def get_value(value):
affiliations = []
if value.get('u'):
recid = None
try:
recid = int(value.get('z'))
except:
pass
affiliations = inspire_dojson_utils.remove_duplicates_from_list(
utils.force_list(value.get('u')))
record = inspire_dojson_utils.get_record_ref(recid, 'institutions')
affiliations = [{
'value': aff,
'record': record
} for aff in affiliations]
person_recid = None
if value.get('x'):
try:
person_recid = int(value.get('x'))
except:
pass
inspire_id = ''
if value.get('i'):
inspire_id = utils.force_list(value.get('i'))[0]
person_record = inspire_dojson_utils.get_record_ref(
person_recid, 'authors')
ret = {
'full_name': value.get('a'),
'role': value.get('e'),
'alternative_name': value.get('q'),
'inspire_id': inspire_id,
'orcid': value.get('j'),
'record': person_record,
'email': value.get('m'),
'affiliations': affiliations,
'profile': {
"__url__":
inspire_dojson_utils.create_profile_url(value.get('x'))
},
'curated_relation': value.get('y', 0) == 1
}
# HACK: This is to workaround broken records where multiple authors
# got meshed up together.
if isinstance(ret['full_name'], (list, tuple)):
import warnings
warnings.warn(
"Record with mashed-up author list! Taking first author: {}".
format(value))
ret['full_name'] = ret['full_name'][0]
return ret
def test_get_record_ref_with_empty_server_name(current_app):
current_app.config = {}
expected = 'http://inspirehep.net/api/record_type/123'
result = get_record_ref(123, 'record_type')
assert expected == result['$ref']
def test_get_record_ref_with_server_name_localhost(current_app):
current_app.config = {'SERVER_NAME': 'localhost:5000'}
expected = 'http://localhost:5000/api/record_type/123'
result = get_record_ref(123, 'record_type')
assert expected == result['$ref']
def test_get_record_ref_with_https_server_name(current_app):
current_app.config = {'SERVER_NAME': 'https://example.com'}
expected = 'https://example.com/api/record_type/123'
result = get_record_ref(123, 'record_type')
assert expected == result['$ref']
def test_get_record_ref_without_record_type_defaults_to_record(current_app):
current_app.config = {}
expected = 'http://inspirehep.net/api/record/123'
result = get_record_ref(123)
assert expected == result['$ref']
def test_get_record_ref_with_server_name_localhost():
config = {'SERVER_NAME': 'localhost:5000'}
with patch.dict(current_app.config, config):
expected = 'http://localhost:5000/api/endpoint/123'
result = get_record_ref(123, 'endpoint')
assert expected == result['$ref']
def test_get_record_ref_with_https_server_name():
config = {'SERVER_NAME': 'https://example.com'}
with patch.dict(current_app.config, config):
expected = 'https://example.com/api/endpoint/123'
result = get_record_ref(123, 'endpoint')
assert expected == result['$ref']
def test_get_record_ref_with_empty_server_name():
config = {}
with patch.dict(current_app.config, config, clear=True):
expected = 'http://inspirehep.net/api/endpoint/123'
result = get_record_ref(123, 'endpoint')
assert expected == result['$ref']
def test_get_record_ref_with_server_name_localhost():
config = {'SERVER_NAME': 'localhost:5000'}
with patch.dict(current_app.config, config):
expected = 'http://localhost:5000/api/endpoint/123'
result = get_record_ref(123, 'endpoint')
assert expected == result['$ref']
def test_get_record_ref_with_https_server_name():
config = {'SERVER_NAME': 'https://example.com'}
with patch.dict(current_app.config, config):
expected = 'https://example.com/api/endpoint/123'
result = get_record_ref(123, 'endpoint')
assert expected == result['$ref']
def test_get_record_ref_without_endpoint_defaults_to_record():
config = {}
with patch.dict(current_app.config, config, clear=True):
expected = 'http://inspirehep.net/api/record/123'
result = get_record_ref(123)
assert expected == result['$ref']
def get_value(value):
affiliations = []
if value.get('u'):
recid = None
try:
recid = int(value.get('z'))
except:
pass
affiliations = inspire_dojson_utils.remove_duplicates_from_list(
utils.force_list(value.get('u')))
record = inspire_dojson_utils.get_record_ref(recid, 'institutions')
affiliations = [{'value': aff, 'record': record} for
aff in affiliations]
person_recid = None
if value.get('x'):
try:
person_recid = int(value.get('x'))
except:
pass
inspire_id = ''
if value.get('i'):
inspire_id = utils.force_list(value.get('i'))[0]
person_record = inspire_dojson_utils.get_record_ref(person_recid,
'authors')
ret = {
'full_name': value.get('a'),
'role': value.get('e'),
'alternative_name': value.get('q'),
'inspire_id': inspire_id,
'inspire_bai': value.get('w'),
'orcid': value.get('j'),
'record': person_record,
'email': value.get('m'),
'affiliations': affiliations,
'profile': {"__url__": inspire_dojson_utils.create_profile_url(
value.get('x')
