Merge branch '13939-unqualified-contents-orders'

[arvados.git] / crunch_scripts / decompress-all.py

diff --git a/crunch_scripts/decompress-all.py b/crunch_scripts/decompress-all.py
index 460425f8b2e06c4ff7a42ca34206231150c98a0f..100ea12239938f8fd8d702b61f97f2990c94a445 100755 (executable)
--- a/crunch_scripts/decompress-all.py
+++ b/crunch_scripts/decompress-all.py
@@ -1,4 +1,7 @@
 #!/usr/bin/env python
+# Copyright (C) The Arvados Authors. All rights reserved.
+#
+# SPDX-License-Identifier: Apache-2.0
 
 #
 # decompress-all.py
@@ -18,6 +21,7 @@ import re
 import subprocess
 import os
 import sys
+import crunchutil.robust_put as robust_put
 
 arvados.job_setup.one_task_per_input_file(if_sequence=0, and_end_task=True,
                                           input_as_path=True)
@@ -32,7 +36,7 @@ outdir = os.path.join(task.tmpdir, "output")
 os.makedirs(outdir)
 os.chdir(outdir)
 
-if infile_parts == None:
+if infile_parts is None:
     print >>sys.stderr, "Failed to parse input filename '%s' as a Keep file\n" % input_file
     sys.exit(1)
 
@@ -40,17 +44,21 @@ cr = arvados.CollectionReader(infile_parts.group(1))
 streamname = infile_parts.group(3)[1:]
 filename = infile_parts.group(4)[1:]
 
-if streamname != None:
+if streamname is not None:
     subprocess.call(["mkdir", "-p", streamname])
     os.chdir(streamname)
 else:
     streamname = '.'
-streamreader = filter(lambda s: s.name() == streamname, cr.all_streams())[0]
-filereader = streamreader.files()[filename]
-rc = subprocess.call(["dtrx", "-r", "-n", "-q", arvados.get_task_param_mount('input')])
-if rc == 0:
-    out = arvados.CollectionWriter()
-    out.write_directory_tree(outdir, max_manifest_depth=0)
-    task.set_output(out.finish())
+
+m = re.match(r'.*\.(gz|Z|bz2|tgz|tbz|zip|rar|7z|cab|deb|rpm|cpio|gem)$', arvados.get_task_param_mount('input'), re.IGNORECASE)
+
+if m is not None:
+    rc = subprocess.call(["dtrx", "-r", "-n", "-q", arvados.get_task_param_mount('input')])
+    if rc == 0:
+        task.set_output(robust_put.upload(outdir))
+    else:
+        sys.exit(rc)
 else:
+    streamreader = filter(lambda s: s.name() == streamname, cr.all_streams())[0]
+    filereader = streamreader.files()[filename]
     task.set_output(streamname + filereader.as_manifest()[1:])