def validate(self, apiobj, method, api, param, safe):
"""Validating all the input parameter as enforced by the WMCore.REST module"""
authz_login_valid()
if method in ['PUT']:
#TODO check optional parameter
#TODO check all the regexp
validate_str("taskname", param, safe, RX_TASKNAME, optional=False)
validate_strlist("outfilelumis", param, safe, RX_LUMILIST)
validate_numlist("outfileruns", param, safe)
if len(safe.kwargs["outfileruns"]) != len(safe.kwargs["outfilelumis"]):
raise InvalidParameter("The number of runs and the number of lumis lists are different")
validate_strlist("inparentlfns", param, safe, RX_PARENTLFN)
validate_str("globalTag", param, safe, RX_GLOBALTAG, optional=True)
validate_str("jobid", param, safe, RX_JOBID, optional=True)
#TODO: for backward compatibility. Get rid of the pandajobid once all jobs using it are done (one month after this gets released)
#(The following five lines can be deleted)
validate_num("pandajobid", param, safe, optional=True)
if bool(safe.kwargs["jobid"]) == bool(safe.kwargs["pandajobid"]):
raise InvalidParameter("Only one among jobid and pandajobid should be set")
#Oracle/cx_oracle/python stack does not like None for numbers, even if they are nullable
if safe.kwargs["pandajobid"] == None:
safe.kwargs["pandajobid"] = 0
validate_num("outsize", param, safe, optional=False)
validate_str("publishdataname", param, safe, RX_PUBLISH, optional=False)
validate_str("appver", param, safe, RX_CMSSW, optional=False)
validate_str("outtype", param, safe, RX_OUTTYPES, optional=False)
validate_str("checksummd5", param, safe, RX_CHECKSUM, optional=False)
validate_str("checksumcksum", param, safe, RX_CHECKSUM, optional=False)
validate_str("checksumadler32", param, safe, RX_CHECKSUM, optional=False)
validate_str("outlocation", param, safe, RX_CMSSITE, optional=False)
validate_str("outtmplocation", param, safe, RX_CMSSITE, optional=False)
validate_str("acquisitionera", param, safe, RX_TASKNAME, optional=False)#TODO Do we really need this?
validate_str("outdatasetname", param, safe, RX_OUTDSLFN, optional=False)#TODO temporary, need to come up with a regex
validate_str("outlfn", param, safe, RX_PARENTLFN, optional=False)
validate_str("outtmplfn", param, safe, RX_PARENTLFN, optional=True)
validate_num("events", param, safe, optional=False)
validate_str("filestate", param, safe, RX_FILESTATE, optional=True)
validate_num("directstageout", param, safe, optional=True)
safe.kwargs["directstageout"] = 'T' if safe.kwargs["directstageout"] else 'F' #'F' if not provided
elif method in ['POST']:
validate_str("taskname", param, safe, RX_TASKNAME, optional=False)
validate_str("outlfn", param, safe, RX_LFN, optional=False)
validate_str("filestate", param, safe, RX_FILESTATE, optional=False)
elif method in ['GET']:
validate_str("taskname", param, safe, RX_TASKNAME, optional=False)
validate_str("filetype", param, safe, RX_OUTTYPES, optional=False)
validate_num("howmany", param, safe, optional=True)
elif method in ['DELETE']:
authz_operator()
validate_str("taskname", param, safe, RX_TASKNAME, optional=True)
validate_str("hours", param, safe, RX_HOURS, optional=True)
if bool(safe.kwargs["taskname"]) == bool(safe.kwargs["hours"]):
raise InvalidParameter("You have to specify a taskname or a number of hours. Files of this task or created before the number of hours"+\
" will be deleted. Only one of the two parameters can be specified.")
