def test_is_marked_returns_false_when_key_does_not_exist():
obj = MockObj({}, {})
eng = MockEng()
is_foo_marked = is_marked('foo')
assert not is_foo_marked(obj, eng)
def test_is_marked():
obj = MockObj({}, {'foo': 'bar'})
eng = MockEng()
is_foo_marked = is_marked('foo')
assert is_foo_marked(obj, eng)
def test_is_marked_returns_false_when_value_is_falsy():
obj = MockObj({}, {'foo': False})
eng = MockEng()
is_foo_marked = is_marked('foo')
assert not is_foo_marked(obj, eng)
def test_is_marked():
obj = MockObj({}, {'foo': 'bar'})
eng = MockEng()
is_foo_marked = is_marked('foo')
assert is_foo_marked(obj, eng)
def test_is_marked_returns_false_when_key_does_not_exist():
obj = MockObj({}, {})
eng = MockEng()
is_foo_marked = is_marked('foo')
assert not is_foo_marked(obj, eng)
def test_is_marked():
obj = StubObj({}, {'foo': 'bar'})
eng = DummyEng()
is_foo_marked = is_marked('foo')
assert is_foo_marked(obj, eng)
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,
validate_record('authors'),
IF_ELSE(
is_marked('is-update'),
SEND_UPDATE_NOTIFICATION,
ASK_FOR_REVIEW + [
IF_ELSE(
is_record_accepted,
(
SEND_TO_LEGACY + NOTIFY_ACCEPTED + [
# TODO: once legacy is out, this should become
# unconditional, and remove the SEND_TO_LEGACY
# steps
IF_NOT(in_production_mode, [store_record]),
] + CLOSE_TICKET_IF_NEEDED),
NOTIFY_NOT_ACCEPTED),
],
),
]
def test_is_marked_returns_false_when_value_is_falsy():
obj = MockObj({}, {'foo': False})
eng = MockEng()
is_foo_marked = is_marked('foo')
assert not is_foo_marked(obj, eng)
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"),
]),
]),
]
class Author(object):
"""Author ingestion workflow for HEPNames/Authors collection."""
name = "Author"
data_type = "authors"
workflow = [
load_from_source_data,
# Make sure schema is set for proper indexing in Holding Pen
set_schema,
validate_record('authors'),
IF_ELSE(
is_marked('is-update'),
[
SEND_TO_LEGACY,
SEND_UPDATE_NOTIFICATION,
],
[
ASK_FOR_REVIEW,
IF_ELSE(is_record_accepted,
([store_record] + SEND_TO_LEGACY + NOTIFY_ACCEPTED +
CLOSE_TICKET_IF_NEEDED), NOTIFY_NOT_ACCEPTED),
],
),
]
keep_new=True
),
]
ADD_INGESTION_MARKS = [
# 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(
is_marked('already-in-holding-pen'),
[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:
# time based filtering (5 days is quite random now).
IF(
previously_rejected(),
[
mark('already-ingested', True),
mark('stop', True),
]
),
]
),
]
DELETE_AND_STOP_IF_NEEDED = [
IF(
is_marked('delete'),
[
update_existing_workflow_object,
# TODO: Wen we get to fix refextract, we can remove the
# following step as the references will be good enough to
# trust
delete_self_and_stop_processing
]
),
IF(
is_marked('stop'),
[stop_processing]
),
]
NOTIFY_ACCEPTED = [
IF(
is_submission,
reply_ticket(
template='literaturesuggest/tickets/user_accepted.html',
context_factory=reply_ticket_context,
),
),
]
NOTIFY_CURATOR_IF_CORE = [
IF_NOT(
is_marked('is-update'),
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',
),
),
),
]
POSTENHANCE_RECORD = [
add_core,
NOTIFY_ACCEPTED = [
IF(
is_submission,
do_not_repeat('reply_ticket_user_submission_accepted')(
reply_ticket(
template='literaturesuggest/tickets/user_accepted.html',
context_factory=reply_ticket_context,
),
),
),
]
NOTIFY_CURATOR_IF_NEEDED = [
IF_NOT(
is_marked('is-update'),
[
IF_ELSE(
jlab_ticket_needed,
do_not_repeat('create_ticket_jlab_curation')(
create_ticket(
template='literaturesuggest/tickets/curation_jlab.html',
queue='HEP_curation_jlab',
context_factory=curation_ticket_context,
ticket_id_key='curation_ticket_id',
),
),
IF(
curation_ticket_needed,
do_not_repeat('create_ticket_curator_core_curation')(
create_ticket(
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),
]),
]),
]
DELETE_AND_STOP_IF_NEEDED = [
IF(
is_marked('delete'),
[
update_existing_workflow_object,
# TODO: Wen we get to fix refextract, we can remove the
# following step as the references will be good enough to
# trust
delete_self_and_stop_processing
]),
IF(is_marked('stop'), [stop_processing]),
]
ENHANCE_RECORD = [
# Article Processing
# ==================
IF(is_arxiv_paper, [
arxiv_fulltext_download,
stop_processing,
]
NOTIFY_ACCEPTED = [
IF(
is_submission,
do_not_repeat('reply_ticket_user_submission_accepted')(reply_ticket(
template='literaturesuggest/tickets/user_accepted.html',
context_factory=reply_ticket_context,
), ),
),
]
NOTIFY_CURATOR_IF_NEEDED = [
IF_NOT(
is_marked('is-update'),
[
IF_ELSE(
jlab_ticket_needed,
do_not_repeat('create_ticket_jlab_curation')(create_ticket(
template='literaturesuggest/tickets/curation_jlab.html',
queue='HEP_curation_jlab',
context_factory=curation_ticket_context,
ticket_id_key='curation_ticket_id',
), ),
IF(
curation_ticket_needed, # if core
IF_ELSE(
check_source_publishing, # if it is coming from publisher, create ticket in hep_publisher queue
do_not_repeat('create_ticket_curator_core_publisher')
(create_ticket(
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")
]
