def get_tags():
#TODO: Error checking
# global d
# d = {}
clarifai_api = ClarifaiApi()
blob_service = BlobService('calhacks', 'mm7EmY+T+MGahePBDSDU5LHpZR5tRXuh4MSco4jFrzHovOPEf06e18c89pxtPIo4NDVhhjSeaQY/FQmKNxjjyA==')
blob_name = request.form['blob_id']
# blob_name = blob_name.decode('utf-8')
blob_service.get_blob_to_path('imagestore', blob_name, 'out.png')
print("checkpoint 1")
i = open ('out.png', 'r')
strd = ""
for line in i:
strd += line.strip()
fname = 'img.png'
with open (fname, 'wb') as f:
f.write (base64.b64decode(strd))
f.close()
f = open (fname, 'rb')
result = clarifai_api.tag_images(f)
st = result['results'][0]['result']['tag']['classes'][0:6]
print(st)
for i in []:#['food', 'nobody', 'still life', 'meal', 'dish', 'plate', 'delicious', 'isolated', 'cutout', 'unhealthy', 'one', 'background']:
while i in st:
st.remove(i)
js = json.dumps(search_terms(st))
print(js)
return js
def submit():
blob_service = BlobService(account_name=ACCOUNT_NAME, account_key=ACCOUNT_KEY)
# Get a SAS signature (read for 24 hours) for the input container save to a string
inputsig = sasUrl(account=ACCOUNT_NAME, key=ACCOUNT_KEY, container=INPUT_CONTAINER, permission='r')
# Get a SAS signature (write for 24 hours) for the output container save to a string
outputsig = sasUrl(account = ACCOUNT_NAME, key = ACCOUNT_KEY, container = OUTPUT_CONTAINER, permission = 'rwl')
# List all the blobs and dump the content to a string
blobs = blob_service.list_blobs(INPUT_CONTAINER)
bloblist = []
for blob in blobs:
bloblist.append(blob.name)
os.environ[SLURMDEMO_INPUTSIG] = inputsig
os.environ[SLURMDEMO_OUTPUTSIG] = outputsig
os.environ[SLURMDEMO_BLOBLIST] = json.dumps(bloblist)
os.environ[SLURMDEMO_INPUTCONTAINER] = INPUT_CONTAINER
os.environ[SLURMDEMO_OUTPUTCONTAINER] = OUTPUT_CONTAINER
os.environ[SLURMDEMO_INPUTACCOUNT] = ACCOUNT_NAME
os.environ[SLURMDEMO_OUTPUTACCOUNT] = ACCOUNT_NAME
# Call sbatch
cli = "sbatch -N 2 -n 2 --array=0-{nb} slurmdemo.sh".format(nb=len(bloblist))
run(cli, showoutput=True)
def _createExternalStore():
from toil.jobStores.azureJobStore import _fetchAzureAccountKey
blobService = BlobService(account_key=_fetchAzureAccountKey(AzureJobStoreTest.accountName),
account_name=AzureJobStoreTest.accountName)
containerName = 'import-export-test-%s' % uuid.uuid4()
blobService.create_container(containerName)
return containerName
def getblob(request):
assert isinstance(request, HttpRequest)
blob_service = BlobService(account_name='araldrift', account_key='YOURKEYGOESHERE')
# http://<storage-account-name>.blob.core.windows.net/<container-name>/<blob-name>
blob_service.get_blob_to_path('flow', 'NARYN.day', './NARYN.day')
# return HttpResponse('ok ye of little faith')
return HttpResponse(json.dumps('i am just a wee bit of json'))
def __create_blob_container(self, storage_acc_name):
sms = self.__get_service_mgmt_object()
# Retrieve the primary key of your storage account
# Maybe the secondary key works too?
storage_acc_key = None
acounts = sms.list_storage_accounts()
for account in acounts:
if account.service_name == storage_acc_name:
storageServiceObj = sms.get_storage_account_keys(account.service_name)
storage_acc_key = storageServiceObj.storage_service_keys.primary
# Create a container
blob_service = BlobService(account_name=storage_acc_name,
account_key=storage_acc_key)
container_name = namesgenerator.get_random_name()
container_name += "container"
blob_service.create_container(container_name)
# This is the url to the container we just created
container_url_template = "http://{}.blob.core.windows.net/{}"
container_url = container_url_template.format(storage_acc_name, container_name)
#print "Created blob container with URL ",container_url
return container_url
def enumerate_objects(container):
blob_service = BlobService(AZURE_ACCOUNT_NAME, AZURE_ACCOUNT_KEY)
blobs = blob_service.list_blobs(container)
items = []
for blob in blobs:
items.append(blob.name)
return items
def gethydrograph(request):
'''
JSON return of a particular hydorgaph by start / stop / station / time interval
In response it will
generate a 404 error if the value is not found
or
return a JSON response with the requested slice.
'''
assert isinstance(request, HttpRequest)
start = request.GET.get('start', None)
end = request.GET.get('end', None)
# Here we can thieve the time parse code from LOS and use this line:
# rd, rh = extract_time(request)
# but for now this is commented out of course
station = request.GET.get('station',None)
interval = request.GET.get('interval',None)
blob_service = BlobService(account_name='araldrift', account_key='XXXXXXXXXXXXXXXXXXXXXXX')
blob_service.get_blob_to_path('flow', 'NARYN.day', './tmp.csv')
f = file('./tmp.csv')
h = []
while True:
line = f.readline()
if line == "": break
h.append(line)
f.close()
json_encoded_result = json.dumps(h)
# Keep the LOS original for reference (see time formatting):
# clean_req = {"rdidx": rdidx,"rdepth": rdepth,"rd": rd.strftime("%Y%m%d"), "rh": rh,"rparm": rparm,}
# clean_req = {"start": start, "end": end, "station": station, "interocitor": interval,}
# json_request = json.dumps(clean_req)
# json_encoded_result = json_request
try:
#####################
#
# This is the commented out LOS code (stripped out)
#
#####################
# Convert the depth to an int.
# depthIndex = int(rdidx)
# Get the depth slice from the dataset.
# ret = getDepthSliceByIndex((rd,rh), depthIndex, rparm, rnorm)
# Serialize out to JSON.
# json_encoded_result = json.dumps(ret)
# Cache the response to blob storage. (Would be great if this was async...)
# cacheResult(request.GET, json_encoded_result)
# Send the view on to the client.
#####################
return HttpResponse(json_encoded_result, content_type="application/json" )
except Exception as a:
return HttpResponseNotFound(content="No dice, either the inputs were out of range, the file couldn't be retrieved, or the winds weren't in your favor.")
def gethydrograph(request):
'''
Returns streamflow data by start / stop / station
In response it will
generate a 404 error if the value is not found
or
return a JSON response with the requested slice or a .csv file by default
'''
assert isinstance(request, HttpRequest)
start = request.GET.get('start', None)
end = request.GET.get('end', None)
station = request.GET.get('station',None)
interval = request.GET.get('interval',None)
jsondat = request.GET.get('jsondat',None)
plot = request.GET.get('plot',None)
#start blob service
stationfile = station + '.day.new'
downloadablefile = station + '_' + start + '_' + end + '.csv'
blob_service = BlobService(account_name='araldrift', account_key='otLzzkwQHQD3xFTQxwxy64PCL6eDINWGjSB7x6Ta2XVw3+3ffI5O2MhAEavf/r8qIW4G/dKrZAVg1R64nK7hDQ==')
blob_service.get_blob_to_path('flow', stationfile, './tmp.csv')
f = file('./tmp.csv')
#read in pandas data and subsetting
d_cols = ["DATE","FLOW"]
d = pd.read_csv('./tmp.csv', sep=" ", names=d_cols)
df = d[(d.DATE >= start) & (d.DATE <= end)]
h = df.to_json(orient='records')
json_encoded_result = json.dumps(h)
df.plot(x='DATE', y='FLOW', figsize=(14,6))
plt.savefig('./plot_test.png')
#h = []
#while True:
# line = f.readline()
# if line == "": break
# h.append(line)
#f.close()
try:
if jsondat in ['TRUE']:
response = HttpResponse(json_encoded_result, content_type="application/json" )
return response
elif plot in ['TRUE']:
image_data = open("./plot_test.png", "rb").read()
response = HttpResponse(image_data, content_type='image/png')
return response
else:
response = HttpResponse(content_type='text/csv')
response['Content-Disposition'] = 'attachment; filename=' +downloadablefile
df.to_csv(response, index=False, lineterminator='\r\n')
return response
except Exception as a:
return HttpResponseNotFound(content="No dice, either the inputs were out of range, the file couldn't be retrieved, or the winds weren't in your favor.")
def upload_from_text(container, content):
filename = str(uuid.uuid4())
blob_service = BlobService(account_name=config.AZURE_STORAGE_NAME, account_key=config.AZURE_STORAGE_KEY)
try:
blob_service.put_block_blob_from_text(container, filename, content)
return generate_blob_url(container, filename)
except:
return ""
def _createExternalStore(self):
from toil.jobStores.azureJobStore import _fetchAzureAccountKey
from azure.storage.blob import BlobService
blobService = BlobService(account_key=_fetchAzureAccountKey(self.accountName),
account_name=self.accountName)
containerName = 'import-export-test-%s' % uuid.uuid4()
blobService.create_container(containerName)
return containerName
class AzureStorage(Storage):
account_name = settings.AZURE_ACCOUNT_NAME
account_key = settings.AZURE_ACCOUNT_KEY
azure_container = settings.AZURE_CONTAINER
def __init__(self, *args, **kwargs):
super(AzureStorage, self).__init__(*args, **kwargs)
self._connection = None
@property
def connection(self):
if self._connection is None:
# Create connection
self._connection = BlobService(self.account_name, self.account_key)
# Create container if needed
containers = [c for c in self._connection.list_containers(prefix=self.azure_container) if c.name == self.azure_container ]
if len(containers) == 0:
self._connection.create_container(self.azure_container, {'origin': 'created by Django web app'}, fail_on_exist=True)
return self._connection
def _open(self, name, mode="rb"):
stream = SimpleUploadedFile(name, None)
self.connection.get_blob_to_file(self.azure_container, name, stream)
stream.seek(0)
return stream
def exists(self, name):
try:
self.connection.get_blob_properties(self.azure_container, name)
except AzureMissingResourceHttpError:
return False
else:
return True
def delete(self, name):
self.connection.delete_blob(self.azure_container, name)
def size(self, name):
properties = self.connection.get_blob_properties(self.azure_container, name)
return properties["content-length"]
def _save(self, name, content):
self.connection.put_block_blob_from_file(self.azure_container, name, content)
return name
def url(self, name):
ap = AccessPolicy(expiry=(timezone.datetime.utcnow() + timezone.timedelta(seconds=600)).strftime('%Y-%m-%dT%H:%M:%SZ'), \
start=(timezone.datetime.utcnow() + timezone.timedelta(seconds=-600)).strftime('%Y-%m-%dT%H:%M:%SZ'), \
permission='r')
sap = SharedAccessPolicy(ap)
sas = SharedAccessSignature(self.account_name, self.account_key)
url = sas.generate_signed_query_string(path=self.azure_container + '/' + name, resource_type='b', shared_access_policy=sap)
return self.connection.make_blob_url(self.azure_container, name) + "?" + url
def uploadFile(sourceFile, destUrl, destKey):
storageparts = split_storage_url(destUrl)
blobservice = BlobService(storageparts[0], destKey)
try:
fh=open(sourceFile, "r")
except:
print "No such file", sourceFile
return
log('uploading ' + str(sourceFile), True)
blobservice.put_page_blob_from_file(storageparts[2], sourceFile, fh, getsize(sourceFile))
def generateImageUrl(request):
account_name = "faceemoji"
account_key = "kaoJiy0T7r6sXyo4wFYKCLgpAXbILKvkloeF+kFpCEUxC+bL9BxGA3WtofVxHcLPn3lMjw/UO/0sS1GCN3/AQw=="
blob_service = BlobService(account_name, account_key)
content = base64.b64decode(request.data)
st = datetime.datetime.fromtimestamp(time.time()).strftime('%Y-%m-%d-%H-%M-%S')
blob_name = hashlib.sha224(st).hexdigest() + 'image.png'
blob_service.put_block_blob_from_bytes('image', blob_name, content)
img_url = blob_service.make_blob_url('image', blob_name)
return img_url
def getblob(request):
assert isinstance(request, HttpRequest)
blob_service = BlobService(account_name='araldrift', account_key='otLzzkwQHQD3xFTQxwxy64PCL6eDINWGjSB7x6Ta2XVw3+3ffI5O2MhAEavf/r8qIW4G/dKrZAVg1R64nK7hDQ==')
# http://<storage-account-name>.blob.core.windows.net/<container-name>/<blob-name>
name = 'test.txt'
fpath = '{0}\{1}'.format(tempfile.gettempdir(),name)
blob_service.get_blob_to_path('flow', 'NARYN.day', fpath)
response = HttpResponse(content_type='text/plain')
response['Content-Disposition'] = 'attachment; filename=test.txt'
blob.Properties.ContentDisposition = "attachment; filename=" + downloadName;
return response
def upload_file_to_azure(in_file, file_name, container_name=settings.AZURE_CONTAINER):
try:
blob_service = BlobService(AZURE_ACCOUNT_NAME, AZURE_ACCOUNT_KEY)
blob_service.put_block_blob_from_path(
container_name=container_name,
blob_name=file_name,
file_path=in_file,
x_ms_blob_content_type='application/octet-stream'
)
except Exception as ex:
print("Failed to upload blob: {0}".format(ex))
def _getUrlForTestFile(cls, size=None):
from toil.jobStores.azureJobStore import _fetchAzureAccountKey
fileName = 'testfile_%s' % uuid.uuid4()
containerName = cls._externalStore()
url = 'wasb://%s@%s.blob.core.windows.net/%s' % (containerName, cls.accountName, fileName)
if size is None:
return url
blobService = BlobService(account_key=_fetchAzureAccountKey(cls.accountName),
account_name=cls.accountName)
content = os.urandom(size)
blobService.put_block_blob_from_text(containerName, fileName, content)
return url, hashlib.md5(content).hexdigest()
class azure_storage_writer (object):
"""storage operation wrapper, desiged for writing logs to storage"""
def __init__(self, account_name, account_key, container, prefix):
self._blob = BlobService(account_name=account_name, account_key=account_key)
self._cur_path = None
self._buf = io.StringIO()
self._prefix = prefix
self._container = container
self._blob.create_container(container)
self._logger = create_timed_rotating_log()
def write_log(self, entity):
path = self._get_path(entity[0])
if (self._cur_path == None):
self._cur_path = path
elif(self._cur_path != path):
self._dump_buf_to_storage()
self._buf.close()
self._buf = io.StringIO()
self._cur_path = path
self._buf.write(entity[1])
self._buf.write("\n")
def close(self):
if (self._cur_path != None):
self._dump_buf_to_storage()
self._buf.close()
def _dump_buf_to_storage(self):
self._logger.info("Begin dump to azure blob")
loop = 0;
while True:
try:
self._blob.put_block_blob_from_text(self._container,self._cur_path, self._buf.getvalue())
break
except AzureHttpError as e:
self._logger.warn("Hit an AzureHttpError " + str(e))
self._logger.warn("Retry times: {0}".format(loop))
loop = loop + 1
if loop >= 3:
raise e
except Exception as e:
self._logger.warn("Hit an Exception " + str(e))
raise e
self._logger.info("Dump to azure blob succeeded.")
def _get_path(self, timestamp):
#timestamp = int(timestamp)
d = datetime.fromtimestamp(int(timestamp))
part = str.format("logs-part-{}.txt", d.minute // 5)
path_str = d.strftime('%Y-%m-%d/%H')
return str.format("{}/{}/{}", self._prefix, path_str, part)
class _BlobStorageFileHandler(object):
def __init__(self,
account_name=None,
account_key=None,
protocol='https',
container='logs',
zip_compression=False,
max_connections=1,
max_retries=5,
retry_wait=1.0):
self.service = BlobService(account_name, account_key, protocol)
self.container_created = False
hostname = gethostname()
self.meta = {'hostname': hostname.replace('_', '-'),
'process': os.getpid()}
self.container = (container % self.meta).lower()
self.meta['hostname'] = hostname
self.zip_compression = zip_compression
self.max_connections = max_connections
self.max_retries = max_retries
self.retry_wait = retry_wait
def put_file_into_storage(self, dirName, fileName):
"""
Ship the outdated log file to the specified blob container.
