def generate(vo):
# we need to call list_replicas before starting to reply
# otherwise the exceptions won't be propagated correctly
first = metalink
# then, stream the replica information
for rfile in list_replicas(dids=dids, schemes=schemes, vo=vo):
if first and metalink:
# first, set the appropriate content type, and stream the header
yield '<?xml version="1.0" encoding="UTF-8"?>\n<metalink xmlns="urn:ietf:params:xml:ns:metalink">\n'
first = False
replicas = []
dictreplica = {}
for rse in rfile['rses']:
for replica in rfile['rses'][rse]:
replicas.append(replica)
dictreplica[replica] = rse
replicas = sort_replicas(dictreplica,
client_location,
selection=select)
if not metalink:
yield dumps(rfile) + '\n'
else:
yield ' <file name="' + rfile['name'] + '">\n'
yield ' <identity>' + rfile['scope'] + ':' + rfile[
'name'] + '</identity>\n'
if rfile['adler32'] is not None:
yield ' <hash type="adler32">' + rfile[
'adler32'] + '</hash>\n'
if rfile['md5'] is not None:
yield ' <hash type="md5">' + rfile[
'md5'] + '</hash>\n'
yield ' <size>' + str(rfile['bytes']) + '</size>\n'
yield f' <glfn name="/atlas/rucio/{rfile["scope"]}:{rfile["name"]}">'
yield '</glfn>\n'
idx = 0
for replica in replicas:
yield ' <url location="' + str(
dictreplica[replica]) + '" priority="' + str(
idx +
1) + '">' + escape(replica) + '</url>\n'
idx += 1
if limit and limit == idx:
break
yield ' </file>\n'
if metalink:
if first:
# if still first output, i.e. there were no replicas
yield '<?xml version="1.0" encoding="UTF-8"?>\n<metalink xmlns="urn:ietf:params:xml:ns:metalink">\n</metalink>\n'
else:
# don't forget to send the metalink footer
yield '</metalink>\n'
def generate():
# first, set the appropriate content type, and stream the header
yield '<?xml version="1.0" encoding="UTF-8"?>\n<metalink xmlns="urn:ietf:params:xml:ns:metalink">\n'
# iteratively stream the XML per file
for rfile in itertools.chain((first, ), replicas_iter):
replicas = []
dictreplica = {}
for rse in rfile['rses']:
for replica in rfile['rses'][rse]:
replicas.append(replica)
dictreplica[replica] = rse
# stream metadata
yield ' <file name="' + rfile['name'] + '">\n'
yield ' <identity>' + rfile['scope'] + ':' + rfile[
'name'] + '</identity>\n'
if rfile['adler32'] is not None:
yield ' <hash type="adler32">' + rfile[
'adler32'] + '</hash>\n'
if rfile['md5'] is not None:
yield ' <hash type="md5">' + rfile['md5'] + '</hash>\n'
yield ' <size>' + str(rfile['bytes']) + '</size>\n'
yield ' <glfn name="/atlas/rucio/%s:%s">' % (
rfile['scope'], rfile['name'])
yield '</glfn>\n'
replicas = sort_replicas(dictreplica,
client_location,
selection=sortby)
# stream URLs
idx = 1
for replica in replicas:
yield ' <url location="' + str(
dictreplica[replica]) + '" priority="' + str(
idx) + '">' + replica + '</url>\n'
idx += 1
yield ' </file>\n'
# don't forget to send the metalink footer
yield '</metalink>\n'
def _list_and_sort_replicas(request_id, issuer, vo):
# we need to call list_replicas before starting to reply
# otherwise the exceptions won't be propagated correctly
for rfile in list_replicas(
dids=dids,
schemes=schemes,
unavailable=unavailable,
request_id=request_id,
ignore_availability=ignore_availability,
all_states=all_states,
rse_expression=rse_expression,
client_location=client_location,
domain=domain,
signature_lifetime=signature_lifetime,
resolve_archives=resolve_archives,
resolve_parents=resolve_parents,
nrandom=nrandom,
updated_after=updated_after,
issuer=issuer,
vo=vo):
# Sort rfile['pfns'] and limit its size according to "limit" parameter
lanreplicas = {}
wanreplicas = {}
for pfn, replica in rfile['pfns'].items():
replica_tuple = (replica['domain'],
replica['priority'], replica['rse'],
replica['client_extract'])
if replica_tuple[0] == 'lan':
lanreplicas[pfn] = replica_tuple
else:
wanreplicas[pfn] = replica_tuple
rfile['pfns'] = dict(
_sorted_with_priorities(
replicas=rfile['pfns'],
# Lan replicas sorted by priority; followed by wan replicas sorted by selection criteria
sorted_pfns=chain(
sorted(lanreplicas.keys(),
key=lambda pfn: lanreplicas[pfn][1]),
sort_replicas(wanreplicas,
client_location,
selection=select)),