def validate(self, apiobj, method, api, param, safe):
"""Validating all the input parameter as enforced by the WMCore.REST module"""
authz_login_valid()
if method in ['PUT']:
#TODO check optional parameter
#TODO check all the regexp
validate_str("taskname", param, safe, RX_WORKFLOW, optional=False)
validate_strlist("outfilelumis", param, safe, RX_LUMILIST)
validate_numlist("outfileruns", param, safe)
if len(safe.kwargs["outfileruns"]) != len(safe.kwargs["outfilelumis"]):
raise InvalidParameter("The number of runs and the number of lumis lists are different")
validate_strlist("inparentlfns", param, safe, RX_PARENTLFN)
validate_str("globalTag", param, safe, RX_GLOBALTAG, optional=True)
validate_num("pandajobid", param, safe, optional=False)
validate_num("outsize", param, safe, optional=False)
validate_str("publishdataname", param, safe, RX_PUBLISH, optional=False)
validate_str("appver", param, safe, RX_CMSSW, optional=False)
validate_str("outtype", param, safe, RX_OUTTYPES, optional=False)
validate_str("checksummd5", param, safe, RX_CHECKSUM, optional=False)
validate_str("checksumcksum", param, safe, RX_CHECKSUM, optional=False)
validate_str("checksumadler32", param, safe, RX_CHECKSUM, optional=False)
validate_str("outlocation", param, safe, RX_CMSSITE, optional=False)
validate_str("outtmplocation", param, safe, RX_CMSSITE, optional=False)
validate_str("acquisitionera", param, safe, RX_WORKFLOW, optional=False)#TODO Do we really need this?
validate_str("outdatasetname", param, safe, RX_OUTDSLFN, optional=False)#TODO temporary, need to come up with a regex
validate_str("outlfn", param, safe, RX_PARENTLFN, optional=False)
validate_str("outtmplfn", param, safe, RX_PARENTLFN, optional=True)
validate_num("events", param, safe, optional=False)
validate_str("filestate", param, safe, RX_FILESTATE, optional=True)
validate_num("directstageout", param, safe, optional=True)
safe.kwargs["directstageout"] = 'T' if safe.kwargs["directstageout"] else 'F' #'F' if not provided
elif method in ['POST']:
validate_str("taskname", param, safe, RX_WORKFLOW, optional=False)
validate_str("outlfn", param, safe, RX_LFN, optional=False)
validate_str("filestate", param, safe, RX_FILESTATE, optional=False)
elif method in ['GET']:
validate_str("taskname", param, safe, RX_WORKFLOW, optional=False)
validate_str("filetype", param, safe, RX_OUTTYPES, optional=False)
elif method in ['DELETE']:
authz_operator()
validate_str("taskname", param, safe, RX_WORKFLOW, optional=True)
validate_str("hours", param, safe, RX_HOURS, optional=True)
if bool(safe.kwargs["taskname"]) == bool(safe.kwargs["hours"]):
raise InvalidParameter("You have to specify a taskname or a number of hours. Files of this task or created before the number of hours"+\
" will be deleted. Only one of the two parameters can be specified.")
def validate(self, apiobj, method, api, param, safe):
"""Validating all the input parameter as enforced by the WMCore.REST module"""
authz_login_valid()
if method in ['PUT']:
#TODO check optional parameter
#TODO check all the regexp
validate_str("taskname", param, safe, RX_TASKNAME, optional=False)
validate_strlist("outfilelumis", param, safe, RX_LUMILIST)
validate_numlist("outfileruns", param, safe)
if len(safe.kwargs["outfileruns"]) != len(
safe.kwargs["outfilelumis"]):
raise InvalidParameter(
"The number of runs and the number of lumis lists are different"
)
validate_strlist("inparentlfns", param, safe, RX_PARENTLFN)
validate_str("globalTag", param, safe, RX_GLOBALTAG, optional=True)
validate_num("pandajobid", param, safe, optional=False)
validate_num("outsize", param, safe, optional=False)
validate_str("publishdataname",
param,
safe,
RX_PUBLISH,
optional=False)
validate_str("appver", param, safe, RX_CMSSW, optional=False)
validate_str("outtype", param, safe, RX_OUTTYPES, optional=False)
validate_str("checksummd5",
param,
safe,
RX_CHECKSUM,
optional=False)
validate_str("checksumcksum",
param,
safe,
RX_CHECKSUM,
optional=False)
validate_str("checksumadler32",
param,
safe,
RX_CHECKSUM,
optional=False)
validate_str("outlocation",
param,
safe,
RX_CMSSITE,
optional=False)
validate_str("outtmplocation",
param,
safe,
RX_CMSSITE,
optional=False)
validate_str("acquisitionera",
param,
safe,
RX_TASKNAME,
optional=False) #TODO Do we really need this?