"""
if not self.container_created:
self.service.create_container(self.container)
self.container_created = True
fd, tmpfile_path = None, ''
try:
file_path = os.path.join(dirName, fileName)
if self.zip_compression:
suffix, content_type = '.zip', 'application/zip'
fd, tmpfile_path = mkstemp(suffix=suffix)
with os.fdopen(fd, 'wb') as f:
with ZipFile(f, 'w', ZIP_DEFLATED) as z:
z.write(file_path, arcname=fileName)
file_path = tmpfile_path
else:
suffix, content_type = '', 'text/plain'
self.service.put_block_blob_from_path(self.container,
fileName + suffix,
file_path,
x_ms_blob_content_type=content_type,
max_connections=self.max_connections,
max_retries=self.max_retries,
retry_wait=self.retry_wait)
finally:
if self.zip_compression and fd:
os.remove(tmpfile_path)
def main():
#continue here
if verbose:
print('Copying')
print('Storage account:',storage_account_name)
print('Storage Key:',storage_account_key)
print('Container Name:',storage_container_name)
print('Input file:',inputfile)
print('Output Blob:',outputblob)
blob_service = BlobService(account_name=storage_account_name, account_key = storage_account_key)
#blob_service.create_container(storage_container_name,x_ms_blob_public_access='containter')
blob_service.put_block_blob_from_path(storage_container_name, outputblob,inputfile,x_ms_blob_content_type="image/jpeg")
def pushToAzureCDN (data): import pickle from azure.storage.blob import BlobService blob_service = BlobService(account_name=azureAccount, account_key=azureAccountKey) blob_service.put_block_blob_from_bytes( azureContainer, azureFile, pickle.dumps(data), content_encoding='application/octet-stream' )
def _prepareTestFile(self, containerName, size=None):
from toil.jobStores.azureJobStore import _fetchAzureAccountKey
from azure.storage.blob import BlobService
fileName = 'testfile_%s' % uuid.uuid4()
url = 'wasb://%s@%s.blob.core.windows.net/%s' % (containerName, self.accountName, fileName)
if size is None:
return url
blobService = BlobService(account_key=_fetchAzureAccountKey(self.accountName),
account_name=self.accountName)
content = os.urandom(size)
blobService.put_block_blob_from_text(containerName, fileName, content)
return url, hashlib.md5(content).hexdigest()
def prepare_storage(settings):
default_storage_account_name = settings["DEFAULT_STORAGE_ACCOUNT_NAME"]
storage_access_key = settings["STORAGE_ACCESS_KEY"]
blob_service = BlobService(default_storage_account_name, storage_access_key)
blob_service.create_container('bosh')
blob_service.create_container(
container_name='stemcell',
x_ms_blob_public_access='blob'
)
# Prepare the table for storing meta datas of storage account and stemcells
table_service = TableService(default_storage_account_name, storage_access_key)
table_service.create_table('stemcells')
def upload_azure_blob(account, account_key, container, filename, file, file_type='file/csv'):
block_blob_service = BlobService(account_name=account, account_key=account_key)
# block_blob_service.put_block_blob_from_path(
# container,
# blockblob,
# file,
# x_ms_blob_content_type='file/csv'
# )
block_blob_service.create_blob_from_stream(container, filename, file)
generator = block_blob_service.list_blobs(container)
for blob in generator:
print(blob.name)
def upload_results():
"""
:return: None
"""
logger = logging.getLogger(__name__)
results_fpath = '/data/wsdm_cup/results/results.tsv'
logger.info('Uploading results from {0}'.format(results_fpath))
blob_service = BlobService(account_name='wsdmcupchallenge',
sas_token=Config.SAS_TOKEN)
blob_service.put_block_blob_from_path(container_name='bletchleypark',
blob_name='results.tsv',
file_path=results_fpath)
logger.info('Done uploading')
return
def uri_get_file(creds, uri, conn=None):
assert uri.startswith('wabs://')
url_tup = urlparse(uri)
if conn is None:
conn = BlobService(creds.account_name, creds.account_key,
sas_token=creds.access_token, protocol='https')
# Determin the size of the target blob
props = conn.get_blob_properties(url_tup.netloc, url_tup.path.lstrip('/'))
blob_size = int(props['content-length'])
ret_size = 0
data = io.BytesIO()
# WABS requires large files to be downloaded in 4MB chunks
while ret_size < blob_size:
ms_range = 'bytes={0}-{1}'.format(ret_size,
ret_size + WABS_CHUNK_SIZE - 1)
while True:
# Because we're downloading in chunks, catch rate limiting and
# connection errors here instead of letting them bubble up to the
# @retry decorator so that we don't have to start downloading the
# whole file over again.
try:
part = conn.get_blob(url_tup.netloc,
url_tup.path.lstrip('/'),
x_ms_range=ms_range)
except EnvironmentError as e:
if e.errno in (errno.EBUSY, errno.ECONNRESET):
logger.warning(
msg="retrying after encountering exception",
detail=("Exception traceback:\n{0}".format(
traceback.format_exception(*sys.exc_info()))),
hint="")
gevent.sleep(30)
else:
raise
else:
break
length = len(part)
ret_size += length
data.write(part)
if length > 0 and length < WABS_CHUNK_SIZE:
break
elif length == 0:
break
return data.getvalue()
def download_data(key):
blob_service = BlobService(account_name='asosdsrecruiting', account_key=key)
blobs = []
marker = None
while True:
batch = blob_service.list_blobs('recruitingdata', marker=marker)
blobs.extend(batch)
if not batch.next_marker:
break
marker = batch.next_marker
for blob in blobs:
file_name = blob.name
print('Downloading: ' + file_name)
blob_service.get_blob_to_path('recruitingdata', file_name, file_name.replace('/', '_') + '.csv')
def __init__(self, accountName, namePrefix, config=None,
jobChunkSize=maxAzureTablePropertySize):
self.jobChunkSize = jobChunkSize
self.keyPath = None
self.account_key = _fetchAzureAccountKey(accountName)
self.accountName = accountName
# Table names have strict requirements in Azure
self.namePrefix = self._sanitizeTableName(namePrefix)
logger.debug("Creating job store with name prefix '%s'" % self.namePrefix)
# These are the main API entrypoints.
self.tableService = TableService(account_key=self.account_key, account_name=accountName)
self.blobService = BlobService(account_key=self.account_key, account_name=accountName)
exists = self._jobStoreExists()
self._checkJobStoreCreation(config is not None, exists, accountName + ":" + self.namePrefix)
# Serialized jobs table
self.jobItems = self._getOrCreateTable(self.qualify('jobs'))
# Job<->file mapping table
self.jobFileIDs = self._getOrCreateTable(self.qualify('jobFileIDs'))
# Container for all shared and unshared files
self.files = self._getOrCreateBlobContainer(self.qualify('files'))
# Stats and logging strings
self.statsFiles = self._getOrCreateBlobContainer(self.qualify('statsfiles'))
# File IDs that contain stats and logging strings
self.statsFileIDs = self._getOrCreateTable(self.qualify('statsFileIDs'))
super(AzureJobStore, self).__init__(config=config)
if self.config.cseKey is not None:
self.keyPath = self.config.cseKey
def save_image_to_azure(profile, url):
try:
response = request('GET', url)
response.raise_for_status()
except ConnectionError:
pass
else:
service = BlobService(
account_name=storagesettings.AZURE_ACCOUNT_NAME,
account_key=storagesettings.AZURE_ACCOUNT_KEY)
service.put_block_blob_from_bytes(
'avatars',
profile.id,
response.content,
x_ms_blob_content_type=response.headers['content-type']
)
def connect(config=False):
import lib.misc as misc
from azure.storage.blob import BlobService
global blob_service, container
# Connect to the cloud service.
if not config: config = misc.config['_private']
container = 'streams'
if not 'azure' in config:
return None, None
if not blob_service:
blob_service = BlobService(config['azure']['storage_account_name'], config['azure']['primary_access_key'])
blob_service.create_container(container, x_ms_blob_public_access='container')
return blob_service, container
def __init__(self, account_name, account_key, container, prefix):
self._blob = BlobService(account_name=account_name, account_key=account_key)
self._cur_path = None
self._buf = io.StringIO()
self._prefix = prefix
self._container = container
self._blob.create_container(container)
self._logger = create_timed_rotating_log()
class AzureJobStore(AbstractJobStore):
"""
A job store that uses Azure's blob store for file storage and
Table Service to store job info with strong consistency."""
@classmethod
def loadOrCreateJobStore(cls, locator, config=None, **kwargs):
account, namePrefix = locator.split(':', 1)
if '--' in namePrefix:
raise ValueError(
"Invalid name prefix '%s'. Name prefixes may not contain "
"%s." % (namePrefix, cls.nameSeparator))
if not cls.containerNameRe.match(namePrefix):
raise ValueError(
"Invalid name prefix '%s'. Name prefixes must contain only digits, "
"hyphens or lower-case letters and must not start or end in a "
"hyphen." % namePrefix)
# Reserve 13 for separator and suffix
if len(namePrefix) > cls.maxContainerNameLen - cls.maxNameLen - len(
cls.nameSeparator):
raise ValueError((
"Invalid name prefix '%s'. Name prefixes may not be longer than 50 "
"characters." % namePrefix))
if '--' in namePrefix:
raise ValueError(
"Invalid name prefix '%s'. Name prefixes may not contain "
"%s." % (namePrefix, cls.nameSeparator))
return cls(account, namePrefix, config=config, **kwargs)
# Dots in container names should be avoided because container names are used in HTTPS bucket
# URLs where the may interfere with the certificate common name. We use a double
# underscore as a separator instead.
#
containerNameRe = re.compile(r'^[a-z0-9](-?[a-z0-9]+)+[a-z0-9]$')
# See https://msdn.microsoft.com/en-us/library/azure/dd135715.aspx
#
minContainerNameLen = 3
maxContainerNameLen = 63
maxNameLen = 10
nameSeparator = 'xx' # Table names must be alphanumeric
# Do not invoke the constructor, use the factory method above.
def __init__(self,
accountName,
namePrefix,
config=None,
jobChunkSize=maxAzureTablePropertySize):
self.jobChunkSize = jobChunkSize
self.keyPath = None
self.account_key = _fetchAzureAccountKey(accountName)
self.accountName = accountName
# Table names have strict requirements in Azure
self.namePrefix = self._sanitizeTableName(namePrefix)
logger.debug("Creating job store with name prefix '%s'" %
self.namePrefix)
# These are the main API entrypoints.
self.tableService = TableService(account_key=self.account_key,
account_name=accountName)
self.blobService = BlobService(account_key=self.account_key,
account_name=accountName)
exists = self._jobStoreExists()
self._checkJobStoreCreation(config is not None, exists,
accountName + ":" + self.namePrefix)
# Serialized jobs table
self.jobItems = self._getOrCreateTable(self.qualify('jobs'))
# Job<->file mapping table
self.jobFileIDs = self._getOrCreateTable(self.qualify('jobFileIDs'))
# Container for all shared and unshared files
self.files = self._getOrCreateBlobContainer(self.qualify('files'))
# Stats and logging strings
self.statsFiles = self._getOrCreateBlobContainer(
self.qualify('statsfiles'))
# File IDs that contain stats and logging strings
self.statsFileIDs = self._getOrCreateTable(
self.qualify('statsFileIDs'))
super(AzureJobStore, self).__init__(config=config)
if self.config.cseKey is not None:
self.keyPath = self.config.cseKey
# Length of a jobID - used to test if a stats file has been read already or not
jobIDLength = len(str(uuid.uuid4()))
def qualify(self, name):
return self.namePrefix + self.nameSeparator + name
def jobs(self):
# How many jobs have we done?
total_processed = 0
for jobEntity in self.jobItems.query_entities_auto():
# Process the items in the page
yield AzureJob.fromEntity(jobEntity)
total_processed += 1
if total_processed % 1000 == 0:
# Produce some feedback for the user, because this can take
# a long time on, for example, Azure
logger.info("Processed %d total jobs" % total_processed)
logger.info("Processed %d total jobs" % total_processed)
def create(self,
command,
memory,
cores,
disk,
preemptable,
predecessorNumber=0):
jobStoreID = self._newJobID()
job = AzureJob(jobStoreID=jobStoreID,
command=command,
memory=memory,
cores=cores,
disk=disk,
preemptable=preemptable,
remainingRetryCount=self._defaultTryCount(),
logJobStoreFileID=None,
predecessorNumber=predecessorNumber)
entity = job.toItem(chunkSize=self.jobChunkSize)
entity['RowKey'] = jobStoreID
self.jobItems.insert_entity(entity=entity)
return job
def exists(self, jobStoreID):
if self.jobItems.get_entity(row_key=jobStoreID) is None:
return False
return True
def load(self, jobStoreID):
jobEntity = self.jobItems.get_entity(row_key=jobStoreID)
if jobEntity is None:
raise NoSuchJobException(jobStoreID)
return AzureJob.fromEntity(jobEntity)
def update(self, job):
self.jobItems.update_entity(
row_key=job.jobStoreID,
entity=job.toItem(chunkSize=self.jobChunkSize))
def delete(self, jobStoreID):
try:
self.jobItems.delete_entity(row_key=jobStoreID)
except AzureMissingResourceHttpError:
# Job deletion is idempotent, and this job has been deleted already
return
filterString = "PartitionKey eq '%s'" % jobStoreID
for fileEntity in self.jobFileIDs.query_entities(filter=filterString):
jobStoreFileID = fileEntity.RowKey
self.deleteFile(jobStoreFileID)
def deleteJobStore(self):
self.jobItems.delete_table()
self.jobFileIDs.delete_table()
self.files.delete_container()
self.statsFiles.delete_container()
self.statsFileIDs.delete_table()
def _jobStoreExists(self):
"""
Checks if job store exists by querying the existence of the statsFileIDs table. Note that
this is the last component that is deleted in deleteJobStore.
