def test_fetch(self):
'''
Parse example dataset
'''
registry = importformats.create_metadata_registry()
client = oaipmh.client.Client(_get_fixture(FIXTURE_DATASET), registry)
record = client.getRecord(identifier=self.TEST_ID, metadataPrefix='oai_dc')
assert record
def test_fetch(self):
'''
Parse example dataset
'''
registry = importformats.create_metadata_registry()
client = oaipmh.client.Client(_get_fixture(FIXTURE_DATASET), registry)
record = client.getRecord(identifier=self.TEST_ID,
metadataPrefix='oai_dc')
assert record
def fetch_stage(self, harvest_object):
'''
The fetch stage will receive a HarvestObject object and will be
responsible for:
- getting the contents of the remote object (e.g. for a CSW server, perform a GetRecordById request).
- saving the content in the provided HarvestObject.
- creating and storing any suitable HarvestObjectErrors that may occur.
- returning True if everything went as expected, False otherwise.
:param harvest_object: HarvestObject object
:returns: True if everything went right, False if errors were found
'''
log.debug("fetch: %s", harvest_object.guid)
# Get metadata content from provider
try:
# Create a OAI-PMH Client
config = self._get_configuration(harvest_object)
registry = self.metadata_registry(config, harvest_object)
client = oaipmh.client.Client(harvest_object.job.source.url,
registry)
# Get source URL
header, metadata, _about = client.getRecord(
identifier=harvest_object.guid, metadataPrefix=self.md_format)
except Exception as e:
import traceback
traceback.print_exc()
self._save_object_error(
'Unable to get metadata from provider: {u}: {e}'.format(
u=harvest_object.source.url, e=e), harvest_object)
return False
if header and header.isDeleted():
return self.on_deleted(harvest_object, header)
# Get contents
try:
content = json.dumps(metadata.getMap())
except Exception as e:
import traceback
traceback.print_exc()
self._save_object_error(
'Unable to get content for package: {u}: {e}'.format(
u=harvest_object.source.url, e=e), harvest_object)
return False
# Save the fetched contents in the HarvestObject
harvest_object.content = content
harvest_object.save()
return True
def _fetch_import_record(self, harvest_object, master_data, client, group):
# The fetch part.
try:
header, metadata, _ = client.getRecord(
metadataPrefix=self.metadata_prefix_value,
identifier=master_data['record'])
except XMLSyntaxError:
log.error('oai_dc XML syntax error: %s' % master_data['record'])
self._save_object_error(
'Syntax error.',
harvest_object, stage='Fetch')
return False
except socket.error:
errno, errstr = sys.exc_info()[:2]
self._save_object_error(
'Socket error OAI-PMH %s, details:\n%s' % (errno, errstr),
harvest_object, stage='Fetch')
return False
except urllib2.URLError:
self._save_object_error(
'Failed to fetch record.',
harvest_object, stage='Fetch')
return False
except httplib.BadStatusLine:
self._save_object_error(
'Bad HTTP response status line.',
harvest_object, stage='Fetch')
return False
if not metadata:
# Assume that there is no metadata and not an error.
# Should this be a cause for retry?
log.warning('No metadata: %s' % master_data['record'])
return False
if 'date' not in metadata.getMap() or not metadata.getMap()['date']:
self._save_object_error(
'Missing date: %s' % master_data['record'],
harvest_object, stage='Fetch')
return False
master_data['record'] = (header.identifier(), metadata.getMap())
# Do not save to database (because we can't json nor pickle _Element).
# The import stage.
# Gather all relevant information into a dictionary.
data = {
'identifier': master_data['record'][0],
'metadata': self._metadata(master_data['record'][1]),
'package_name': self._package_name_from_identifier(master_data['record'][0]),
'package_url': master_data['record'][1]['source'][0] if master_data['record'][1]['source'] else ''
}
return oai_dc2ckan(data, oai_dc_reader._namespaces, group, harvest_object)
def fetch_stage(self, harvest_object):
'''
The fetch stage will receive a HarvestObject object and will be
responsible for:
- getting the contents of the remote object (e.g. for a CSW server, perform a GetRecordById request).
- saving the content in the provided HarvestObject.