limit=limit))
yield rfile
def _list_and_sort_replicas(vo):
# we need to call list_replicas before starting to reply
# otherwise the exceptions won't be propagated correctly
for rfile in list_replicas(dids=dids, schemes=schemes, vo=vo):
replicas = []
dictreplica = {}
for rse in rfile['rses']:
for replica in rfile['rses'][rse]:
replicas.append(replica)
dictreplica[replica] = rse
replicas = sort_replicas(dictreplica,
client_location,
selection=select)
rfile['pfns'] = dict(
_sorted_with_priorities(rfile['pfns'],
replicas,
limit=limit))
yield rfile
def generate(request_id, issuer, vo):
# we need to call list_replicas before starting to reply
# otherwise the exceptions won't be propagated correctly
first = metalink
for rfile in list_replicas(dids=dids, schemes=schemes,
unavailable=unavailable,
request_id=request_id,
ignore_availability=ignore_availability,
all_states=all_states,
rse_expression=rse_expression,
client_location=client_location,
domain=domain, signature_lifetime=signature_lifetime,
resolve_archives=resolve_archives,
resolve_parents=resolve_parents,
updated_after=updated_after,
issuer=issuer,
vo=vo):
# in first round, set the appropriate content type, and stream the header
if first and metalink:
yield '<?xml version="1.0" encoding="UTF-8"?>\n<metalink xmlns="urn:ietf:params:xml:ns:metalink">\n'
first = False
if not metalink:
yield dumps(rfile, cls=APIEncoder) + '\n'
else:
replicas = []
dictreplica = {}
for replica in rfile['pfns'].keys():
replicas.append(replica)
dictreplica[replica] = (rfile['pfns'][replica]['domain'],
rfile['pfns'][replica]['priority'],
rfile['pfns'][replica]['rse'],
rfile['pfns'][replica]['client_extract'])
yield ' <file name="' + rfile['name'] + '">\n'
if 'parents' in rfile and rfile['parents']:
yield ' <parents>\n'
for parent in rfile['parents']:
yield ' <did>' + parent + '</did>\n'
yield ' </parents>\n'
yield ' <identity>' + rfile['scope'] + ':' + rfile['name'] + '</identity>\n'
if rfile['adler32'] is not None:
yield ' <hash type="adler32">' + rfile['adler32'] + '</hash>\n'
if rfile['md5'] is not None:
yield ' <hash type="md5">' + rfile['md5'] + '</hash>\n'
yield ' <size>' + str(rfile['bytes']) + '</size>\n'
policy_schema = config_get('policy', 'schema', raise_exception=False, default='generic')
yield f' <glfn name="/{policy_schema}/rucio/{rfile["scope"]}:{rfile["name"]}"></glfn>\n'
lanreplicas = [replica for replica, v in dictreplica.items() if v[0] == 'lan']
# sort lan by priority
lanreplicas.sort(key=lambda rep: dictreplica[rep][1])
replicas = lanreplicas + sort_replicas({k: v for k, v in dictreplica.items() if v[0] != 'lan'}, client_location, selection=select)
for idx, replica in enumerate(replicas, start=1):
yield ' <url location="' + str(dictreplica[replica][2]) \
+ '" domain="' + str(dictreplica[replica][0]) \
+ '" priority="' + str(idx) \
+ '" client_extract="' + str(dictreplica[replica][3]).lower() \
+ '">' + escape(replica) + '</url>\n'
if limit and limit == idx:
break
yield ' </file>\n'
if metalink:
if first:
# if still first output, i.e. there were no replicas
yield '<?xml version="1.0" encoding="UTF-8"?>\n<metalink xmlns="urn:ietf:params:xml:ns:metalink">\n</metalink>\n'
else:
# don't forget to send the metalink footer
yield '</metalink>\n'
def get(self, scope_name):
"""
---
summary: Header redirect
description: Get the header redirect.
tags:
- Redirect
parameters:
- name: scope_name
in: path
description: The data identifier (scope)/(name).
schema:
type: string
style: simple
- name: ip
in: query
description: The client ip.
schema:
type: string
style: simple
required: false
- name: fqdn
in: query
schema:
type: string
style: simple
required: false
- name: site
in: query
schema:
type: string
style: simple
required: false
- name: schemes
in: query
schema:
type: array
style: simple
required: false
- name: select
in: query
schema:
type: string
style: simple
required: false
- name: sort
in: query
schema:
type: string
style: simple
required: false
- name: rse
in: query
schema:
type: string
style: simple
required: false
responses:
303:
description: OK
content:
application/json:
schema:
description: The redirect url.
