def get_payload(startkey=None, endkey={}, startkey_docid=None, from_date=None, until_date=None, key=None, keys=None, any_tags=None, identity=None):
payload = {}
payload["startkey"] = startkey
payload["endkey"] = endkey
if key:
payload["key"] = key
if keys:
payload["keys"] = keys
if startkey_docid:
payload["startkey_docid"] = startkey_docid
if any_tags:
payload['any_tags'] = any_tags
if identity:
payload['identity'] = identity
if from_date and isinstance(from_date, datetime.datetime):
from lr.lib import helpers as h
payload["from_date"] = h.convertToISO8601Zformat(from_date)
if until_date and isinstance(until_date, datetime.datetime):
payload["until_date"] = h.convertToISO8601Zformat(until_date)
return payload
def list_identifiers(self,h, body ,params, verb = 'GET'):
data = self.get_base_response(verb,body)
if params.has_key('from'):
data['request']['from'] = params['from']
if params.has_key('until'):
data['request']['until'] = params['until']
from_date, until_date = self._test_time_params(params)
data['listidentifiers'] = []
base_response = json.dumps(data).split('[')
self._getServiceDocment(False)
resumption_token = None
if self.enable_flow_control and params.has_key('resumption_token'):
resumption_token = rt.parse_token(self.service_id,params['resumption_token'])
yield base_response[0] +'['
first = True;
count = 0
lastID = None
lastKey = None
for d in h.list_identifiers(from_date,until_date,resumption_token=resumption_token, limit=self.limit):
count += 1
lastID = d['id']
lastKey = d['key']
if not first:
yield ',\n'
first = False
return_value = {"header":{'identifier':d['id'], 'datestamp':helpers.convertToISO8601Zformat(datetime.today()) ,'status':'active'}}
yield json.dumps(return_value)
if self.enable_flow_control and self.limit <= count:
token = rt.get_token(serviceid=self.service_id,startkey=lastKey,endkey=helpers.convertToISO8601Zformat(until_date),startkey_docid=lastID,from_date=helpers.convertToISO8601Zformat(from_date),until_date=helpers.convertToISO8601Zformat(until_date))
resp = base_response[1]
yield resp[:-1] +(',"resumption_token":"%s"' %token) +resp[-1:]
else:
yield base_response[1]
def getrecord():
data = self.get_base_response(verb, body)
by_doc_ID = self._check_bool_param(params, 'by_doc_ID')
by_resource_ID = self._check_bool_param(params, 'by_resource_ID')
if not params.has_key(self.REQUESTID) and not params.has_key(
self.REQUESTID.lower()):
data['OK'] = False
data['error'] = 'badArgument'
return json.dumps(data)
if by_doc_ID and by_resource_ID:
data['OK'] = False
data['error'] = 'badArgument'
return json.dumps(data)
if params.has_key(self.REQUESTID):
request_id = params[self.REQUESTID]
elif params.has_key(self.REQUESTID.lower()):
request_id = params[self.REQUESTID.lower()]
if by_doc_ID:
document = h.get_record(request_id)
if document is not None:
records = map(
lambda doc: {
"header": {
'identifier':
doc['_id'],
'datestamp':
helpers.convertToISO8601Zformat(datetime.today(
)),
'status':
'active'
},
'resource_data': doc
}, [document])
else:
records = []
else:
request_id = unquote_plus(request_id)
records = map(
lambda doc: {
"header": {
'identifier':
doc['_id'],
'datestamp':
helpers.convertToISO8601Zformat(datetime.today()),
'status':
'active'
},
'resource_data': doc
}, h.get_records_by_resource(request_id))
if len(records) == 0:
data['OK'] = False
data['error'] = 'idDoesNotExist'
return json.dumps(data)
data['getrecord'] = {'record': records}
data['request']['identifier'] = request_id
data['request']['by_doc_ID'] = by_doc_ID
data['request']['by_resource_ID'] = by_resource_ID
return json.dumps(data)
def getrecord():
data = self.get_base_response(verb, body)
by_doc_ID = params.has_key('by_doc_ID') and ast.literal_eval(
str(params['by_doc_ID']))
by_resource_ID = params.has_key(
'by_resource_ID') and ast.literal_eval(
str(params['by_resource_ID']))
if not params.has_key('request_id'):
data['OK'] = False
data['error'] = 'badArgument'
return json.dumps(data)
if by_doc_ID and by_resource_ID:
data['OK'] = False
