class Author(object):
"""Author ingestion workflow for HEPNames/Authors collection."""
name = "Author"
data_type = "authors"
workflow = [
# Make sure schema is set for proper indexing in Holding Pen
set_schema,
# Emit record signals to receive metadata enrichment
emit_record_signals,
IF_ELSE(is_marked('is-update'), [
send_robotupload(marcxml_processor=hepnames2marc,
mode="holdingpen"),
create_ticket(
template="authors/tickets/curator_update.html",
queue="Authors_cor_user",
context_factory=update_ticket_context,
),
], [
create_ticket(template="authors/tickets/curator_new.html",
queue="Authors_add_user",
context_factory=new_ticket_context),
reply_ticket(template="authors/tickets/user_new.html",
context_factory=reply_ticket_context,
keep_new=True),
halt_record(action="author_approval",
message="Accept submission?"),
IF_ELSE(is_record_accepted, [
send_robotupload(marcxml_processor=hepnames2marc,
mode="insert"),
reply_ticket(template="authors/tickets/user_accepted.html",
context_factory=reply_ticket_context),
close_ticket(ticket_id_key="ticket_id"),
IF(curation_ticket_needed, [
create_ticket(
template="authors/tickets/curation_needed.html",
queue="AUTHORS_curation",
context_factory=curation_ticket_context,
ticket_id_key="curation_ticket_id"),
]),
], [
close_ticket(ticket_id_key="ticket_id"),
]),
]),
]
def test_create_ticket_calls_tickets_create_with_template(
mock_create_ticket_with_template, mock_user):
mock_user.query.get.return_value = MockUser('*****@*****.**')
data = {
'titles': [
{
'title': 'Partial Symmetries of Weak Interactions'
},
],
}
template = 'template_path'
extra_data = {'recid': '1'}
obj = MockObj(data, extra_data)
eng = MockEng()
_create_ticket = create_ticket(template=template)
_create_ticket(obj, eng)
mock_create_ticket_with_template.assert_called_with(
'Test', '*****@*****.**', template, {}, None, extra_data['recid'])
def test_create_ticket_handles_unicode_when_in_production_mode(g_i, r_t, user):
user.query.get.return_value = MockUser('*****@*****.**')
r_t.return_value = u'φοο'
g_i.return_value = StubRTInstance()
config = {'PRODUCTION_MODE': True}
with patch.dict(current_app.config, config):
data = {}
extra_data = {}
files = MockFiles({})
obj = MockObj(data, extra_data, files)
eng = MockEng()
_create_ticket = create_ticket(None)
assert _create_ticket(obj, eng) is None
assert obj.extra_data == {'ticket_id': 1}
assert u'Ticket 1 created:\nφοο' in obj.log._info.getvalue()
def test_create_ticket_calls_tickets_create_with_template(mock_create_ticket_with_template, mock_user):
mock_user.query.get.return_value = MockUser('*****@*****.**')
data = {
'titles': [
{'title': 'Partial Symmetries of Weak Interactions'},
],
}
template = 'template_path'
extra_data = {'recid': '1'}
obj = MockObj(data, extra_data)
eng = MockEng()
_create_ticket = create_ticket(template=template)
_create_ticket(obj, eng)
mock_create_ticket_with_template.assert_called_with(
'Test',
'*****@*****.**',
template,
{},
None,
extra_data['recid']
)
def test_create_ticket_handles_unicode_when_not_in_production_mode(r_t, user):
user.query.get.return_value = MockUser('*****@*****.**')
r_t.return_value = u'φοο'
config = {'PRODUCTION_MODE': False}
with patch.dict(current_app.config, config):
data = {}
extra_data = {}
files = MockFiles({})
obj = MockObj(data, extra_data, files)
eng = MockEng()
_create_ticket = create_ticket(None)
expected = (
u'Was going to create ticket: None\n\nφοο\n\n'