type: string
401:
description: Invalid Auth Token
404:
description: Rse or did not found
"""
headers = self.get_headers()
try:
scope, name = parse_scope_name(scope_name,
extract_vo(request.headers))
except ValueError as error:
return generate_http_error_flask(400, error, headers=headers)
try:
client_ip = request.headers.get('X-Forwarded-For',
default=request.remote_addr)
client_location = {
'ip': request.args.get('ip', default=client_ip),
'fqdn': request.args.get('fqdn', default=None),
'site': request.args.get('site', default=None),
}
# use the default HTTP protocols if no scheme is given
schemes = request.args.getlist('schemes') or [
'davs', 'https', 's3'
]
sortby = request.args.get('select', default='random')
sortby = request.args.get('sort', default=sortby)
rse = request.args.get('rse', default=None)
site = request.args.get('site', default=None)
# correctly forward the schemes and select to potential metalink followups
cleaned_url = request.environ.get('REQUEST_URI').split('?')[0]
headers.set(
'Link',
f'<{cleaned_url}/metalink?schemes={",".join(schemes)}&select={sortby}>; rel=describedby; type="application/metalink+xml"'
)
# get vo if given
vo = extract_vo(request.headers)
replicas = list(
list_replicas(dids=[{
'scope': scope,
'name': name,
'type': 'FILE'
}],
schemes=schemes,
client_location=client_location,
vo=vo))
selected_url = None
for r in replicas:
if r['rses']:
dictreplica = {}
if rse:
if rse in r['rses'] and r['rses'][rse]:
selected_url = r['rses'][rse][0]
else:
return 'no redirection possible - no valid RSE for HTTP redirection found', 404, headers
else:
for rep in r['rses']:
for replica in r['rses'][rep]:
# since this is HTTP-only redirection, and to ensure compatibility with as many http clients as possible
# forcibly replacement davs and s3 URLs to https
replica = replica.replace(
'davs://',
'https://').replace('s3://', 'https://')
dictreplica[replica] = rep
if not dictreplica:
return 'no redirection possible - no valid RSE for HTTP redirection found', 404, headers
elif site:
rep = site_selector(dictreplica, site, vo)
if rep:
selected_url = rep[0]
else:
return 'no redirection possible - no valid RSE for HTTP redirection found', 404, headers
else:
rep = sort_replicas(dictreplica,
client_location,
selection=sortby)
selected_url = rep[0]
if selected_url:
response = redirect(selected_url, code=303)
response.headers.extend(headers)
return response
return 'no redirection possible - file does not exist', 404, headers
except ReplicaNotFound as error:
return generate_http_error_flask(404, error, headers=headers)
def get(self, scope_name):
"""
Header Redirect
.. :quickref: HeaderRedirector; Header redirect.
:param scope_name: data identifier (scope)/(name).
:resheader Content-Type: application/metalink+xml'.
:status 303: Redirect.
:status 401: Invalid Auth Token.
:status 404: RSE Not Found.
:status 404: DID Not Found.
:status 500: Internal Error.
"""
headers = Headers()
headers.set('Access-Control-Allow-Origin',
request.environ.get('HTTP_ORIGIN'))
headers.set('Access-Control-Allow-Headers',
request.environ.get('HTTP_ACCESS_CONTROL_REQUEST_HEADERS'))
headers.set('Access-Control-Allow-Methods', '*')
headers.set('Access-Control-Allow-Credentials', 'true')
try:
scope, name = parse_scope_name(
scope_name, request.headers.get('X-Rucio-VO', default='def'))
except ValueError as error:
return generate_http_error_flask(400,
'ValueError',
error.args[0],
headers=headers)
except Exception as error:
logging.exception("Internal Error")
return str(error), 500, headers
try:
# use the default HTTP protocols if no scheme is given
select, rse, site, schemes = 'random', None, None, [
'davs', 'http', 'https'
]
client_ip = request.headers.get('X-Forwarded-For',
default=request.remote_addr)
client_location = {'ip': client_ip, 'fqdn': None, 'site': None}
if request.query_string:
query_string = request.query_string.decode(encoding='utf-8')
params = parse_qs(query_string)
if 'select' in params:
select = params['select'][0]
if 'sort' in params:
select = params['sort'][0]
if 'rse' in params:
rse = params['rse'][0]
if 'site' in params:
site = params['site'][0]
if 'schemes' in params:
schemes = params['schemes'][0]
else:
schemes = ['davs', 'https', 's3']
if 'ip' in params:
client_location['ip'] = params['ip'][0]
if 'fqdn' in params:
client_location['fqdn'] = params['fqdn'][0]
if 'site' in params:
client_location['site'] = params['site'][0]
# correctly forward the schemes and select to potential metalink followups
cleaned_url = request.environ.get('REQUEST_URI').split('?')[0]
if isinstance(schemes, list):
headers.set(
'Link',
'<%s/metalink?schemes=%s&select=%s>; rel=describedby; type="application/metalink+xml"'
% (cleaned_url, ','.join(schemes), select))
else:
headers.set(
'Link',
'<%s/metalink?schemes=%s&select=%s>; rel=describedby; type="application/metalink+xml"'