data['error'] = 'badArgument'
return json.dumps(data)
request_id = params['request_id']
if by_doc_ID:
records = map(
lambda doc: {
'record': {
"header": {
'identifier':
doc.id,
'datestamp':
helpers.convertToISO8601Zformat(datetime.today(
)),
'status':
'active'
}
},
'resource_data': doc
}, [h.get_record(request_id)])
else:
records = map(
lambda doc: {
'record': {
"header": {
'identifier':
doc.id,
'datestamp':
helpers.convertToISO8601Zformat(datetime.today(
)),
'status':
'active'
}
},
'resource_data': doc
}, h.get_records_by_resource(request_id))
data['getrecord'] = {'record': records}
return json.dumps(data)
def list_identifiers(self, h, body, params, verb='GET'):
data = self.get_base_response(verb, body)
if params.has_key('from'):
data['request']['from'] = params['from']
if params.has_key('until'):
data['request']['until'] = params['until']
from_date, until_date = self._test_time_params(params)
data['listidentifiers'] = []
base_response = json.dumps(data).split('[')
yield base_response[0] + '['
first = True
for id in h.list_identifiers(from_date, until_date):
if not first:
yield ','
first = False
return_value = {
"header": {
'identifier': id,
'datestamp':
helpers.convertToISO8601Zformat(datetime.today()),
'status': 'active'
}
}
yield json.dumps(return_value)
yield base_response[1]
def get_payload(startkey=None, endkey={}, startkey_docid=None, from_date=None, until_date=None):
payload = {}
payload["startkey"] = startkey
payload["endkey"] = endkey
if startkey_docid:
payload["startkey_docid"] = startkey_docid
if from_date and isinstance(from_date, datetime.datetime):
from lr.lib import helpers as h
payload["from_date"] = h.convertToISO8601Zformat(from_date)
if until_date and isinstance(until_date, datetime.datetime):
payload["until_date"] = h.convertToISO8601Zformat(until_date)
return payload
def get_base_response(self, verb, body):
return {
'OK': True,
'error': '',
'responseDate': helpers.convertToISO8601Zformat(datetime.today()),
'request': {
'verb': verb,
'HTTP_request': body
}
}
def get_base_response(self, verb, body):
return {
'OK':True,
'error':'',
'responseDate':helpers.convertToISO8601Zformat(datetime.today()),
'request':{
'verb':verb,
'HTTP_request': body
}
}
def get_payload(startkey=None,
endkey={},
startkey_docid=None,
from_date=None,
until_date=None):
payload = {}
payload["startkey"] = startkey
payload["endkey"] = endkey
if startkey_docid:
payload["startkey_docid"] = startkey_docid
if from_date and isinstance(from_date, datetime.datetime):
from lr.lib import helpers as h
payload["from_date"] = h.convertToISO8601Zformat(from_date)
if until_date and isinstance(until_date, datetime.datetime):
payload["until_date"] = h.convertToISO8601Zformat(until_date)
return payload
def debug_map(doc):
data = {
'record': {
"header": {
'identifier':
doc.id,
'datestamp':
helpers.convertToISO8601Zformat(datetime.today()),
'status':
'active'
},
'resource_data': doc
}
}
return data
def get_payload(startkey=None,
endkey={},
startkey_docid=None,
from_date=None,
until_date=None,
key=None,
keys=None,
any_tags=None,
identity=None):
payload = {}
payload["startkey"] = startkey
payload["endkey"] = endkey
if key:
payload["key"] = key
if keys:
payload["keys"] = keys
if startkey_docid:
payload["startkey_docid"] = startkey_docid
if any_tags:
payload['any_tags'] = any_tags
if identity:
payload['identity'] = identity
if from_date and isinstance(from_date, datetime.datetime):
from lr.lib import helpers as h
payload["from_date"] = h.convertToISO8601Zformat(from_date)
if until_date and isinstance(until_date, datetime.datetime):
payload["until_date"] = h.convertToISO8601Zformat(until_date)
return payload
def list_identifiers(self,h,body ,params, verb = 'GET'):
data = self.get_base_response(verb,body)
if params.has_key('from'):
data['request']['from'] = params['from']
if params.has_key('until'):
data['request']['until'] = params['until']
from_date, until_date = self._test_time_params(params)
data['listidentifiers'] = []