"""
for attempt in retry_azure():
with attempt:
try:
table = self.tableService.query_tables(
table_name=self.qualify('statsFileIDs'))
return table is not None
except AzureMissingResourceHttpError as e:
if e.status_code == 404:
return False
else:
raise
def getEnv(self):
return dict(AZURE_ACCOUNT_KEY=self.account_key)
@classmethod
def _readFromUrl(cls, url, writable):
blobService, containerName, blobName = cls._extractBlobInfoFromUrl(url)
blobService.get_blob_to_file(containerName, blobName, writable)
@classmethod
def _writeToUrl(cls, readable, url):
blobService, containerName, blobName = cls._extractBlobInfoFromUrl(url)
blobService.put_block_blob_from_file(containerName, blobName, readable)
blobService.get_blob(containerName, blobName)
@staticmethod
def _extractBlobInfoFromUrl(url):
"""
:return: (blobService, containerName, blobName)
"""
def invalidUrl():
raise RuntimeError("The URL '%s' is invalid" % url.geturl())
netloc = url.netloc.split('@')
if len(netloc) != 2:
invalidUrl()
accountEnd = netloc[1].find('.blob.core.windows.net')
if accountEnd == -1:
invalidUrl()
containerName, accountName = netloc[0], netloc[1][0:accountEnd]
blobName = url.path[1:] # urlparse always includes a leading '/'
blobService = BlobService(
account_key=_fetchAzureAccountKey(accountName),
account_name=accountName)
return blobService, containerName, blobName
@classmethod
def _supportsUrl(cls, url, export=False):
return url.scheme.lower() == 'wasb' or url.scheme.lower() == 'wasbs'
def writeFile(self, localFilePath, jobStoreID=None):
jobStoreFileID = self._newFileID()
self.updateFile(jobStoreFileID, localFilePath)
self._associateFileWithJob(jobStoreFileID, jobStoreID)
return jobStoreFileID
def updateFile(self, jobStoreFileID, localFilePath):
with open(localFilePath) as read_fd:
with self._uploadStream(jobStoreFileID, self.files) as write_fd:
while True:
buf = read_fd.read(self._maxAzureBlockBytes)
write_fd.write(buf)
if len(buf) == 0:
break
def readFile(self, jobStoreFileID, localFilePath):
try:
with self._downloadStream(jobStoreFileID, self.files) as read_fd:
with open(localFilePath, 'w') as write_fd:
while True:
buf = read_fd.read(self._maxAzureBlockBytes)
write_fd.write(buf)
if not buf: break
except AzureMissingResourceHttpError:
raise NoSuchFileException(jobStoreFileID)
def deleteFile(self, jobStoreFileID):
try:
self.files.delete_blob(blob_name=jobStoreFileID)
self._dissociateFileFromJob(jobStoreFileID)
except AzureMissingResourceHttpError:
pass
def fileExists(self, jobStoreFileID):
# As Azure doesn't have a blob_exists method (at least in the
# python API) we just try to download the metadata, and hope
# the metadata is small so the call will be fast.
try:
self.files.get_blob_metadata(blob_name=jobStoreFileID)
return True
except AzureMissingResourceHttpError:
return False
@contextmanager
def writeFileStream(self, jobStoreID=None):
# TODO: this (and all stream methods) should probably use the
# Append Blob type, but that is not currently supported by the
# Azure Python API.
jobStoreFileID = self._newFileID()
with self._uploadStream(jobStoreFileID, self.files) as fd:
yield fd, jobStoreFileID
self._associateFileWithJob(jobStoreFileID, jobStoreID)
@contextmanager
def updateFileStream(self, jobStoreFileID):
with self._uploadStream(jobStoreFileID,
self.files,
checkForModification=True) as fd:
yield fd
def getEmptyFileStoreID(self, jobStoreID=None):
jobStoreFileID = self._newFileID()
self.files.put_blob(blob_name=jobStoreFileID,
blob='',
x_ms_blob_type='BlockBlob')
self._associateFileWithJob(jobStoreFileID, jobStoreID)
return jobStoreFileID
@contextmanager
def readFileStream(self, jobStoreFileID):
if not self.fileExists(jobStoreFileID):
raise NoSuchFileException(jobStoreFileID)
with self._downloadStream(jobStoreFileID, self.files) as fd:
yield fd
@contextmanager
def writeSharedFileStream(self, sharedFileName, isProtected=None):
assert self._validateSharedFileName(sharedFileName)
sharedFileID = self._newFileID(sharedFileName)
with self._uploadStream(sharedFileID,
self.files,
encrypted=isProtected) as fd:
yield fd
@contextmanager
def readSharedFileStream(self, sharedFileName):
assert self._validateSharedFileName(sharedFileName)
sharedFileID = self._newFileID(sharedFileName)
if not self.fileExists(sharedFileID):
raise NoSuchFileException(sharedFileID)
with self._downloadStream(sharedFileID, self.files) as fd:
yield fd
def writeStatsAndLogging(self, statsAndLoggingString):
# TODO: would be a great use case for the append blobs, once implemented in the Azure SDK
jobStoreFileID = self._newFileID()
encrypted = self.keyPath is not None
if encrypted:
statsAndLoggingString = encryption.encrypt(statsAndLoggingString,
self.keyPath)
self.statsFiles.put_block_blob_from_text(
blob_name=jobStoreFileID,
text=statsAndLoggingString,
x_ms_meta_name_values=dict(encrypted=str(encrypted)))
self.statsFileIDs.insert_entity(entity={'RowKey': jobStoreFileID})
def readStatsAndLogging(self, callback, readAll=False):
suffix = '_old'
numStatsFiles = 0
for entity in self.statsFileIDs.query_entities():
jobStoreFileID = entity.RowKey
hasBeenRead = len(jobStoreFileID) > self.jobIDLength
if not hasBeenRead:
with self._downloadStream(jobStoreFileID,
self.statsFiles) as fd:
callback(fd)
# Mark this entity as read by appending the suffix
self.statsFileIDs.insert_entity(
entity={'RowKey': jobStoreFileID + suffix})
self.statsFileIDs.delete_entity(row_key=jobStoreFileID)
numStatsFiles += 1
elif readAll:
# Strip the suffix to get the original ID
jobStoreFileID = jobStoreFileID[:-len(suffix)]
with self._downloadStream(jobStoreFileID,
self.statsFiles) as fd:
callback(fd)
numStatsFiles += 1
return numStatsFiles
_azureTimeFormat = "%Y-%m-%dT%H:%M:%SZ"
def getPublicUrl(self, jobStoreFileID):
try:
self.files.get_blob_properties(blob_name=jobStoreFileID)
except AzureMissingResourceHttpError:
raise NoSuchFileException(jobStoreFileID)
# Compensate of a little bit of clock skew
startTimeStr = (datetime.utcnow() - timedelta(minutes=5)).strftime(
self._azureTimeFormat)
endTime = datetime.utcnow() + self.publicUrlExpiration
endTimeStr = endTime.strftime(self._azureTimeFormat)
sap = SharedAccessPolicy(
AccessPolicy(startTimeStr, endTimeStr,
BlobSharedAccessPermissions.READ))
sas_token = self.files.generate_shared_access_signature(
blob_name=jobStoreFileID, shared_access_policy=sap)
return self.files.make_blob_url(
blob_name=jobStoreFileID) + '?' + sas_token
def getSharedPublicUrl(self, sharedFileName):
jobStoreFileID = self._newFileID(sharedFileName)
return self.getPublicUrl(jobStoreFileID)
def _newJobID(self):
# raw UUIDs don't work for Azure property names because the '-' character is disallowed.
return str(uuid.uuid4()).replace('-', '_')
# A dummy job ID under which all shared files are stored.
sharedFileJobID = uuid.UUID('891f7db6-e4d9-4221-a58e-ab6cc4395f94')
def _newFileID(self, sharedFileName=None):
if sharedFileName is None:
ret = str(uuid.uuid4())
else:
ret = str(uuid.uuid5(self.sharedFileJobID, str(sharedFileName)))
return ret.replace('-', '_')
def _associateFileWithJob(self, jobStoreFileID, jobStoreID=None):
if jobStoreID is not None:
self.jobFileIDs.insert_entity(entity={
'PartitionKey': jobStoreID,
'RowKey': jobStoreFileID
})
def _dissociateFileFromJob(self, jobStoreFileID):
entities = self.jobFileIDs.query_entities(filter="RowKey eq '%s'" %
jobStoreFileID)
if entities:
assert len(entities) == 1
jobStoreID = entities[0].PartitionKey
self.jobFileIDs.delete_entity(partition_key=jobStoreID,
row_key=jobStoreFileID)
def _getOrCreateTable(self, tableName):
# This will not fail if the table already exists.
for attempt in retry_azure():
with attempt:
self.tableService.create_table(tableName)
return AzureTable(self.tableService, tableName)
def _getOrCreateBlobContainer(self, containerName):
for attempt in retry_azure():
with attempt:
self.blobService.create_container(containerName)
return AzureBlobContainer(self.blobService, containerName)
def _sanitizeTableName(self, tableName):
"""
Azure table names must start with a letter and be alphanumeric.
This will never cause a collision if uuids are used, but
otherwise may not be safe.
"""
return 'a' + filter(lambda x: x.isalnum(), tableName)
# Maximum bytes that can be in any block of an Azure block blob
# https://github.com/Azure/azure-storage-python/blob/4c7666e05a9556c10154508335738ee44d7cb104/azure/storage/blob/blobservice.py#L106
_maxAzureBlockBytes = 4 * 1024 * 1024
@contextmanager
def _uploadStream(self,
jobStoreFileID,
container,
checkForModification=False,
encrypted=None):
"""
:param encrypted: True to enforce encryption (will raise exception unless key is set),
False to prevent encryption or None to encrypt if key is set.
"""
if checkForModification:
try:
expectedVersion = container.get_blob_properties(
blob_name=jobStoreFileID)['etag']
except AzureMissingResourceHttpError:
expectedVersion = None
if encrypted is None:
encrypted = self.keyPath is not None
elif encrypted:
if self.keyPath is None:
raise RuntimeError(
'Encryption requested but no key was provided')
maxBlockSize = self._maxAzureBlockBytes
if encrypted:
# There is a small overhead for encrypted data.
maxBlockSize -= encryption.overhead
readable_fh, writable_fh = os.pipe()
with os.fdopen(readable_fh, 'r') as readable:
with os.fdopen(writable_fh, 'w') as writable:
def reader():
blockIDs = []
try:
while True:
buf = readable.read(maxBlockSize)
if len(buf) == 0:
# We're safe to break here even if we never read anything, since
# putting an empty block list creates an empty blob.
break
if encrypted:
buf = encryption.encrypt(buf, self.keyPath)
blockID = self._newFileID()
container.put_block(blob_name=jobStoreFileID,
block=buf,
blockid=blockID)
blockIDs.append(blockID)
except:
# This is guaranteed to delete any uncommitted
# blocks.
container.delete_blob(blob_name=jobStoreFileID)
raise
if checkForModification and expectedVersion is not None:
# Acquire a (60-second) write lock,
leaseID = container.lease_blob(
blob_name=jobStoreFileID,
x_ms_lease_action='acquire')['x-ms-lease-id']
# check for modification,
blobProperties = container.get_blob_properties(
blob_name=jobStoreFileID)
if blobProperties['etag'] != expectedVersion:
container.lease_blob(blob_name=jobStoreFileID,
x_ms_lease_action='release',
x_ms_lease_id=leaseID)
raise ConcurrentFileModificationException(
jobStoreFileID)
# commit the file,
container.put_block_list(blob_name=jobStoreFileID,
block_list=blockIDs,
x_ms_lease_id=leaseID,
x_ms_meta_name_values=dict(
encrypted=str(encrypted)))
# then release the lock.
container.lease_blob(blob_name=jobStoreFileID,
x_ms_lease_action='release',
x_ms_lease_id=leaseID)
else:
# No need to check for modification, just blindly write over whatever
# was there.
container.put_block_list(blob_name=jobStoreFileID,
block_list=blockIDs,
x_ms_meta_name_values=dict(
encrypted=str(encrypted)))
thread = ExceptionalThread(target=reader)
thread.start()
yield writable
# The writable is now closed. This will send EOF to the readable and cause that
# thread to finish.
thread.join()
@contextmanager
def _downloadStream(self, jobStoreFileID, container):
# The reason this is not in the writer is so we catch non-existant blobs early
blobProps = container.get_blob_properties(blob_name=jobStoreFileID)
encrypted = strict_bool(blobProps['x-ms-meta-encrypted'])
if encrypted and self.keyPath is None:
raise AssertionError(
'Content is encrypted but no key was provided.')
readable_fh, writable_fh = os.pipe()
with os.fdopen(readable_fh, 'r') as readable:
with os.fdopen(writable_fh, 'w') as writable:
def writer():
try:
chunkStartPos = 0
fileSize = int(blobProps['Content-Length'])
while chunkStartPos < fileSize:
chunkEndPos = chunkStartPos + self._maxAzureBlockBytes - 1
buf = container.get_blob(
blob_name=jobStoreFileID,
x_ms_range="bytes=%d-%d" %
(chunkStartPos, chunkEndPos))
if encrypted:
buf = encryption.decrypt(buf, self.keyPath)
writable.write(buf)
chunkStartPos = chunkEndPos + 1
finally:
# Ensure readers aren't left blocking if this thread crashes.
# This close() will send EOF to the reading end and ultimately cause the
# yield to return. It also makes the implict .close() done by the enclosing
# "with" context redundant but that should be ok since .close() on file
# objects are idempotent.
writable.close()
thread = ExceptionalThread(target=writer)
thread.start()
yield readable
thread.join()
def connection(self):
if self._connection is None:
self._connection = BlobService(self.account_name, self.account_key)
return self._connection
def gethydrograph(request):
'''
Returns streamflow data by start / stop / station
In response it will
generate a 404 error if the value is not found
or
return a JSON response with the requested slice or a .csv file by default
'''
assert isinstance(request, HttpRequest)
start = request.GET.get('start', None)
end = request.GET.get('end', None)
station = request.GET.get('station', None)
interval = request.GET.get('interval', None)
jsondat = request.GET.get('jsondat', None)
plot = request.GET.get('plot', None)
#start blob service
stationfile = station + '.day.new'
downloadablefile = station + '_' + start + '_' + end + '.csv'
blob_service = BlobService(
account_name='araldrift',
account_key=
'otLzzkwQHQD3xFTQxwxy64PCL6eDINWGjSB7x6Ta2XVw3+3ffI5O2MhAEavf/r8qIW4G/dKrZAVg1R64nK7hDQ=='
)
blob_service.get_blob_to_path('flow', stationfile, './tmp.csv')
f = file('./tmp.csv')
#read in pandas data and subsetting
d_cols = ["DATE", "FLOW"]
d = pd.read_csv('./tmp.csv', sep=" ", names=d_cols)
df = d[(d.DATE >= start) & (d.DATE <= end)]
h = df.to_json(orient='records')
json_encoded_result = json.dumps(h)
df.plot(x='DATE', y='FLOW', figsize=(14, 6))
plt.savefig('./plot_test.png')
#h = []
#while True:
# line = f.readline()
# if line == "": break
# h.append(line)
#f.close()
try:
if jsondat in ['TRUE']:
response = HttpResponse(json_encoded_result,
content_type="application/json")
return response
elif plot in ['TRUE']:
image_data = open("./plot_test.png", "rb").read()
response = HttpResponse(image_data, content_type='image/png')
return response
else:
response = HttpResponse(content_type='text/csv')
response[
'Content-Disposition'] = 'attachment; filename=' + downloadablefile
df.to_csv(response, index=False, lineterminator='\r\n')
return response
except Exception as a:
return HttpResponseNotFound(
content=
"No dice, either the inputs were out of range, the file couldn't be retrieved, or the winds weren't in your favor."