% (cleaned_url, schemes, select))
schemes = [schemes] # list_replicas needs a list
# get vo if given
vo = request.headers.get('X-Rucio-VO', default='def')
replicas = [
r for r in list_replicas(dids=[{
'scope': scope,
'name': name,
'type': 'FILE'
}],
schemes=schemes,
client_location=client_location,
vo=vo)
]
selected_url = None
for r in replicas:
if r['rses']:
dictreplica = {}
if rse:
if rse in r['rses'] and r['rses'][rse]:
selected_url = r['rses'][rse][0]
else:
return 'no redirection possible - no valid RSE for HTTP redirection found', 404, headers
else:
for rep in r['rses']:
for replica in r['rses'][rep]:
# since this is HTTP-only redirection, and to ensure compatibility with as many http clients as possible
# forcibly replacement davs and s3 URLs to https
replica = replica.replace(
'davs://',
'https://').replace('s3://', 'https://')
dictreplica[replica] = rep
if not dictreplica:
return 'no redirection possible - no valid RSE for HTTP redirection found', 404, headers
elif site:
rep = site_selector(dictreplica, site, vo)
if rep:
selected_url = rep[0]
else:
return 'no redirection possible - no valid RSE for HTTP redirection found', 404, headers
else:
rep = sort_replicas(dictreplica,
client_location,
selection=select)
selected_url = rep[0]
if selected_url:
response = redirect(selected_url, code=303)
response.headers.extend(headers)
return response
return 'no redirection possible - file does not exist', 404, headers
except ReplicaNotFound as error:
return generate_http_error_flask(404,
'ReplicaNotFound',
error.args[0],
headers=headers)
except RucioException as error:
return generate_http_error_flask(500,
error.__class__.__name__,
error.args[0],
headers=headers)
except Exception as error:
logging.exception("Internal Error")
return str(error), 500, headers
def generate(request_id, issuer, vo):
# we need to call list_replicas before starting to reply
# otherwise the exceptions won't be propagated correctly
first = True
for rfile in list_replicas(
dids=dids,
schemes=schemes,
unavailable=unavailable,
request_id=request_id,
ignore_availability=ignore_availability,
all_states=all_states,
rse_expression=rse_expression,
client_location=client_location,
domain=domain,
signature_lifetime=signature_lifetime,
resolve_archives=resolve_archives,
resolve_parents=resolve_parents,
nrandom=nrandom,
updated_after=updated_after,
issuer=issuer,
vo=vo):
# Sort rfile['pfns'] and limit its size according to "limit" parameter
lanreplicas = {}
wanreplicas = {}
for pfn, replica in rfile['pfns'].items():
replica_tuple = (replica['domain'],
replica['priority'], replica['rse'],
replica['client_extract'])
if replica_tuple[0] == 'lan':
lanreplicas[pfn] = replica_tuple
else:
wanreplicas[pfn] = replica_tuple
rfile['pfns'] = dict(
_sorted_with_priorities(
replicas=rfile['pfns'],
# Lan replicas sorted by priority; followed by wan replicas sorted by selection criteria
sorted_pfns=chain(
sorted(lanreplicas.keys(),
key=lambda pfn: lanreplicas[pfn][1]),
sort_replicas(wanreplicas,
client_location,
selection=select)),
limit=limit))
if not metalink:
yield dumps(rfile, cls=APIEncoder) + '\n'
else:
# in first round, set the appropriate content type, and stream the header
if first:
yield '<?xml version="1.0" encoding="UTF-8"?>\n<metalink xmlns="urn:ietf:params:xml:ns:metalink">\n'
first = False
yield ' <file name="' + rfile['name'] + '">\n'
if 'parents' in rfile and rfile['parents']:
yield ' <parents>\n'
for parent in rfile['parents']:
yield ' <did>' + parent + '</did>\n'
yield ' </parents>\n'
yield ' <identity>' + rfile['scope'] + ':' + rfile[
'name'] + '</identity>\n'
if rfile['adler32'] is not None:
yield ' <hash type="adler32">' + rfile[
'adler32'] + '</hash>\n'
if rfile['md5'] is not None:
yield ' <hash type="md5">' + rfile[
'md5'] + '</hash>\n'
yield ' <size>' + str(rfile['bytes']) + '</size>\n'
policy_schema = config_get('policy',
'schema',
raise_exception=False,
default='generic')
yield f' <glfn name="/{policy_schema}/rucio/{rfile["scope"]}:{rfile["name"]}"></glfn>\n'
for pfn, replica in rfile['pfns'].items():
yield ' <url location="' + str(replica['rse']) \
+ '" domain="' + str(replica['domain']) \
+ '" priority="' + str(replica['priority']) \
+ '" client_extract="' + str(replica['client_extract']).lower() \
+ '">' + escape(pfn) + '</url>\n'
yield ' </file>\n'
if metalink:
if first:
# if still first output, i.e. there were no replicas
yield '<?xml version="1.0" encoding="UTF-8"?>\n<metalink xmlns="urn:ietf:params:xml:ns:metalink">\n</metalink>\n'
else:
# don't forget to send the metalink footer
yield '</metalink>\n'
def get(self, scope_name):
"""
Header Redirect
.. :quickref: HeaderRedirector; Header redirect.
:param scope_name: data identifier (scope)/(name).
:resheader Content-Type: application/metalink+xml'.
:status 303: Redirect.
:status 401: Invalid Auth Token.