base_response = json.dumps(data).split('[')
yield base_response[0] +'['
first = True;
for id in h.list_identifiers(from_date,until_date):
if not first:
yield ','
first = False
return_value = {"header":{'identifier':id, 'datestamp':helpers.convertToISO8601Zformat(datetime.today()) ,'status':'active'}}
yield json.dumps(return_value)
yield base_response[1]
def list_records(self, h , body , params, verb = 'GET' ):
data = self.get_base_response(verb,body)
if params.has_key('from'):
data['request']['from'] = params['from']
if params.has_key('until'):
data['request']['until'] = params['until']
from_date, until_date = self._test_time_params(params)
data['listrecords'] = []
self._getServiceDocment(False)
resumption_token = None
count = 0
lastID = None
lastKey = None
if self.enable_flow_control and params.has_key('resumption_token'):
resumption_token = rt.parse_token(self.service_id,params['resumption_token'])
base_response = json.dumps(data).split('[')
yield base_response[0] +'['
def debug_map(doc):
data ={'record':{"header":{'identifier':doc['_id'], 'datestamp':helpers.convertToISO8601Zformat(datetime.today()),'status':'active'},'resource_data':doc}}
return data
if from_date > until_date:
data['OK'] = False
data['error'] = 'badArgument'
else:
first = True
for data in h.list_records(from_date,until_date,resumption_token=resumption_token, limit=self.limit):
lastID = data['id']
lastKey = data['key']
doc = data['doc']
count += 1
if not first:
yield ',\n'
first = False
yield json.dumps(debug_map(doc))
if self.enable_flow_control and self.limit <= count:
token = rt.get_token(serviceid=self.service_id,startkey=lastKey,endkey=helpers.convertToISO8601Zformat(until_date),startkey_docid=lastID,from_date=helpers.convertToISO8601Zformat(from_date),until_date=helpers.convertToISO8601Zformat(until_date))
resp = base_response[1]
yield resp[:-1] +(',"resumption_token":"%s"' %token) +resp[-1:]
else:
yield base_response[1]
def getrecord():
data = self.get_base_response(verb,body)
by_doc_ID = self._check_bool_param(params,'by_doc_ID')
by_resource_ID = self._check_bool_param(params,'by_resource_ID')
if not params.has_key(self.REQUESTID):
data['OK'] = False
data['error'] = 'badArgument'
return json.dumps(data)
if by_doc_ID and by_resource_ID:
data['OK'] = False
data['error'] = 'badArgument'
return json.dumps(data)
request_id = params[self.REQUESTID]
if by_doc_ID:
document = h.get_record(request_id)
if document is not None:
records = map(lambda doc: {"header":{'identifier':doc['_id'], 'datestamp':helpers.convertToISO8601Zformat(datetime.today()),'status':'active'},'resource_data':doc},[document])
else:
records = []
else:
records = map(lambda doc: {"header":{'identifier':doc['_id'], 'datestamp':helpers.convertToISO8601Zformat(datetime.today()),'status':'active'},'resource_data':doc},h.get_records_by_resource(request_id))
if len(records) == 0:
abort(500,'idDoesNotExist')
data['getrecord'] ={
'record': records
}
data['request']['identifier'] = request_id
data['request']['by_doc_ID'] = by_doc_ID
data['request']['by_resource_ID'] = by_resource_ID
return json.dumps(data)
def debug_map(doc):
data ={'record':{"header":{'identifier':doc['_id'], 'datestamp':helpers.convertToISO8601Zformat(datetime.today()),'status':'active'},'resource_data':doc}}
return data
def _handleOAIRequest(self, format='html'):
t_req = request._current_obj()
t_res = response._current_obj()
enable_flow_control = False
fc_id_limit = None
fc_doc_limit = None
service_id = None
serviceDoc = h.getServiceDocument(appConfig['lr.oaipmh.docid'])
if serviceDoc != None:
if 'service_id' in serviceDoc:
service_id = serviceDoc['service_id']
if 'service_data' in serviceDoc:
serviceData = serviceDoc['service_data']
if 'flow_control' in serviceData:
enable_flow_control = serviceData['flow_control']
if enable_flow_control and 'id_limit' in serviceData:
fc_id_limit = serviceData['id_limit']
elif enable_flow_control:
fc_id_limit = 100
if enable_flow_control and 'doc_limit' in serviceData:
fc_doc_limit = serviceData['doc_limit']
