def main():
verbose = '-v' in sys.argv
recids = perform_request_search(p='-035:spirestex -035:inspiretex', cc='HEP')
print "Found %s records to assign texkeys" % len(recids)
processed = []
to_process = []
for count, recid in enumerate(recids):
if count % 300 == 0:
print 'done %s of %s' % (count, len(recids))
if verbose:
print "processing ", recid
# Check that the record does not have already a texkey
has_texkey = False
recstruct = get_record(recid)
for instance in record_get_field_instances(recstruct, tag="035", ind1="", ind2=""):
try:
provenance = field_get_subfield_values(instance, "9")[0]
except IndexError:
provenance = ""
try:
value = field_get_subfield_values(instance, "z")[0]
except IndexError:
value = ""
provenances = ["SPIRESTeX", "INSPIRETeX"]
if provenance in provenances and value:
has_texkey = True
print "INFO: Record %s has already texkey %s" % (recid, value)
if not has_texkey:
TexKeySeq = TexkeySeq()
new_texkey = ""
try:
new_texkey = TexKeySeq.next_value(recid)
except TexkeyNoAuthorError:
print "WARNING: Record %s has no first author or collaboration" % recid
continue
xml = create_xml(recid, new_texkey)
processed.append(recid)
to_process.append(xml)
if len(to_process) == 500:
process_chunk(to_process)
to_process = []
if to_process:
process_chunk(to_process)
# Finally, index all the records processed
if processed:
submit_bibindex_task(processed)
def check_record(record, texkey_field="035__a", extra_subfields=()):
"""
Add a tex key to a record, checking that it doesn't have one already.
"""
tag = texkey_field[:3]
ind1, ind2, subfield = texkey_field[3:]
provenances = list(record.iterfield(texkey_field[:5] + "9"))
if len(provenances) and provenances[0][1] in ("SPIRESTeX", "INSPIRETeX"):
for _, val in record.iterfield(texkey_field[:5] + "z"):
if val:
return # Record already has a texkey
if len(list(record.iterfield(texkey_field))) == 0:
try:
texkey = TexkeySeq().next_value(bibrecord=record)
except TexkeyNoAuthorError:
record.warn("No first author or collaboration")
return
subfields_to_add = [(subfield, texkey)] + map(tuple, extra_subfields)
record_add_field(record,
tag=tag,
ind1=ind1,
ind2=ind2,
subfields=subfields_to_add)
record.set_amended("Added Tex key '%s' to field %s" %
(texkey, texkey_field))
def test_get_next_texkey_no_author(self):
""" Generate an error while getting a texkey with no author """
texkey_seq = TexkeySeq()
self.assertRaises(TexkeyNoAuthorError,
texkey_seq.next_value,
xml_record=get_sample_texkey(3))
def check_record(record,
texkey_field="035__a",
provenances=None,
extra_subfields=None):
"""
Add a texkey to a record, checking that it doesn't have one already.
"""
if provenances is None:
provenances = ["SPIRESTeX", "INSPIRETeX"]
for provenance in provenances:
for _, val in record.iterfield(texkey_field,
subfield_filter=('9', provenance)):
if len(val) > 6: # texkey composition <name>:\d{4}<randchars>
return
try:
texkey = TexkeySeq().next_value(bibrecord=record)
except TexkeyNoAuthorError:
record.warn("No first author or collaboration")
return
except TexkeyNoYearError:
record.warn("No suitable year/date info")
return
if extra_subfields is None:
extra_subfields = [('9', 'INSPIRETeX')]
tag = texkey_field[:3]
ind1, ind2, subfield = texkey_field[3:]
subfields_to_add = [(subfield, texkey)] + map(tuple, extra_subfields)
record_add_field(record,
tag=tag,
ind1=ind1,
ind2=ind2,
subfields=subfields_to_add)
record.set_amended("Added Tex key '%s' to field %s" %
(texkey, texkey_field))
def test_get_next_texkey3(self):
""" Generate the third texkey """
texkey_seq = TexkeySeq()
self.texkey3 = texkey_seq.next_value(xml_record=get_sample_texkey(2))
self.assertEqual(self.texkey3[:-3], 'ATLAS:2012')
def test_get_next_texkey2(self):
""" Generate the second texkey """
texkey_seq = TexkeySeq()
self.texkey2 = texkey_seq.next_value(xml_record=get_sample_texkey(1))
self.assertEqual(self.texkey2[:-3], 'Broekhoven-Fiene:2012')
def test_get_next_texkey1(self):
""" Generate the first texkey """
texkey_seq = TexkeySeq()
self.texkey1 = texkey_seq.next_value(xml_record=get_sample_texkey(0))
self.assertEqual(self.texkey1[:-3], 'Boyle:2012')
def test_get_next_texkey3(self):
""" Generate the third texkey """
texkey_seq = TexkeySeq()
self.texkey3 = texkey_seq.next_value(xml_record=get_sample_texkey(2))
self.assertEqual(self.texkey3[:-3], 'ATLAS:2012')
def test_get_next_texkey2(self):
""" Generate the second texkey """
texkey_seq = TexkeySeq()
self.texkey2 = texkey_seq.next_value(xml_record=get_sample_texkey(1))
self.assertEqual(self.texkey2[:-3], 'Broekhoven-Fiene:2012')
def test_get_next_texkey1(self):
""" Generate the first texkey """
texkey_seq = TexkeySeq()
self.texkey1 = texkey_seq.next_value(xml_record=get_sample_texkey(0))
self.assertEqual(self.texkey1[:-3], 'Boyle:2012')