:status 404: RSE Not Found.
:status 404: DID Not Found.
"""
headers = self.get_headers()
try:
scope, name = parse_scope_name(scope_name, request.headers.get('X-Rucio-VO', default='def'))
except ValueError as error:
return generate_http_error_flask(400, error, headers=headers)
try:
client_ip = request.headers.get('X-Forwarded-For', default=request.remote_addr)
client_location = {
'ip': request.args.get('ip', default=client_ip),
'fqdn': request.args.get('fqdn', default=None),
'site': request.args.get('site', default=None),
}
# use the default HTTP protocols if no scheme is given
schemes = request.args.getlist('schemes') or ['davs', 'https', 's3']
sortby = request.args.get('select', default='random')
sortby = request.args.get('sort', default=sortby)
rse = request.args.get('rse', default=None)
site = request.args.get('site', default=None)
# correctly forward the schemes and select to potential metalink followups
cleaned_url = request.environ.get('REQUEST_URI').split('?')[0]
headers.set('Link', f'<{cleaned_url}/metalink?schemes={",".join(schemes)}&select={sortby}>; rel=describedby; type="application/metalink+xml"')
# get vo if given
vo = request.headers.get('X-Rucio-VO', default='def')
replicas = list(
list_replicas(
dids=[{'scope': scope, 'name': name, 'type': 'FILE'}],
schemes=schemes,
client_location=client_location,
vo=vo
)
)
selected_url = None
for r in replicas:
if r['rses']:
dictreplica = {}
if rse:
if rse in r['rses'] and r['rses'][rse]:
selected_url = r['rses'][rse][0]
else:
return 'no redirection possible - no valid RSE for HTTP redirection found', 404, headers
else:
for rep in r['rses']:
for replica in r['rses'][rep]:
# since this is HTTP-only redirection, and to ensure compatibility with as many http clients as possible
# forcibly replacement davs and s3 URLs to https
replica = replica.replace('davs://', 'https://').replace('s3://', 'https://')
dictreplica[replica] = rep
if not dictreplica:
return 'no redirection possible - no valid RSE for HTTP redirection found', 404, headers
elif site:
rep = site_selector(dictreplica, site, vo)
if rep:
selected_url = rep[0]
else:
return 'no redirection possible - no valid RSE for HTTP redirection found', 404, headers
else:
rep = sort_replicas(dictreplica, client_location, selection=sortby)
selected_url = rep[0]
if selected_url:
response = redirect(selected_url, code=303)
response.headers.extend(headers)
return response
return 'no redirection possible - file does not exist', 404, headers
except ReplicaNotFound as error:
return generate_http_error_flask(404, error, headers=headers)
def GET(self, scope, name):
"""
List all replicas for data identifiers.
HTTP Success:
200 OK
HTTP Error:
401 Unauthorized
406 Not Acceptable
500 InternalError
:returns: A dictionary containing all replicas information.
:returns: A metalink description of replicas if metalink(4)+xml is specified in Accept:
"""
metalink = False
if ctx.env.get('HTTP_ACCEPT') is not None:
tmp = ctx.env.get('HTTP_ACCEPT').split(',')
if 'application/metalink4+xml' in tmp:
metalink = True
dids, schemes, select, limit = [{
'scope': scope,
'name': name
}], None, None, None
if ctx.query:
params = parse_qs(ctx.query[1:])
if 'schemes' in params:
schemes = params['schemes']
if 'select' in params:
select = params['select'][0]
if 'limit' in params:
limit = int(params['limit'][0])
client_ip = ctx.env.get('HTTP_X_FORWARDED_FOR')
if client_ip is None:
client_ip = ctx.ip
client_location = {'ip': client_ip, 'fqdn': None, 'site': None}
# Resolve all reasonable protocols when doing metalink for maximum access possibilities
if metalink and schemes is None:
schemes = SUPPORTED_PROTOCOLS
try:
# we need to call list_replicas before starting to reply
# otherwise the exceptions won't be propagated correctly
__first = True
header(
'Content-Type', 'application/metalink4+xml'
if metalink else 'application/x-json-stream')
for rfile in list_replicas(dids=dids,
schemes=schemes,
vo=ctx.env.get('vo')):
# in first round, set the appropriate content type, and stream the header
if __first:
if metalink:
yield '<?xml version="1.0" encoding="UTF-8"?>\n<metalink xmlns="urn:ietf:params:xml:ns:metalink">\n'
__first = False
# ... then, stream the replica information
replicas = []
dictreplica = {}
for rse in rfile['rses']:
for replica in rfile['rses'][rse]:
replicas.append(replica)
dictreplica[replica] = rse
replicas = sort_replicas(dictreplica,
client_location,
selection=select)
if not metalink:
yield dumps(rfile) + '\n'
else:
yield ' <file name="' + rfile['name'] + '">\n'
yield ' <identity>' + rfile['scope'] + ':' + rfile[
'name'] + '</identity>\n'
if rfile['adler32'] is not None:
yield ' <hash type="adler32">' + rfile[
'adler32'] + '</hash>\n'
if rfile['md5'] is not None:
yield ' <hash type="md5">' + rfile['md5'] + '</hash>\n'
yield ' <size>' + str(rfile['bytes']) + '</size>\n'
yield ' <glfn name="/atlas/rucio/%s:%s">' % (
rfile['scope'], rfile['name'])
yield '</glfn>\n'
idx = 0
for replica in replicas:
yield ' <url location="' + str(
dictreplica[replica]) + '" priority="' + str(
idx + 1) + '">' + escape(replica) + '</url>\n'
idx += 1
if limit and limit == idx:
break
yield ' </file>\n'
if metalink:
if __first:
# ensure complete metalink on success without any content
yield '<?xml version="1.0" encoding="UTF-8"?>\n<metalink xmlns="urn:ietf:params:xml:ns:metalink">\n</metalink>\n'
else:
# if metalink start was already sent, always send the end
yield '</metalink>\n'
except DataIdentifierNotFound as error:
raise generate_http_error(404, 'DataIdentifierNotFound',
error.args[0])
except RucioException as error:
raise generate_http_error(500, error.__class__.__name__,
error.args[0])
except Exception as error:
print(format_exc())
raise InternalError(error)
def POST(self):
"""
List all replicas for data identifiers.
