1 files changed, 26 insertions, 33 deletions
diff --git a/morphlib/bins.py b/morphlib/bins.py
index c5bacc26..2e8ba0b3 100644
--- a/morphlib/bins.py
+++ b/morphlib/bins.py
@@ -78,8 +78,12 @@ if sys.version_info < (2, 7, 3): # pragma: no cover
                 raise ExtractError("could not change owner")
     tarfile.TarFile.chown = fixed_chown
 
-def create_chunk(rootdir, chunkdir, include, dump_memory_profile=None):
-    '''Create a chunk from the contents of a directory.'''
+def create_chunk(rootdir, f, include, dump_memory_profile=None):
+    '''Create a chunk from the contents of a directory.
+    
+    ``f`` is an open file handle, to which the tar file is written.
+
+    '''
 
     dump_memory_profile = dump_memory_profile or (lambda msg: None)
 
@@ -87,42 +91,31 @@ def create_chunk(rootdir, chunkdir, include, dump_memory_profile=None):
     # chunk artifact. This is useful to avoid problems from smallish
     # clock skew. It needs to be recent enough, however, that GNU tar
     # does not complain about an implausibly old timestamp.
-    normalized_timestamp = (683074800, 683074800)
+    normalized_timestamp = 683074800
 
     dump_memory_profile('at beginning of create_chunk')
-
-    def check_parent(name, paths):
-        parent = os.path.dirname(name)
-        if parent:
-            path = os.path.join(rootdir, parent)
-            if parent != rootdir and path not in paths:
-                paths.append(path)
-            check_parent(parent, paths)
-
-    def filter_contents(dirname, filenames):
-        paths = [os.path.join(rootdir, relname) for relname in include]
-        for name in include:
-            check_parent(name, paths)
-
-        return [f for f in filenames if os.path.join(dirname, f) not in paths]
-
-    logging.debug('Copying artifact into %s.' % chunkdir)
-    shutil.copytree(rootdir, chunkdir,
-                    symlinks=True, ignore=filter_contents)
-
-    path_triplets = [(relname, os.path.join(chunkdir, relname),
-                      os.path.join(rootdir, relname))
-                     for relname in include]
-    for relname, filename, orig in path_triplets:
+    
+    path_pairs = [(relname, os.path.join(rootdir, relname))
+                  for relname in include]
+    tar = tarfile.open(fileobj=f, mode='w')
+    for relname, filename in path_pairs:
         # Normalize mtime for everything.
-        if not os.path.islink(filename):
-            os.utime(filename, normalized_timestamp)
+        tarinfo = tar.gettarinfo(filename,
+                                 arcname=relname)
+        tarinfo.ctime = normalized_timestamp
+        tarinfo.mtime = normalized_timestamp
+        if tarinfo.isreg():
+            with open(filename, 'rb') as f:
+                tar.addfile(tarinfo, fileobj=f)
+        else:
+            tar.addfile(tarinfo)
+    tar.close()
 
-    for relname, filename, orig in reversed(path_triplets):
-        if os.path.isdir(orig) and not os.path.islink(orig):
+    for relname, filename in reversed(path_pairs):
+        if os.path.isdir(filename) and not os.path.islink(filename):
             continue
         else:
-            os.remove(orig)
+            os.remove(filename)
     dump_memory_profile('after removing in create_chunks')
 
 
@@ -216,7 +209,7 @@ def unpack_binary_from_file(f, dirname):  # pragma: no cover
         tf.close()
 
 
-def unpack_binary(filename, dirname): # pragma: no cover
+def unpack_binary(filename, dirname):
     with open(filename, "rb") as f:
         unpack_binary_from_file(f, dirname)