Add support for tar.gz archive files to download from download_from_google_storage

upload_to_google_storage.py

Index: upload_to_google_storage.py
diff --git a/upload_to_google_storage.py b/upload_to_google_storage.py
index 4cf9d1a6e84970e428fd039a3b8b151ab00ef64f..44bd1bfce0a097c73965452dc275f3a382c20a43 100755
--- a/upload_to_google_storage.py
+++ b/upload_to_google_storage.py
@@ -12,6 +12,7 @@ import Queue
 import re
 import stat
 import sys
+import tarfile
 import threading
 import time
 
@@ -206,12 +207,38 @@ def upload_to_google_storage(
 
   return max_ret_code
 

[Ryan Tseng, 2015/06/25 17:19:14]

top level function, 2 newlines

[ricow1, 2015/06/26 06:18:05]

Done.

+def create_archives(dirs):
+  archive_names = []
+  for name in dirs:
+    tarname = '%s.tar.gz' % name
+    with tarfile.open(tarname, 'w:gz') as tar:
+      tar.add(name)
+    archive_names.append(tarname)
+  return archive_names
+

[Ryan Tseng, 2015/06/25 17:19:14]

2 newlines

[ricow1, 2015/06/26 06:18:05]

Done.

+def validate_archive_dirs(dirs):
+  # We don't allow .. in paths in our archives.
+  if any(map(lambda x: '..' in x, dirs)):
+    return False
+  # We only allow dirs.
+  if any(map(lambda x: not os.path.isdir(x), dirs)):
+    return False
+  # We don't allow sym links in our archives.
+  if any(map(os.path.islink, dirs)):
+    return False
+  # We required that the subdirectories we are archiving are all just below
+  # cwd.
+  return not any(map(lambda x: x not in next(os.walk('.'))[1], dirs))
+

[Ryan Tseng, 2015/06/25 17:19:14]

remove 1 newline so there are 2 total

[ricow1, 2015/06/26 06:18:05]

Done.

+
 
 def main():
   parser = optparse.OptionParser(USAGE_STRING)
   parser.add_option('-b', '--bucket',
                     help='Google Storage bucket to upload to.')
   parser.add_option('-e', '--boto', help='Specify a custom boto file.')
+  parser.add_option('-z', '--archive', action='store_true',
+                    help='Archive directory as a tar.gz file')
   parser.add_option('-f', '--force', action='store_true',
                     help='Force upload even if remote file exists.')
   parser.add_option('-g', '--gsutil_path', default=GSUTIL_DEFAULT_PATH,
@@ -235,6 +262,16 @@ def main():
   # Enumerate our inputs.
   input_filenames = get_targets(args, parser, options.use_null_terminator)
 
+

[Ryan Tseng, 2015/06/25 17:19:14]

remove newline (inside function, just 1 is enough)

[ricow1, 2015/06/26 06:18:05]

Done.

+  if options.archive:
+    if validate_archive_dirs(input_filenames):

[Ryan Tseng, 2015/06/25 17:19:15]

if not validate_archive_dirs()?

[ricow1, 2015/06/26 06:18:05]

Done.

+      parser.error('Only directories just below cwd are valid entries when '
+                   'using the --archive argument. Entries can not contain .. '
+                   ' and entries can not be symlinks. Entries was %s' %
+                    input_filenames)
+      return 1
+    input_filenames = create_archives(input_filenames)
+
   # Make sure we can find a working instance of gsutil.
   if os.path.exists(GSUTIL_DEFAULT_PATH):
     gsutil = Gsutil(GSUTIL_DEFAULT_PATH, boto_path=options.boto)