elif enable_flow_control:
fc_doc_limit = 100
o = oaipmh()
def GetRecord(params):
try:
from lr.mustache.oaipmh import GetRecord as must_GetRecord
identifier = params["identifier"]
if params["by_doc_ID"] == True:
resolver = OAIPMHDocumentResolver()
single_doc = o.get_record(params["identifier"])
if single_doc is not None:
docList = [resolver.process({"doc": single_doc})]
else:
docList = []
else:
docList = o.get_records_by_resource(params["identifier"])
doc_idx = 0
valid_docs = 0
mustache = must_GetRecord()
for doc in docList:
if doc is not None:
doc_idx += 1
if "payload_schema" in doc and params[
"metadataPrefix"] in map(
lambda x: o_mod.getMetadataPrefix(x),
doc["payload_schema"]
) and OAIPMHDocumentResolver.PAYLOAD_ERROR not in doc:
valid_docs += 1
if valid_docs == 1:
part = mustache.prefix(**self._initMustache(
args=params, req=t_req))
yield h.fixUtf8(
self._returnResponse(part, res=t_res))
part = mustache.doc(doc)
yield h.fixUtf8(
self._returnResponse(part, res=t_res))
if doc_idx == 0:
raise IdDoesNotExistError(params['verb'], req=t_req)
elif valid_docs == 0:
raise CannotDisseminateFormatError(params['verb'],
req=t_req)
else:
yield h.fixUtf8(
self._returnResponse(mustache.suffix(), res=t_res))
except oaipmherrors.Error as e:
from lr.mustache.oaipmh import Error as err_stache
err = err_stache()
yield h.fixUtf8(self._returnResponse(err.xml(e), res=t_res))
def ListGeneric(params, showDocs=False, record_limit=None):
if not showDocs:
from lr.mustache.oaipmh import ListIdentifiers as must_ListID
mustache = must_ListID()
else:
from lr.mustache.oaipmh import ListRecords as must_ListRec
mustache = must_ListRec()
try:
doc_index = 0
err_count = 0
metadataPrefix = params["metadataPrefix"]
from_date = params["from"]
until_date = params["until"]
doc_err = None
rendered_init = False
resumptionToken = None if "resumptionToken" not in params else params[
'resumptionToken']
records = o.list_identifiers_or_records(metadataPrefix,
from_date=from_date,
until_date=until_date,
rt=resumptionToken,
fc_limit=record_limit,
include_docs=showDocs)
for ident in records:
doc_index += 1
doc_err = False
if OAIPMHDocumentResolver.PAYLOAD_ERROR in ident:
err_count += 1
doc_err = True
log.debug(
"Payload Error detected, doc_index: {0}, err_count: {1}"
.format(doc_index, err_count))
if doc_index - err_count == 1:
rendered_init = True
part = mustache.prefix(
**self._initMustache(args=params, req=t_req))
yield h.fixUtf8(self._returnResponse(part, res=t_res))
if doc_err is False and (record_limit is None
or doc_index <= record_limit):
part = mustache.doc(ident)
yield h.fixUtf8(part)
elif enable_flow_control:
from lr.lib import resumption_token
if doc_index - err_count > 0 and doc_index > record_limit:
opts = o.list_opts(
metadataPrefix,
h.convertToISO8601UTC(ident["node_timestamp"]),
until_date)
opts["startkey_docid"] = ident["doc_ID"]
token = resumption_token.get_token(
serviceid=service_id,
from_date=from_date,
until_date=until_date,
**opts)
part = mustache.resumptionToken(token)
yield h.fixUtf8(part)
break
elif doc_index - err_count == 0 and doc_index > record_limit:
opts = o.list_opts(
metadataPrefix,
h.convertToISO8601UTC(ident["node_timestamp"]),
until_date)
opts["startkey_docid"] = ident["doc_ID"]
payload = resumption_token.get_payload(
from_date=from_date,
until_date=until_date,
**opts)
records = o.list_identifiers_or_records(
metadataPrefix,
from_date=from_date,
until_date=until_date,
rt=payload,
fc_limit=record_limit,
include_docs=showDocs)
doc_index = 0
err_count = 0
if doc_index == 0 and err_count == 0:
raise NoRecordsMatchError(params['verb'], req=t_req)
elif (doc_index - err_count) == 0:
raise CannotDisseminateFormatError(params['verb'],
req=t_req)
else:
if enable_flow_control and doc_index <= record_limit:
yield h.fixUtf8(mustache.resumptionToken())
yield h.fixUtf8(mustache.suffix())
except oaipmherrors.Error as e:
if not rendered_init:
from lr.mustache.oaipmh import Error as err_stache
err = err_stache()
yield h.fixUtf8(self._returnResponse(err.xml(e),
