.py

@@ -1,4 +1,4 @@
-metadata(description="Read-only implementation of Python's tarfile.", version="0.4.0")
+metadata(description="Read-only implementation of Python's tarfile.", version="0.4.1")
 
 # Originally written by Paul Sokolovsky.
 

micropython/utarfile/utarfile/__init__.py

@@ -4,9 +4,10 @@
 
 # Minimal set of tar header fields for reading.
 # http://www.gnu.org/software/tar/manual/html_node/Standard.html
+# The "size" entry is 11 (not 12) to implicitly cut off the null terminator.
 _TAR_HEADER = {
     "name": (uctypes.ARRAY | 0, uctypes.UINT8 | 100),
-    "size": (uctypes.ARRAY | 124, uctypes.UINT8 | 12),
+    "size": (uctypes.ARRAY | 124, uctypes.UINT8 | 11),
 }
 
 DIRTYPE = const("dir")