)
class AzureConnector():
def __init__(self, config):
tree = ET.parse('SharedConfig.xml')
self.myMachineName = tree.find('.//Instance').get("id")
self.sms = ServiceManagementService(
subscription_id=config.get("azure", "subscription_id"),
cert_file=config.get("azure", "cert_file")
);
self.bus_service = ServiceBusService(
service_namespace=config.get("azure", "bus_namespace"),
shared_access_key_name=config.get("azure", "bus_shared_access_key_name"),
shared_access_key_value=config.get("azure", "bus_shared_access_key_value"))
self.command_queue = config.get("azure", "commandQueuePath")
for tries in range(1,10):
try:
self.bus_service.create_queue(self.command_queue)
break
except:
print "Esperando"
self.status_topic = config.get("azure", "statusTopicPath")
self.bus_service.create_queue(self.status_topic)
self.storage = BlobService(account_name=config.get("azure", "account_name"),
account_key=config.get("azure", "account_key"))
self.algo_storage_name = config.get("azure", "algorithm_storage_name")
self.storage.create_container(self.algo_storage_name, fail_on_exist=False)
self.proj_storage_name = config.get("azure", "project_storage_name")
self.storage.create_container(self.proj_storage_name, fail_on_exist=False)
def check_new_tasks(self):
for tries in range(1,2):
try:
message = self.bus_service.receive_queue_message(self.command_queue, peek_lock=False, timeout=60)
break
except:
message = None
if message is None or message.body is None:
return None
job_description = json.loads(message.body.replace('/AzureBlobStorage/', ''))
command = CommandMetadata(
command_id = job_description["command_id"],
algorithm_directory = job_description["algorithm_prfx"],
project_prfx = job_description["project_prfx"],
project_input_files = job_description["project_input_files"],
algorithm_executable_name = job_description["algorithm_executable_name"],
algorithm_parameters = job_description["algorithm_parameters"],
sent_timestamp = datetime.datetime.strptime(job_description["sent_timestamp"], "%d/%m/%Y %H:%M:%S"),
machine_size=job_description["machine_size"])
# Retornar dados sobre o comando consumido da fila
return command
# Não há nada na fila
return None
def list_algo_files(self, prfx):
list = self.storage.list_blobs(container_name=self.algo_storage_name, prefix=prfx)
result = []
for blob in list:
result.append(blob.name)
return result
def download_algo_zip(self, algorithm_bin_file, tmp_file):
print "download_algo_zip(algorithm_bin_file="+algorithm_bin_file+", tmp_file="+tmp_file+")"
for tries in range(1,5):
try:
self.storage.get_blob_to_path(self.algo_storage_name, algorithm_bin_file, tmp_file,
open_mode='wb', snapshot=None, x_ms_lease_id=None,
progress_callback=None)
break
except Exception as e:
if tries == 5:
print("Muitos erros de conexão. Operação abortada.")
else:
print("Erro de conexão com serviço. Retentando..." + e.__str__())
def download_file_to_project(self, project_name, blob_name, dir):
print "download_file_to_project(project_name="+project_name+", blob_name="+blob_name+", dir="+dir+")"
for tries in range(1,5):
try:
self.storage.get_blob_to_path(self.proj_storage_name,
os.path.join(project_name,blob_name),
os.path.join(dir,os.path.join(project_name,blob_name)),
open_mode='wb', snapshot=None, x_ms_lease_id=None,
progress_callback=None)
break
except Exception as e:
if tries == 5:
print("Muitos erros de conexão. Operação abortada.")
else:
print("Erro de conexão com serviço. Retentando..." + e.__str__())
def download_file_to_project(self, project_name, blob_name, dir):
print "download_file_to_project(project_name="+project_name+", blob_name="+blob_name+", dir="+dir+")"
for tries in range(1,5):
try:
self.storage.get_blob_to_path(self.proj_storage_name,
os.path.join(project_name,blob_name),
os.path.join(dir,os.path.join(project_name,blob_name)),
open_mode='wb', snapshot=None, x_ms_lease_id=None,
progress_callback=None)
break
except Exception as e:
if tries == 5:
print("Muitos erros de conexão. Operação abortada.")
else:
print("Erro de conexão com serviço. Retentando..." + e.__str__())
def upload_proj_file(self, project_name, blob_name, dir):
print "upload_proj_file(project_name="+project_name+", blob_name="+blob_name+", dir="+dir+")"
if blob_name[0] == '/':
blob_name = blob_name[1:]
for tries in range(1,5):
try:
self.storage.put_block_blob_from_path(self.proj_storage_name,
os.path.join(project_name,blob_name),
os.path.join(dir,os.path.join(project_name,blob_name)))
break
except Exception as e:
if tries == 5:
print("Muitos erros de conexão. Operação abortada.")
else:
print("Erro de conexão com serviço. Retentando..." + e.__str__())
def download_file_to_algo(self, blob_name, dir):
print "download_file_to_algo(blob_name="+blob_name+", dir="+dir+")"
for tries in range(1,5):
try:
self.storage.get_blob_to_path(container_name=self.algo_storage_name,
blob_name=os.path.join(blob_name),
file_path=os.path.join(dir,blob_name),
open_mode='wb', snapshot=None, x_ms_lease_id=None,
progress_callback=None)
break
except Exception as e:
if tries == 5:
print("Muitos erros de conexão. Operação abortada.")
else:
print("Erro de conexão com serviço. Retentando..." + e.__str__())
def send_status(self, main_status):
for tries in range(1,5):
try:
self.bus_service.send_topic_message(topic_name=self.status_topic,
message=Message(main_status.encode('utf-8')))
break
except Exception as e:
if tries == 5:
print("Muitos erros de conexão. Operação abortada.")
else:
print("Erro de conexão com serviço. Retentando..." + e.__str__())
def shutdown_myself(self):
# A máquina virtual irá cometer suicídio.
print("Removendo máquina virtual da nuvem...")
for tries in range(1,5):
try:
self.sms.delete_deployment(
service_name=self.myMachineName,
deployment_name=self.myMachineName, delete_vhd=True)
exit(0)
break
except Exception as e:
if tries == 5:
print("Muitos erros de conexão. Operação abortada.")
else:
print("Erro de conexão com serviço. Retentando..." + e.__str__())
def _cleanUpExternalStore(self, containerName):
from toil.jobStores.azureJobStore import _fetchAzureAccountKey
from azure.storage.blob import BlobService
blobService = BlobService(account_key=_fetchAzureAccountKey(self.accountName),
account_name=self.accountName)
blobService.delete_container(containerName)
def service(self):
return BlobService(account_name=self.account,
account_key=_fetchAzureAccountKey(self.account))
def __init__(self):
self.storage_account = getenv('STORAGE_ACCOUNT')
self.blob_service = BlobService(self.storage_account,
getenv('STORAGE_KEY'))
class BlobSource(DataSource):
def __init__(self):
self.storage_account = getenv('STORAGE_ACCOUNT')
self.blob_service = BlobService(self.storage_account,
getenv('STORAGE_KEY'))
def load(self, sparkContext, container, path):
path = ('/' if path[0] != '/' else '') + path
uri = 'wasb://%s@%s.blob.core.windows.net%s' % (
container, self.storage_account, path)
print 'Loading from %s' % uri
return sparkContext.textFile(uri)
def download(self, container, path):
print 'Downloading blob from %s/%s' % (container, path)
self.blob_service.get_blob_to_path(container, path, path)
print 'Downloaded blob to ' + path
def saveAsJson(self, payload, container, path):
path = path.lstrip('/')
print path
print 'Saving to %s/%s' % (container, path)
json_string = json.dumps(payload, ensure_ascii=False).encode('utf-8')
try:
self.blob_service.put_blob(
container,
path,
json_string,
'BlockBlob',
x_ms_blob_cache_control='max-age=3600',
x_ms_blob_content_type='application/json')
except Exception as e:
print 'Failed to save %s/%s: %s' % (container, path, str(e))
raise
def saveAsText(self, rdd, container, path):
path = path.lstrip('/')
path = '/' + path
print 'Saving rdd to %s%s' % (container, path)
uri = 'wasb://%s@%s.blob.core.windows.net%s' % (
container, self.storage_account, path)
try:
rdd.saveAsTextFile(uri)
except Exception as e:
print 'Failed to save %s%s: %s' % (container, path, str(e))
raise
def deleteAllBut(self, container, exceptFolderName):
print 'deleteAllBut called'
try:
bloblistingresult = self.blob_service.list_blobs(container)
for i in bloblistingresult:
print i.name
if not exceptFolderName in i.name:
try:
print 'deleting'
self.blob_service.delete_blob(container, i.name)
print 'deleted'
except Exception as e:
print 'Failed to delete %s/%s: %s' % (container,
i.name, str(e))
raise
except Exception as e:
print 'Failed to list things in %s: %s' % (container, str(e))
raise
class AzureFS(LoggingMixIn, Operations):
"""Azure Blob Storage filesystem"""
blobs = None
containers = dict() # <cname, dict(stat:dict,
#files:None|dict<fname, stat>)
fds = dict() # <fd, (path, bytes, dirty)>
fd = 0
def __init__(self, account, key):
self.blobs = BlobService(account, key)
self.rebuild_container_list()
def convert_to_epoch(self, date):
"""Converts Tue, 31 Jul 2012 07:17:34 GMT format to epoch"""
return int(time.mktime(time.strptime(date, TIME_FORMAT)))
def rebuild_container_list(self):
cmap = dict()
cnames = set()
for c in self.blobs.list_containers():
date = c.properties.last_modified
cstat = dict(st_mode=(S_IFDIR | 0755),
st_uid=getuid(),
st_size=0,
st_mtime=self.convert_to_epoch(date))
cname = c.name
cmap['/' + cname] = dict(stat=cstat, files=None)
cnames.add(cname)
cmap['/'] = dict(files={},
stat=dict(st_mode=(S_IFDIR | 0755),
st_uid=getuid(),
st_size=0,
st_mtime=int(time.time())))
self.containers = cmap # destroys fs tree cache resistant to misses
def _parse_path(self, path): # returns </dir, file(=None)>
if path.count('/') > 1: # file
return str(path[:path.rfind('/')]), str(path[path.rfind('/') + 1:])
else: # dir
pos = path.rfind('/', 1)
if pos == -1:
return path, None
else:
return str(path[:pos]), None
def parse_container(self, path):
base_container = path[1:] # /abc/def/g --> abc
if base_container.find('/') > -1:
base_container = base_container[:base_container.find('/')]
return str(base_container)
def _get_dir(self, path, contents_required=False):
if not self.containers:
self.rebuild_container_list()
if path in self.containers and not (contents_required and \
self.containers[path]['files'] is None):
return self.containers[path]
cname = self.parse_container(path)
if '/' + cname not in self.containers:
raise FuseOSError(ENOENT)
else:
if self.containers['/' + cname]['files'] is None:
# fetch contents of container
log.info("------> CONTENTS NOT FOUND: %s" % cname)
blobs = self.blobs.list_blobs(cname)
dirstat = dict(st_mode=(S_IFDIR | 0755),
st_size=0,
st_uid=getuid(),
st_mtime=time.time())
if self.containers['/' + cname]['files'] is None:
self.containers['/' + cname]['files'] = dict()
for f in blobs:
blob_name = f.name
blob_date = f.properties.last_modified
blob_size = long(f.properties.content_length)
node = dict(st_mode=(S_IFREG | 0644),
st_size=blob_size,
st_mtime=self.convert_to_epoch(blob_date),
st_uid=getuid())
if blob_name.find('/') == -1: # file just under container
self.containers['/' + cname]['files'][blob_name] = node
return self.containers['/' + cname]
return None
def _get_file(self, path):
d, f = self._parse_path(path)
dir = self._get_dir(d, True)
if dir is not None and f in dir['files']:
return dir['files'][f]
def getattr(self, path, fh=None):
d, f = self._parse_path(path)
if f is None:
dir = self._get_dir(d)
return dir['stat']
else:
file = self._get_file(path)
if file:
return file
raise FuseOSError(ENOENT)
# FUSE
def mkdir(self, path, mode):
if path.count('/') <= 1: # create on root
name = path[1:]
if not 3 <= len(name) <= 63:
log.error("Container names can be 3 through 63 chars long.")
raise FuseOSError(ENAMETOOLONG)
if name is not name.lower():
log.error("Container names cannot contain uppercase \
characters.")
raise FuseOSError(EACCES)
if name.count('--') > 0:
log.error('Container names cannot contain consecutive \
dashes (-).')
raise FuseOSError(EAGAIN)
#TODO handle all "-"s must be preceded by letter or numbers
#TODO starts with only letter or number, can contain letter, nr,'-'
resp = self.blobs.create_container(name)
if resp:
self.rebuild_container_list()
log.info("CONTAINER %s CREATED" % name)
else:
raise FuseOSError(EACCES)
log.error("Invalid container name or container already \
exists.")
