def populate_cnums():
"""Populate table seqSTORE with the cnums present in CONFERENCE records."""
# First get all records from conference collection
conf_records = perform_request_search(cc="Conferences",
p="111__g:C*", rg=0)
for recid in conf_records:
cnum = record_get_field_value(
get_bibrecord(recid), tag="111", ind1="", ind2="", code="g")
if cnum:
if not _cnum_exists(cnum):
_insert_cnum(cnum)
print("cnum %s from record %s inserted" % (cnum, recid))
def _next_value(self, recid=None, xml_record=None, start_date=None):
"""
Returns the next cnum for the given recid
@param recid: id of the record where the cnum will be generated
@type recid: int
@param xml_record: record in xml format
@type xml_record: string
@param start_date: use given start date
@type start_date: string
@return: next cnum for the given recid. Format is Cyy-mm-dd.[.1n]
@rtype: string
@raises ConferenceNoStartDateError: No date information found in the
given recid
"""
bibrecord = None
if recid is None and xml_record is not None:
bibrecord = create_record(xml_record)[0]
elif recid is not None:
bibrecord = get_bibrecord(recid)
if start_date is None and bibrecord is not None:
start_date = record_get_field_value(bibrecord,
tag="111",
ind1="",
ind2="",
code="x")
if not start_date:
raise ConferenceNoStartDateError
base_cnum = "C" + start_date[2:]
record_cnums = self._get_record_cnums(base_cnum)
if not record_cnums:
new_cnum = base_cnum
else:
# Get the max current revision, cnums are in format Cyy-mm-dd,
# Cyy-mm-dd.1, Cyy-mm-dd.2
highest_revision = max([0] + [
int(rev[0].split('.')[1])
for rev in record_cnums if '.' in rev[0]
])
new_cnum = base_cnum + '.' + str(highest_revision + 1)
return new_cnum
def _next_value(self, recid=None, xml_record=None, start_date=None):
"""
Returns the next cnum for the given recid
@param recid: id of the record where the cnum will be generated
@type recid: int
@param xml_record: record in xml format
@type xml_record: string
@param start_date: use given start date
@type start_date: string
@return: next cnum for the given recid. Format is Cyy-mm-dd.[.1n]
@rtype: string
@raises ConferenceNoStartDateError: No date information found in the
given recid
"""
bibrecord = None
if recid is None and xml_record is not None:
bibrecord = create_record(xml_record)[0]
elif recid is not None:
bibrecord = get_bibrecord(recid)
if start_date is None and bibrecord is not None:
start_date = record_get_field_value(bibrecord,
tag="111",
ind1="",
ind2="",
code="x")
if not start_date:
raise ConferenceNoStartDateError
base_cnum = "C" + start_date[2:]
record_cnums = self._get_record_cnums(base_cnum)
if not record_cnums:
new_cnum = base_cnum
else:
# Get the max current revision, cnums are in format Cyy-mm-dd,
# Cyy-mm-dd.1, Cyy-mm-dd.2
highest_revision = max([0] + [int(rev[0].split('.')[1]) for rev in record_cnums if '.' in rev[0]])
new_cnum = base_cnum + '.' + str(highest_revision + 1)
return new_cnum
def _create_ticket(recid, bibcatalog_system, queue):
subject = "Refs for #%s" % recid
if CFG_INSPIRE_SITE:
# Add report number in the subjecet
report_number = ""
record = get_bibrecord(recid)
in_core = False
for collection_tag in record_get_field_instances(record, "980"):
for collection in field_get_subfield_values(collection_tag, 'a'):
if collection == 'CORE':
in_core = True
if collection == 'arXiv':
# Do not create tickets for arxiv papers
# Tickets for arxiv papers are created in bibcatelog
write_message("arXiv paper", verbose=1)
return
# Only create tickets for HEP
if not in_core:
write_message("not in hep", verbose=1)
return
# Do not create tickets for old records
creation_date = run_sql("""SELECT creation_date FROM bibrec
WHERE id = %s""", [recid])[0][0]
if creation_date < datetime.now() - timedelta(days=30*4):
return
for report_tag in record_get_field_instances(record, "037"):
for category in field_get_subfield_values(report_tag, 'c'):
if category.startswith('astro-ph'):
write_message("astro-ph", verbose=1)
# We do not curate astro-ph
return
for report_number in field_get_subfield_values(report_tag, 'a'):
subject += " " + report_number
break
text = '%s/record/edit/#state=edit&recid=%s' % (CFG_SITE_SECURE_URL,
recid)
bibcatalog_system.ticket_submit(subject=subject,
queue=queue,
text=text,
recordid=recid)
def enumerate_records(records):
"""
Given an array of record IDs this function will yield a
triplet of the count (starting from 0), the record ID and
the record object.