validate_str(
"outdatasetname", param, safe, RX_OUTDSLFN,
optional=False) #TODO temporary, need to come up with a regex
validate_str("outlfn", param, safe, RX_PARENTLFN, optional=False)
validate_str("outtmplfn", param, safe, RX_PARENTLFN, optional=True)
validate_num("events", param, safe, optional=False)
validate_str("filestate", param, safe, RX_FILESTATE, optional=True)
validate_num("directstageout", param, safe, optional=True)
safe.kwargs["directstageout"] = 'T' if safe.kwargs[
"directstageout"] else 'F' #'F' if not provided
elif method in ['POST']:
validate_str("taskname", param, safe, RX_TASKNAME, optional=False)
validate_str("outlfn", param, safe, RX_LFN, optional=False)
validate_str("filestate",
param,
safe,
RX_FILESTATE,
optional=False)
elif method in ['GET']:
validate_str("taskname", param, safe, RX_TASKNAME, optional=False)
validate_str("filetype", param, safe, RX_OUTTYPES, optional=False)
elif method in ['DELETE']:
authz_operator()
validate_str("taskname", param, safe, RX_TASKNAME, optional=True)
validate_str("hours", param, safe, RX_HOURS, optional=True)
if bool(safe.kwargs["taskname"]) == bool(safe.kwargs["hours"]):
raise InvalidParameter("You have to specify a taskname or a number of hours. Files of this task or created before the number of hours"+\
" will be deleted. Only one of the two parameters can be specified.")
def validate(self, apiobj, method, api, param, safe):
"""Validating all the input parameter as enforced by the WMCore.REST module"""
authz_login_valid()
if method in ['PUT']:
#TODO check optional parameter
#TODO check all the regexp
validate_str("taskname", param, safe, RX_TASKNAME, optional=False)
validate_strlist("outfilelumis", param, safe, RX_LUMILIST)
validate_numlist("outfileruns", param, safe)
if len(safe.kwargs["outfileruns"]) != len(
safe.kwargs["outfilelumis"]):
raise InvalidParameter(
"The number of runs and the number of lumis lists are different"
)
validate_strlist("inparentlfns", param, safe, RX_PARENTLFN)
validate_str("globalTag", param, safe, RX_GLOBALTAG, optional=True)
validate_str("jobid", param, safe, RX_JOBID, optional=True)
#TODO: for backward compatibility. Get rid of the pandajobid once all jobs using it are done (one month after this gets released)
#(The following five lines can be deleted)
validate_num("pandajobid", param, safe, optional=True)
if bool(safe.kwargs["jobid"]) == bool(safe.kwargs["pandajobid"]):
raise InvalidParameter(
"Only one among jobid and pandajobid should be set")
#Oracle/cx_oracle/python stack does not like None for numbers, even if they are nullable
if safe.kwargs["pandajobid"] == None:
safe.kwargs["pandajobid"] = 0
validate_num("outsize", param, safe, optional=False)
validate_str("publishdataname",
param,
safe,
RX_PUBLISH,
optional=False)
validate_str("appver", param, safe, RX_CMSSW, optional=False)
validate_str("outtype", param, safe, RX_OUTTYPES, optional=False)
validate_str("checksummd5",
param,
safe,
RX_CHECKSUM,
optional=False)
validate_str("checksumcksum",
param,
safe,
RX_CHECKSUM,
optional=False)
validate_str("checksumadler32",
param,
safe,
RX_CHECKSUM,
optional=False)
validate_str("outlocation",
param,
safe,
RX_CMSSITE,
optional=False)
validate_str("outtmplocation",
param,
safe,
RX_CMSSITE,
optional=False)
validate_str("acquisitionera",
param,
safe,
RX_TASKNAME,
optional=False) #TODO Do we really need this?