res=t_res))
else:
from lr.mustache.oaipmh import ErrorOnly as err_stache
err = err_stache()
yield h.fixUtf8(
self._returnResponse(err.xml(e) + mustache.suffix(),
res=t_res))
except:
log.exception("Unknown Error Occurred")
def ListIdentifiers(params):
return ListGeneric(params, False, fc_id_limit)
def ListRecords(params):
return ListGeneric(params, True, fc_doc_limit)
# def ListRecords(params):
# try:
# from lr.mustache.oaipmh import ListRecords as must_ListRec
#
# doc_index = 0
# mustache = must_ListRec()
# for record in o.list_records(params["metadataPrefix"],from_date=params["from"], until_date=params["until"] ):
# doc_index += 1
# log.debug(json.dumps(record))
# if doc_index == 1:
# part = mustache.prefix(**self._initMustache(args=params, req=t_req))
# yield self._returnResponse(part, res=t_res)
#
# part = mustache.doc(record)
# yield self._returnResponse(part, res=t_res)
#
#
# if doc_index == 0:
# raise NoRecordsMatchError(params['verb'], req=t_req)
# else:
# yield mustache.suffix()
#
# except oaipmherrors.Error as e:
# from lr.mustache.oaipmh import Error as err_stache
# err = err_stache()
# yield self._returnResponse(err.xml(e), res=t_res)
# except:
# log.exception("Unable to render template")
def Identify(params=None):
body = ""
try:
self._initRender(params, ctx=c, req=t_req)
c.identify = o.identify()
body = render("/oaipmh-Identify.mako")
except Exception as e:
raise BadVerbError()
return self._returnResponse(body, res=t_res)
def ListMetadataFormats(params):
body = ""
try:
self._initRender(params, ctx=c, req=t_req)
fmts = o.list_metadata_formats(identity=params["identifier"],
by_doc_ID=params["by_doc_ID"])
if len(fmts) == 0:
raise NoMetadataFormats(params["verb"])
c.formats = fmts
body = render("/oaipmh-ListMetadataFormats.mako")
return self._returnResponse(body, res=t_res)
except Error as e:
raise e
def ListSets(params=None):
raise NoSetHierarchyError(verb)
def NotYetSupported(params=None):
raise BadVerbError()
switch = {
'GetRecord': GetRecord,
'ListRecords': ListRecords,
'ListIdentifiers': ListIdentifiers,
'Identify': Identify,
'ListMetadataFormats': ListMetadataFormats,
'ListSets': ListSets
}
try:
params = self._parseParams(flow_control=enable_flow_control,
serviceid=service_id)
# If this is a special case where we are actually using OAI interface to serve basic harvest
if params.has_key("metadataPrefix") and params[
"metadataPrefix"] == "LR_JSON_0.10.0":
if params.has_key("identifier") == True:
params[self.REQUESTID] = params["identifier"]
if params.has_key("from") and isinstance(
params["from"], datetime):
params["from"] = h.convertToISO8601Zformat(params["from"])
if params.has_key("until") and isinstance(
params["until"], datetime):
params["until"] = h.convertToISO8601Zformat(
params["until"])
return HarvestController.harvest(self, params, request.body,
params['verb'].lower())
verb = params['verb']
response.headers['Content-Type'] = "text/xml; charset=utf-8"
return switch[verb](params)
except Error as e:
from lr.mustache.oaipmh import Error as err_stache
err = err_stache()
return self._returnResponse(err.xml(e), res=t_res)
def listGeneral(self, h , body , params, includeDocs,verb = 'GET'):
data = self.get_base_response(verb,body)
try:
from_date, until_date = self._test_time_params(params)
except Exception as ex:
log.error(ex)
data['OK'] = False
data['error'] = 'badArgument'
yield json.dumps(data)
return
data['request']['from'] = from_date
data['request']['until'] = until_date
if from_date > until_date:
data['OK'] = False
data['error'] = 'badArgument'
yield json.dumps(data)
else:
self._getServiceDocment(includeDocs)
resumption_token = None
count = 0
lastID = None
lastKey = None
if self.enable_flow_control and params.has_key('resumption_token'):
resumption_token = rt.parse_token(self.service_id,params['resumption_token'])
if includeDocs:
data['listrecords'] = []
viewResults = h.list_records(from_date,until_date,resumption_token=resumption_token, limit=self.limit)
