def test_halt_record_accepts_custom_action():
obj = MockObj({}, {})
eng = MockEng()
foo_action_halt_record = halt_record(action='foo')
assert foo_action_halt_record(obj, eng) is None
assert eng.action == 'foo'
def test_halt_record_accepts_custom_msg():
obj = MockObj({}, {})
eng = MockEng()
bar_message_halt_record = halt_record(message='bar')
assert bar_message_halt_record(obj, eng) is None
assert eng.msg == 'bar'
def test_halt_record_accepts_custom_msg():
obj = MockObj({}, {})
eng = MockEng()
bar_message_halt_record = halt_record(message='bar')
assert bar_message_halt_record(obj, eng) is None
assert eng.msg == 'bar'
def test_halt_record_accepts_custom_action():
obj = MockObj({}, {})
eng = MockEng()
foo_action_halt_record = halt_record(action='foo')
assert foo_action_halt_record(obj, eng) is None
assert eng.action == 'foo'
def test_halt_record():
obj = MockObj({}, {'halt_action': 'foo', 'halt_message': 'bar'})
eng = MockEng()
default_halt_record = halt_record()
assert default_halt_record(obj, eng) is None
assert eng.action == 'foo'
assert eng.msg == 'bar'
def test_halt_record():
obj = MockObj({}, {'halt_action': 'foo', 'halt_message': 'bar'})
eng = MockEng()
default_halt_record = halt_record()
assert default_halt_record(obj, eng) is None
assert eng.action == 'foo'
assert eng.msg == 'bar'
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"),
]),
]),
]
NOTIFY_ALREADY_EXISTING
)
)
]
HALT_FOR_APPROVAL_IF_NEW_OR_STOP_IF_NOT_RELEVANT = [
preserve_root,
IF_ELSE(
is_marked('is-update'),
[
merge_articles,
IF(
has_conflicts,
halt_record(
action='merge_approval',
message='Submission halted for merging conflicts.'
),
),
mark('approved', True),
mark('merged', True),
],
IF_ELSE(
is_marked('auto-approved'),
mark('approved', True),
[
IF_NOT(
is_record_relevant,
[
reject_record('Article automatically rejected'),
mark('approved', False),
save_workflow,
),
]
ASK_FOR_REVIEW = [
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?",
),
]
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'),
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")
]
]
ASK_FOR_REVIEW = [
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?",
),
]
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'),
is_record_relevant,
[
reject_record('Article automatically rejected'),
mark('approved', False),
save_workflow,
stop_processing,
],
),
IF_ELSE(
is_marked('is-update'),
[
merge_articles,
IF(
has_conflicts,
halt_record(
action='merge_approval',
message='Submission halted for merging conflicts.'
),
),
mark('approved', True),
mark('merged', True),
],
IF_ELSE(
is_marked('auto-approved'),
mark('approved', True),
halt_record(
action="hep_approval",
message="Submission halted for curator approval.",
)
),
),
]
),
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
# =====================================
]
CHECK_IF_SUBMISSION_AND_ASK_FOR_APPROVAL = [
IF_ELSE(is_record_relevant, [
halt_record(
action="hep_approval",
message="Submission halted for curator approval.",
)
], [reject_record("Article automatically rejected"), stop_processing]),
]
NOTIFY_NOT_ACCEPTED = [
IF(is_submission, [reply_ticket(context_factory=reply_ticket_context)])
]
NOTIFY_ALREADY_EXISTING = [
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"),
# [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
# =====================================
]
CHECK_IF_SUBMISSION_AND_ASK_FOR_APPROVAL = [
IF_ELSE(
is_record_relevant,
[halt_record(
action="hep_approval",
message="Submission halted for curator approval.",
)],
[
reject_record("Article automatically rejected"),
stop_processing
]
),
]
NOTIFY_NOT_ACCEPTED = [
IF(
is_submission,
[reply_ticket(context_factory=reply_ticket_context)]
)
]
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
# =====================================
]
CHECK_IF_SUBMISSION_AND_ASK_FOR_APPROVAL = [
IF_ELSE(
is_record_relevant, [halt_record(action="hep_approval")],
[reject_record("Article automatically rejected"), stop_processing]),
]
NOTIFY_NOT_ACCEPTED = [
IF(is_submission, [reply_ticket(context_factory=reply_ticket_context)])
]
NOTIFY_ALREADY_EXISTING = [
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"),
]