validate_str(
"outdatasetname", param, safe, RX_OUTDSLFN,
optional=False) #TODO temporary, need to come up with a regex
validate_str("outlfn", param, safe, RX_PARENTLFN, optional=False)
validate_str("outtmplfn", param, safe, RX_PARENTLFN, optional=True)
validate_num("events", param, safe, optional=False)
validate_str("filestate", param, safe, RX_FILESTATE, optional=True)
validate_num("directstageout", param, safe, optional=True)
safe.kwargs["directstageout"] = 'T' if safe.kwargs[
"directstageout"] else 'F' #'F' if not provided
elif method in ['POST']:
validate_str("taskname", param, safe, RX_TASKNAME, optional=False)
validate_str("outlfn", param, safe, RX_LFN, optional=False)
validate_str("filestate",
param,
safe,
RX_FILESTATE,
optional=False)
elif method in ['GET']:
validate_str("taskname", param, safe, RX_TASKNAME, optional=False)
validate_str("filetype", param, safe, RX_OUTTYPES, optional=False)
validate_num("howmany", param, safe, optional=True)
validate_strlist("lfn", param, safe, RX_LFN)
elif method in ['DELETE']:
authz_operator()
validate_str("taskname", param, safe, RX_TASKNAME, optional=True)
validate_str("hours", param, safe, RX_HOURS, optional=True)
if bool(safe.kwargs["taskname"]) == bool(safe.kwargs["hours"]):
raise InvalidParameter("You have to specify a taskname or a number of hours. Files of this task or created before the number of hours"+\
" will be deleted. Only one of the two parameters can be specified.")
def get(self, subresource, objecttype, taskname, username, tarballname): # pylint: disable=redefined-builtin
"""
:arg str subresource: the specific information to be accessed;
"""
authenticatedUserName = cherrypy.request.user[
'login'] # the username of who's calling
# a bit of code common to 3 subresource's: validate args and prepare the s3_objectKey inside the bucket
if subresource in ['upload', 'retrieve', 'download']:
if not objecttype:
raise MissingParameter("objecttype is missing")
if objecttype == 'sandbox':
if not tarballname:
raise MissingParameter("tarballname is missing")
ownerName = authenticatedUserName if subresource == 'upload' else username
# sandbox goes in bucket/username/sandboxes/
objectPath = ownerName + '/sandboxes/' + tarballname
else:
if not taskname:
raise MissingParameter("takskname is missing")
ownerName = getUsernameFromTaskname(taskname)
# task related files go in bucket/username/taskname/
objectPath = ownerName + '/' + taskname + '/' + objecttype
s3_objectKey = fromNewBytesToString(objectPath)
if subresource == 'upload':
# returns a dictionary with the information to upload a file with a POST
# via a "PreSigned URL". It can return an empty string '' as URL to indicate that
# a sandbox upload request refers to an existing object with same name
# WMCore REST does not allow to return None
authz_operator(username=ownerName, group='crab3', role='operator')
if objecttype == 'sandbox':
# we only upload same sandbox once
alreadyThere = False
try:
# from https://stackoverflow.com/a/38376288
self.s3_client.head_object(Bucket=self.s3_bucket,
Key=s3_objectKey)
alreadyThere = True
except ClientError:
pass
if alreadyThere:
return ["", {}] # this tells client not to upload
expiration = 60 * 60 # 1 hour is good for retries and debugging
try:
response = self.s3_client.generate_presigned_post(
self.s3_bucket, s3_objectKey, ExpiresIn=expiration)
