X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/1c5b0ee281a30b25bc622565dac2df75f99e4863..50ab35e7c81a9c1363da289bf76424d669bdf80c:/crunch_scripts/decompress-all.py

diff --git a/crunch_scripts/decompress-all.py b/crunch_scripts/decompress-all.py
index 8a0fb6f8d7..50d11f4d97 100755
--- a/crunch_scripts/decompress-all.py
+++ b/crunch_scripts/decompress-all.py
@@ -1,9 +1,24 @@
 #!/usr/bin/env python
 
+#
+# decompress-all.py
+#
+# Decompress all compressed files in the collection using the "dtrx" tool and
+# produce a new collection with the contents.  Uncompressed files
+# are passed through.
+#
+# input:
+# A collection at script_parameters["input"]
+#
+# output:
+# A manifest of the uncompressed contents of the input collection.
+
 import arvados
 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)
@@ -12,27 +27,35 @@ task = arvados.current_task()
 
 input_file = task['parameters']['input']
 
-result = re.match(r"(^[a-f0-9]{32}\+\d+)(\+\S+)*(/.*)(/[^/]+)$", input_file)
+infile_parts = re.match(r"(^[a-f0-9]{32}\+\d+)(\+\S+)*(/.*)?(/[^/]+)$", input_file)
 
 outdir = os.path.join(task.tmpdir, "output")
 os.makedirs(outdir)
 os.chdir(outdir)
 
-if result != None:
-    cr = arvados.CollectionReader(result.group(1))
-    streamname = result.group(3)[1:]
-    filename = result.group(4)[1:]
+if infile_parts is None:
+    print >>sys.stderr, "Failed to parse input filename '%s' as a Keep file\n" % input_file
+    sys.exit(1)
+
+cr = arvados.CollectionReader(infile_parts.group(1))
+streamname = infile_parts.group(3)[1:]
+filename = infile_parts.group(4)[1:]
 
+if streamname is not None:
     subprocess.call(["mkdir", "-p", streamname])
     os.chdir(streamname)
-    streamreader = filter(lambda s: s.name() == streamname, cr.all_streams())[0]
-    filereader = streamreader.files()[filename]
+else:
+    streamname = '.'
+
+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:
-        out = arvados.CollectionWriter()
-        out.write_directory_tree(outdir, max_manifest_depth=0)
-        task.set_output(out.finish())
+        task.set_output(robust_put.upload(outdir))
     else:
-        task.set_output(streamname + filereader.as_manifest()[1:])
+        sys.exit(rc)
 else:
-    sys.exit(1)
+    streamreader = filter(lambda s: s.name() == streamname, cr.all_streams())[0]
+    filereader = streamreader.files()[filename]
+    task.set_output(streamname + filereader.as_manifest()[1:])