u'To: [email protected] Queue: Test'
)
assert _create_ticket(obj, eng) is None
assert expected == obj.log._info.getvalue()
wait_webcoll,
)
from inspirehep.modules.workflows.utils import do_not_repeat
from inspirehep.modules.literaturesuggest.tasks import (
curation_ticket_needed,
reply_ticket_context,
new_ticket_context,
curation_ticket_context,
)
NOTIFY_SUBMISSION = [
do_not_repeat('create_ticket_curator_new_submission')(
create_ticket(
template="literaturesuggest/tickets/curator_submitted.html",
queue="HEP_add_user",
context_factory=new_ticket_context,
ticket_id_key="ticket_id"
),
),
do_not_repeat('reply_ticket_user_new_submission')(
reply_ticket(
template="literaturesuggest/tickets/user_submitted.html",
context_factory=reply_ticket_context,
keep_new=True
),
)
]
CHECK_AUTO_APPROVE = [
IF_ELSE(
is_submission,
)
SEND_TO_LEGACY = [
send_robotupload(mode="insert"),
]
NOTIFY_ACCEPTED = [
reply_ticket(template="authors/tickets/user_accepted.html",
context_factory=reply_ticket_context),
close_ticket(ticket_id_key="ticket_id"),
]
CLOSE_TICKET_IF_NEEDED = [
IF(curation_ticket_needed, [
create_ticket(template="authors/tickets/curation_needed.html",
queue="AUTHORS_curation",
context_factory=curation_ticket_context,
ticket_id_key="curation_ticket_id"),
]),
]
NOTIFY_NOT_ACCEPTED = [
close_ticket(ticket_id_key="ticket_id"),
]
SEND_UPDATE_NOTIFICATION = [
send_robotupload(mode="holdingpen", callback_url=None),
create_ticket(
template="authors/tickets/curator_update.html",
queue="Authors_cor_user",
context_factory=update_ticket_context,
),
reply_ticket,
send_to_legacy,
)
from inspirehep.modules.workflows.utils import do_not_repeat
from inspirehep.modules.literaturesuggest.tasks import (
curation_ticket_needed,
check_source_publishing,
reply_ticket_context,
new_ticket_context,
curation_ticket_context,
)
NOTIFY_SUBMISSION = [
do_not_repeat('create_ticket_curator_new_submission')(create_ticket(
template="literaturesuggest/tickets/curator_submitted.html",
queue="HEP_add_user",
context_factory=new_ticket_context,
ticket_id_key="ticket_id"), ),
do_not_repeat('reply_ticket_user_new_submission')(reply_ticket(
template="literaturesuggest/tickets/user_submitted.html",
context_factory=reply_ticket_context,
keep_new=True), )
]
CHECK_AUTO_APPROVE = [
IF_ELSE(
is_submission,
mark('auto-approved', False),
IF_ELSE(
auto_approve,
[
send_robotupload(mode="insert"),
]
NOTIFY_ACCEPTED = [
do_not_repeat('reply_ticket_author_submission_accepted')(reply_ticket(
template="authors/tickets/user_accepted.html",
context_factory=reply_ticket_context)),
do_not_repeat('close_ticket_author_submission_accepted')(
close_ticket(ticket_id_key="ticket_id")),
]
CLOSE_TICKET_IF_NEEDED = [
IF(curation_ticket_needed, [
do_not_repeat('create_ticket_author_submission_curation_needed')(
create_ticket(template="authors/tickets/curation_needed.html",
queue="AUTHORS_curation",
context_factory=curation_ticket_context,
ticket_id_key="curation_ticket_id")),
]),
]
NOTIFY_NOT_ACCEPTED = [
close_ticket(ticket_id_key="ticket_id"),
]
SEND_UPDATE_NOTIFICATION = [
send_robotupload(mode="holdingpen", callback_url=None),
do_not_repeat('create_ticket_author_submission_curator_update')(
create_ticket(
template="authors/tickets/curator_update.html",
queue="Authors_cor_user",
context_factory=update_ticket_context,
class Article(object):
"""Article ingestion workflow for Literature collection."""
name = "HEP"
data_type = "hep"