- creating and storing any suitable HarvestObjectErrors that may occur.
- returning True if everything went as expected, False otherwise.
:param harvest_object: HarvestObject object
:returns: True if everything went right, False if errors were found
'''
log.debug("fetch: %s", harvest_object.guid)
# Get metadata content from provider
try:
# Create a OAI-PMH Client
config = self._get_configuration(harvest_object)
registry = self.metadata_registry(config, harvest_object)
client = oaipmh.client.Client(harvest_object.job.source.url, registry)
# Get source URL
header, metadata, _about = client.getRecord(identifier=harvest_object.guid, metadataPrefix=self.md_format)
except Exception as e:
import traceback
traceback.print_exc()
self._save_object_error('Unable to get metadata from provider: {u}: {e}'.format(
u=harvest_object.source.url, e=e), harvest_object)
return False
if header and header.isDeleted():
return self.on_deleted(harvest_object, header)
# Get contents
try:
content = json.dumps(metadata.getMap())
except Exception as e:
import traceback
traceback.print_exc()
self._save_object_error('Unable to get content for package: {u}: {e}'.format(
u=harvest_object.source.url, e=e), harvest_object)
return False
# Save the fetched contents in the HarvestObject
harvest_object.content = content
harvest_object.save()
return True
def getrecord():
client.getRecord(identifier=self.TEST_ID, metadataPrefix='oai_dc')
def fetch_stage(self, harvest_object):
'''
The fetch stage will receive a HarvestObject object and will be
responsible for:
- getting the contents of the remote object (e.g. for a CSW server,
perform a GetRecordById request).
- saving the content in the provided HarvestObject.
- creating and storing any suitable HarvestObjectErrors that may
occur.
- returning True if everything went as expected, False otherwise.
:param harvest_object: HarvestObject object
:returns: True if everything went right, False if errors were found
'''
log.debug("in fetch stage: %s" % harvest_object.guid)
try:
self._set_config(harvest_object.job.source.config)
registry = self._create_metadata_registry()
client = oaipmh.client.Client(harvest_object.job.source.url,
registry,
self.credentials,
force_http_get=self.force_http_get)
record = None
try:
log.debug("Load %s with metadata prefix '%s'" %
(harvest_object.guid, self.md_format))
self._before_record_fetch(harvest_object)
record = client.getRecord(identifier=harvest_object.guid,
metadataPrefix=self.md_format)
self._after_record_fetch(record)
log.debug('record found!')
except:
log.exception('getRecord failed for %s' % harvest_object.guid)
self._save_object_error(
'Get record failed for %s!' % harvest_object.guid,
harvest_object)
return False
header, metadata, _ = record
log.debug('metadata %s' % metadata)
log.debug('header %s' % header)
try:
metadata_modified = header.datestamp().isoformat()
except:
metadata_modified = None
try:
content_dict = metadata.getMap()
content_dict['set_spec'] = header.setSpec()
if metadata_modified:
content_dict['metadata_modified'] = metadata_modified
log.debug(content_dict)
content = json.dumps(content_dict)
except:
log.exception('Dumping the metadata failed!')
self._save_object_error('Dumping the metadata failed!',
harvest_object)
return False
harvest_object.content = content
harvest_object.save()
except Exception as e:
log.exception(e)
self._save_object_error(
('Exception in fetch stage for %s: %r / %s' %
(harvest_object.guid, e, traceback.format_exc())),
harvest_object)
return False
return True
def test_fetch(url, record_id, fmt):
registry = importformats.create_metadata_registry()
client = oaipmh.client.Client(url, registry)
record = client.getRecord(identifier=record_id, metadataPrefix=fmt)
return record
def getrecord():
client.getRecord(identifier=self.TEST_ID, metadataPrefix='oai_dc')
def fetch_stage(self, harvest_object):
'''
The fetch stage will receive a HarvestObject object and will be
responsible for:
- getting the contents of the remote object (e.g. for a CSW server,
perform a GetRecordById request).
- saving the content in the provided HarvestObject.
- creating and storing any suitable HarvestObjectErrors that may
occur.
- returning True if everything went as expected, False otherwise.
