Scripts to download files from google storage based on sha1 sums

download_from_google_storage.py

+#!/usr/bin/env python
+# Copyright (c) 2012 The Chromium Authors. All rights reserved.
+# Use of this source code is governed by a BSD-style license that can be
+# found in the LICENSE file.
+
+"""Download files from Google Storage based on SHA1 sums."""
+
+
+import optparse
+import os
+import Queue
+import re
+import sys
+import threading
+import time
+
+import gstools
+
+
+GSUTIL_DEFAULT_PATH = os.path.join(
+    os.path.dirname(os.path.abspath(__file__)),
+    'third_party', 'gsutil', 'gsutil')
+
+
+def enumerate_work_queue(input_filename, work_queue, options):

[M-A Ruel, 2013/02/28 14:53:56]

Please pass directory, recursive, output as 3 separate arguments.

[Ryan Tseng, 2013/03/01 02:41:35]

Done.

+  work_queue_size = 0

[M-A Ruel, 2013/02/28 14:53:56]

if not directory:
  work_queue.put((input_filename, options.output))
  return 1

then the rest of the code.

[Ryan Tseng, 2013/03/01 02:41:35]

Done.

+  if options.directory:
+    for root, dirs, files in os.walk(input_filename):
+      if not options.recursive:
+        for item in dirs[:]:
+          dirs.remove(item)
+      else:
+        for exclude in ['.svn', '.git']:
+          if exclude in dirs:
+            dirs.remove(exclude)
+      for filename in files:
+        full_path = os.path.join(root, filename)
+        if full_path.endswith('.sha1'):
+          with open(full_path, 'rb') as f:
+            sha1_match = re.match('^([A-Za-z0-9]{40})$', f.read(1024).rstrip())
+          if sha1_match:
+            work_queue.put(
+                (sha1_match.groups(1)[0], full_path.replace('.sha1', '')))
+            work_queue_size += 1
+          else:
+            print >> sys.stderr, 'No sha1 sum found in %s.' % filename

[M-A Ruel, 2013/02/28 14:53:56]

Is this an error or something to safely ignore?

[Ryan Tseng, 2013/03/01 02:41:35]

Hm. This should probably throw an error and stop the program.  But I can see
people wanting to ignore this error. I'll add a flag.

+  else:
+    work_queue.put((input_filename, options.output))
+    work_queue_size += 1
+  return work_queue_size
+
+
+def _downloader_worker_thread(thread_num, q, options, base_url, gsutil, out_q):
+  while True:
+    input_sha1_sum, output_filename = q.get()
+    if input_sha1_sum is None:
+      out_q.put('Thread %d is done' % thread_num)
+      return
+    if os.path.exists(output_filename) and not options.force:
+      if gstools.GetSHA1(output_filename) == input_sha1_sum:
+        out_q.put('File %s exists and SHA1 sum (%s) matches. Skipping.' % (

[M-A Ruel, 2013/02/28 14:53:56]

out_q.put(
    'File ...

[Ryan Tseng, 2013/03/01 02:41:35]

Done.

+            output_filename , input_sha1_sum))
+        continue
+    # Check if file exists.
+    file_url = '%s/%s' % (base_url, input_sha1_sum)
+    if gsutil.check_call('ls', file_url)[0] != 0:
+      out_q.put('File %s for %s does not exist, skipping.' % (
+          file_url, output_filename))
+      continue
+    # Fetch the file.
+    out_q.put('Downloading %s to %s...' % (file_url, output_filename))
+    code = gsutil.call('cp', '-q', file_url, output_filename)
+    if code != 0:
+      out_q.put(gsutil.stderr)
+      return code
+
+
+def download_from_google_storage(input_filename, base_url, gsutil, options):

[M-A Ruel, 2013/02/28 14:53:56]

Accept each option individually instead.

[Ryan Tseng, 2013/03/01 02:41:35]

Done.

+  # Start up all the worker threads.
+  all_threads = []
+  download_timer = time.time()
+  stdout_queue = Queue.Queue()
+  work_queue = Queue.Queue()
+  for thread_num in range(options.num_threads):
+    t = threading.Thread(
+        target=_downloader_worker_thread,
+        args=[thread_num, work_queue, options, base_url,
+              gsutil.clone(), stdout_queue])
+    t.daemon = True
+    t.start()
+    all_threads.append(t)
+
+  # Enumerate our work queue.
+  work_queue_size = enumerate_work_queue(input_filename, work_queue, options)
+  for _ in all_threads:
+    work_queue.put((None, None))  # Used to tell worker threads to stop.
+
+  # Wait for all downloads to finish.
+  while True:

[M-A Ruel, 2013/02/28 14:53:56]

while not any(t.is_alive() for t in all_threads) and not stdout_queue.empty():
  print stdout_queue.get()

[Ryan Tseng, 2013/03/01 02:41:35]

Done.  That's a useful keyword :O
Added some stuff because there was a race condition where the "Thread finished"
gets queued, printed, and the looped continued, checked that the thread was
alive, blocked on get(), and then the last thread died after that.