HTTP Success:
200 OK
HTTP Error:
401 Unauthorized
406 Not Acceptable
500 InternalError
:returns: A dictionary containing all replicas information, either as JSON stream or metalink4.
"""
metalink = False
if ctx.env.get('HTTP_ACCEPT') is not None:
tmp = ctx.env.get('HTTP_ACCEPT').split(',')
if 'application/metalink4+xml' in tmp:
metalink = True
client_ip = ctx.env.get('HTTP_X_FORWARDED_FOR')
if client_ip is None:
client_ip = ctx.ip
dids, schemes, select, unavailable, limit = [], None, None, False, None
ignore_availability, rse_expression, all_states, domain = False, None, False, None
signature_lifetime, resolve_archives, resolve_parents = None, True, False
updated_after = None
client_location = {'ip': client_ip, 'fqdn': None, 'site': None}
json_data = data()
try:
params = parse_response(json_data)
if 'dids' in params:
dids = params['dids']
if 'schemes' in params:
schemes = params['schemes']
if 'unavailable' in params:
unavailable = params['unavailable']
ignore_availability = True
if 'all_states' in params:
all_states = params['all_states']
if 'rse_expression' in params:
rse_expression = params['rse_expression']
if 'client_location' in params:
client_location.update(params['client_location'])
if 'sort' in params:
select = params['sort']
if 'domain' in params:
domain = params['domain']
if 'resolve_archives' in params:
resolve_archives = params['resolve_archives']
if 'resolve_parents' in params:
resolve_parents = params['resolve_parents']
if 'signature_lifetime' in params:
signature_lifetime = params['signature_lifetime']
else:
# hardcoded default of 10 minutes if config is not parseable
signature_lifetime = config_get('credentials',
'signature_lifetime',
raise_exception=False,
default=600)
if 'updated_after' in params:
if isinstance(params['updated_after'], (int, float)):
# convert from epoch time stamp to datetime object
updated_after = datetime.utcfromtimestamp(
params['updated_after'])
else:
# attempt UTC format '%Y-%m-%dT%H:%M:%S' conversion
updated_after = datetime.strptime(params['updated_after'],
'%Y-%m-%dT%H:%M:%S')
except ValueError:
raise generate_http_error(400, 'ValueError',
'Cannot decode json parameter list')
if ctx.query:
params = parse_qs(ctx.query[1:])
if 'select' in params:
select = params['select'][0]
if 'limit' in params:
limit = params['limit'][0]
if 'sort' in params:
select = params['sort']
# Resolve all reasonable protocols when doing metalink for maximum access possibilities
if metalink and schemes is None:
schemes = SUPPORTED_PROTOCOLS
try:
# we need to call list_replicas before starting to reply
# otherwise the exceptions won't be propagated correctly
__first = True
header(
'Content-Type', 'application/metalink4+xml'
if metalink else 'application/x-json-stream')
for rfile in list_replicas(dids=dids,
schemes=schemes,
unavailable=unavailable,
request_id=ctx.env.get('request_id'),
ignore_availability=ignore_availability,
all_states=all_states,
rse_expression=rse_expression,
client_location=client_location,
domain=domain,
signature_lifetime=signature_lifetime,
resolve_archives=resolve_archives,
resolve_parents=resolve_parents,
updated_after=updated_after,
issuer=ctx.env.get('issuer'),
vo=ctx.env.get('vo')):
# in first round, set the appropriate content type, and stream the header
if __first:
if metalink:
yield '<?xml version="1.0" encoding="UTF-8"?>\n<metalink xmlns="urn:ietf:params:xml:ns:metalink">\n'
__first = False
# ... then, stream the replica information
if not metalink:
yield dumps(rfile, cls=APIEncoder) + '\n'
else:
replicas = []
dictreplica = {}
for replica in rfile['pfns'].keys():
replicas.append(replica)
dictreplica[replica] = (
rfile['pfns'][replica]['domain'],
rfile['pfns'][replica]['priority'],
rfile['pfns'][replica]['rse'],
rfile['pfns'][replica]['client_extract'])
yield ' <file name="' + rfile['name'] + '">\n'
if 'parents' in rfile and rfile['parents']:
yield ' <parents>\n'
for parent in rfile['parents']:
yield ' <did>' + parent + '</did>\n'
yield ' </parents>\n'
yield ' <identity>' + rfile['scope'] + ':' + rfile[
'name'] + '</identity>\n'
if rfile['adler32'] is not None:
yield ' <hash type="adler32">' + rfile[
'adler32'] + '</hash>\n'
if rfile['md5'] is not None:
yield ' <hash type="md5">' + rfile['md5'] + '</hash>\n'
yield ' <size>' + str(rfile['bytes']) + '</size>\n'
yield ' <glfn name="/%s/rucio/%s:%s"></glfn>\n' % (
config_get(
'policy',
'schema',
raise_exception=False,
