gsutil: Parallel-safe, specify target, add clean.

gsutil.py

Index: gsutil.py
diff --git a/gsutil.py b/gsutil.py
index 53589a27bc48e97261d47f3d22a18b9ea1a4d88c..8a11c050651fe15dbc1ae9771357c5fc9573519f 100755
--- a/gsutil.py
+++ b/gsutil.py
@@ -14,6 +14,7 @@ import os
 import shutil
 import subprocess
 import sys
+import time
 import urllib2
 import zipfile
 
@@ -26,7 +27,6 @@ DEFAULT_BIN_DIR = os.path.join(THIS_DIR, 'external_bin', 'gsutil')
 DEFAULT_FALLBACK_GSUTIL = os.path.join(
     THIS_DIR, 'third_party', 'gsutil', 'gsutil')
 
-
 class InvalidGsutilError(Exception):
   pass
 
@@ -73,22 +73,35 @@ def check_gsutil(gsutil_bin):
       [sys.executable, gsutil_bin, 'version'],
       stdout=subprocess.PIPE, stderr=subprocess.STDOUT) == 0
 
-def ensure_gsutil(version, target):
+def ensure_gsutil(version, target, clean):
   bin_dir = os.path.join(target, 'gsutil_%s' % version)
   gsutil_bin = os.path.join(bin_dir, 'gsutil', 'gsutil')
-  if os.path.isfile(gsutil_bin) and check_gsutil(gsutil_bin):
+  if not clean and os.path.isfile(gsutil_bin) and check_gsutil(gsutil_bin):
     # Everything is awesome! we're all done here.
     return gsutil_bin
 
-  if os.path.isdir(bin_dir):
-    # Clean up if we're redownloading a corrupted gsutil.
-    shutil.rmtree(bin_dir)
-  cache_dir = os.path.join(target, '.cache_dir')
+  # Clean up if we're redownloading a corrupted gsutil.
+  suffix = '%d_%d' % (time.time(), os.getpid())

[vapier, 2015/09/17 19:31:00]

this isn't parallel safe.  you cannot rely on PID's being unique (pid namespaces
for example).  if you need a tempdir, then use a proper library like
tempfile.NamedTemporaryFile().

[dnj, 2015/09/17 19:51:47]

Using "tempfile.mkdtemp" now.

+  try:
+    cleanup_path = '%s.%s' % (bin_dir, suffix)
+    shutil.move(bin_dir, cleanup_path)
+    shutil.rmtree(cleanup_path)
+  except IOError:
+    # Directory was not present.
+    pass
+
+  cache_dir = os.path.join(target, '.cache_dir.%s' % (suffix,))

[Ryan Tseng, 2015/09/17 19:26:45]

this would be more aptly called "tempdir" now?

maybe also clean up lingering cache_dirs older than 1 hour.

[dnj, 2015/09/17 19:51:46]

Acknowledged.

   if not os.path.isdir(cache_dir):
     os.makedirs(cache_dir)
   target_zip_filename = download_gsutil(version, cache_dir)
   with zipfile.ZipFile(target_zip_filename, 'r') as target_zip:
-    target_zip.extractall(bin_dir)
+    target_zip.extractall(cache_dir)
+
+  try:
+    shutil.move(cache_dir, bin_dir)
+  except IOError:
+    # Something else did this in parallel.
+    pass
 
   # Final check that the gsutil bin is okay.  This should never fail.
   if not check_gsutil(gsutil_bin):
@@ -97,9 +110,9 @@ def ensure_gsutil(version, target):
   return gsutil_bin
 
 
-def run_gsutil(force_version, fallback, target, args):
+def run_gsutil(force_version, fallback, target, args, clean=False):
   if force_version:
-    gsutil_bin = ensure_gsutil(force_version, target)
+    gsutil_bin = ensure_gsutil(force_version, target, clean)
   else:
     gsutil_bin = fallback
   cmd = [sys.executable, gsutil_bin] + args
@@ -107,10 +120,18 @@ def run_gsutil(force_version, fallback, target, args):
 
 
 def parse_args():
+  bin_dir = os.environ.get('DEPOT_TOOLS_GSUTIL_BIN_DIR')

[Ryan Tseng, 2015/09/17 19:26:45]

get(key, default)

[dnj, 2015/09/17 19:51:47]

Done.

+  if not bin_dir:
+    bin_dir = DEFAULT_BIN_DIR
+
   parser = argparse.ArgumentParser()
   parser.add_argument('--force-version', default='4.13')
+  parser.add_argument('--clean', action='store_true',

[Ryan Tseng, 2015/09/17 19:26:45]

What do you see the use case for this being?

[dnj, 2015/09/17 19:51:47]

Useful for testing massively in parallel.

+      help='Clear any existing gsutil package, forcing a new download.')
   parser.add_argument('--fallback', default=DEFAULT_FALLBACK_GSUTIL)
-  parser.add_argument('--target', default=DEFAULT_BIN_DIR)
+  parser.add_argument('--target', default=bin_dir,
+      help='The target directory to download/store a gsutil version in. '
+           '(default is %(default)s).')
   parser.add_argument('args', nargs=argparse.REMAINDER)
 
   args, extras = parser.parse_known_args()
@@ -118,12 +139,13 @@ def parse_args():
     args.args.pop(0)
   if extras:
     args.args = extras + args.args
-  return args.force_version, args.fallback, args.target, args.args
+  return args
 
 
 def main():
-  force_version, fallback, target, args = parse_args()
-  return run_gsutil(force_version, fallback, target, args)
+  args = parse_args()
+  return run_gsutil(args.force_version, args.fallback, args.target, args.args,
+                    clean=args.clean)
 
 if __name__ == '__main__':
   sys.exit(main())