def _send_robotupload(obj, eng):
from invenio_workflows.errors import WorkflowError
from inspire.utils.robotupload import make_robotupload_marcxml
combined_callback_url = os.path.join(cfg["CFG_SITE_URL"], callback_url)
model = eng.workflow_definition.model(obj)
sip = model.get_latest_sip()
marcxml = sip.package
result = make_robotupload_marcxml(
url=url,
marcxml=marcxml,
callback_url=combined_callback_url,
mode=mode,
nonce=obj.id
)
if "[INFO]" not in result.text:
if "cannot use the service" in result.text:
# IP not in the list
obj.log.error("Your IP is not in "
"CFG_BATCHUPLOADER_WEB_ROBOT_RIGHTS "
"on host")
obj.log.error(result.text)
txt = "Error while submitting robotupload: {0}".format(result.text)
raise WorkflowError(txt, eng.uuid, obj.id)
else:
obj.log.info("Robotupload sent!")
obj.log.info(result.text)
eng.halt("Waiting for robotupload: {0}".format(result.text))
obj.log.info("end of upload")
def _send_robotupload(obj, eng):
from invenio.base.globals import cfg
from invenio.modules.workflows.errors import WorkflowError
from inspire.utils.robotupload import make_robotupload_marcxml
callback_url = os.path.join(cfg["CFG_SITE_URL"],
"callback/workflows/robotupload")
marcxml = obj.get_extra_data().get("marcxml")
if not marcxml:
obj.log.error("No MARCXML found in extra data.")
result = make_robotupload_marcxml(
url=None,
marcxml=marcxml,
callback_url=callback_url,
mode=mode,
nonce=obj.id
)
if "[INFO]" not in result.text:
if "cannot use the service" in result.text:
# IP not in the list
obj.log.error("Your IP is not in "
"CFG_BATCHUPLOADER_WEB_ROBOT_RIGHTS "
"on host")
obj.log.error(result.text)
txt = "Error while submitting robotupload: {0}".format(result.text)
raise WorkflowError(txt, eng.uuid, obj.id)
else:
obj.log.info("Robotupload sent!")
obj.log.info(result.text)
if mode != "holdingpen":
eng.halt("Waiting for robotupload: {0}".format(result.text))
obj.log.info("end of upload")
def test_robotupload_success(self):
from inspire.utils.robotupload import make_robotupload_marcxml
httpretty.register_uri(
httpretty.POST,
"http://localhost:4000/batchuploader/robotupload/insert",
body="[INFO] bibupload batchupload --insert /dummy/file/path\n",
status=200
)
valid_marcxml = "<record></record>"
response = make_robotupload_marcxml(
"http://localhost:4000",
valid_marcxml,
)
self.assertEqual(response.status_code, 200)
self.assertTrue("[INFO] bibupload batchupload" in response.text)
def test_robotupload_bad_xml(self):
from inspire.utils.robotupload import make_robotupload_marcxml
httpretty.register_uri(
httpretty.POST,
"http://localhost:4000/batchuploader/robotupload/insert",
body="[ERROR] MARCXML is not valid.\n",
status=400
)
invalid_marcxml = "record></record>"
response = make_robotupload_marcxml(
"http://localhost:4000",
invalid_marcxml,
)
self.assertEqual(response.status_code, 400)
self.assertTrue("not valid" in response.text)
def _send_robotupload_oaiharvest(obj, eng):
from invenio.modules.records.api import Record
from inspire.utils.robotupload import make_robotupload_marcxml
sequence_id = random.randrange(1, 60000)
arguments = obj.extra_data.get("repository", {}).get("arguments", {})
default_args = []
default_args.extend(['-I', str(sequence_id)])
if arguments.get('u_name', ""):
default_args.extend(['-N', arguments.get('u_name', "")])
if arguments.get('u_priority', 5):
default_args.extend(['-P', str(arguments.get('u_priority', 5))])
extract_path = os.path.join(
cfg['CFG_TMPSHAREDDIR'],
str(eng.uuid)
)
if not os.path.exists(extract_path):
os.makedirs(extract_path)
callback_url = os.path.join(cfg["CFG_SITE_URL"],
"callback/workflows/continue")
marcxml = Record(obj.data.dumps()).legacy_export_as_marc()
result = make_robotupload_marcxml(
url=url,
marcxml=marcxml,
callback_url=callback_url,
mode='insert',
nonce=obj.id
)
if "[INFO]" not in result.text:
if "cannot use the service" in result.text:
# IP not in the list
obj.log.error("Your IP is not in "
"CFG_BATCHUPLOADER_WEB_ROBOT_RIGHTS "
"on host")
obj.log.error(result.text)
from invenio.modules.workflows.errors import WorkflowError
txt = "Error while submitting robotupload: {0}".format(result.text)
raise WorkflowError(txt, eng.uuid, obj.id)
else:
obj.log.info("Robotupload sent!")
obj.log.info(result.text)
eng.halt("Waiting for robotupload: {0}".format(result.text))
obj.log.info("end of upload")
def test_robotupload_success_append(self):
"""Test proper handling when good MARCXML is sent."""