+    num_alive_threads = 0
+    for t in all_threads:
+      if t.is_alive():
+        num_alive_threads += 1
+    if num_alive_threads == 0 and stdout_queue.empty():
+      break
+    line = stdout_queue.get()
+    print line
+    
+
+  print 'Success.'
+  print 'Downloading %d files took %1f second(s)' % (
+      work_queue_size, time.time() - download_timer)
+  return 0
+
+
+def main(args):
+  usage = ('usage: %prog [options] target\nTarget must be:\n'
+           '(default) a sha1 sum ([A-Za-z0-9]{40}).\n(-s or --sha1_file) a '
+           '.sha1 file, containing a sha1 sum on the first line. (-d or '
+           '--directory) A directory to scan for .sha1 files. ')
+  parser = optparse.OptionParser(usage)
+  parser.add_option('-o', '--output',
+                    help='Specify the output file name. Defaults to:\n'
+                         '(a) Given a SHA1 hash, the name is the SHA1 hash.\n'
+                         '(b) Given a .sha1 file or directory, the name will '
+                         'match (.*).sha1.')
+  parser.add_option('-b', '--bucket',
+                    help='Google Storage bucket to fetch from.')
+  parser.add_option('-e', '--boto',
+                    help='Specify a custom boto file.')
+  parser.add_option('-c', '--no_resume', action='store_true',
+                    help='Resume download if file is partially downloaded.')
+  parser.add_option('-f', '--force', action='store_true',
+                    help='Force download even if local file exists.')
+  parser.add_option('-r', '--recursive', action='store_true',
+                    help='Scan folders recursively for .sha1 files. '
+                         'Must be used with -d/--directory')
+  parser.add_option('-t', '--num_threads', default=1, type='int',
+                    help='Number of downloader threads to run.')
+  parser.add_option('-d', '--directory', action='store_true',
+                    help='The target is a directory.  '
+                         'Cannot be used with -s/--sha1_file.')
+  parser.add_option('-s', '--sha1_file', action='store_true',
+                    help='The target is a file containing a sha1 sum.  '
+                         'Cannot be used with -d/--directory.')
+
+  (options, args) = parser.parse_args()
+  if not args:
+    parser.error('Missing target.')
+  if len(args) > 1:
+    parser.error('Too many targets.')
+  if not options.bucket:
+    parser.error('Missing bucket.  Specify bucket with --bucket.')
+  if options.sha1_file and options.directory:
+    parser.error('Both --directory and --sha1_file are specified, '
+                 'can only specify one.')
+  elif options.recursive and not options.directory:
+    parser.error('--recursive specified but --directory not specified.')
+  elif options.output and options.directory:
+    parser.error('--directory is specified, so --output has no effect.')
+  else:
+    input_filename = args[0]
+
+  # Set output filename if not specified.
+  if not options.output and not options.directory:
+    if not options.sha1_file:
+      # Target is a sha1 sum, so output filename would also be the sha1 sum.
+      options.output = input_filename
+    elif options.sha1_file:
+      # Target is a .sha1 file.
+      if not input_filename.endswith('.sha1'):
+        parser.error('--sha1_file is specified, but the input filename '
+                     'does not end with .sha1, and no --output is specified. '
+                     'Either make sure the input filename has a .sha1 '
+                     'extension, or specify --output.')
+      options.output = input_filename[:-5]
+    else:
+      raise NotImplementedError('Unreachable state.')
+
+  # Check if output file already exists.
+  if not options.directory and not options.force and not options.no_resume:
+    if os.path.exists(options.output):
+      parser.error('Output file %s exists and --no_resume is specified.'
+                   % options.output)
+
+  # Make sure we can find a working instance of gsutil.
+  if os.path.exists(GSUTIL_DEFAULT_PATH):
+    gsutil = gstools.Gsutil(GSUTIL_DEFAULT_PATH)
+  else:
+    print >> sys.stderr, ('gsutil not found in %s, bad depot_tools checkout?' %
+                          GSUTIL_DEFAULT_PATH)
+    return 1
+
+  # Check we have a valid bucket with valid permissions.
+  base_url, code = gstools.CheckBucketPermissions(options.bucket, gsutil)
+  if code:
+    return code
+
+  return download_from_google_storage(input_filename, base_url, gsutil, options)
+
+
+if __name__ == '__main__':
+  sys.exit(main(sys.argv))