def bootstrap_oai(self, endpoint): #todo update granularity
"""bootstraps OAI-PMH Source"""
startdate = self.config['fromdate']
self.logger.debug("Connecting to OAI-Endpoint %s" % endpoint)
self.client = Client(endpoint, self.config['limit'],
self.config['checkurl'])
try:
no_records = 0
for i, record in enumerate(
self.client.listRecords(startdate,
delay=self.config['delay_time'])):
no_records += self.process_record(record, init=True)
self.lastcheckdate = record.responseDate()
self.logger.info(
"Finished adding %d initial resources with checkdate: %s" %
((no_records * 2), self.lastcheckdate))
except URLError, e:
self.logger.error("URLError: %s" % (e))
def bootstrap_oai(self,endpoint): #todo update granularity
"""bootstraps OAI-PMH Source"""
startdate=self.config['fromdate']
self.logger.debug("Connecting to OAI-Endpoint %s" % endpoint)
self.client=Client(endpoint,self.config['limit'],self.config['checkurl'])
try:
no_records=0
for i,record in enumerate(self.client.listRecords(startdate,delay=self.config['delay_time'])):
no_records+=self.process_record(record,init=True)
self.lastcheckdate=record.responseDate()
self.logger.info("Finished adding %d initial resources with checkdate: %s" % ((no_records*2),self.lastcheckdate))
except URLError, e:
self.logger.error("URLError: %s" % (e))
class Source(Observable):
"""A source contains a list of resources and changes over time"""
RESOURCE_PATH = "/resources"
STATIC_FILE_PATH = os.path.join(os.path.dirname(__file__), "static")
TEMP_FILE_PATH = os.path.join(os.path.dirname(__file__), "temp")
def __init__(self, config, hostname, port):
"""Initalize the source"""
super(Source, self).__init__()
self.logger = logging.getLogger('source')
self.config = config
self.logger.info("Source config: %s " % self.config)
self.hostname = hostname
self.port = port
self.max_res_id = 1
self._repository = {} # {basename, {timestamp, size}}
self.inventory_builder = None # The inventory builder implementation
self.changememory = None # The change memory implementation
self.no_events = 0
self.oaimapping = {} #oai
self.client = None #oai
self.lastcheckdate = dateutil_parser.parse(
config['fromdate'].strftime("%Y-%m-%d %H:%SZ")) #oai
##### Source capabilities #####
def add_inventory_builder(self, inventory_builder):
"""Adds an inventory builder implementation"""
self.inventory_builder = inventory_builder
@property
def has_inventory_builder(self):
"""Returns True in the Source has an inventory builder"""
return bool(self.inventory_builder is not None)
def add_changememory(self, changememory):
"""Adds a changememory implementation"""
self.changememory = changememory
@property
def has_changememory(self):
"""Returns True if a source maintains a change memory"""
return bool(self.changememory is not None)
##### Bootstrap Source ######
def bootstrap(self):
"""Bootstrap the source with a set of resources"""
self.logger.info("Bootstrapping source")
if self.has_changememory: self.changememory.bootstrap()
if self.has_inventory_builder: self.inventory_builder.bootstrap()
self._log_stats()
##### Source data accessors #####
@property
def base_uri(self):
"""Returns the base URI of the source (e.g., http://localhost:8888)"""
return "http://" + self.hostname + ":" + str(self.port)
@property
def resource_count(self):
"""The number of resources in the source's repository"""
return len(self._repository)
@property
def resources(self):
"""Iterates over resources and yields resource objects"""
for basename in self._repository.keys():
resource = self.resource(basename)
if resource is None:
self.logger.error("Cannot create resource %s " % basename + \
"because source object has been deleted.")
else:
yield resource
@property
def random_resource(self):
"""Returns a single random resource"""
rand_res = self.random_resources()
if len(rand_res) == 1:
return rand_res[0]
else:
return None
def resource(self, basename):
"""Creates and returns a resource object from internal resource
repository. Repositoy values are copied into the object."""
if not self._repository.has_key(basename): return None
uri = basename
timestamp = self._repository[basename]['timestamp']
return Resource(uri=uri, timestamp=timestamp)
def random_resources(self, number=1):
"Return a random set of resources, at most all resources"
if number > len(self._repository):
number = len(self._repository)
rand_basenames = random.sample(self._repository.keys(), number)
return [self.resource(basename) for basename in rand_basenames]
# Private Methods
def _create_resource(self,
basename=None,
identifier=None,
timestamp=time.time(),
notify_observers=True,
oai=True):
"""Create a new resource, add it to the source, notify observers."""