@param record: Array of record IDs
@type record: int
@yield: tuple (count, recordId, record structure (dict))
"""
for i, recid in enumerate(records):
record = get_bibrecord(int(recid))
if not record:
write_message("Error: could not load record '%s'." % (recid,))
continue
yield i, int(recid), AmendableRecord(record)
def load_records_from_id(records):
"""
Given a record tuple of record id and last updated/created date,
this function will yield a tuple with the record id replaced with
a record structure iterativly.
@param record: tuple of (recid, date-string) Ex: (1, 2012-12-12 12:12:12)
@type record: tuple
@yield: tuple of (record structure (dict), date-string)
"""
for recid, date in records:
record = get_bibrecord(int(recid))
if not record:
write_message("Error: could not load record %s" % (recid,))
continue
yield record, date
def populate_cnums():
"""Populate table seqSTORE with the cnums present in CONFERENCE records."""
# First get all records from conference collection
conf_records = perform_request_search(cc="Conferences",
p="111__g:C*",
rg=0)
for recid in conf_records:
cnum = record_get_field_value(get_bibrecord(recid),
tag="111",
ind1="",
ind2="",
code="g")
if cnum:
if not _cnum_exists(cnum):
_insert_cnum(cnum)
print("cnum %s from record %s inserted" % (cnum, recid))
def create_ticket(recid, bibcatalog_system, queue=CFG_REFEXTRACT_TICKET_QUEUE):
write_message('bibcatalog_system %s' % bibcatalog_system, verbose=1)
write_message('queue %s' % queue, verbose=1)
if bibcatalog_system and queue:
subject = "Refs for #%s" % recid
# Add report number in the subjecet
report_number = ""
record = get_bibrecord(recid)
in_hep = False
for collection_tag in record_get_field_instances(record, "980"):
for collection in field_get_subfield_values(collection_tag, 'a'):
if collection == 'HEP':
in_hep = True
# Only create tickets for HEP
if not in_hep:
write_message("not in hep", verbose=1)
return
for report_tag in record_get_field_instances(record, "037"):
for category in field_get_subfield_values(report_tag, 'c'):
if category.startswith('astro-ph'):
write_message("astro-ph", verbose=1)
# We do not curate astro-ph
return
for report_number in field_get_subfield_values(report_tag, 'a'):
subject += " " + report_number
break
text = '%s/record/edit/#state=edit&recid=%s' % (CFG_SITE_SECURE_URL, \
recid)
bibcatalog_system.ticket_submit(subject=subject,
queue=queue,
text=text,
recordid=recid)
def _next_value(self, recid=None, xml_record=None, bibrecord=None):
"""
Returns the next texkey for the given recid
@param recid: id of the record where the texkey will be generated
@type recid: int
@param xml_record: record in xml format
@type xml_record: string
@return: next texkey for the given recid.