else:
raise FuseOSError(ENOSYS) # TODO support 2nd+ level mkdirs
def rmdir(self, path):
if path.count('/') == 1:
c_name = path[1:]
resp = self.blobs.delete_container(c_name)
if resp:
if path in self.containers:
del self.containers[path]
else:
raise FuseOSError(EACCES)
else:
raise FuseOSError(ENOSYS) # TODO support 2nd+ level mkdirs
def create(self, path, mode):
node = dict(st_mode=(S_IFREG | mode),
st_size=0,
st_nlink=1,
st_uid=getuid(),
st_mtime=time.time())
d, f = self._parse_path(path)
if not f:
log.error("Cannot create files on root level: /")
raise FuseOSError(ENOSYS)
dir = self._get_dir(d, True)
if not dir:
raise FuseOSError(EIO)
dir['files'][f] = node
return self.open(path, data='') # reusing handler provider
def open(self, path, flags=0, data=None):
if data == None: # download contents
c_name = self.parse_container(path)
f_name = path[path.find('/', 1) + 1:]
try:
data = self.blobs.get_blob(c_name, f_name)
except AzureMissingResourceHttpError:
dir = self._get_dir('/' + c_name, True)
if f_name in dir['files']:
del dir['files'][f_name]
raise FuseOSError(ENOENT)
except AzureException as e:
log.error("Read blob failed HTTP %d" % e.code)
raise FuseOSError(EAGAIN)
self.fd += 1
self.fds[self.fd] = (path, data, False)
return self.fd
def flush(self, path, fh=None):
if not fh:
raise FuseOSError(EIO)
else:
if fh not in self.fds:
raise FuseOSError(EIO)
path = self.fds[fh][0]
data = self.fds[fh][1]
dirty = self.fds[fh][2]
if not dirty:
return 0 # avoid redundant write
d, f = self._parse_path(path)
c_name = self.parse_container(path)
if data is None:
data = ''
try:
if len(data) < 64 * 1024 * 1024: # 64 mb
self.blobs.put_blob(c_name, f, data, 'BlockBlob')
else:
# divide file by blocks and upload
block_size = 8 * 1024 * 1024
num_blocks = int(math.ceil(len(data) * 1.0 / block_size))
rd = str(random.randint(1, 1e8))
block_ids = list()
for i in range(num_blocks):
part = data[i * block_size:min((i + 1) *
block_size, len(data))]
block_id = base64.encodestring(
'%s_%s' % (rd, (8 - len(str(i))) * '0' + str(i)))
self.blobs.put_block(c_name, f, part, block_id)
block_ids.append(block_id)
self.blobs.put_block_list(c_name, f, block_ids)
except AzureException:
raise FuseOSError(EAGAIN)
dir = self._get_dir(d, True)
if not dir or f not in dir['files']:
raise FuseOSError(EIO)
# update local data
dir['files'][f]['st_size'] = len(data)
dir['files'][f]['st_mtime'] = time.time()
self.fds[fh] = (path, data, False) # mark as not dirty
return 0
def release(self, path, fh=None):
if fh is not None and fh in self.fds:
del self.fds[fh]
def truncate(self, path, length, fh=None):
return 0 # assume done, no need
def write(self, path, data, offset, fh=None):
if not fh or fh not in self.fds:
raise FuseOSError(ENOENT)
else:
d = self.fds[fh][1]
if d is None:
d = ""
self.fds[fh] = (self.fds[fh][0], d[:offset] + data, True)
return len(data)
def unlink(self, path):
c_name = self.parse_container(path)
d, f = self._parse_path(path)
try:
self.blobs.delete_blob(c_name, f)
_dir = self._get_dir(path, True)
if _dir and f in _dir['files']:
del _dir['files'][f]
return 0
except AzureMissingResourceHttpError:
raise FuseOSError(ENOENT)
except Exception as e:
raise FuseOSError(EAGAIN)
def readdir(self, path, fh):
if path == '/':
return ['.', '..'] + [x[1:] for x in self.containers.keys() \
if x is not '/']
dir = self._get_dir(path, True)
if not dir:
raise FuseOSError(ENOENT)
return ['.', '..'] + dir['files'].keys()
def read(self, path, size, offset, fh):
if not fh or fh not in self.fds:
raise FuseOSError(ENOENT)
f_name = path[path.find('/', 1) + 1:]
c_name = path[1:path.find('/', 1)]
try:
data = self.blobs.get_blob(c_name, f_name)
self.fds[fh] = (self.fds[fh][0], data, False)
return data[offset:offset + size]
except URLError, e:
if e.code == 404:
raise FuseOSError(ENOENT)
elif e.code == 403:
raise FUSEOSError(EPERM)
else:
log.error("Read blob failed HTTP %d" % e.code)
raise FuseOSError(EAGAIN)
data = self.fds[fh][1]
if data is None:
data = ""
return data[offset:offset + size]
def _save(self, name, content):
blob_service = BlobService(account_name=accountName,
account_key=accountKey)
import mimetypes
small_content = content
content.open()
content_type = None
if hasattr(content.file, 'content_type'):
content_type = content.file.content_type
else:
content_type = mimetypes.guess_type(name)[0]
content_str = content.read()
blob_service.put_blob(
'videos',
name,
content_str,
x_ms_blob_type='BlockBlob',
x_ms_blob_content_type=content_type,
x_ms_blob_cache_control=
'public, max-age=3600, s-maxage=86400' #cache in the browser for 1 hr, on the edge for 24 hrs
)
content.close()
# if "avatars" in name: #creating and saving thumbnail
# small_image_name = name
# small_image_name = string.replace(small_image_name, "avatars", "thumbnails")
# thumbnail = StringIO.StringIO()
# size = 22, 22
# image = small_content.file
# image = Image.open(image)
# small_image = image.resize(size, Image.ANTIALIAS)
# small_image.save(thumbnail,'JPEG',quality=70, optimize=True)
# img = InMemoryUploadedFile(thumbnail, None, 'small.jpg', 'image/jpeg', thumbnail.len, None)
# small_content.file = img
# small_content.open()
# stream = small_content.read()
# blob_service.put_blob(
# 'pictures',
# small_image_name,
# stream,
# x_ms_blob_type='BlockBlob',
# x_ms_blob_content_type=content_type,
# x_ms_blob_cache_control ='public, max-age=604800, s-maxage=604800' #cache in the browser and on the edge for 7 days
# )
# small_content.close()
# elif "photos" in name:
# small_image_name = name
# small_image_name = string.replace(small_image_name, "photos", "thumbnails")
# thumbnail = StringIO.StringIO()
# #size = 40, 40
# height = 38
# image = small_content.file
# image = Image.open(image)
# wpercent = (height/float(image.size[1]))
# bsize = int((float(image.size[0])*float(wpercent)))
# small_image = image.resize((bsize,height), PIL.Image.ANTIALIAS)
# small_image.save(thumbnail,'JPEG',quality=70, optimize=True)
# img = InMemoryUploadedFile(thumbnail, None, 'small.jpg', 'image/jpeg', thumbnail.len, None)
# small_content.file = img
# small_content.open()
# stream = small_content.read()
# blob_service.put_blob(
# 'pictures',
# small_image_name,
# stream,
# x_ms_blob_type='BlockBlob',
# x_ms_blob_content_type=content_type,
# x_ms_blob_cache_control ='public, max-age=3600, s-maxage=86400' #cache in the browser for 1 hr, on the edge for 24 hrs
# )
# small_content.close()
# else:
# pass
return name
def __init__(self, account, key):
self.blobs = BlobService(account, key)
self.rebuild_container_list()
def Main(argv=sys.argv):
logging.basicConfig(level=logging.INFO)
try:
argv = FLAGS(argv) # parse flags
except flags.FlagsError as e:
logging.error('%s\nUsage: %s ARGS\n%s', e, sys.argv[0], FLAGS)
sys.exit(1)
if FLAGS.bucket is None:
raise ValueError('Must specify a valid bucket for this test.')
logging.info('Storage provider is %s, bucket is %s, scenario is %s',
FLAGS.storage_provider, FLAGS.bucket, FLAGS.scenario)
host_to_connect = None
if FLAGS.host is not None:
logging.info('Will use user-specified host endpoint: %s', FLAGS.host)
host_to_connect = FLAGS.host
if FLAGS.storage_provider == 'AZURE':
if FLAGS.azure_key is None or FLAGS.azure_account is None:
raise ValueError('Must specify azure account and key')
else:
global _AZURE_BLOB_SERVICE
_AZURE_BLOB_SERVICE = BlobService(FLAGS.azure_account,
FLAGS.azure_key)
# There are DNS lookup issues with the provider Azure when doing
# "high" number of concurrent requests using multiple threads. The error
# came from getaddrinfo() called by the azure python library. By reducing
# the concurrent thread count to 10 or below, the issue can be mitigated.
# If we lower the thread count, we need to lower the total object count
# too so the time to write these object remains short
global LIST_CONSISTENCY_THREAD_COUNT
LIST_CONSISTENCY_THREAD_COUNT = 10
global LIST_CONSISTENCY_OBJECT_COUNT
LIST_CONSISTENCY_OBJECT_COUNT = 1000
storage_schema = STORAGE_TO_SCHEMA_DICT[FLAGS.storage_provider]
if FLAGS.scenario == 'OneByteRW':
return OneByteRWBenchmark(storage_schema, host_to_connect)
elif FLAGS.scenario == 'ListConsistency':
list_latency = {}
list_inconsistency_window = {}
inconsistent_list_count = {}
for scenario in [
LIST_AFTER_WRITE_SCENARIO, LIST_AFTER_UPDATE_SCENARIO
]:
list_latency[scenario] = []
list_inconsistency_window[scenario] = []
inconsistent_list_count[scenario] = 0.0
logging.info('Running list consistency tests for %d iterations...',
FLAGS.iterations)
for _ in range(FLAGS.iterations):
result = ListConsistencyBenchmark(storage_schema, host_to_connect)
# Analyze the result for both scenarios.
for scenario in [
LIST_AFTER_WRITE_SCENARIO, LIST_AFTER_UPDATE_SCENARIO
]:
result_consistent = '%s%s' % (scenario,
LIST_RESULT_SUFFIX_CONSISTENT)
if result_consistent in result:
if result[result_consistent]:
list_latency[scenario].append(
result['%s%s' %
(scenario, LIST_RESULT_SUFFIX_LATENCY)])
else:
inconsistent_list_count[scenario] += 1
list_inconsistency_window[scenario].append(
result['%s%s' %
(scenario,
LIST_RESULT_SUFFIX_INCONSISTENCY_WINDOW)])
# All iterations completed, ready to print out final stats.
logging.info('\n\nFinal stats:')
for scenario in [
LIST_AFTER_WRITE_SCENARIO, LIST_AFTER_UPDATE_SCENARIO
]:
logging.info(
'%s consistency percentage: %f', scenario, 100 *
(1 - inconsistent_list_count[scenario] / FLAGS.iterations))
if len(list_inconsistency_window[scenario]) > 0:
logging.info(
'%s inconsistency window: %s', scenario,
json.dumps(PercentileCalculator(
list_inconsistency_window[scenario]),
sort_keys=True))
if len(list_latency[scenario]) > 0:
logging.info(
'%s latency: %s', scenario,
json.dumps(PercentileCalculator(list_latency[scenario]),
sort_keys=True))
return 0
elif FLAGS.scenario == 'SingleStreamThroughput':
return SingleStreamThroughputBenchmark(storage_schema, host_to_connect)
elif FLAGS.scenario == 'CleanupBucket':
return CleanupBucket(storage_schema)
def apathetic_container_delete(container_name, *args, **kwargs):
conn = BlobService(*args, **kwargs)
conn.delete_container(container_name)
return conn
from azure.storage.blob import BlobService
source_account = BlobService(
account_name='cngssd',
account_key=
'DVvOtpOVW71er9ztR3mooJk4Zc3ZNovW9YV3qu4Y6bkN0eCfHutpcNVXW6gtpfolRk4CcAlmftz/+SDwm2BQag=='
)
dest_account = BlobService(
account_name='testcng',
account_key=
'piWr6zleZ1sL8aopv5Y4NRYyrVWaW2/QrXcPpsxRec4IxtEoR1IyRmZCkbdyq50Bfu0qidF8SicQahdM+OExvg=='
)
# list blobs
source_blobs = []
source_marker = None
while True:
batch = source_account.list_blobs('vhds', marker=source_marker)
source_blobs.extend(batch)
if not batch.next_marker:
break
source_marker = batch.next_marker
for blob in source_blobs:
print(blob.name)
dest_blobs = []
dest_marker = None
while True:
batch = dest_account.list_blobs('vhds', marker=dest_marker)
dest_blobs.extend(batch)
if not batch.next_marker:
# -*- coding: utf-8 -*-
"""
Created on Fri Oct 09 16:36:56 2015
@author: justin.malinchak
"""
# List blobs in container
from azure.storage.blob import BlobService
blob_service = BlobService(
account_name='portalvhdss5m831rhl98hj',
account_key=
'Z1MliCYE7p9Ks9kYQoGeM4V99hODtiJL82BVi/zIm06jLYh7n0tV8YaZHzITKixMwUUmjJ1Vp05XrgHG+gXFlg=='
)
blobs = []
marker = None
while True:
batch = blob_service.list_blobs('mycontainer', marker=marker)
blobs.extend(batch)
if not batch.next_marker:
break
marker = batch.next_marker
for blob in blobs:
bname = blob.name
print('')
print(bname)
print('')
bpathname = 'C:\\Batches\\$Work\\' + bname
blob_service.get_blob_to_path('mycontainer', bname, bpathname)
from sklearn import metrics
from __future__ import division
from sklearn import linear_model
from azureml import services
# ###Step 1. Read in the Data from blob
# In[64]:
#Connection String
CONTAINERNAME = 'test1'
STORAGEACCOUNTNAME = 'weigstoragefordsvm'
STORAGEACCOUNTKEY = 'FUyNCM83pY4K2srBfZv4yDr6ru7d+BfbmHPPtucqS7EIgvUSQBG4zPkznpCuClWVOMitAQXG3aJFbvuD7mBkhQ=='
BLOBNAME = 'demo_ex_9_stratified_1_1000_copy.csv'
blob_service = BlobService(account_name=STORAGEACCOUNTNAME,
account_key=STORAGEACCOUNTKEY)
#Read in as text
t1 = time.time()
data = blob_service.get_blob_to_text(CONTAINERNAME, BLOBNAME).split("\n")
t2 = time.time()
print(("It takes %s seconds to read in " + BLOBNAME) % (t2 - t1))
#Add column names and separate columns
colnames = [
'medallion', 'hack_license', 'vendor_id', 'rate_code',
'store_and_fwd_flag', 'pickup_datetime', 'dropoff_datetime',
'passenger_count', 'trip_time_in_secs', 'trip_distance',
'pickup_longitude', 'pickup_latitude', 'dropoff_longitude',
'dropoff_latitude', 'payment_type', 'fare_amount', 'surcharge', 'mta_tax',
'tolls_amount', 'total_amount', 'tip_amount', 'tipped', 'tip_class',
class AzureJobStore(AbstractJobStore):
"""
A job store that uses Azure's blob store for file storage and Table Service to store job info
with strong consistency.
