+ for f in CollectionReader(zipball).all_files():
+ if not re.search('\.zip$', f.name()):
+ raise Exception("zipball_extract cannot handle filename %s"
+ % f.name())
+ zip_filename = os.path.join(path, os.path.basename(f.name()))
+ zip_file = open(zip_filename, 'wb')
+ while True:
+ buf = f.read(2**20)
+ if len(buf) == 0:
+ break
+ zip_file.write(buf)
+ zip_file.close()
+
+ p = subprocess.Popen(["unzip",
+ "-q", "-o",
+ "-d", path,
+ zip_filename],
+ stdout=None,
+ stdin=None, stderr=sys.stderr,
+ shell=False, close_fds=True)
+ p.wait()
+ if p.returncode != 0:
+ lockfile.close()
+ raise Exception("unzip exited %d" % p.returncode)
+ os.unlink(zip_filename)
+ os.symlink(zipball, os.path.join(path, '.locator'))
+ tld_extracts = filter(lambda f: f != '.locator', os.listdir(path))
+ lockfile.close()
+ if len(tld_extracts) == 1:
+ return os.path.join(path, tld_extracts[0])
+ return path
+
+ @staticmethod
+ def collection_extract(collection, path, files=[], decompress=True):
+ """Retrieve a collection from Keep and extract it to a local
+ directory. Return the absolute path where the collection was
+ extracted.
+
+ collection -- collection locator
+ path -- where to extract: absolute, or relative to job tmp
+ """
+ if not re.search('^/', path):
+ path = os.path.join(current_job().tmpdir, path)
+ lockfile = open(path + '.lock', 'w')
+ fcntl.flock(lockfile, fcntl.LOCK_EX)
+ try:
+ os.stat(path)
+ except OSError:
+ os.mkdir(path)
+ already_have_it = False
+ try:
+ if os.readlink(os.path.join(path, '.locator')) == collection:
+ already_have_it = True
+ except OSError:
+ pass
+
+ # emulate "rm -f" (i.e., if the file does not exist, we win)
+ try:
+ os.unlink(os.path.join(path, '.locator'))
+ except OSError:
+ if os.path.exists(os.path.join(path, '.locator')):
+ os.unlink(os.path.join(path, '.locator'))
+
+ files_got = []
+ for s in CollectionReader(collection).all_streams():
+ stream_name = s.name()
+ for f in s.all_files():
+ if (files == [] or
+ ((f.name() not in files_got) and
+ (f.name() in files or
+ (decompress and f.decompressed_name() in files)))):
+ outname = f.decompressed_name() if decompress else f.name()
+ files_got += [outname]
+ if os.path.exists(os.path.join(path, stream_name, outname)):
+ continue
+ util.mkdir_dash_p(os.path.dirname(os.path.join(path, stream_name, outname)))
+ outfile = open(os.path.join(path, stream_name, outname), 'wb')
+ for buf in (f.readall_decompressed() if decompress
+ else f.readall()):