self._repository[basename] = {'timestamp': timestamp}
change = ResourceChange(resource=self.resource(basename),
changetype="CREATED")
if notify_observers:
self.notify_observers(change)
self.logger.debug("Event: %s" % repr(change))
# add metadata resource url
if oai:
self._create_resource(
basename=self.client.endpoint +
"?verb=GetRecord&metadataPrefix=oai_dc&identifier=" +
identifier,
timestamp=timestamp,
notify_observers=notify_observers,
oai=False)
self.oaimapping[identifier] = basename
def _update_resource(self, basename, identifier, timestamp, oai=True):
"""Update a resource, notify observers."""
self._repository[basename] = {'timestamp': timestamp}
change = ResourceChange(resource=self.resource(basename),
changetype="UPDATED")
self.notify_observers(change)
self.logger.debug("Event: %s" % repr(change))
# update metadata resource url
if oai:
self._update_resource(
self.client.endpoint +
"?verb=GetRecord&metadataPrefix=oai_dc&identifier=" +
identifier,
identifier,
timestamp,
oai=False)
def _delete_resource(self,
identifier,
timestamp,
notify_observers=True,
oai=True):
"""Delete a given resource, notify observers."""
basename = None
if oai:
basename = self.oaimapping[identifier]
del self.oaimapping[identifier]
# delete metadata resource url
self._delete_resource(identifier,
timestamp,
notify_observers=notify_observers,
oai=False)
else:
basename = self.client.endpoint + "?verb=GetRecord&metadataPrefix=oai_dc&identifier=" + identifier
res = self.resource(basename)
del self._repository[basename]
res.timestamp = timestamp
if notify_observers:
change = ResourceChange(resource=res, changetype="DELETED")
self.notify_observers(change)
self.logger.debug("Event: %s" % repr(change))
def bootstrap_oai(self, endpoint): #todo update granularity
"""bootstraps OAI-PMH Source"""
startdate = self.config['fromdate']
self.logger.debug("Connecting to OAI-Endpoint %s" % endpoint)
self.client = Client(endpoint, self.config['limit'],
self.config['checkurl'])
try:
no_records = 0
for i, record in enumerate(
self.client.listRecords(startdate,
delay=self.config['delay_time'])):
no_records += self.process_record(record, init=True)
self.lastcheckdate = record.responseDate()
self.logger.info(
"Finished adding %d initial resources with checkdate: %s" %
((no_records * 2), self.lastcheckdate))
except URLError, e:
self.logger.error("URLError: %s" % (e))
except NoRecordsException as e:
self.logger.info("No new records found: %s" % e)
class Source(Observable):
"""A source contains a list of resources and changes over time"""
RESOURCE_PATH = "/resources"
STATIC_FILE_PATH = os.path.join(os.path.dirname(__file__), "static")
TEMP_FILE_PATH = os.path.join(os.path.dirname(__file__), "temp")
def __init__(self, config, hostname, port):
"""Initalize the source"""
super(Source, self).__init__()
self.logger = logging.getLogger('source')
self.config = config
self.logger.info("Source config: %s " % self.config)
self.hostname = hostname
self.port = port
self.max_res_id = 1
self._repository = {} # {basename, {timestamp, size}}
self.inventory_builder = None # The inventory builder implementation
self.changememory = None # The change memory implementation
self.no_events = 0
self.oaimapping = {} #oai
self.client=None #oai
self.lastcheckdate=dateutil_parser.parse(config['fromdate'].strftime("%Y-%m-%d %H:%SZ")) #oai
##### Source capabilities #####
def add_inventory_builder(self, inventory_builder):
"""Adds an inventory builder implementation"""
self.inventory_builder = inventory_builder
@property
def has_inventory_builder(self):
"""Returns True in the Source has an inventory builder"""
return bool(self.inventory_builder is not None)
def add_changememory(self, changememory):
"""Adds a changememory implementation"""
self.changememory = changememory
@property
def has_changememory(self):
"""Returns True if a source maintains a change memory"""
return bool(self.changememory is not None)
##### Bootstrap Source ######
def bootstrap(self):
"""Bootstrap the source with a set of resources"""
self.logger.info("Bootstrapping source")
if self.has_changememory: self.changememory.bootstrap()
if self.has_inventory_builder: self.inventory_builder.bootstrap()
self._log_stats()
##### Source data accessors #####
@property
def base_uri(self):
"""Returns the base URI of the source (e.g., http://localhost:8888)"""
return "http://" + self.hostname + ":" + str(self.port)
@property
def resource_count(self):
"""The number of resources in the source's repository"""
return len(self._repository)
@property
def resources(self):
"""Iterates over resources and yields resource objects"""
for basename in self._repository.keys():
resource = self.resource(basename)
if resource is None:
self.logger.error("Cannot create resource %s " % basename + \
"because source object has been deleted.")