"""
# Dots in container names should be avoided because container names are used in HTTPS bucket
# URLs where the may interfere with the certificate common name. We use a double underscore
# as a separator instead.
#
containerNameRe = re.compile(r'^[a-z0-9](-?[a-z0-9]+)+[a-z0-9]$')
# See https://msdn.microsoft.com/en-us/library/azure/dd135715.aspx
#
minContainerNameLen = 3
maxContainerNameLen = 63
maxNameLen = 10
nameSeparator = 'xx' # Table names must be alphanumeric
# Length of a jobID - used to test if a stats file has been read already or not
jobIDLength = len(str(uuid.uuid4()))
def __init__(self, locator, jobChunkSize=maxAzureTablePropertySize):
super(AzureJobStore, self).__init__()
accountName, namePrefix = locator.split(':', 1)
if '--' in namePrefix:
raise ValueError(
"Invalid name prefix '%s'. Name prefixes may not contain %s." %
(namePrefix, self.nameSeparator))
if not self.containerNameRe.match(namePrefix):
raise ValueError(
"Invalid name prefix '%s'. Name prefixes must contain only digits, "
"hyphens or lower-case letters and must not start or end in a "
"hyphen." % namePrefix)
# Reserve 13 for separator and suffix
if len(namePrefix) > self.maxContainerNameLen - self.maxNameLen - len(
self.nameSeparator):
raise ValueError((
"Invalid name prefix '%s'. Name prefixes may not be longer than 50 "
"characters." % namePrefix))
if '--' in namePrefix:
raise ValueError(
"Invalid name prefix '%s'. Name prefixes may not contain "
"%s." % (namePrefix, self.nameSeparator))
self.locator = locator
self.jobChunkSize = jobChunkSize
self.accountKey = _fetchAzureAccountKey(accountName)
self.accountName = accountName
# Table names have strict requirements in Azure
self.namePrefix = self._sanitizeTableName(namePrefix)
# These are the main API entry points.
self.tableService = TableService(account_key=self.accountKey,
account_name=accountName)
self.blobService = BlobService(account_key=self.accountKey,
account_name=accountName)
# Serialized jobs table
self.jobItems = None
# Job<->file mapping table
self.jobFileIDs = None
# Container for all shared and unshared files
self.files = None
# Stats and logging strings
self.statsFiles = None
# File IDs that contain stats and logging strings
self.statsFileIDs = None
@property
def keyPath(self):
return self.config.cseKey
def initialize(self, config):
if self._jobStoreExists():
raise JobStoreExistsException(self.locator)
logger.debug("Creating job store at '%s'" % self.locator)
self._bind(create=True)
super(AzureJobStore, self).initialize(config)
def resume(self):
if not self._jobStoreExists():
raise NoSuchJobStoreException(self.locator)
logger.debug("Using existing job store at '%s'" % self.locator)
self._bind(create=False)
super(AzureJobStore, self).resume()
def destroy(self):
self._bind()
for name in 'jobItems', 'jobFileIDs', 'files', 'statsFiles', 'statsFileIDs':
resource = getattr(self, name)
if resource is not None:
if isinstance(resource, AzureTable):
resource.delete_table()
elif isinstance(resource, AzureBlobContainer):
resource.delete_container()
else:
assert False
setattr(self, name, None)
def _jobStoreExists(self):
"""
Checks if job store exists by querying the existence of the statsFileIDs table. Note that
this is the last component that is deleted in :meth:`.destroy`.
"""
for attempt in retry_azure():
with attempt:
try:
table = self.tableService.query_tables(
table_name=self._qualify('statsFileIDs'))
except AzureMissingResourceHttpError as e:
if e.status_code == 404:
return False
else:
raise
else:
return table is not None
def _bind(self, create=False):
table = self._bindTable
container = self._bindContainer
for name, binder in (('jobItems', table), ('jobFileIDs', table),
('files', container), ('statsFiles', container),
('statsFileIDs', table)):
if getattr(self, name) is None:
setattr(self, name, binder(self._qualify(name), create=create))
def _qualify(self, name):
return self.namePrefix + self.nameSeparator + name.lower()
def jobs(self):
# How many jobs have we done?
total_processed = 0
for jobEntity in self.jobItems.query_entities_auto():
# Process the items in the page
yield AzureJob.fromEntity(jobEntity)
total_processed += 1
if total_processed % 1000 == 0:
# Produce some feedback for the user, because this can take
# a long time on, for example, Azure
logger.debug("Processed %d total jobs" % total_processed)
logger.debug("Processed %d total jobs" % total_processed)
def create(self, jobNode):
jobStoreID = self._newJobID()
job = AzureJob.fromJobNode(jobNode, jobStoreID,
self._defaultTryCount())
entity = job.toItem(chunkSize=self.jobChunkSize)
entity['RowKey'] = EntityProperty('Edm.String', jobStoreID)
self.jobItems.insert_entity(entity=entity)
return job
def exists(self, jobStoreID):
if self.jobItems.get_entity(row_key=bytes(jobStoreID)) is None:
return False
return True
def load(self, jobStoreID):
jobEntity = self.jobItems.get_entity(row_key=bytes(jobStoreID))
if jobEntity is None:
raise NoSuchJobException(jobStoreID)
return AzureJob.fromEntity(jobEntity)
def update(self, job):
self.jobItems.update_entity(
row_key=bytes(job.jobStoreID),
entity=job.toItem(chunkSize=self.jobChunkSize))
def delete(self, jobStoreID):
try:
self.jobItems.delete_entity(row_key=bytes(jobStoreID))
except AzureMissingResourceHttpError:
# Job deletion is idempotent, and this job has been deleted already
return
filterString = "PartitionKey eq '%s'" % jobStoreID
for fileEntity in self.jobFileIDs.query_entities(filter=filterString):
jobStoreFileID = fileEntity.RowKey
self.deleteFile(jobStoreFileID)
def getEnv(self):
return dict(AZURE_ACCOUNT_KEY=self.accountKey)
class BlobInfo(namedtuple('BlobInfo', ('account', 'container', 'name'))):
@property
@memoize
def service(self):
return BlobService(account_name=self.account,
account_key=_fetchAzureAccountKey(self.account))
@classmethod
def getSize(cls, url):
blob = cls._parseWasbUrl(url)
blobProps = blob.service.get_blob_properties(blob.container, blob.name)
return int(blobProps['content-length'])
@classmethod
def _readFromUrl(cls, url, writable):
blob = cls._parseWasbUrl(url)
for attempt in retry_azure():
with attempt:
blob.service.get_blob_to_file(container_name=blob.container,
blob_name=blob.name,
stream=writable)
@classmethod
def _writeToUrl(cls, readable, url):
blob = cls._parseWasbUrl(url)
blob.service.put_block_blob_from_file(container_name=blob.container,
blob_name=blob.name,
stream=readable)
@classmethod
def _parseWasbUrl(cls, url):
"""
:param urlparse.ParseResult url: x
:rtype: AzureJobStore.BlobInfo
"""
assert url.scheme in ('wasb', 'wasbs')
try:
container, account = url.netloc.split('@')
except ValueError:
raise InvalidImportExportUrlException(url)
suffix = '.blob.core.windows.net'
if account.endswith(suffix):
account = account[:-len(suffix)]
else:
raise InvalidImportExportUrlException(url)
assert url.path[0] == '/'
return cls.BlobInfo(account=account,
container=container,
name=url.path[1:])
@classmethod
def _supportsUrl(cls, url, export=False):
return url.scheme.lower() in ('wasb', 'wasbs')
def writeFile(self, localFilePath, jobStoreID=None):
jobStoreFileID = self._newFileID()
self.updateFile(jobStoreFileID, localFilePath)
self._associateFileWithJob(jobStoreFileID, jobStoreID)
return jobStoreFileID
def updateFile(self, jobStoreFileID, localFilePath):
with open(localFilePath) as read_fd:
with self._uploadStream(jobStoreFileID, self.files) as write_fd:
while True:
buf = read_fd.read(self._maxAzureBlockBytes)
write_fd.write(buf)
if len(buf) == 0:
break
def readFile(self, jobStoreFileID, localFilePath):
try:
with self._downloadStream(jobStoreFileID, self.files) as read_fd:
with open(localFilePath, 'w') as write_fd:
while True:
buf = read_fd.read(self._maxAzureBlockBytes)
write_fd.write(buf)
if not buf:
break
except AzureMissingResourceHttpError:
raise NoSuchFileException(jobStoreFileID)
def deleteFile(self, jobStoreFileID):
try:
self.files.delete_blob(blob_name=bytes(jobStoreFileID))
self._dissociateFileFromJob(jobStoreFileID)
except AzureMissingResourceHttpError:
pass
def fileExists(self, jobStoreFileID):
# As Azure doesn't have a blob_exists method (at least in the
# python API) we just try to download the metadata, and hope
# the metadata is small so the call will be fast.
try:
self.files.get_blob_metadata(blob_name=bytes(jobStoreFileID))
return True
except AzureMissingResourceHttpError:
return False
@contextmanager
def writeFileStream(self, jobStoreID=None):
# TODO: this (and all stream methods) should probably use the
# Append Blob type, but that is not currently supported by the
# Azure Python API.
jobStoreFileID = self._newFileID()
with self._uploadStream(jobStoreFileID, self.files) as fd:
yield fd, jobStoreFileID
self._associateFileWithJob(jobStoreFileID, jobStoreID)
@contextmanager
def updateFileStream(self, jobStoreFileID):
with self._uploadStream(jobStoreFileID,
self.files,
checkForModification=True) as fd:
yield fd
def getEmptyFileStoreID(self, jobStoreID=None):
jobStoreFileID = self._newFileID()
with self._uploadStream(jobStoreFileID, self.files) as _:
pass
self._associateFileWithJob(jobStoreFileID, jobStoreID)
return jobStoreFileID
@contextmanager
def readFileStream(self, jobStoreFileID):
if not self.fileExists(jobStoreFileID):
raise NoSuchFileException(jobStoreFileID)
with self._downloadStream(jobStoreFileID, self.files) as fd:
yield fd
@contextmanager
def writeSharedFileStream(self, sharedFileName, isProtected=None):
assert self._validateSharedFileName(sharedFileName)
sharedFileID = self._newFileID(sharedFileName)
with self._uploadStream(sharedFileID,
self.files,
encrypted=isProtected) as fd:
yield fd
@contextmanager
def readSharedFileStream(self, sharedFileName):
assert self._validateSharedFileName(sharedFileName)
sharedFileID = self._newFileID(sharedFileName)
if not self.fileExists(sharedFileID):
raise NoSuchFileException(sharedFileID)
with self._downloadStream(sharedFileID, self.files) as fd:
yield fd
def writeStatsAndLogging(self, statsAndLoggingString):
# TODO: would be a great use case for the append blobs, once implemented in the Azure SDK
jobStoreFileID = self._newFileID()
encrypted = self.keyPath is not None
if encrypted:
statsAndLoggingString = encryption.encrypt(statsAndLoggingString,
self.keyPath)
self.statsFiles.put_block_blob_from_text(
blob_name=bytes(jobStoreFileID),
text=statsAndLoggingString,
x_ms_meta_name_values=dict(encrypted=str(encrypted)))
self.statsFileIDs.insert_entity(entity={'RowKey': jobStoreFileID})
def readStatsAndLogging(self, callback, readAll=False):
suffix = '_old'
numStatsFiles = 0
for attempt in retry_azure():
with attempt:
for entity in self.statsFileIDs.query_entities():
jobStoreFileID = entity.RowKey
hasBeenRead = len(jobStoreFileID) > self.jobIDLength
if not hasBeenRead:
with self._downloadStream(jobStoreFileID,
self.statsFiles) as fd:
callback(fd)
# Mark this entity as read by appending the suffix
self.statsFileIDs.insert_entity(
entity={'RowKey': jobStoreFileID + suffix})
self.statsFileIDs.delete_entity(
row_key=bytes(jobStoreFileID))
numStatsFiles += 1
elif readAll:
# Strip the suffix to get the original ID
jobStoreFileID = jobStoreFileID[:-len(suffix)]
with self._downloadStream(jobStoreFileID,
self.statsFiles) as fd:
callback(fd)
numStatsFiles += 1
return numStatsFiles
_azureTimeFormat = "%Y-%m-%dT%H:%M:%SZ"
def getPublicUrl(self, jobStoreFileID):
try:
self.files.get_blob_properties(blob_name=bytes(jobStoreFileID))
except AzureMissingResourceHttpError:
raise NoSuchFileException(jobStoreFileID)
# Compensate of a little bit of clock skew
startTimeStr = (datetime.utcnow() - timedelta(minutes=5)).strftime(
self._azureTimeFormat)
endTime = datetime.utcnow() + self.publicUrlExpiration
endTimeStr = endTime.strftime(self._azureTimeFormat)
sap = SharedAccessPolicy(
AccessPolicy(startTimeStr, endTimeStr,
BlobSharedAccessPermissions.READ))
sas_token = self.files.generate_shared_access_signature(
blob_name=bytes(jobStoreFileID), shared_access_policy=sap)
return self.files.make_blob_url(
blob_name=bytes(jobStoreFileID)) + '?' + sas_token
def getSharedPublicUrl(self, sharedFileName):
jobStoreFileID = self._newFileID(sharedFileName)
return self.getPublicUrl(jobStoreFileID)
def _newJobID(self):
# raw UUIDs don't work for Azure property names because the '-' character is disallowed.
return str(uuid.uuid4()).replace('-', '_')
# A dummy job ID under which all shared files are stored.
sharedFileJobID = uuid.UUID('891f7db6-e4d9-4221-a58e-ab6cc4395f94')
def _newFileID(self, sharedFileName=None):
if sharedFileName is None:
ret = bytes(uuid.uuid4())
else:
ret = bytes(uuid.uuid5(self.sharedFileJobID,
bytes(sharedFileName)))
return ret.replace('-', '_')
def _associateFileWithJob(self, jobStoreFileID, jobStoreID=None):
if jobStoreID is not None:
self.jobFileIDs.insert_entity(
entity={
'PartitionKey': EntityProperty('Edm.String', jobStoreID),
'RowKey': EntityProperty('Edm.String', jobStoreFileID)
})
def _dissociateFileFromJob(self, jobStoreFileID):
entities = self.jobFileIDs.query_entities(filter="RowKey eq '%s'" %
jobStoreFileID)
if entities:
assert len(entities) == 1
jobStoreID = entities[0].PartitionKey
self.jobFileIDs.delete_entity(partition_key=bytes(jobStoreID),
row_key=bytes(jobStoreFileID))
def _bindTable(self, tableName, create=False):
for attempt in retry_azure():
with attempt:
try:
tables = self.tableService.query_tables(
table_name=tableName)
except AzureMissingResourceHttpError as e:
if e.status_code != 404:
raise
else:
if tables:
assert tables[0].name == tableName
return AzureTable(self.tableService, tableName)
if create:
self.tableService.create_table(tableName)
return AzureTable(self.tableService, tableName)
else:
return None
def _bindContainer(self, containerName, create=False):
for attempt in retry_azure():
with attempt:
try:
self.blobService.get_container_properties(containerName)
except AzureMissingResourceHttpError as e:
if e.status_code == 404:
if create:
self.blobService.create_container(containerName)
else:
return None
else:
raise
return AzureBlobContainer(self.blobService, containerName)
def _sanitizeTableName(self, tableName):
"""
Azure table names must start with a letter and be alphanumeric.