from inspire.utils.robotupload import make_robotupload_marcxml
httpretty.register_uri(
httpretty.POST,
"http://localhost:4000/batchuploader/robotupload/append",
body="[INFO] bibupload batchupload --append /dummy/file/path\n",
status=200
)
valid_marcxml = "<record></record>"
response = make_robotupload_marcxml(
"http://localhost:4000",
valid_marcxml,
mode="append",
)
self.assertEqual(response.status_code, 200)
self.assertTrue("[INFO] bibupload batchupload" in response.text)
def test_robotupload_callback_url(self):
from inspire.utils.robotupload import make_robotupload_marcxml
body = (
"[INFO] bibupload batchupload --insert /some/path"
"--callback-url http://localhost"
)
httpretty.register_uri(
httpretty.POST,
"http://localhost:4000/batchuploader/robotupload/insert",
body=body,
status=200
)
valid_marcxml = "<record></record>"
response = make_robotupload_marcxml(
"http://localhost:4000",
valid_marcxml,
callback_url="http://localhost",
)
self.assertEqual(response.status_code, 200)
self.assertTrue("--callback-url http://localhost" in response.text)
def _send_robotupload_deposit(obj, eng):
from invenio.modules.workflows.errors import WorkflowError
from inspire.utils.robotupload import make_robotupload_marcxml
callback_url = os.path.join(cfg["CFG_SITE_URL"],
"callback/workflows/robotupload")
deposition = Deposition(obj)
sip = deposition.get_latest_sip(deposition.submitted)
if not sip:
raise WorkflowError("No sip found", eng.uuid, obj.id)
if not deposition.submitted:
sip.seal()
deposition.update()
marcxml = sip.package
result = make_robotupload_marcxml(
url=url,
marcxml=marcxml,
callback_url=callback_url,
mode='insert',
nonce=obj.id
)
if "[INFO]" not in result.text:
if "cannot use the service" in result.text:
# IP not in the list
obj.log.error("Your IP is not in "
"CFG_BATCHUPLOADER_WEB_ROBOT_RIGHTS "
"on host")
obj.log.error(result.text)
txt = "Error while submitting robotupload: {0}".format(result.text)
raise WorkflowError(txt, eng.uuid, obj.id)
else:
obj.log.info("Robotupload sent!")
obj.log.info(result.text)
eng.halt("Waiting for robotupload: {0}".format(result.text))
obj.log.info("end of upload")
def _send_robotupload(obj, eng):
from invenio.modules.deposit.models import Deposition
from invenio.modules.workflows.errors import WorkflowError
from inspire.utils.robotupload import make_robotupload_marcxml
from invenio.base.globals import cfg
d = Deposition(obj)
sip = d.get_latest_sip(d.submitted)
if not sip:
raise WorkflowError("No sip found", eng.uuid, obj.id)
if not d.submitted:
sip.seal()
d.update()
if url is None:
base_url = cfg.get("CFG_ROBOTUPLOAD_SUBMISSION_BASEURL")
callback_url = os.path.join(cfg["CFG_SITE_URL"],
"callback/workflows/robotupload")
obj.log.info("Sending Robotupload to {0} with callback {1}".format(
base_url, callback_url))
result = make_robotupload_marcxml(url=base_url,
marcxml=sip.package,
callback_url=callback_url,
nonce=obj.id)
if "[INFO]" not in result.text:
if "cannot use the service" in result.text:
# IP not in the list
obj.log.error("Your IP is not in "
"CFG_BATCHUPLOADER_WEB_ROBOT_RIGHTS "
"on host")
obj.log.error(result.text)
from invenio.modules.workflows.errors import WorkflowError
txt = "Error while submitting robotupload: {0}".format(result.text)
raise WorkflowError(txt, eng.uuid, obj.id)
else:
obj.log.info("Robotupload sent!")
obj.log.info(result.text)
eng.halt("Waiting for robotupload: {0}".format(result.text))
def _update(obj, eng):
import dictdiffer
from lxml import objectify, etree
from invenio.base.globals import cfg
from invenio_workflows.utils import convert_marcxml_to_bibfield
from invenio_records.api import Record
from inspire.utils.robotupload import make_robotupload_marcxml
try:
recid = obj.extra_data["recid"]
except KeyError:
obj.log.error("Cannot locate record ID")
return
callback_url = os.path.join(cfg["CFG_SITE_URL"],
"callback/workflows/continue")
search_url = "%s?p=recid:%s&of=xm" % (cfg["WORKFLOWS_MATCH_REMOTE_SERVER_URL"], recid)
prod_data = objectify.parse(search_url)
# remove controlfields
root = prod_data.getroot()
record = root['record']
while True:
try:
record.remove(record['controlfield'])
except AttributeError:
break
prod_data = etree.tostring(record)
prod_data = convert_marcxml_to_bibfield(prod_data, model=["hep"])
new_data = dict(obj.data.dumps(clean=True))
prod_data = dict(prod_data.dumps(clean=True))
updated_keys = []
diff = dictdiffer.diff(prod_data, new_data)
for diff_type, new_key, content in diff:
if diff_type == 'add':
if new_key:
if isinstance(new_key, list):
# ['subject_terms', 0]
updated_keys.append(new_key[0])
else:
# 'subject_terms'
updated_keys.append(new_key)
else:
# content must be list of new adds
for key in content:
updated_keys.append(key)
updates = dictdiffer.patch(diff, new_data)
for key in updates.keys():
if key not in updated_keys:
del updates[key]
if updates:
updates['recid'] = recid
marcxml = Record(updates).legacy_export_as_marc()
result = make_robotupload_marcxml(
url=url,
marcxml=marcxml,
callback_url=callback_url,
mode='correct',
nonce=obj.id
)
if "[INFO]" not in result.text:
if "cannot use the service" in result.text:
# IP not in the list
obj.log.error("Your IP is not in "
"CFG_BATCHUPLOADER_WEB_ROBOT_RIGHTS "
"on host")
obj.log.error(result.text)
from invenio_workflows.errors import WorkflowError
txt = "Error while submitting robotupload: {0}".format(result.text)
raise WorkflowError(txt, eng.uuid, obj.id)
else:
obj.log.info("Robotupload sent!")
obj.log.info(result.text)
eng.halt("Waiting for robotupload: {0}".format(result.text))
obj.log.info("end of upload")
else:
obj.log.info("No updates to do.")