default='generic'), rfile['scope'], rfile['name'])
lanreplicas = [
replica for replica, v in dictreplica.items()
if v[0] == 'lan'
]
replicas = lanreplicas + sort_replicas(
{
k: v
for k, v in dictreplica.items() if v[0] != 'lan'
},
client_location,
selection=select)
idx = 1
for replica in replicas:
yield ' <url location="' + str(dictreplica[replica][2]) \
+ '" domain="' + str(dictreplica[replica][0]) \
+ '" priority="' + str(idx) \
+ '" client_extract="' + str(dictreplica[replica][3]).lower() \
+ '">' + escape(replica) + '</url>\n'
if limit and limit == idx:
break
idx += 1
yield ' </file>\n'
if metalink:
if __first:
# ensure complete metalink on success without any content
yield '<?xml version="1.0" encoding="UTF-8"?>\n<metalink xmlns="urn:ietf:params:xml:ns:metalink">\n</metalink>\n'
else:
# if metalink start was already sent, always send the end
yield '</metalink>\n'
except DataIdentifierNotFound as error:
raise generate_http_error(404, 'DataIdentifierNotFound',
error.args[0])
except RucioException as error:
raise generate_http_error(500, error.__class__.__name__,
error.args[0])
except Exception as error:
print(format_exc())
raise InternalError(error)
def GET(self, scope, name):
"""
Metalink redirect
HTTP Success:
200 OK
HTTP Error:
401 Unauthorized
500 InternalError
404 Notfound
406 Not Acceptable
:param scope: The scope name of the file.
:param name: The name of the file.
"""
header('Access-Control-Allow-Origin', ctx.env.get('HTTP_ORIGIN'))
header('Access-Control-Allow-Headers',
ctx.env.get('HTTP_ACCESS_CONTROL_REQUEST_HEADERS'))
header('Access-Control-Allow-Methods', '*')
header('Access-Control-Allow-Credentials', 'true')
dids, schemes, select = [{
'scope': scope,
'name': name
}], ['http', 'https', 'root', 'gsiftp', 'srm', 'davs'], None
# set the correct client IP
client_ip = ctx.env.get('HTTP_X_FORWARDED_FOR')
if client_ip is None:
client_ip = ctx.ip
client_location = {'ip': client_ip, 'fqdn': None, 'site': None}
if ctx.query:
params = parse_qs(ctx.query[1:])
if 'schemes' in params:
schemes = params['schemes']
if 'select' in params:
select = params['select'][0]
if 'sort' in params:
select = params['sort'][0]
if 'ip' in params:
client_location['ip'] = params['ip'][0]
if 'fqdn' in params:
client_location['fqdn'] = params['fqdn'][0]
if 'site' in params:
client_location['site'] = params['site'][0]
# get vo if given
vo = ctx.env.get('HTTP_X_RUCIO_VO', 'def')
try:
tmp_replicas = [
rep for rep in list_replicas(dids=dids,
schemes=schemes,
client_location=client_location,
vo=vo)
]
if not tmp_replicas:
raise ReplicaNotFound(
'no redirection possible - cannot find the DID')
# first, set the appropriate content type, and stream the header
header('Content-Type', 'application/metalink4+xml')
yield '<?xml version="1.0" encoding="UTF-8"?>\n<metalink xmlns="urn:ietf:params:xml:ns:metalink">\n'
# iteratively stream the XML per file
for rfile in tmp_replicas:
replicas = []
dictreplica = {}
for rse in rfile['rses']:
for replica in rfile['rses'][rse]:
replicas.append(replica)
dictreplica[replica] = rse
# stream metadata
yield ' <file name="' + rfile['name'] + '">\n'
yield ' <identity>' + rfile['scope'] + ':' + rfile[
'name'] + '</identity>\n'
if rfile['adler32'] is not None:
yield ' <hash type="adler32">' + rfile[
'adler32'] + '</hash>\n'
if rfile['md5'] is not None:
yield ' <hash type="md5">' + rfile['md5'] + '</hash>\n'
yield ' <size>' + str(rfile['bytes']) + '</size>\n'
yield ' <glfn name="/atlas/rucio/%s:%s">' % (rfile['scope'],
rfile['name'])
yield '</glfn>\n'
replicas = sort_replicas(dictreplica,
client_location,
selection=select)
# stream URLs
idx = 1
for replica in replicas:
yield ' <url location="' + str(
dictreplica[replica]) + '" priority="' + str(
idx) + '">' + replica + '</url>\n'
idx += 1
yield ' </file>\n'
# don't forget to send the metalink footer
yield '</metalink>\n'
except DataIdentifierNotFound as error:
raise generate_http_error(404, 'DataIdentifierNotFound',
error.args[0])
except ReplicaNotFound as error:
raise generate_http_error(404, 'ReplicaNotFound', error.args[0])
except RucioException as error:
raise generate_http_error(500, error.__class__.__name__,
error.args[0])
except Exception as error:
print(format_exc())
raise InternalError(error)
def GET(self, scope, name):
"""
Header Redirect
HTTP Success:
303 See Other
HTTP Error:
401 Unauthorized
500 InternalError
404 Notfound
:param scope: The scope name of the file.