This will never cause a collision if uuids are used, but
otherwise may not be safe.
"""
return 'a' + ''.join([x for x in tableName if x.isalnum()])
# Maximum bytes that can be in any block of an Azure block blob
# https://github.com/Azure/azure-storage-python/blob/4c7666e05a9556c10154508335738ee44d7cb104/azure/storage/blob/blobservice.py#L106
_maxAzureBlockBytes = 4 * 1024 * 1024
@contextmanager
def _uploadStream(self,
jobStoreFileID,
container,
checkForModification=False,
encrypted=None):
"""
:param encrypted: True to enforce encryption (will raise exception unless key is set),
False to prevent encryption or None to encrypt if key is set.
"""
if checkForModification:
try:
expectedVersion = container.get_blob_properties(
blob_name=bytes(jobStoreFileID))['etag']
except AzureMissingResourceHttpError:
expectedVersion = None
if encrypted is None:
encrypted = self.keyPath is not None
elif encrypted:
if self.keyPath is None:
raise RuntimeError(
'Encryption requested but no key was provided')
maxBlockSize = self._maxAzureBlockBytes
if encrypted:
# There is a small overhead for encrypted data.
maxBlockSize -= encryption.overhead
store = self
class UploadPipe(WritablePipe):
def readFrom(self, readable):
blockIDs = []
try:
while True:
buf = readable.read(maxBlockSize)
if len(buf) == 0:
# We're safe to break here even if we never read anything, since
# putting an empty block list creates an empty blob.
break
if encrypted:
buf = encryption.encrypt(buf, store.keyPath)
blockID = store._newFileID()
container.put_block(blob_name=bytes(jobStoreFileID),
block=buf,
blockid=blockID)
blockIDs.append(blockID)
except:
with panic(log=logger):
# This is guaranteed to delete any uncommitted blocks.
container.delete_blob(blob_name=bytes(jobStoreFileID))
if checkForModification and expectedVersion is not None:
# Acquire a (60-second) write lock,
leaseID = container.lease_blob(
blob_name=bytes(jobStoreFileID),
x_ms_lease_action='acquire')['x-ms-lease-id']
# check for modification,
blobProperties = container.get_blob_properties(
blob_name=bytes(jobStoreFileID))
if blobProperties['etag'] != expectedVersion:
container.lease_blob(blob_name=bytes(jobStoreFileID),
x_ms_lease_action='release',
x_ms_lease_id=leaseID)
raise ConcurrentFileModificationException(
jobStoreFileID)
# commit the file,
container.put_block_list(
blob_name=bytes(jobStoreFileID),
block_list=blockIDs,
x_ms_lease_id=leaseID,
x_ms_meta_name_values=dict(encrypted=str(encrypted)))
# then release the lock.
container.lease_blob(blob_name=bytes(jobStoreFileID),
x_ms_lease_action='release',
x_ms_lease_id=leaseID)
else:
# No need to check for modification, just blindly write over whatever
# was there.
container.put_block_list(
blob_name=bytes(jobStoreFileID),
block_list=blockIDs,
x_ms_meta_name_values=dict(encrypted=str(encrypted)))
with UploadPipe() as writable:
yield writable
@contextmanager
def _downloadStream(self, jobStoreFileID, container):
# The reason this is not in the writer is so we catch non-existant blobs early
blobProps = container.get_blob_properties(
blob_name=bytes(jobStoreFileID))
encrypted = strict_bool(blobProps['x-ms-meta-encrypted'])
if encrypted and self.keyPath is None:
raise AssertionError(
'Content is encrypted but no key was provided.')
outer_self = self
class DownloadPipe(ReadablePipe):
def writeTo(self, writable):
chunkStart = 0
fileSize = int(blobProps['Content-Length'])
while chunkStart < fileSize:
chunkEnd = chunkStart + outer_self._maxAzureBlockBytes - 1
buf = container.get_blob(blob_name=bytes(jobStoreFileID),
x_ms_range="bytes=%d-%d" %
(chunkStart, chunkEnd))
if encrypted:
buf = encryption.decrypt(buf, outer_self.keyPath)
writable.write(buf)
chunkStart = chunkEnd + 1
with DownloadPipe() as readable:
yield readable
# upload a file to Azure Blob Storage
from azure.storage.blob import BlobService
# Add your details here
blob_service = BlobService(account_name="", account_key="")
blob_service.put_block_blob_from_path("container", "remote-name.jpg",
"localfile.jpg")
class AzureIOStore(IOStore):
"""
A class that lets you get input from and send output to Azure Storage.
"""
def __init__(self, account_name, container_name, name_prefix=""):
"""
Make a new AzureIOStore that reads from and writes to the given
container in the given account, adding the given prefix to keys. All
paths will be interpreted as keys or key prefixes.
If the name prefix does not end with a trailing slash, and is not empty,
one will be added automatically.
Account keys are retrieved from the AZURE_ACCOUNT_KEY environment
variable or from the ~/.toilAzureCredentials file, as in Toil itself.
"""
# Make sure azure libraries actually loaded
assert (have_azure)
self.account_name = account_name
self.container_name = container_name
self.name_prefix = name_prefix
if self.name_prefix != "" and not self.name_prefix.endswith("/"):
# Make sure it has the trailing slash required.
self.name_prefix += "/"
# Sneak into Toil and use the same keys it uses
self.account_key = toil.jobStores.azureJobStore._fetchAzureAccountKey(
self.account_name)
# This will hold out Azure blob store connection
self.connection = None
def __getstate__(self):
"""
Return the state to use for pickling. We don't want to try and pickle
an open Azure connection.
"""
return (self.account_name, self.account_key, self.container_name,
self.name_prefix)
def __setstate__(self, state):
"""
Set up after unpickling.
"""
self.account_name = state[0]
self.account_key = state[1]
self.container_name = state[2]
self.name_prefix = state[3]
self.connection = None
def __connect(self):
"""
Make sure we have an Azure connection, and set one up if we don't.
"""
if self.connection is None:
RealtimeLogger.debug("Connecting to account {}, using "
"container {} and prefix {}".format(
self.account_name, self.container_name,
self.name_prefix))
# Connect to the blob service where we keep everything
self.connection = BlobService(account_name=self.account_name,
account_key=self.account_key)
@backoff
def read_input_file(self, input_path, local_path):
"""
Get input from Azure.
"""
self.__connect()
RealtimeLogger.debug("Loading {} from AzureIOStore".format(input_path))
# Download the blob. This is known to be synchronous, although it can
# call a callback during the process.
self.connection.get_blob_to_path(self.container_name,
self.name_prefix + input_path,
local_path)
def list_input_directory(self,
input_path,
recursive=False,
with_times=False):
"""
Loop over fake /-delimited directories on Azure. The prefix may or may
not not have a trailing slash; if not, one will be added automatically.
Returns the names of files and fake directories in the given input fake
directory, non-recursively.
If with_times is specified, will yield (name, time) pairs including
modification times as datetime objects. Times on directories are None.
"""
self.__connect()
RealtimeLogger.info(
"Enumerating {} from AzureIOStore".format(input_path))
# Work out what the directory name to list is
fake_directory = self.name_prefix + input_path
if fake_directory != "" and not fake_directory.endswith("/"):
# We have a nonempty prefix, and we need to end it with a slash
fake_directory += "/"
# This will hold the marker that we need to send back to get the next
# page, if there is one. See <http://stackoverflow.com/a/24303682>
marker = None
# This holds the subdirectories we found; we yield each exactly once if
# we aren't recursing.
subdirectories = set()
while True:
# Get the results from Azure. We don't use delimiter since Azure
# doesn't seem to provide the placeholder entries it's supposed to.
result = self.connection.list_blobs(self.container_name,
prefix=fake_directory,
marker=marker)
RealtimeLogger.info("Found {} files".format(len(result)))
for blob in result:
# Yield each result's blob name, but directory names only once
# Drop the common prefix
relative_path = blob.name[len(fake_directory):]
if (not recursive) and "/" in relative_path:
# We found a file in a subdirectory, and we aren't supposed
# to be recursing.
subdirectory, _ = relative_path.split("/", 1)
if subdirectory not in subdirectories:
# It's a new subdirectory. Yield and remember it
subdirectories.add(subdirectory)
if with_times:
yield subdirectory, None
else:
yield subdirectory
else:
# We found an actual file
if with_times:
mtime = blob.properties.last_modified
if isinstance(mtime, datetime.datetime):
# Make sure we're getting proper localized datetimes
# from the new Azure Storage API.
assert (mtime.tzinfo is not None and
mtime.tzinfo.utcoffset(mtime) is not None)
else:
# Convert mtime from a string as in the old API.
mtime = dateutil.parser.parse(mtime).replace(
tzinfo=dateutil.tz.tzutc())
yield relative_path, mtime
else:
yield relative_path
# Save the marker
marker = result.next_marker
if not marker:
break
@backoff
def write_output_file(self, local_path, output_path):
"""
Write output to Azure. Will create the container if necessary.
"""
self.__connect()
RealtimeLogger.debug("Saving {} to AzureIOStore".format(output_path))
try:
# Make the container
self.connection.create_container(self.container_name)
except azure.WindowsAzureConflictError:
# The container probably already exists
pass
# Upload the blob (synchronously)
# TODO: catch no container error here, make the container, and retry
self.connection.put_block_blob_from_path(
self.container_name, self.name_prefix + output_path, local_path)
@backoff
def exists(self, path):
"""
Returns true if the given input or output file exists in Azure already.
"""
self.__connect()
marker = None
while True:
try:
# Make the container
self.connection.create_container(self.container_name)
except azure.WindowsAzureConflictError:
# The container probably already exists
pass
# Get the results from Azure.
result = self.connection.list_blobs(self.container_name,
prefix=self.name_prefix + path,
marker=marker)
for blob in result:
# Look at each blob
if blob.name == self.name_prefix + path:
# Found it
return True
# Save the marker
marker = result.next_marker
if not marker:
break
return False
@backoff
def get_mtime(self, path):
"""
Returns the modification time of the given blob if it exists, or None
otherwise.
"""
self.__connect()
marker = None
while True:
# Get the results from Azure.
result = self.connection.list_blobs(self.container_name,
prefix=self.name_prefix + path,
marker=marker)
for blob in result:
# Look at each blob
if blob.name == self.name_prefix + path:
# Found it
mtime = blob.properties.last_modified
if isinstance(mtime, datetime.datetime):
# Make sure we're getting proper localized datetimes
# from the new Azure Storage API.
assert (mtime.tzinfo is not None
and mtime.tzinfo.utcoffset(mtime) is not None)
else:
# Convert mtime from a string as in the old API.
mtime = dateutil.parser.parse(mtime).replace(
tzinfo=dateutil.tz.tzutc())
return mtime
# Save the marker
marker = result.next_marker
if not marker:
break
return None
@backoff
def get_size(self, path):
"""
Returns the size in bytes of the given blob if it exists, or None
otherwise.
"""
self.__connect()
marker = None
while True:
# Get the results from Azure.
result = self.connection.list_blobs(self.container_name,
prefix=self.name_prefix + path,
marker=marker)
for blob in result:
# Look at each blob
if blob.name == self.name_prefix + path:
# Found it
size = blob.properties.content_length
return size
# Save the marker
marker = result.next_marker
if not marker:
break
return None
class AzureStorage(Storage):
account_name = settings.AZURE_ACCOUNT_NAME
account_key = settings.AZURE_ACCOUNT_KEY
azure_container = settings.AZURE_CONTAINER
def __init__(self, *args, **kwargs):
super(AzureStorage, self).__init__(*args, **kwargs)
self._connection = None
@property
def connection(self):
if self._connection is None:
# Create connection
self._connection = BlobService(self.account_name, self.account_key)
# Create container if needed
containers = [
c for c in self._connection.list_containers(
prefix=self.azure_container)
if c.name == self.azure_container
]
if len(containers) == 0:
self._connection.create_container(
self.azure_container,
{'origin': 'created by Django web app'},
fail_on_exist=True)
return self._connection
def _open(self, name, mode="rb"):
stream = SimpleUploadedFile(name, None)
self.connection.get_blob_to_file(self.azure_container, name, stream)
stream.seek(0)
return stream
def exists(self, name):
try:
self.connection.get_blob_properties(self.azure_container, name)
except AzureMissingResourceHttpError:
return False
else:
return True
def delete(self, name):
self.connection.delete_blob(self.azure_container, name)
def size(self, name):
properties = self.connection.get_blob_properties(
self.azure_container, name)
return properties["content-length"]
def _save(self, name, content):
self.connection.put_block_blob_from_file(self.azure_container, name,
content)
return name
def url(self, name):
ap = AccessPolicy(expiry=(timezone.datetime.utcnow() + timezone.timedelta(seconds=600)).strftime('%Y-%m-%dT%H:%M:%SZ'), \
start=(timezone.datetime.utcnow() + timezone.timedelta(seconds=-600)).strftime('%Y-%m-%dT%H:%M:%SZ'), \
permission='r')
sap = SharedAccessPolicy(ap)
sas = SharedAccessSignature(self.account_name, self.account_key)
url = sas.generate_signed_query_string(path=self.azure_container +
'/' + name,
resource_type='b',
shared_access_policy=sap)
return self.connection.make_blob_url(self.azure_container,
name) + "?" + url
def bs(self):
if not self._bs:
self._bs = BlobService(self.get_storage_name(), self.get_key())
return self._bs
USERID = 'ENTER DB USERID'
PASSWORD = '******'
DB_DRIVER = 'SQL Server Native Client 11.0'
# Specify the Azure Storage Account name where you will have a private blob to copy in the CSV file
STORAGEACCOUNTNAME = "ENTER AZURE STORAGE ACCOUNT NAME"
# Sepcify the storage account key.
# You can retrieve it "Primary Access Key" found on Azure portal Storage account blade by clicking on the "Key" icon.
# More info: https://azure.microsoft.com/en-us/documentation/articles/storage-create-storage-account/#manage-your-storage-access-keys
STORAGEKEY = "ENTER STORAGE ACCOUNT KEY "
#Read dataset
#Dataset is read from a public blob and copied to a private blob to locad it into SQL DW via Polybase
f = urllib2.urlopen('https://cahandson.blob.core.windows.net/nyctaxi/nyctaxipoint1pct.csv')
taxisample = f.read()
blob_service = BlobService(account_name=STORAGEACCOUNTNAME, account_key=STORAGEKEY)
blob_service.create_container('nyctaxinb')
blob_service.put_block_blob_from_bytes(
'nyctaxinb',
'nyctaxipoint1pct.csv',
taxisample
)
# Construct the SQL DW Connection string
driver = 'DRIVER={' + DB_DRIVER + '}'
server = 'SERVER=' + SERVER_NAME
database = 'DATABASE=' + DATABASE_NAME
uid = 'UID=' + USERID
pwd = 'PWD=' + PASSWORD
CONNECTION_STRING = ';'.join([driver,server,database,uid,pwd, 'Encrypt=yes;TrustServerCertificate=no'])
print CONNECTION_STRING
from azure.storage.blob import BlobService
import datetime
import string
import base64
import urllib
from static.app_keys import blob_account_name, blob_account_key
#get accountName and accountKey from app_keys module
accountName = blob_account_name
accountKey = blob_account_key
#create the blob_service object which connects to the Azure Storage account
blob_service = BlobService(accountName, accountKey)
#flag variable to verify upload
uploaded = False
#uploadBlob takes in the username which is used for the storage container name
#file is the file to be uploaded
#filename is concatenated onto the URL for user readability
#token and secret are used for oAuth verification which must happen at every step.
