def PrepareToUpload(local_root, dest_dir):
"""Check necessary conditions required for uploading.
Args:
local_root: The local directory containing directories to upload.
dest_dir: The remote directory to upload to.
Returns:
existing_dest_paths: A set of filenames for existing files in the
destination directory.
gs_api: The gsutil.GsutilApi object.
Raises:
BadTimeToUploadError: Internet is not available.
ValueError: local_root is not a valid path.
"""
if not os.path.isdir(local_root):
raise ValueError('Cannot find local directory %s.' % local_root)
if not HasInternet():
raise BadTimeToUploadError('No internet connection detected.')
gs_api = gsutil.GsutilApi()
try:
existing_dest_paths = set(gs_api.List(dest_dir))
except httplib2.ServerNotFoundError:
raise BadTimeToUploadError('Internet has become unavailable.')
return existing_dest_paths, gs_api
def RemoveRemoteFiles():
gs_api = gsutil.GsutilApi()
filenames = gs_api.List(
gcloud_util.GcsPath(CLOUD_BASE_DIR, CLOUD_LOG_PATH))
for filename in filenames:
gs_api.DeleteFile(filename)
self.assertFalse(
gs_api.List(gcloud_util.GcsPath(CLOUD_BASE_DIR,
CLOUD_LOG_PATH)))
def __init__(self, folder, base_params):
self._folder = folder
self.label = 'Wind Database'
# Pull down info from the cloud about files in this database set.
gsutil_api = gsutil.GsutilApi()
cloud_path = os.path.join(
'gs://gcp-public-data-makani-deps/deps/turbsim_databases', folder,
'h5_files')
self._databases = gsutil_api.List(cloud_path)
self._base_params = base_params
def DownloadDatabase(database_file, local_path):
"""Downloads TurbSim file to the specified local directory, if needed."""
if os.path.exists(local_path):
print 'TurbSim database already exists locally, skipping download.'
else:
gsutil_api = gsutil.GsutilApi()
# TODO: Set up a class for handling more autoselection tasks so
# that we don't have to pass as much back and forth,
# or requery the folder list?
cloud_path = os.path.join(GetCloudBasePath(),
GetOnlineFolder(database_file), 'h5_files',
database_file)
gsutil_api.Copy(cloud_path, local_path, False)
def GetOnlineFolder(database_file):
"""Extracts source id of the file and returns matching folder name."""
assert CheckTurbsimFileName(database_file), (
'File name does not follow TurbSim naming convention')
online_folder_id = ('-' + database_file[16:19] + '-')
gsutil_api = gsutil.GsutilApi()
online_paths = gsutil_api.List(GetCloudBasePath())
online_folder = [
os.path.basename(os.path.normpath(path)) for path in online_paths
if online_folder_id in path
]
assert len(online_folder) == 1, (
'Turbsim online folder identification failed.')
return online_folder[0]
def AssertRemoteFiles():
gs_api = gsutil.GsutilApi()
filenames = gs_api.List(
gcloud_util.GcsPath(CLOUD_BASE_DIR, CLOUD_LOG_PATH))
prefix = os.path.join(CLOUD_BASE_DIR, CLOUD_LOG_PATH)
self.assertEqual(
set(filenames), {
os.path.join(prefix,
'w7/logs/logs1-folder/subfolder/dummy.json'),
os.path.join(
prefix,
'w7/logs/logs1-folder/another_subfolder/dummy.json'),
os.path.join(
prefix,
'w7/logs/logs1-folder/one_more_subfolder/dummy.json'),
os.path.join(prefix, 'w7/logs/w7-2013.h5'),
os.path.join(prefix, 'M600A/logs/m600.h5'),
os.path.join(prefix, 'M600A/logs/dummy.json'),
os.path.join(prefix, 'w7/logs/logs1-folder/dummy.json'),
os.path.join(prefix, 'w7/logs/logs1-folder/dummy.txt'),
})
def IterFilesFromCloud(path_prefix, regex_str):
"""Iterate through files in a cloud path recursively.
The function downloads matching files in the cloud one by one.
Downloaded files are removed after each iteration.
Args:
path_prefix: Path, or prefix, to the cloud storage. E.g., gs://bucket/dir/.
regex_str: Regular expression to match the file from the beginning.
Yields:
The full cloud path to the file, and the path to the downloaded file.
"""
gs_api = gsutil.GsutilApi()
filenames = [
f for f in gs_api.List(path_prefix)
if f.endswith('.h5') and not f.endswith('-format.h5')
]
if regex_str:
regex = re.compile(regex_str)
filenames = [
f for f in filenames if regex.match(gcloud_util.GcsBasename(f))
]
if not filenames:
print 'Found no files matching the criteria.'
return
for full_cloud_path in sorted(filenames):
temp_fp = tempfile.NamedTemporaryFile(suffix='.h5', delete=False)
temp_fp.close()
print '------------------- %s ------------' % full_cloud_path
print 'Downloading %s...' % full_cloud_path
gs_api.Copy(full_cloud_path, temp_fp.name, overwrite=True)
yield (full_cloud_path, temp_fp.name)
os.remove(temp_fp.name)
def AssertNoRemoteFiles():
gs_api = gsutil.GsutilApi()
filenames = gs_api.List(
gcloud_util.GcsPath(CLOUD_BASE_DIR, CLOUD_LOG_PATH))
self.assertEqual(filenames, [])