:param harvest_object: HarvestObject object
:returns: True if everything went right, False if errors were found
'''
log.debug("in fetch stage: %s" % harvest_object.guid)
try:
self._set_config(harvest_object.job.source.config)
registry = self._create_metadata_registry()
client = oaipmh.client.Client(
harvest_object.job.source.url,
registry,
self.credentials,
force_http_get=self.force_http_get
)
record = None
try:
log.debug(
"Load %s with metadata prefix '%s'" %
(harvest_object.guid, self.md_format)
)
self._before_record_fetch(harvest_object)
record = client.getRecord(
identifier=harvest_object.guid,
metadataPrefix=self.md_format
)
self._after_record_fetch(record)
log.debug('record found!')
except:
log.exception('getRecord failed for %s' % harvest_object.guid)
self._save_object_error(
'Get record failed for %s!' % harvest_object.guid,
harvest_object
)
return False
header, metadata, _ = record
log.debug('metadata %s' % metadata)
log.debug('header %s' % header)
try:
metadata_modified = header.datestamp().isoformat()
except:
metadata_modified = None
try:
content_dict = metadata.getMap()
content_dict['set_spec'] = header.setSpec()
if metadata_modified:
content_dict['metadata_modified'] = metadata_modified
log.debug(content_dict)
content = json.dumps(content_dict)
except:
log.exception('Dumping the metadata failed!')
self._save_object_error(
'Dumping the metadata failed!',
harvest_object
)
return False
harvest_object.content = content
harvest_object.save()
except Exception, e:
log.exception(e)
self._save_object_error(
(
'Exception in fetch stage for %s: %r / %s'
% (harvest_object.guid, e, traceback.format_exc())
),
harvest_object
)
return False
def test_fetch(url, record_id, fmt):
registry = importformats.create_metadata_registry()
client = oaipmh.client.Client(url, registry)
record = client.getRecord(identifier=record_id, metadataPrefix=fmt)
return record
def fetch_stage(self, harvest_object):
'''
The fetch stage will receive a HarvestObject object and will be
responsible for:
- getting the contents of the remote object (e.g. for a CSW server,
perform a GetRecordById request).
- saving the content in the provided HarvestObject.
- creating and storing any suitable HarvestObjectErrors that may
occur.
- returning True if everything went as expected, False otherwise.
:param harvest_object: HarvestObject object
:returns: True if everything went right, False if errors were found
'''
log.debug("HDR: Fetch url %s" % harvest_object.job.source.url)
try:
self._set_config(harvest_object.job.source.config)
# Registry creation is dependant on job.source.config
# because of differentiation possibilities in
# namespaces for equal md_prefix.
log.debug('Application: ' + self.md_application)
log.debug('Md_format: ' + self.md_format)
log.debug('AddInfo: ' + self.additional_info)
# EPOS - trick to collect extra info via GFZ - solely intended for harvesting of GFZ
log.debug('Extra citation info URL: ' +
self.collect_extra_info_from_gfz)
registry = self._create_metadata_registry()
client = oaipmh.client.Client(
harvest_object.job.source.url,
registry,
self.credentials,
force_http_get=self.force_http_get
)
record = None
try:
self._before_record_fetch(harvest_object)
record = client.getRecord(
identifier=harvest_object.guid,
metadataPrefix=self.md_format
)
self._after_record_fetch(record)
except Exception:
log.exception('getRecord failed')
self._save_object_error('Get record failed!', harvest_object)
return False
header, metadata, _ = record
log.debug(record)
try:
metadata_modified = header.datestamp().isoformat()
except Exception:
metadata_modified = None
try:
content_dict = metadata.getMap()
# HDR? required still?
content_dict['set_spec'] = header.setSpec()
if metadata_modified:
content_dict['metadata_modified'] = metadata_modified
content = json.dumps(content_dict,
ensure_ascii=False,
encoding="utf-8")
except Exception:
log.exception('Dumping the metadata failed!')
self._save_object_error(
'Dumping the metadata failed!',
harvest_object
)
return False
harvest_object.content = content
harvest_object.save()
except Exception:
log.exception('Something went wrong 1!')