def uploadBlob(username, file, filename):
#try:
global uploaded
username = username.lower()
returnList = []
#decode base64 image string
decodedFile = file.decode("base64")
blob_service.create_container(username, x_ms_blob_public_access="container")
#get current datetime in UTC for a completely unique identifier
import string
from unidecode import unidecode
from urllib import parse
from azure.storage.blob import BlobService
from datetime import datetime
import animesources
shows = []
with open('title-map.json') as titlemap_file:
titlemap = json.load(titlemap_file)
with open('multi-season.json') as multiseason_file:
multiseason = json.load(multiseason_file)
with open('azure.json') as azure_file:
azure_storage = json.load(azure_file)
azure_blob = BlobService(account_name=azure_storage['account'],
account_key=azure_storage['key'])
with open('proxies.json') as proxies_file:
proxy_data = json.load(proxies_file)
proxy = proxy_data['uk']
sources = [
animesources.Crunchyroll(titlemap, multiseason, 'uk', proxy),
animesources.Funimation(titlemap, multiseason, 'gb', proxy),
animesources.Netflix(titlemap, multiseason, 'uk', proxy),
animesources.Daisuki(titlemap, multiseason, 'uk', proxy),
animesources.Viewster(titlemap, multiseason, 'uk', proxy),
animesources.Animax(titlemap, multiseason, 'uk', proxy),
animesources.HiDive(titlemap, multiseason, 'uk', proxy)
]
for source in sources:
source.UpdateShowList(shows)
print(source.GetName() + ': ' + str(len(shows)))
print "=== Converting Images ==="
command = "ffmpeg -i {0} -r 25 -vf scale=640:-1 -q:v 9 -an -f image2 {1}".format(
filename, image_file_name)
os.system(command)
# if there is no video key return
if len(sys.argv) <= 2:
exit()
print "=== Uploading to Azure ==="
video_key = sys.argv[2]
# walk all files in dir and push to bucket
key = raw_input("Please enter azure vidoepath blob storage key: ")
blob_service = BlobService(account_name='videopathmobilefiles',
account_key=key)
basepath = os.path.dirname(os.path.abspath(__file__)) + "/" + output_folder
container_name = video_key.lower()
blob_service.create_container(container_name,
x_ms_blob_public_access='container')
# collect files for uploading
filepaths = []
for path, subdirs, files in os.walk(basepath):
for name in files:
# don't upload hidden files
if name[0] == ".":
continue
pathname = os.path.join(path, name)
filepaths.append(pathname)
def do_lzop_get(creds, url, path, decrypt, do_retry=True):
"""
Get and decompress a WABS URL
This streams the content directly to lzop; the compressed version
is never stored on disk.
"""
assert url.endswith('.lzo'), 'Expect an lzop-compressed file'
assert url.startswith('wabs://')
conn = BlobService(creds.account_name,
creds.account_key,
sas_token=creds.access_token,
protocol='https')
def log_wal_fetch_failures_on_error(exc_tup, exc_processor_cxt):
def standard_detail_message(prefix=''):
return (prefix + ' There have been {n} attempts to fetch wal '
'file {url} so far.'.format(n=exc_processor_cxt, url=url))
typ, value, tb = exc_tup
del exc_tup
# Screen for certain kinds of known-errors to retry from
if issubclass(typ, socket.error):
socketmsg = value[1] if isinstance(value, tuple) else value
logger.info(
msg='Retrying fetch because of a socket error',
detail=standard_detail_message(
"The socket error's message is '{0}'.".format(socketmsg)))
else:
# For all otherwise untreated exceptions, report them as a
# warning and retry anyway -- all exceptions that can be
# justified should be treated and have error messages
# listed.
logger.warning(
msg='retrying WAL file fetch from unexpected exception',
detail=standard_detail_message(
'The exception type is {etype} and its value is '
'{evalue} and its traceback is {etraceback}'.format(
etype=typ,
evalue=value,
etraceback=''.join(traceback.format_tb(tb)))))
# Help Python GC by resolving possible cycles
del tb
def download():
with files.DeleteOnError(path) as decomp_out:
with get_download_pipeline(PIPE, decomp_out.f, decrypt) as pl:
g = gevent.spawn(write_and_return_error, url, conn, pl.stdin)
try:
# Raise any exceptions guarded by
# write_and_return_error.
exc = g.get()
if exc is not None:
raise exc
except AzureMissingResourceHttpError:
# Short circuit any re-try attempts under certain race
# conditions.
pl.abort()
logger.warning(
msg=('could no longer locate object while '
'performing wal restore'),
detail=('The absolute URI that could not be '
'located is {url}.'.format(url=url)),
hint=('This can be normal when Postgres is trying '
'to detect what timelines are available '
'during restoration.'))
decomp_out.remove_regardless = True
return False
logger.info(
msg='completed download and decompression',
detail='Downloaded and decompressed "{url}" to "{path}"'.
format(url=url, path=path))
return True
if do_retry:
download = retry(
retry_with_count(log_wal_fetch_failures_on_error))(download)
return download()
class Azure(object):
'''
A class used to connect to the Azure storage and
upload/download files using blob storage
'''
def __init__(self, params={}):
'''
Constructor for the Azure object
'''
if "user" in params:
self.user = params["user"]
else:
self.user = None
if "key" in params:
self.key = params["key"]
else:
self.key = None
def connect(self, host, port, user, password, secure):
'''
Connect to the Azure service with given user and key
@param user - username to use to connect to
@param key - key to use to connect
'''
kwargs = {}
err = None
if not host is None:
kwargs["host_base"] = "." + host
if not user is None:
kwargs["account_name"] = user
elif not self.user is None:
kwargs["account_name"] = self.user
if not password is None:
kwargs["account_key"] = password
elif not self.key is None:
kwargs["account_key"] = self.key
kwargs["protocol"] = "https" if secure else "http"
try:
self.service = BlobService(**kwargs)
except Exception as e:
err = e.message
self.service = None
if self.service is None:
raise OsakaException("Failed to connect to Azure:" +
("" if err is None else err))
@classmethod
def getSchemes(clazz):
'''
Returns a list of schemes this handler handles
Note: handling the scheme of another handler produces unknown results
@returns list of handled schemes
'''
return ["azure", "azures"]
def close(self):
'''
Close this service
'''
pass
def put(self, path, url):
'''
Put a file up to the cloud
@param path - path to upload
@param url - path in cloud to upload too
'''
if os.path.isdir(path):
return walk(self.put, path, url)
cont, blob = get_container_and_path(urlparse.urlparse(url).path)
self.service.create_container(cont)
self.service.put_block_blob_from_path(cont, blob, path)
return True
def get(self, url, dest):
'''
Get file(s) from the cloud
@param url - url on cloud to pull down (on cloud)
@param dest - dest to download too
'''
cont, blob = get_container_and_path(urlparse.urlparse(url).path)
for b in self.service.list_blobs(cont, prefix=blob):
destination = os.path.join(dest, os.path.relpath(
b.name, blob)) if blob != b.name else dest
if not os.path.exists(os.path.dirname(destination)):
os.mkdir(os.path.dirname(destination))
self.service.get_blob_to_path(cont, b.name, destination)
return True
def rm(self, url):
'''
Remove this url and all children urls
@param url - url to remove
'''
cont, blob = get_container_and_path(urlparse.urlparse(url).path)
for b in self.service.list_blobs(cont, prefix=blob):
self.service.delete_blob(cont, b.name)
return True
def uri_put_file(creds, uri, fp, content_type=None):
assert fp.tell() == 0
assert uri.startswith('wabs://')
def log_upload_failures_on_error(exc_tup, exc_processor_cxt):
def standard_detail_message(prefix=''):
return (prefix + ' There have been {n} attempts to upload '
'file {url} so far.'.format(n=exc_processor_cxt, url=uri))
typ, value, tb = exc_tup
del exc_tup
# Screen for certain kinds of known-errors to retry from
if issubclass(typ, socket.error):
socketmsg = value[1] if isinstance(value, tuple) else value
logger.info(
msg='Retrying upload because of a socket error',
detail=standard_detail_message(
"The socket error's message is '{0}'.".format(socketmsg)))
else:
# For all otherwise untreated exceptions, report them as a
# warning and retry anyway -- all exceptions that can be
# justified should be treated and have error messages
# listed.
logger.warning(
msg='retrying file upload from unexpected exception',
detail=standard_detail_message(
'The exception type is {etype} and its value is '
'{evalue} and its traceback is {etraceback}'.format(
etype=typ,
evalue=value,
etraceback=''.join(traceback.format_tb(tb)))))
# Help Python GC by resolving possible cycles
del tb
# Because we're uploading in chunks, catch rate limiting and
# connection errors which occur for each individual chunk instead of
# failing the whole file and restarting.
@retry(retry_with_count(log_upload_failures_on_error))
def upload_chunk(chunk, block_id):
if isinstance(chunk, str):
chunk = chunk.encode('utf-8')
check_sum = base64.b64encode(md5(chunk).digest()).decode('utf-8')
conn.put_block(url_tup.netloc,
url_tup.path.lstrip('/'),
chunk,
block_id,
content_md5=check_sum)
url_tup = urlparse(uri)
kwargs = dict(x_ms_blob_type='BlockBlob')
if content_type is not None:
kwargs['x_ms_blob_content_type'] = content_type
conn = BlobService(creds.account_name,
creds.account_key,
sas_token=creds.access_token,
protocol='https')
conn.put_blob(url_tup.netloc, url_tup.path.lstrip('/'), b'', **kwargs)
# WABS requires large files to be uploaded in 4MB chunks
block_ids = []
length, index = 0, 0
pool_size = os.getenv('WABS_UPLOAD_POOL_SIZE', 5)
p = gevent.pool.Pool(size=pool_size)
while True:
data = fp.read(WABS_CHUNK_SIZE)
if data:
length += len(data)
block_id = base64.b64encode(
str(index).encode('utf-8')).decode('utf-8')
p.wait_available()
p.spawn(upload_chunk, data, block_id)
block_ids.append(block_id)
index += 1
else:
p.join()
break
conn.put_block_list(url_tup.netloc, url_tup.path.lstrip('/'), block_ids)
# To maintain consistency with the S3 version of this function we must
# return an object with a certain set of attributes. Currently, that set
# of attributes consists of only 'size'
return _Key(size=len(data))
def getanalysis(request):
assert isinstance(request, HttpRequest)
latstart = request.GET.get('latstart', None)
latend = request.GET.get('latend', None)
lonstart = request.GET.get('lonstart', None)
lonend = request.GET.get('lonend', None)
sea = request.GET.get('season', None)
#start SSH
ssh = paramiko.SSHClient()
blob_service = BlobService(
account_name='araldrift',
account_key=
'otLzzkwQHQD3xFTQxwxy64PCL6eDINWGjSB7x6Ta2XVw3+3ffI5O2MhAEavf/r8qIW4G/dKrZAVg1R64nK7hDQ=='
)
blob_service.get_blob_to_path('security', 'id_rsa', './id_rsa')
privkey = paramiko.RSAKey.from_private_key_file(filename='./id_rsa',
password='******')
ssh.set_missing_host_key_policy(paramiko.AutoAddPolicy())
try:
ssh.connect('40.112.209.249',
username='******',
password='******',
allow_agent=False,
pkey=None,
key_filename=None,
timeout=10,
look_for_keys=False,
compress=False)
except paramiko.SSHException:
return HttpResponse()
quit()
#stdin,stdout,stderr = ssh.exec_command("ls /etc/")
cmd = '/home/araldif/anaconda3/bin/python /datadrive/from_webapp/xarray_analysis.py ' + latstart + ' ' + latend + ' ' + lonstart + ' ' + lonend + ' ' + sea
#cmd = '/datadrive/from_webapp/xarray_analysis.py'
#cmd = 'python /datadrive/from_webapp/test.py ' + name
stdin, stdout, stderr = ssh.exec_command(cmd)
h = []
for line in stderr.readlines():
h.append(line)
stderr.close()
ssh.close()
try:
imageoutfile1 = 'prec_' + str(sea) + '_' + str(latstart) + '_' + str(
latend) + '_' + str(lonstart) + '_' + str(lonend) + '.png'
imageoutfile2 = './' + imageoutfile1
blob_service = BlobService(
account_name='araldrift',
account_key=
'otLzzkwQHQD3xFTQxwxy64PCL6eDINWGjSB7x6Ta2XVw3+3ffI5O2MhAEavf/r8qIW4G/dKrZAVg1R64nK7hDQ=='
)
blob_service.get_blob_to_path('flow', imageoutfile1, imageoutfile2)
image_data = open(imageoutfile2, "rb").read()
response = HttpResponse(image_data, content_type='image/png')
return response
except:
return HttpResponse(h, content_type='text/plain')
while not successful:
try:
blobService.put_blob(HEATMAP_CONTAINER,
row[KEY_FIELD],
json.dumps(tileTimePeriodResultSet),
"BlockBlob",
x_ms_blob_cache_control="max-age=3600",
x_ms_blob_content_type="application/json")
successful = True
except:
print "error putting heatmap: ", sys.exc_info()[0]
continue
yield None
blobService = BlobService(account_name=os.environ["LOCATION_STORAGE_ACCOUNT"],
account_key=os.environ["LOCATION_STORAGE_KEY"])
blobService.create_container(HEATMAP_CONTAINER)
blobService.set_container_acl(HEATMAP_CONTAINER,
x_ms_blob_public_access='container')
def check_config():
if not "LOCATION_STORAGE_ACCOUNT" in os.environ:
print "Required environment variable LOCATION_STORAGE_ACCOUNT missing."
if not "LOCATION_STORAGE_KEY" in os.environ:
print "Required environment variable LOCATION_STORAGE_KEY missing."
if not "LOCATIONS_ROOT" in os.environ:
print "Required environment variable LOCATIONS_ROOT missing."