else:
yield resource
@property
def random_resource(self):
"""Returns a single random resource"""
rand_res = self.random_resources()
if len(rand_res) == 1:
return rand_res[0]
else:
return None
def resource(self, basename):
"""Creates and returns a resource object from internal resource
repository. Repositoy values are copied into the object."""
if not self._repository.has_key(basename): return None
uri = basename
timestamp = self._repository[basename]['timestamp']
return Resource(uri = uri, timestamp = timestamp)
def random_resources(self, number = 1):
"Return a random set of resources, at most all resources"
if number > len(self._repository):
number = len(self._repository)
rand_basenames = random.sample(self._repository.keys(), number)
return [self.resource(basename) for basename in rand_basenames]
# Private Methods
def _create_resource(self, basename = None, identifier = None, timestamp=time.time(), notify_observers = True, oai = True):
"""Create a new resource, add it to the source, notify observers."""
self._repository[basename] = {'timestamp': timestamp}
change = ResourceChange(resource = self.resource(basename),
changetype = "CREATED")
if notify_observers:
self.notify_observers(change)
self.logger.debug("Event: %s" % repr(change))
# add metadata resource url
if oai:
self._create_resource(basename=self.client.endpoint+"?verb=GetRecord&metadataPrefix=oai_dc&identifier="+identifier,timestamp=timestamp,notify_observers=notify_observers,oai=False)
self.oaimapping[identifier]=basename;
def _update_resource(self, basename, identifier, timestamp, oai = True):
"""Update a resource, notify observers."""
self._repository[basename] = {'timestamp': timestamp}
change = ResourceChange(
resource = self.resource(basename),
changetype = "UPDATED")
self.notify_observers(change)
self.logger.debug("Event: %s" % repr(change))
# update metadata resource url
if oai:
self._update_resource(self.client.endpoint+"?verb=GetRecord&metadataPrefix=oai_dc&identifier="+identifier,identifier,timestamp,oai=False)
def _delete_resource(self, identifier, timestamp, notify_observers = True, oai = True):
"""Delete a given resource, notify observers."""
basename=None
if oai:
basename=self.oaimapping[identifier]
del self.oaimapping[identifier]
# delete metadata resource url
self._delete_resource(identifier,timestamp,notify_observers=notify_observers,oai=False)
else:
basename=self.client.endpoint+"?verb=GetRecord&metadataPrefix=oai_dc&identifier="+identifier
res = self.resource(basename)
del self._repository[basename]
res.timestamp = timestamp
if notify_observers:
change = ResourceChange(resource = res, changetype = "DELETED")
self.notify_observers(change)
self.logger.debug("Event: %s" % repr(change))
def bootstrap_oai(self,endpoint): #todo update granularity
"""bootstraps OAI-PMH Source"""
startdate=self.config['fromdate']
self.logger.debug("Connecting to OAI-Endpoint %s" % endpoint)
self.client=Client(endpoint,self.config['limit'],self.config['checkurl'])
try:
no_records=0
for i,record in enumerate(self.client.listRecords(startdate,delay=self.config['delay_time'])):
no_records+=self.process_record(record,init=True)
self.lastcheckdate=record.responseDate()
self.logger.info("Finished adding %d initial resources with checkdate: %s" % ((no_records*2),self.lastcheckdate))
except URLError, e:
self.logger.error("URLError: %s" % (e))
except NoRecordsException as e:
self.logger.info("No new records found: %s" % e)