self._save_object_error(
'Exception in fetch stage',
harvest_object
)
return False
return True
def _fetch_import_record(self, harvest_object, master_data, client, group):
# The fetch part.
metadataPrefixes = []
if('metadata_formats' in self.config):
metadataPrefixes = self.config['metadata_formats']
if self.metadata_prefix_value not in metadataPrefixes:
metadataPrefixes.append(self.metadata_prefix_value)
data = {'metadata': {}, 'package_xml_save' : {}, 'package_resource' : {}}
data['identifier'] = master_data['record']
data['package_name'] = self._package_name_from_identifier(data['identifier'])
data['package_url'] = '%s?verb=GetRecord&identifier=%s&%s=%s' % (
harvest_object.job.source.url,
data['identifier'],
self.metadata_prefix_key,
self.metadata_prefix_value
)
for mdp in metadataPrefixes:
try:
header, metadata, _ = client.getRecord(metadataPrefix=mdp,
identifier=master_data['record'])
except XMLSyntaxError:
self._add_retry(harvest_object)
log.error('XML syntax error: %s' % master_data['record'])
self._save_object_error('Syntax error.', harvest_object, stage='Fetch')
if (mdp == self.metadata_prefix_value):
return False
else: continue
except socket.error:
self._add_retry(harvest_object)
errno, errstr = sys.exc_info()[:2]
self._save_object_error('Socket error OAI-PMH %s, details:\n%s' % (errno, errstr),
harvest_object,
stage='Fetch')
if (mdp == self.metadata_prefix_value):
return False
else: continue
except urllib2.URLError:
self._add_retry(harvest_object)
self._save_object_error('Failed to fetch record.', harvest_object, stage='Fetch')
if (mdp == self.metadata_prefix_value):
return False
else: continue
except httplib.BadStatusLine:
self._add_retry(harvest_object)
self._save_object_error('Bad HTTP response status line.', harvest_object, stage='Fetch')
if (mdp == self.metadata_prefix_value):
return False
else: continue
if not metadata:
# Assume that there is no metadata and not an error.
# Should this be a cause for retry?
log.warning('No metadata: %s' % master_data['record'])
#return False
# if 'date' not in metadata.getMap() or not metadata.getMap()['date']:
# self._add_retry(harvest_object)
# self._save_object_error('Missing date: %s' % master_data['record'], harvest_object, stage='Fetch')
# return False
#master_data['record'] = (header.identifier(), metadata.getMap())
# Do not save to database (because we can't json nor pickle _Element).
# The import stage.
# Gather all relevant information into a dictionary.
data['metadata'][mdp] = metadata.getMap()
try:
nowstr = datetime.datetime.now().strftime('%Y-%m-%dT%H:%M:%S.%f')
#fix for identifiers containing '/' char
esc_identifier = data['identifier'].replace('/','-');
label = '%s/%s-%s.xml' % (nowstr, esc_identifier,mdp)
resource_url = '%s?verb=GetRecord&identifier=%s&%s=%s' % (
harvest_object.job.source.url,
data['identifier'],
self.metadata_prefix_key,
mdp
)
f = urllib2.urlopen(resource_url)
x = f.read()
fileurl = pylons.configuration.config['ckan.site_url'] + pylons.configuration.config['ckan.api_url'] + h.url_for('storage_file', label=label) #quick fix for ckan in non-root url
data['package_xml_save'][mdp] = {
'label': label,
'xml': x
}
data['package_resource'][mdp] = {
'url': fileurl,
'description': 'Original ' + mdp + ' metadata record',
'format': 'xml',
'size': len(x)
}
except (urllib2.HTTPError, urllib2.URLError):
self._add_retry(harvest_object)
self._save_object_error('Could not get original metadata record!',
harvest_object, stage='Import')
if (mdp == self.metadata_prefix_value):
return False
else: continue
except socket.error:
self._add_retry(harvest_object)
errno, errstr = sys.exc_info()[:2]
self._save_object_error(
'Socket error original metadata record %s, details:\n%s' % (errno, errstr),
harvest_object, stage='Import')
if (mdp == self.metadata_prefix_value):
return False
else: continue
return oai_dc2ckan(data, kata_oai_dc_reader._namespaces, group, harvest_object)