:param name: The name of the file.
"""
header('Access-Control-Allow-Origin', ctx.env.get('HTTP_ORIGIN'))
header('Access-Control-Allow-Headers',
ctx.env.get('HTTP_ACCESS_CONTROL_REQUEST_HEADERS'))
header('Access-Control-Allow-Methods', '*')
header('Access-Control-Allow-Credentials', 'true')
try:
# use the default HTTP protocols if no scheme is given
select, rse, site, schemes = 'random', None, None, [
'davs', 'http', 'https'
]
client_ip = ctx.env.get('HTTP_X_FORWARDED_FOR')
if client_ip is None:
client_ip = ctx.ip
client_location = {'ip': client_ip, 'fqdn': None, 'site': None}
if ctx.query:
params = parse_qs(ctx.query[1:])
if 'select' in params:
select = params['select'][0]
if 'sort' in params:
select = params['sort'][0]
if 'rse' in params:
rse = params['rse'][0]
if 'site' in params:
site = params['site'][0]
if 'schemes' in params:
schemes = params['schemes'][0]
else:
schemes = ['davs', 'https', 's3']
if 'ip' in params:
client_location['ip'] = params['ip'][0]
if 'fqdn' in params:
client_location['fqdn'] = params['fqdn'][0]
if 'site' in params:
client_location['site'] = params['site'][0]
# correctly forward the schemes and select to potential metalink followups
cleaned_url = ctx.env.get('REQUEST_URI').split('?')[0]
if isinstance(schemes, list):
header(
'Link',
'<%s/metalink?schemes=%s&select=%s>; rel=describedby; type="application/metalink+xml"'
% (cleaned_url, ','.join(schemes), select))
else:
header(
'Link',
'<%s/metalink?schemes=%s&select=%s>; rel=describedby; type="application/metalink+xml"'
% (cleaned_url, schemes, select))
schemes = [schemes] # list_replicas needs a list
# get vo if given
vo = ctx.env.get('HTTP_X_RUCIO_VO', 'def')
replicas = [
r for r in list_replicas(dids=[{
'scope': scope,
'name': name,
'type': 'FILE'
}],
schemes=schemes,
client_location=client_location,
vo=vo)
]
selected_url = None
for r in replicas:
if r['rses']:
dictreplica = {}
if rse:
if rse in r['rses'] and r['rses'][rse]:
selected_url = r['rses'][rse][0]
else:
raise ReplicaNotFound(
'no redirection possible - no valid RSE for HTTP redirection found'
)
else:
for rep in r['rses']:
for replica in r['rses'][rep]:
# since this is HTTP-only redirection, and to ensure compatibility with as many http clients as possible
# forcibly replacement davs and s3 URLs to https
replica = replica.replace(
'davs://',
'https://').replace('s3://', 'https://')
dictreplica[replica] = rep
if not dictreplica:
raise ReplicaNotFound(
'no redirection possible - no valid RSE for HTTP redirection found'
)
elif site:
rep = site_selector(dictreplica, site, vo)
if rep:
selected_url = rep[0]
else:
raise ReplicaNotFound(
'no redirection possible - no valid RSE for HTTP redirection found'
)
else:
rep = sort_replicas(dictreplica,
client_location,
selection=select)
selected_url = rep[0]
if selected_url:
raise seeother(selected_url)
raise ReplicaNotFound(
'no redirection possible - file does not exist')
except seeother:
raise
except ReplicaNotFound as error:
raise generate_http_error(404, 'ReplicaNotFound', error.args[0])
except RucioException as error:
raise generate_http_error(500, error.__class__.__name__,
error.args[0])
except Exception as error:
print(format_exc())
raise InternalError(error)