)},
'curated_relation': value.get('y', 0) == 1
}
# HACK: This is to workaround broken records where multiple authors
# got meshed up together.
if isinstance(ret['full_name'], (list, tuple)):
import warnings
warnings.warn("Record with mashed-up author list! Taking first author: {}".format(value))
ret['full_name'] = ret['full_name'][0]
return ret
def test_get_record_ref_with_empty_server_name():
config = {}
with patch.dict(current_app.config, config, clear=True):
expected = 'http://inspirehep.net/api/endpoint/123'
result = get_record_ref(123, 'endpoint')
assert expected == result['$ref']
def test_get_record_ref_without_endpoint_defaults_to_record():
config = {}
with patch.dict(current_app.config, config, clear=True):
expected = 'http://inspirehep.net/api/record/123'
result = get_record_ref(123)
assert expected == result['$ref']
def related_experiments(self, key, value):
"""Related experiments."""
try:
recid = int(value.get('0'))
except (TypeError, ValueError):
recid = None
return {
'name': value.get('a'),
'record': inspire_dojson_utils.get_record_ref(recid, 'experiments')
}
def related_experiments(self, key, value):
"""Related experiments."""
try:
recid = int(value.get('0'))
except (TypeError, ValueError):
recid = None
return {
'name': value.get('a'),
'record': inspire_dojson_utils.get_record_ref(recid, 'experiments')
}
def test_get_record_ref_server_name(current_app):
# Test empty SERVER_NAME.
current_app.config = {}
ref = utils.get_record_ref(123, 'record_type')
assert ref['$ref'].startswith('http://inspirehep.net')
# Test dev SERVER_NAME.
current_app.config = {'SERVER_NAME': 'localhost:5000'}
ref = utils.get_record_ref(123, 'record_type')
assert ref['$ref'].startswith('http://localhost:5000')
# Test http prod SERVER_NAME.
current_app.config = {'SERVER_NAME': 'http://inspirehep.net'}
ref = utils.get_record_ref(123, 'record_type')
assert ref['$ref'].startswith('http://inspirehep.net')
# Test https prod SERVER_NAME.
current_app.config = {'SERVER_NAME': 'https://inspirehep.net'}
ref = utils.get_record_ref(123, 'record_type')
assert ref['$ref'].startswith('https://inspirehep.net')
def get_value(value):
recid = None
if '0' in value:
try:
recid = int(value.get('0'))
except:
pass
return {
'value': value.get('g'),
'record': inspire_dojson_utils.get_record_ref(recid, 'experiments')
}
def test_get_record_ref_server_name(current_app):
# Test empty SERVER_NAME.
current_app.config = {}
ref = utils.get_record_ref(123, 'record_type')
assert ref['$ref'].startswith('http://inspirehep.net')
# Test dev SERVER_NAME.
current_app.config = {'SERVER_NAME': 'localhost:5000'}
ref = utils.get_record_ref(123, 'record_type')
assert ref['$ref'].startswith('http://localhost:5000')
# Test http prod SERVER_NAME.
current_app.config = {'SERVER_NAME': 'http://inspirehep.net'}
ref = utils.get_record_ref(123, 'record_type')
assert ref['$ref'].startswith('http://inspirehep.net')
# Test https prod SERVER_NAME.
current_app.config = {'SERVER_NAME': 'https://inspirehep.net'}
ref = utils.get_record_ref(123, 'record_type')
assert ref['$ref'].startswith('https://inspirehep.net')
def succeeding_entry(self, key, value):
"""Succeeding Entry."""