# this returns a dictionary like:
# {'url': u'https://s3.cern.ch/bucket1',
# 'fields': {'policy': u'eyJjb ... jEzWiJ9', # policy is a 164-char-long string
# 'AWSAccessKeyId': u'5d4270f1e022442783646c34cf552d55',
# 'key': objectPath, 'signature': u'pm58cUqxNQHBZXS1B/Er6P89IhU='}}
# need to build a single URL string to return
preSignedUrl = response
except ClientError as e:
raise ExecutionError("Connection to s3.cern.ch failed:\n%s" %
str(e))
# somehow it does not work to return preSignedUrl as a single object
return [preSignedUrl['url'], preSignedUrl['fields']]
if subresource == 'download':
authz_operator(username=ownerName, group='crab3', role='operator')
if subresource == 'sandbox' and not username:
raise MissingParameter("username is missing")
# returns a PreSignedUrl to download the file within the expiration time
expiration = 60 * 60 # 1 hour default is good for retries and debugging
if objecttype in ['debugfiles', 'clientlog', 'twlog']:
expiration = 60 * 60 * 24 * 30 # for logs make url valid as long as we keep files (1 month)
try:
response = self.s3_client.generate_presigned_url(
'get_object',
Params={
'Bucket': self.s3_bucket,
'Key': s3_objectKey
},
ExpiresIn=expiration)
preSignedUrl = response
except ClientError as e:
raise ExecutionError("Connection to s3.cern.ch failed:\n%s" %
str(e))
return preSignedUrl
if subresource == 'retrieve':
# download from S3 into a temporary file, read it, and return content to caller
authz_operator(username=ownerName, group='crab3', role='operator')
tempFile = '/tmp/boto.' + uuid.uuid4().hex
try:
self.s3_client.download_file(self.s3_bucket, s3_objectKey,
tempFile)
except ClientError as e:
raise ExecutionError("Connection to s3.cern.ch failed:\n%s" %
str(e))
with open(tempFile) as f:
txt = f.read()
os.remove(tempFile)
return txt
if subresource == 'list':
# list all files (aka objects, aka keys in S3 lingo) for a given usermame
# if arg objecttype is present, returns only the file names for that objecttype
if not username:
raise MissingParameter('username is missing')
# In S3 we always need to retrieve all keys even if some filtering/compression
# will be applied before reporting, since there is a limit of 1K key per call,
# multiple calls will be needed, S3 paginators make that easy
# https://boto3.amazonaws.com/v1/documentation/api/latest/guide/paginators.html
# We use S3 prefix to limit retrieved list to a user, since in our buckets
# file keys always have the form <username>/... see:
# https://github.com/dmwm/CRABServer/wiki/CRABCache-replacement-with-S3#bucket-organization and
# https://docs.aws.amazon.com/AmazonS3/latest/userguide/using-prefixes.html
#
fileNames = []
paginator = self.s3_client.get_paginator('list_objects_v2')
user = fromNewBytesToString(username)
operation_parameters = {'Bucket': self.s3_bucket, 'Prefix': user}
page_iterator = paginator.paginate(**operation_parameters)
for page in page_iterator:
namesInPage = [
item['Key'].lstrip(user + '/') for item in page['Contents']
]
fileNames += namesInPage
if objecttype:
filteredFileNames = [f for f in fileNames if objecttype in f]
fileNames = filteredFileNames
return fileNames
if subresource == 'used':
# return space used by username, in MBytes (rounded to integer)
if not username:
raise MissingParameter('username is missing')
paginator = self.s3_client.get_paginator('list_objects_v2')
user = fromNewBytesToString(username)
operation_parameters = {'Bucket': self.s3_bucket, 'Prefix': user}