debug_map = lambda doc: {'record':{"header":{'identifier':doc['id'], 'datestamp':doc['key']+"Z",'status':'active'},'resource_data':doc['doc']}}
else:
data['listidentifiers'] = []
viewResults = h.list_identifiers(from_date,until_date,resumption_token=resumption_token, limit=self.limit)
debug_map = lambda doc:{"header":{'identifier':doc['id'], 'datestamp':doc['key']+"Z",'status':'active'}}
base_response = json.dumps(data).split('[')
yield base_response[0] +'['
first = True
for data in viewResults:
lastID = data['id']
lastKey = data['key']
count += 1
if not first:
yield ',\n'
first = False
yield json.dumps(debug_map(data))
if self.enable_flow_control and self.limit <= count:
token = rt.get_token(serviceid=self.service_id,startkey=lastKey,endkey=helpers.convertToISO8601Zformat(until_date),startkey_docid=lastID,from_date=helpers.convertToISO8601Zformat(from_date),until_date=helpers.convertToISO8601Zformat(until_date))
resp = base_response[1]
yield resp[:-1] +(',"resumption_token":"%s"' %token) +resp[-1:]
elif self.limit > count:
resp = base_response[1]
yield resp[:-1] +(',"resumption_token":"%s"' %'null') +resp[-1:]
else:
yield base_response[1]
def _handleOAIRequest(self, format='html'):
t_req = request._current_obj()
t_res = response._current_obj()
enable_flow_control = False
fc_id_limit = None
fc_doc_limit = None
service_id = None
serviceDoc = h.getServiceDocument(appConfig['lr.oaipmh.docid'])
if serviceDoc != None:
if 'service_id' in serviceDoc:
service_id = serviceDoc['service_id']
if 'service_data' in serviceDoc:
serviceData = serviceDoc['service_data']
if 'flow_control' in serviceData:
enable_flow_control = serviceData['flow_control']
if enable_flow_control and 'id_limit' in serviceData:
fc_id_limit = serviceData['id_limit']
elif enable_flow_control:
fc_id_limit = 100
if enable_flow_control and 'doc_limit' in serviceData:
fc_doc_limit = serviceData['doc_limit']
elif enable_flow_control:
fc_doc_limit = 100
o = oaipmh()
def GetRecord(params):
try:
from lr.mustache.oaipmh import GetRecord as must_GetRecord
identifier = params["identifier"]
if params["by_doc_ID"] == True:
resolver = OAIPMHDocumentResolver()
single_doc = o.get_record(params["identifier"])
if single_doc is not None:
docList = [resolver.process({ "doc": single_doc })]
else:
docList = []
else:
docList = o.get_records_by_resource(params["identifier"])
doc_idx = 0
valid_docs = 0
mustache = must_GetRecord()
for doc in docList:
if doc is not None:
doc_idx += 1
if "payload_schema" in doc and params["metadataPrefix"] in map(lambda x: o_mod.getMetadataPrefix(x), doc["payload_schema"]) and OAIPMHDocumentResolver.PAYLOAD_ERROR not in doc:
valid_docs += 1
if valid_docs == 1:
part = mustache.prefix(**self._initMustache(args=params, req=t_req))
yield h.fixUtf8(self._returnResponse(part, res=t_res))
part = mustache.doc(doc)
yield h.fixUtf8(self._returnResponse(part, res=t_res))
if doc_idx == 0:
raise IdDoesNotExistError(params['verb'], req=t_req)
elif valid_docs == 0:
raise CannotDisseminateFormatError(params['verb'], req=t_req)
else:
yield h.fixUtf8(self._returnResponse(mustache.suffix(), res=t_res))
except oaipmherrors.Error as e:
from lr.mustache.oaipmh import Error as err_stache
err = err_stache()
yield h.fixUtf8(self._returnResponse(err.xml(e), res=t_res))
def ListGeneric(params, showDocs=False, record_limit=None):
if not showDocs:
from lr.mustache.oaipmh import ListIdentifiers as must_ListID
mustache = must_ListID()
else:
from lr.mustache.oaipmh import ListRecords as must_ListRec
mustache = must_ListRec()
try:
doc_index = 0
err_count = 0
metadataPrefix=params["metadataPrefix"]
from_date=params["from"]
until_date=params["until"]
doc_err = None
rendered_init = False
resumptionToken = None if "resumptionToken" not in params else params['resumptionToken']
records = o.list_identifiers_or_records(metadataPrefix,
from_date=from_date,
until_date=until_date,
rt=resumptionToken,
fc_limit=record_limit,
include_docs=showDocs )
for ident in records:
doc_index += 1
doc_err = False