if isinstance(value, (tuple, list)):
# Too bad: there can only be one succeeding entry.
value = value[0]
return {
'relationship_code': value.get('r'),
'record': inspire_dojson_utils.get_record_ref(
value.get('w'), 'literature'),
'isbn': value.get('z'),
}
def get_value(value):
recid = None
if '0' in value:
try:
recid = int(value.get('0'))
except:
pass
return {
'value': value.get('g'),
'record':
inspire_dojson_utils.get_record_ref(recid, 'experiments')
}
def institution(self, key, value):
"""Institution info."""
curated_relation = False
recid = None
if value.get('z') and value.get('z').isdigit():
curated_relation = True
recid = int(value.get('z'))
return {
'curated_relation': curated_relation,
'record': inspire_dojson_utils.get_record_ref(recid, 'isntitutions'),
'name': value.get('a'),
}
def institution(self, key, value):
"""Institution info."""
curated_relation = False
recid = None
if value.get('z') and value.get('z').isdigit():
curated_relation = True
recid = int(value.get('z'))
return {
'curated_relation': curated_relation,
'record': inspire_dojson_utils.get_record_ref(recid, 'isntitutions'),
'name': value.get('a'),
}
def publication_info(self, key, value):
"""Publication info about record."""
def get_int_value(val):
if val:
out = utils.force_list(val)[0]
if out.isdigit():
out = int(out)
return out
return None
year = get_int_value(value.get('y'))
parent_recid = get_int_value(value.get('0'))
journal_recid = get_int_value(value.get('1'))
conference_recid = get_int_value(value.get('2'))
parent_record = inspire_dojson_utils.get_record_ref(parent_recid,
'literature')
conference_record = inspire_dojson_utils.get_record_ref(conference_recid,
'conferences')
journal_record = inspire_dojson_utils.get_record_ref(journal_recid,
'journals')
res = {
'parent_record': parent_record,
'conference_record': conference_record,
'journal_record': journal_record,
'page_artid': value.get('c'),
'journal_issue': value.get('n'),
'conf_acronym': value.get('o'),
'journal_title': value.get('p'),
'reportnumber': value.get('r'),
'confpaper_info': value.get('t'),
'journal_volume': value.get('v'),
'cnum': utils.force_list(value.get('w')),
'pubinfo_freetext': value.get('x'),
'year': year,
'isbn': value.get('z'),
'note': value.get('m'),
}
return res
def thesis(self, key, value):
"""Get Thesis Information."""
curated_relation = False
if value.get('z'):
curated_relation = True
return {
'defense_date': value.get('a'),
'degree_type': value.get('b'),
'university': value.get('c'),
'date': value.get('d'),
'record': inspire_dojson_utils.get_record_ref(value.get('z'),
'institutions'),
'curated_relation': curated_relation,
}
def add_book_info(record, blob):
"""Add link to the appropriate book record."""
collections = []
if 'collections' in record:
for c in record.get('collections', ''):
if c.get('primary', ''):
collections.append(c.get('primary').lower())
if 'bookchapter' in collections:
pubinfos = force_list(blob.get("773__", []))
for pubinfo in pubinfos:
if pubinfo.get('0'):
record['book'] = {
'record': inspire_dojson_utils.get_record_ref(
int(force_list(pubinfo.get('0'))[0]), 'literature')
}
def add_book_info(record, blob):
"""Add link to the appropriate book record."""