workflow = [
# Make sure schema is set for proper indexing in Holding Pen
set_schema,
# Emit record signals to receive metadata enrichment
emit_record_signals,
# Query locally or via legacy search API to see if article
# is already ingested and this is an update
IF(article_exists, [
mark('match-found', True),
]),
IF_ELSE(
is_submission,
[
# Article matching for submissions
# ================================
IF(pending_in_holding_pen, [
mark('already-in-holding-pen', True),
]),
# Special RT integration for submissions
# ======================================
create_ticket(
template="literaturesuggest/tickets/curator_submitted.html",
queue="HEP_add_user",
context_factory=new_ticket_context,
ticket_id_key="ticket_id"),
reply_ticket(
template="literaturesuggest/tickets/user_submitted.html",
context_factory=reply_ticket_context,
keep_new=True),
],
[
# Article matching for non-submissions
# ====================================
# Query holding pen to see if we already have this article ingested
#
# NOTE on updates:
# If the same article has been harvested before and the
# ingestion has been completed, process is continued
# to allow for updates.
IF(pending_in_holding_pen, [
mark('already-in-holding-pen', True),
mark('delete', True),
]),
IF(
is_arxiv_paper,
[
# FIXME: This filtering step should be removed when this
# workflow includes arXiv CORE harvesting
IF(already_harvested, [
mark('already-ingested', True),
mark('stop', True),
]),
# FIXME: This filtering step should be removed when:
# old previously rejected records are treated
# differently e.g. good auto-reject heuristics or better
# time based filtering (5 days is quite random now).
IF(previously_rejected(), [
mark('already-ingested', True),
mark('stop', True),
]),
]),
IF(is_marked('delete'),
[update_old_object, delete_self_and_stop_processing]),
IF(is_marked('stop'), [stop_processing]),
]),
#
# Article Processing
# ==================
IF(is_arxiv_paper, [
arxiv_fulltext_download,
arxiv_plot_extract,
arxiv_refextract,
arxiv_author_list("authorlist2marcxml.xsl"),
]),
extract_journal_info,
classify_paper(
taxonomy="HEPont.rdf",
only_core_tags=False,
spires=True,
with_author_keywords=True,
),
filter_core_keywords,
guess_categories,
IF(is_experimental_paper, [
guess_experiments,
]),
guess_keywords,
# Predict action for a generic HEP paper based only on title
# and abstract.
guess_coreness, # ("arxiv_skip_astro_title_abstract.pickle)
# Check if we shall halt or auto-reject
# =====================================
# NOTE: User submissions are always relevant
IF_ELSE(is_record_relevant, [
halt_record(action="hep_approval"),
], [reject_record("Article automatically rejected"), stop_processing]),
IF_ELSE(is_record_accepted, [
IF(article_exists, [
IF_ELSE(is_submission, [
reject_record('Article was already found on INSPIRE'),
stop_processing,
reply_ticket(
template=
"literaturesuggest/tickets/user_rejected_exists.html",
context_factory=reply_ticket_context),
close_ticket(ticket_id_key="ticket_id"),
], [
halt_record(action="merge_approval"),
]),
]),
add_core,
add_note_entry,
filter_keywords,
user_pdf_get,
IF_ELSE(shall_push_remotely, [
IF_ELSE(article_exists, [
prepare_update_payload(extra_data_key="update_payload"),
send_robotupload(marcxml_processor=hep2marc,
mode="correct",
extra_data_key="update_payload"),
], [
send_robotupload(marcxml_processor=hep2marc,
mode="insert"),
])
], [store_record]),
IF(is_submission, [
IF(curation_ticket_needed, [
create_ticket(
template="literaturesuggest/tickets/curation_core.html",
queue="HEP_curation",
context_factory=curation_ticket_context,
ticket_id_key="curation_ticket_id")
]),
reply_ticket(
template="literaturesuggest/tickets/user_accepted.html",
context_factory=reply_ticket_context),
]),
], [
IF(is_submission,
[reply_ticket(context_factory=reply_ticket_context)])
]),
close_ticket(ticket_id_key="ticket_id")
]
]
NOTIFY_ACCEPTED = [
reply_ticket(
template="authors/tickets/user_accepted.html",
context_factory=reply_ticket_context),
close_ticket(ticket_id_key="ticket_id"),
]
CLOSE_TICKET_IF_NEEDED = [
IF(curation_ticket_needed, [
create_ticket(
template="authors/tickets/curation_needed.html",
queue="AUTHORS_curation",
context_factory=curation_ticket_context,
ticket_id_key="curation_ticket_id"
),
]),
]
NOTIFY_NOT_ACCEPTED = [
close_ticket(ticket_id_key="ticket_id"),
]
SEND_UPDATE_NOTIFICATION = [
send_robotupload(
mode="holdingpen",
callback_url=None
)
from inspirehep.modules.literaturesuggest.tasks import (
curation_ticket_needed,
reply_ticket_context,
new_ticket_context,
curation_ticket_context,
)
NOTIFY_SUBMISSION = [
# Special RT integration for submissions
# ======================================
create_ticket(
template="literaturesuggest/tickets/curator_submitted.html",
queue="HEP_add_user",
context_factory=new_ticket_context,
ticket_id_key="ticket_id"
),
reply_ticket(
template="literaturesuggest/tickets/user_submitted.html",
context_factory=reply_ticket_context,
keep_new=True
),
]
ADD_INGESTION_MARKS = [
# Article matching for non-submissions
# ====================================
# Query holding pen to see if we already have this article ingested
#