if OAIPMHDocumentResolver.PAYLOAD_ERROR in ident:
err_count += 1
doc_err = True
log.debug("Payload Error detected, doc_index: {0}, err_count: {1}".format(doc_index, err_count))
if doc_index - err_count == 1:
rendered_init = True
part = mustache.prefix(**self._initMustache(args=params, req=t_req))
yield h.fixUtf8(self._returnResponse(part, res=t_res))
if doc_err is False and (record_limit is None or doc_index <= record_limit):
part = mustache.doc(ident)
yield h.fixUtf8(part)
elif enable_flow_control:
from lr.lib import resumption_token
if doc_index - err_count > 0 and doc_index > record_limit:
opts = o.list_opts(metadataPrefix, h.convertToISO8601UTC(ident["node_timestamp"]), until_date)
opts["startkey_docid"] = ident["doc_ID"]
token = resumption_token.get_token(serviceid=service_id, from_date=from_date, until_date=until_date, **opts)
part = mustache.resumptionToken(token)
yield h.fixUtf8(part)
break
elif doc_index - err_count == 0 and doc_index > record_limit:
opts = o.list_opts(metadataPrefix, h.convertToISO8601UTC(ident["node_timestamp"]), until_date)
opts["startkey_docid"] = ident["doc_ID"]
payload = resumption_token.get_payload(from_date=from_date, until_date=until_date, **opts)
records = o.list_identifiers_or_records(metadataPrefix,
from_date=from_date,
until_date=until_date,
rt=payload,
fc_limit=record_limit,
include_docs=showDocs )
doc_index = 0
err_count = 0
if doc_index == 0 and err_count == 0:
raise NoRecordsMatchError(params['verb'], req=t_req)
elif (doc_index - err_count) == 0:
raise CannotDisseminateFormatError(params['verb'], req=t_req)
else:
if enable_flow_control and doc_index <= record_limit:
yield h.fixUtf8(mustache.resumptionToken())
yield h.fixUtf8(mustache.suffix())
except oaipmherrors.Error as e:
if not rendered_init:
from lr.mustache.oaipmh import Error as err_stache
err = err_stache()
yield h.fixUtf8(self._returnResponse(err.xml(e), res=t_res))
else:
from lr.mustache.oaipmh import ErrorOnly as err_stache
err = err_stache()
yield h.fixUtf8(self._returnResponse(err.xml(e)+mustache.suffix(), res=t_res))
except:
log.exception("Unknown Error Occurred")
def ListIdentifiers(params):
return ListGeneric(params, False, fc_id_limit)
def ListRecords(params):
return ListGeneric(params, True, fc_doc_limit)
# def ListRecords(params):
# try:
# from lr.mustache.oaipmh import ListRecords as must_ListRec
#
# doc_index = 0
# mustache = must_ListRec()
# for record in o.list_records(params["metadataPrefix"],from_date=params["from"], until_date=params["until"] ):
# doc_index += 1
# log.debug(json.dumps(record))
# if doc_index == 1:
# part = mustache.prefix(**self._initMustache(args=params, req=t_req))
# yield self._returnResponse(part, res=t_res)
#
# part = mustache.doc(record)
# yield self._returnResponse(part, res=t_res)
#
#
# if doc_index == 0:
# raise NoRecordsMatchError(params['verb'], req=t_req)
# else:
# yield mustache.suffix()
#
# except oaipmherrors.Error as e:
# from lr.mustache.oaipmh import Error as err_stache
# err = err_stache()
# yield self._returnResponse(err.xml(e), res=t_res)
# except:
# log.exception("Unable to render template")
def Identify(params=None):
body = ""
try:
self._initRender(params, ctx=c, req=t_req)
c.identify = o.identify()
body = render("/oaipmh-Identify.mako")
except Exception as e:
raise BadVerbError()
return self._returnResponse(body, res=t_res)
def ListMetadataFormats(params):
body = ""
try:
self._initRender(params, ctx=c, req=t_req)
fmts = o.list_metadata_formats(identity=params["identifier"], by_doc_ID=params["by_doc_ID"])
if len(fmts) == 0:
raise NoMetadataFormats(params["verb"])
c.formats = fmts
body = render("/oaipmh-ListMetadataFormats.mako")
return self._returnResponse(body, res=t_res)
except Error as e:
raise e
def ListSets(params=None):
raise NoSetHierarchyError(verb)
def NotYetSupported(params=None):
raise BadVerbError()
switch = {
'GetRecord': GetRecord,
'ListRecords': ListRecords,
'ListIdentifiers': ListIdentifiers,
'Identify': Identify,
'ListMetadataFormats': ListMetadataFormats,