@rtype: string
@raises TexkeyNoAuthorError: No main author (100__a) or collaboration
(710__g) in the given recid
"""
if recid is None and xml_record is not None:
bibrecord = create_record(xml_record)[0]
elif bibrecord is None:
bibrecord = get_bibrecord(recid)
main_author = record_get_field_value(bibrecord,
tag="100",
ind1="",
ind2="",
code="a")
if not main_author:
# Try with collaboration name
main_author = record_get_field_value(bibrecord,
tag="710",
ind1="",
ind2="",
code="g")
main_author = "".join([
p for p in main_author.split() if p.lower() != "collaboration"
])
if not main_author:
# Try with corporate author
main_author = record_get_field_value(bibrecord,
tag="100",
ind1="",
ind2="",
code="a")
if not main_author:
raise TexkeyNoAuthorError
# Remove utf-8 special characters
main_author = unidecode(main_author.decode('utf-8'))
try:
texkey_first_part = main_author.split(',')[0].replace(" ", "")
except KeyError:
texkey_first_part = ""
year = record_get_field_value(bibrecord,
tag="269",
ind1="",
ind2="",
code="c")
if not year:
year = record_get_field_value(bibrecord,
tag="260",
ind1="",
ind2="",
code="c")
if not year:
year = record_get_field_value(bibrecord,
tag="773",
ind1="",
ind2="",
code="y")
if not year:
year = record_get_field_value(bibrecord,
tag="502",
ind1="",
ind2="",
code="d")
if not year:
raise TexkeyNoYearError
try:
texkey_second_part = year.split("-")[0]
except KeyError:
texkey_second_part = ""
texkey_third_part = _texkey_random_chars(recid)
texkey = texkey_first_part + ":" + texkey_second_part + texkey_third_part
tries = 0
while self._value_exists(texkey) and tries < TEXKEY_MAXTRIES:
# Key is already in the DB, generate a new one
texkey_third_part = _texkey_random_chars(recid, use_random=True)
texkey = texkey_first_part + ":" + texkey_second_part + texkey_third_part
tries += 1
return texkey
def _next_value(self, recid=None, xml_record=None, bibrecord=None):
"""Return the next texkey for the given recid.
:param recid: id of the record where the texkey will be generated
:type recid: int
:param xml_record: record in xml format
:type xml_record: string
:return: next texkey for the given recid.
:rtype: string
:raises TexkeyNoAuthorError: No main author (100__a) or collaboration
(710__g) in the given recid
"""
if recid is None and xml_record is not None:
bibrecord = create_record(xml_record)[0]
elif bibrecord is None:
bibrecord = get_bibrecord(recid)
main_author = record_get_field_value(bibrecord,
tag="100",
ind1="",
ind2="",
code="a")
if not main_author:
# Try with collaboration name
main_author = record_get_field_value(bibrecord,
tag="710",
ind1="",
ind2="",
code="g")
main_author = "".join([p for p in main_author.split()
if p.lower() != "collaboration"])
if not main_author:
# Try with corporate author
main_author = record_get_field_value(bibrecord,
tag="100",
ind1="",
ind2="",
code="a")
if not main_author:
raise TexkeyNoAuthorError
# Remove utf-8 special characters
main_author = unidecode(main_author.decode('utf-8'))
try:
texkey_first_part = main_author.split(',')[0].replace(" ", "")
except KeyError:
texkey_first_part = ""
year = record_get_field_value(bibrecord,
tag="269",
ind1="",
ind2="",
code="c")
if not year:
year = record_get_field_value(bibrecord,
tag="260",
ind1="",
ind2="",
code="c")
if not year:
year = record_get_field_value(bibrecord,
tag="773",
ind1="",
ind2="",
code="y")
if not year:
year = record_get_field_value(bibrecord,
tag="502",
ind1="",
ind2="",
code="d")
if not year:
raise TexkeyNoYearError
try:
texkey_second_part = year.split("-")[0]
except KeyError:
texkey_second_part = ""
texkey_third_part = _texkey_random_chars(recid)
texkey = texkey_first_part + ":" + \
texkey_second_part + texkey_third_part
tries = 0
while self._value_exists(texkey) and tries < TEXKEY_MAXTRIES:
# Key is already in the DB, generate a new one
texkey_third_part = _texkey_random_chars(recid, use_random=True)
texkey = texkey_first_part + ":" + \
texkey_second_part + texkey_third_part
tries += 1
return texkey