collections = []
if 'collections' in record:
for c in record.get('collections', ''):
if c.get('primary', ''):
collections.append(c.get('primary').lower())
if 'bookchapter' in collections:
pubinfos = force_list(blob.get("773__", []))
for pubinfo in pubinfos:
if pubinfo.get('0'):
record['book'] = {
'record':
inspire_dojson_utils.get_record_ref(
int(force_list(pubinfo.get('0'))[0]), 'literature')
}
def spires_sysnos(self, key, value):
"""Old SPIRES number and new_recid from 970."""
value = utils.force_list(value)
sysnos = []
new_recid = None
for val in value:
if 'a' in val:
# Only append if there is something
sysnos.append(val.get('a'))
elif 'd' in val:
new_recid = val.get('d')
if new_recid is not None:
# FIXME we are currently using the default /record API. Which might
# resolve to a 404 response.
self['new_record'] = inspire_dojson_utils.get_record_ref(new_recid)
return sysnos or None
def spires_sysnos(self, key, value):
"""Old SPIRES number and new_recid from 970."""
value = utils.force_list(value)
sysnos = []
new_recid = None
for val in value:
if 'a' in val:
# Only append if there is something
sysnos.append(val.get('a'))
elif 'd' in val:
new_recid = val.get('d')
if new_recid is not None:
# FIXME we are currently using the default /record API. Which might
# resolve to a 404 response.
self['new_record'] = inspire_dojson_utils.get_record_ref(new_recid)
return sysnos or None
def thesis_supervisor(self, key, value):
"""Get thesis supervisor from object."""
value = utils.force_list(value)
out = []
for val in value:
out.append({
"full_name":
val.get('full_name'),
"record":
inspire_dojson_utils.get_record_ref(val.get('external_id'),
'literature'),
"affiliation": {
"value": val.get('affiliation')
},
})
return out
def accelerator_experiments(self, key, value):
"""The accelerator/experiment related to this record."""
recid = None
curated_relation = False
if '0' in value:
try:
recid = int(value.get('0'))
except (TypeError, ValueError, AttributeError):
pass
if recid:
curated_relation = True
return {
'record': inspire_dojson_utils.get_record_ref(recid, 'experiments'),
'accelerator': value.get('a'),
'experiment': value.get('e'),
'curated_relation': curated_relation
}
def thesis(self, key, value):
"""Get Thesis Information."""
curated_relation = False
if value.get('z'):
curated_relation = True
return {
'defense_date':
value.get('a'),
'degree_type':
value.get('b'),
'university':
value.get('c'),
'date':
value.get('d'),
'record':
inspire_dojson_utils.get_record_ref(value.get('z'), 'institutions'),
'curated_relation':
curated_relation,
}
def positions(self, key, value):
"""Position information.
Accepted values for 371__r:
+ senior
+ junior
+ staff
+ visitor
+ postdoc
+ phd
+ masters
+ undergrad
In dates field you can put months in addition to years. In this case you
have to follow the convention `mth-year`. For example: `10-2012`.
"""
curated_relation = False
recid = None
status = ''
recid_status = utils.force_list(value.get('z'))
if recid_status:
for val in recid_status:
if val.lower() == 'current':
status = val
elif type(val) is int:
recid = val
curated_relation = True
inst = {'name': value.get('a'),
'record': inspire_dojson_utils.get_record_ref(recid,
'institutions')}
return {
'institution': inst if inst['name'] else None,
'rank': value.get('r'),
'start_date': value.get('s'),
'end_date': value.get('t'),
'email': value.get('m'),
'old_email': value.get('o'),
'status': status,
'curated_relation': curated_relation,
}
def get_value(value):
recid = None
number = ''
year = ''
if '0' in value:
try:
recid = int(value.get('0'))
except:
pass
if 'o' in value:
try:
number = int(value.get('o'))
except:
pass
if 'y' in value:
try:
year = int(value.get('y'))
except:
pass
return {
'record': inspire_dojson_utils.get_record_ref(recid, 'literature'),
'texkey': value.get('1'),
'doi': value.get('a'),
'collaboration': utils.force_list(value.get('c')),
'editors': value.get('e'),
'authors': utils.force_list(value.get('h')),
'misc': utils.force_list(value.get('m')),
'number': number,
'isbn': value.get('i'),
'publisher': utils.force_list(value.get('p')),
'maintitle': value.get('q'),
'report_number': utils.force_list(value.get('r')),
'title': utils.force_list(value.get('t')),