'ListSets': ListSets
}
try:
params = self._parseParams(flow_control=enable_flow_control, serviceid=service_id)
# If this is a special case where we are actually using OAI interface to serve basic harvest
if params.has_key("metadataPrefix") and params["metadataPrefix"] == "LR_JSON_0.10.0":
if params.has_key("identifier") == True:
params[self.REQUESTID] = params["identifier"]
if params.has_key("from") and isinstance(params["from"], datetime):
params["from"] = h.convertToISO8601Zformat(params["from"])
if params.has_key("until") and isinstance(params["until"], datetime):
params["until"] = h.convertToISO8601Zformat(params["until"])
return HarvestController.harvest(self, params, request.body, params['verb'].lower())
verb = params['verb']
response.headers['Content-Type'] = "text/xml; charset=utf-8"
return switch[verb](params)
except Error as e:
from lr.mustache.oaipmh import Error as err_stache
err = err_stache()
return self._returnResponse(err.xml(e), res=t_res)
def getrecord():
data = self.get_base_response(verb,body)
by_doc_ID = params.has_key('by_doc_ID') and ast.literal_eval(str(params['by_doc_ID']))
by_resource_ID = params.has_key('by_resource_ID') and ast.literal_eval(str(params['by_resource_ID']))
if not params.has_key('request_id'):
data['OK'] = False
data['error'] = 'badArgument'
return json.dumps(data)
if by_doc_ID and by_resource_ID:
data['OK'] = False
data['error'] = 'badArgument'
return json.dumps(data)
request_id = params['request_id']
if by_doc_ID:
records = map(lambda doc: {'record':{"header":{'identifier':doc.id, 'datestamp':helpers.convertToISO8601Zformat(datetime.today()),'status':'active'}},'resource_data':doc},[h.get_record(request_id)])
else:
records = map(lambda doc: {'record':{"header":{'identifier':doc.id, 'datestamp':helpers.convertToISO8601Zformat(datetime.today()),'status':'active'}},'resource_data':doc},h.get_records_by_resource(request_id))
data['getrecord'] ={
'record': records
}
return json.dumps(data)
def listGeneral(self, h, body, params, includeDocs, verb='GET'):
data = self.get_base_response(verb, body)
try:
from_date, until_date = self._test_time_params(params)
except:
data['OK'] = False
data['error'] = 'badArgument'
yield json.dumps(data)
return
data['request']['from'] = from_date
data['request']['until'] = until_date
if from_date > until_date:
data['OK'] = False
data['error'] = 'badArgument'
yield json.dumps(data)
else:
self._getServiceDocment(includeDocs)
resumption_token = None
count = 0
lastID = None
lastKey = None
if self.enable_flow_control and params.has_key('resumption_token'):
resumption_token = rt.parse_token(self.service_id,
params['resumption_token'])
if includeDocs:
data['listrecords'] = []
viewResults = h.list_records(from_date,
until_date,
resumption_token=resumption_token,
limit=self.limit)
debug_map = lambda doc: {
'record': {
"header": {
'identifier': doc['id'],
'datestamp': doc['key'] + "Z",
'status': 'active'
},
'resource_data': doc['doc']
}
}
else:
data['listidentifiers'] = []
viewResults = h.list_identifiers(
from_date,
until_date,
resumption_token=resumption_token,
limit=self.limit)
debug_map = lambda doc: {
"header": {
'identifier': doc['id'],
'datestamp': doc['key'] + "Z",
'status': 'active'
}
}
base_response = json.dumps(data).split('[')
yield base_response[0] + '['
first = True
for data in viewResults:
lastID = data['id']
lastKey = data['key']
count += 1
if not first:
yield ',\n'
first = False
yield json.dumps(debug_map(data))
if self.enable_flow_control and self.limit <= count:
token = rt.get_token(
serviceid=self.service_id,
startkey=lastKey,
endkey=helpers.convertToISO8601Zformat(until_date),
startkey_docid=lastID,
from_date=helpers.convertToISO8601Zformat(from_date),
until_date=helpers.convertToISO8601Zformat(until_date))
resp = base_response[1]
yield resp[:-1] + (',"resumption_token":"%s"' %
token) + resp[-1:]
elif self.limit > count:
resp = base_response[1]
yield resp[:-1] + (',"resumption_token":"%s"' %
'null') + resp[-1:]
else:
yield base_response[1]