page_iterator = paginator.paginate(**operation_parameters)
# S3 records object size in bytes, see:
# https://boto3.amazonaws.com/v1/documentation/api/latest/reference/services/s3.html#S3.Client.list_objects_v2
usedBytes = 0
for page in page_iterator:
for item in page['Contents']:
usedBytes += item['Size']
usedMBytes = usedBytes // 1024 // 1024
# WMCore REST wants to return lists
return [usedMBytes]
def validate(self, apiobj, method, api, param, safe):
"""Validating all the input parameter as enforced by the WMCore.REST module"""
authz_login_valid()
if method in ['PUT']:
validate_str("taskname", param, safe, RX_TASKNAME, optional=False)
validate_strlist("outfilelumis", param, safe, RX_LUMILIST)
validate_strlist("outfileruns", param, safe, RX_RUNS)
if len(safe.kwargs["outfileruns"]) != len(
safe.kwargs["outfilelumis"]):
raise InvalidParameter(
"The number of runs and the number of lumis lists are different"
)
validate_strlist("inparentlfns", param, safe, RX_PARENTLFN)
# inparentlfns will be inserted in Oracle as CLOB, so it must be a string
safe.kwargs['inparentlfns'] = str(safe.kwargs['inparentlfns'])
validate_str("globalTag", param, safe, RX_GLOBALTAG, optional=True)
validate_str("jobid", param, safe, RX_JOBID, optional=True)
safe.kwargs["pandajobid"] = 0
validate_num("outsize", param, safe, optional=False)
validate_str("publishdataname",
param,
safe,
RX_PUBLISH,
optional=False)
validate_str("appver", param, safe, RX_CMSSW, optional=False)
validate_str("outtype", param, safe, RX_OUTTYPES, optional=False)
validate_str("checksummd5",
param,
safe,
RX_CHECKSUM,
optional=False)
validate_num("checksumcksum", param, safe, optional=False)
validate_str("checksumadler32",
param,
safe,
RX_CHECKSUM,
optional=False)
validate_str("outlocation",
param,
safe,
RX_CMSSITE,
optional=False)
validate_str("outtmplocation",
param,
safe,
RX_CMSSITE,
optional=False)
validate_str("acquisitionera",
param,
safe,
RX_TASKNAME,
optional=False)
validate_str("outdatasetname",
param,
safe,
RX_OUTDSLFN,
optional=False)
# need to use RX_PARENTLFN becasue same API is also used for input metadata
validate_str("outlfn", param, safe, RX_PARENTLFN, optional=False)
validate_str("outtmplfn", param, safe, RX_LFN, optional=True)
validate_num("events", param, safe, optional=False)
validate_str("filestate", param, safe, RX_FILESTATE, optional=True)
validate_num("directstageout", param, safe, optional=True)
safe.kwargs["directstageout"] = 'T' if safe.kwargs[
"directstageout"] else 'F' #'F' if not provided
elif method in ['POST']:
validate_str("taskname", param, safe, RX_TASKNAME, optional=False)
validate_str("outlfn", param, safe, RX_LFN, optional=False)
validate_str("filestate",
param,
safe,
RX_FILESTATE,
optional=False)
elif method in ['GET']:
validate_str("taskname", param, safe, RX_TASKNAME, optional=False)
validate_str("filetype", param, safe, RX_OUTTYPES, optional=False)
validate_num("howmany", param, safe, optional=True)
validate_strlist("lfn", param, safe, RX_LFN)
elif method in ['DELETE']:
authz_operator()
validate_str("taskname", param, safe, RX_TASKNAME, optional=True)
validate_str("hours", param, safe, RX_HOURS, optional=True)
if bool(safe.kwargs["taskname"]) == bool(safe.kwargs["hours"]):
raise InvalidParameter("You have to specify a taskname or a number of hours. Files of this task or created before the number of hours"+\
" will be deleted. Only one of the two parameters can be specified.")