'url': utils.force_list(value.get('u')),
'journal_pubnote': utils.force_list(value.get('s')),
'raw_reference': utils.force_list(value.get('x')),
'year': year,
}
def get_value(value):
recid = None
number = ''
year = ''
if '0' in value:
try:
recid = int(value.get('0'))
except:
pass
if 'o' in value:
try:
number = int(value.get('o'))
except:
pass
if 'y' in value:
try:
year = int(value.get('y'))
except:
pass
return {
'record': inspire_dojson_utils.get_record_ref(recid, 'literature'),
'texkey': value.get('1'),
'doi': value.get('a'),
'collaboration': utils.force_list(value.get('c')),
'editors': value.get('e'),
'authors': utils.force_list(value.get('h')),
'misc': utils.force_list(value.get('m')),
'number': number,
'isbn': value.get('i'),
'publisher': utils.force_list(value.get('p')),
'maintitle': value.get('q'),
'report_number': utils.force_list(value.get('r')),
'title': utils.force_list(value.get('t')),
'url': utils.force_list(value.get('u')),
'journal_pubnote': utils.force_list(value.get('s')),
'raw_reference': utils.force_list(value.get('x')),
'year': year,
}
def _self_url(self, key, value):
"""Url of the record itself."""
self['control_number'] = value
return inspire_dojson_utils.get_record_ref(value, index)
def create_author(profile):
"""Create a new author profile based on a given signature.
The method receives a dictionary representing an author.
Based on the values, it creates a dictionary in the invenio_records format.
After all the fields are processed, the method calls create_record
from invenio_records.api to put the new record.
:param profile:
A signature representing an author's to be created as a profile.
Example:
profile = {u'affiliations': [{u'value': u'Yerevan Phys. Inst.'}],
u'alternative_name': None,
u'curated_relation': False,
u'email': None,
u'full_name': u'Chatrchyan, Serguei',
u'inspire_id': None,
u'orcid': None,
u'profile': u'',
u'recid': None,
u'role': None,
u'uuid': u'd63537a8-1df4-4436-b5ed-224da5b5028c'}
:return:
A recid, where the new profile can be accessed.
Example:
"1234"
"""
name = profile.get('full_name')
# Template of an initial record.
record = {'collections': [{'primary': 'HEPNAMES'}],
'name': {'value': name},
'$schema': _get_author_schema()}
# The author's email address.
# Unfortunately the method will not correlate a given e-mail address
# with an affiliation.
if 'email' in profile:
email = profile.get('email')
record['positions'] = []
record['positions'].append({'email': email})
# The author can be a member of more than one affiliation.
if 'affiliations' in profile:
affiliations = profile.get('affiliations')
if 'positions' not in record:
record['positions'] = []
for affiliation in affiliations:
name = affiliation.get('value')
recid = affiliation.get('recid', None)
if recid:
record['positions'].append(
{'institution': {'name': name, 'recid': recid}})
else:
record['positions'].append(
{'institution': {'name': name}})
# FIXME: The method should also collect the useful data
# from the publication, like category field, subject,
# etc.
# Disconnect the signal on insert of a new record.
after_record_insert.disconnect(append_new_record_to_queue)
# Create a new author profile.
record = InspireRecord.create(record, id_=None)
# Create Inspire recid.
record_pid = inspire_recid_minter(record.id, record)
# Extend the new record with Inspire recid and self key.
record['control_number'] = record_pid.pid_value
record['self'] = inspire_dojson_utils.get_record_ref(
record_pid.pid_value, 'authors')
# Apply the changes.
record.commit()
db.session.commit()
# Reconnect the disconnected signal.
after_record_insert.connect(append_new_record_to_queue)
# Report.
logger.info("Created profile: %s", record_pid.pid_value)
# Return the recid of new profile to which signatures will point to.
return record_pid.pid_value
def test_get_record_ref_with_record_type():
ref = utils.get_record_ref(123, 'record_type')
assert ref['$ref'].endswith('/api/record_type/123')
assert utils.get_record_ref(None, 'record_type') == None
def test_get_record_ref_without_recid_returns_none():
assert get_record_ref(None, 'endpoint') is None
def create_author(profile):
"""Create a new author profile based on a given signature.
The method receives a dictionary representing an author.
Based on the values, it creates a dictionary in the invenio_records format.
After all the fields are processed, the method calls create_record
from invenio_records.api to put the new record.
:param profile:
A signature representing an author's to be created as a profile.
Example:
profile = {u'affiliations': [{u'value': u'Yerevan Phys. Inst.'}],
u'alternative_name': None,
u'curated_relation': False,
u'email': None,
u'full_name': u'Chatrchyan, Serguei',
u'inspire_id': None,
u'orcid': None,
u'profile': u'',
u'recid': None,
u'role': None,
u'uuid': u'd63537a8-1df4-4436-b5ed-224da5b5028c'}
:return:
A recid, where the new profile can be accessed.
Example:
"1234"
"""
name = profile.get('full_name')
# Template of an initial record.
record = {'collections': [{'primary': 'HEPNAMES'}],
'name': {'value': name},
'$schema': _get_author_schema()}
# The author's email address.
# Unfortunately the method will not correlate a given e-mail address
# with an affiliation.
if 'email' in profile:
email = profile.get('email')
record['positions'] = []
record['positions'].append({'email': email})
# The author can be a member of more than one affiliation.
if 'affiliations' in profile:
affiliations = profile.get('affiliations')
if 'positions' not in record:
record['positions'] = []
for affiliation in affiliations:
name = affiliation.get('value')
recid = affiliation.get('recid', None)
if recid:
record['positions'].append(
{'institution': {'name': name, 'recid': recid}})
else:
record['positions'].append(
{'institution': {'name': name}})
# FIXME: The method should also collect the useful data
# from the publication, like category field, subject,
# etc.
# Disconnect the signal on insert of a new record.
after_record_insert.disconnect(append_new_record_to_queue)
# Create a new author profile.
record = Record.create(record, id_=None)
# Create Inspire recid.
record_pid = inspire_recid_minter(record.id, record)
# Extend the new record with Inspire recid and self key.
record['control_number'] = record_pid.pid_value
record['self'] = inspire_dojson_utils.get_record_ref(
record_pid.pid_value, 'authors')
# Apply the changes.
record.commit()
db.session.commit()
# Add the record to Elasticsearch.
indexer = RecordIndexer()
indexer.index_by_id(record_pid.object_uuid)
# Reconnect the disconnected signal.
after_record_insert.connect(append_new_record_to_queue)
# Report.
logger.info("Created profile: %s", record_pid.pid_value)
# Return the recid of new profile to which signatures will point to.
return record_pid.pid_value
def test_get_record_ref_without_recid_returns_none():
assert get_record_ref(None, 'record_type') is None
def _self_url(self, key, value):
"""Url of the record itself."""
self['control_number'] = value
return inspire_dojson_utils.get_record_ref(value, index)
def test_get_record_ref_without_recid_returns_none():
assert get_record_ref(None, 'endpoint') is None
def test_get_record_ref_with_record_type():
ref = utils.get_record_ref(123, 'record_type')
assert ref['$ref'].endswith('/api/record_type/123')
assert utils.get_record_ref(None, 'record_type') == None
def test_get_record_ref_default():
ref = utils.get_record_ref(123)
assert ref['$ref'].endswith('/api/record/123')
def deleted_records(self, key, value):
"""Recid of deleted record this record is master for."""
# FIXME we are currently using the default /record API. Which might
# resolve to a 404 response.
return inspire_dojson_utils.get_record_ref(value.get('a'))
def deleted_records(self, key, value):
"""Recid of deleted record this record is master for."""
# FIXME we are currently using the default /record API. Which might
# resolve to a 404 response.
return inspire_dojson_utils.get_record_ref(value.get('a'))
def test_get_record_ref_default():
ref = utils.get_record_ref(123)
assert ref['$ref'].endswith('/api/record/123')
