def test_arxiv_derive_inspire_categories():
schema = load_schema('hep')
arxiv_eprints_schema = schema['properties']['arxiv_eprints']
inspire_categories_schema = schema['properties']['inspire_categories']
data = {
'arxiv_eprints': [
{
'categories': [
'nucl-th',
],
'value': '1605.03898',
},
],
} # literature/1458300
extra_data = {}
assert validate(data['arxiv_eprints'], arxiv_eprints_schema) is None
obj = MockObj(data, extra_data)
eng = MockEng()
assert arxiv_derive_inspire_categories(obj, eng) is None
expected = [
{
'source': 'arxiv',
'term': 'Theory-Nucl',
},
]
result = obj.data['inspire_categories']
assert validate(result, inspire_categories_schema) is None
assert expected == result
def test_date_started_and_date_completed_from_046():
schema = load_schema('experiments')
subschema_date_started = schema['properties']['date_started']
subschema_date_completed = schema['properties']['date_started']
snippet = (
'<record>'
' <datafield tag="046" ind1=" " ind2=" ">'
' <subfield code="s">1996</subfield>'
' </datafield>'
' <datafield tag="046" ind1=" " ind2=" ">'
' <subfield code="t">2002</subfield>'
' </datafield>'
' <datafield tag="046" ind1=" " ind2=" ">'
' <subfield code="x">yes</subfield>'
' </datafield>'
'</record>'
) # record/1108324
result = experiments.do(create_record(snippet))
assert validate(result['date_started'], subschema_date_started) is None
assert validate(result['date_completed'], subschema_date_completed) is None
assert result['date_started'] == '1996'
assert result['date_completed'] == '2002'
def test_extract_journal_info_handles_year_an_empty_string():
schema = load_schema('hep')
subschema = schema['properties']['publication_info']
data = {
'publication_info': [
{'pubinfo_freetext': (
'The Astrophysical Journal, 838:134 (16pp), 2017 April 1')},
],
}
extra_data = {}
assert validate(data['publication_info'], subschema) is None
obj = MockObj(data, extra_data)
eng = MockEng()
assert extract_journal_info(obj, eng) is None
assert validate(obj.data['publication_info'], subschema) is None
assert obj.data['publication_info'] == [
{
'artid': '134',
'journal_title': 'Astrophys. J.',
'journal_volume': '838',
'page_start': '134',
'pubinfo_freetext': (
'The Astrophysical Journal, 838:134 (16pp), 2017 April 1'),
},
]
def test_date_closed_from_046__i_and_046__l_an_url():
schema = load_schema('jobs')
subschema_deadline_date = schema['properties']['deadline_date']
subschema_urls = schema['properties']['urls']
snippet = (
'<record>'
' <datafield tag="046" ind1=" " ind2=" ">'
' <subfield code="i">2012-06-01</subfield>'
' </datafield>'
' <datafield tag="046" ind1=" " ind2=" ">'
' <subfield code="l">http://www.pma.caltech.edu/physics-search</subfield>'
' </datafield>'
'</record>'
) # record/963314
expected_deadline_date = '2012-06-01'
expected_urls = [
{'value': 'http://www.pma.caltech.edu/physics-search'},
]
result = jobs.do(create_record(snippet))
assert validate(result['deadline_date'], subschema_deadline_date) is None
assert expected_deadline_date == result['deadline_date']
assert validate(result['urls'], subschema_urls) is None
assert expected_urls == result['urls']
def test_date_closed_from_046__i_and_046__l_an_email():
schema = load_schema('jobs')
subschema_deadline_date = schema['properties']['deadline_date']
subschema_reference_email = schema['properties']['reference_email']
snippet = (
'<record>'
' <datafield tag="046" ind1=" " ind2=" ">'
' <subfield code="l">[email protected]</subfield>'
' </datafield>'
' <datafield tag="046" ind1=" " ind2=" ">'
' <subfield code="i">8888</subfield>'
' </datafield>'
'</record>'
) # record/1089529
expected_deadline_date = '8888'
expected_reference_email = ['*****@*****.**']
result = jobs.do(create_record(snippet))
assert validate(result['deadline_date'], subschema_deadline_date) is None
assert expected_deadline_date == result['deadline_date']
assert validate(result['reference_email'], subschema_reference_email) is None
assert expected_reference_email == result['reference_email']
def test_extract_journal_info():
schema = load_schema('hep')
subschema = schema['properties']['publication_info']
data = {
'publication_info': [
{'pubinfo_freetext': 'J. Math. Phys. 55, 082102 (2014)'},
],
}
extra_data = {}
assert validate(data['publication_info'], subschema) is None
obj = MockObj(data, extra_data)
eng = MockEng()
assert extract_journal_info(obj, eng) is None
assert validate(obj.data['publication_info'], subschema) is None
assert obj.data['publication_info'] == [
{
'artid': '082102',
'journal_title': 'J. Math. Phys.',
'journal_volume': '55',
'pubinfo_freetext': 'J. Math. Phys. 55, 082102 (2014)',
'year': 2014,
}
]
def test_languages_from_041__a_handles_multiple_languages_in_one_a():
schema = load_schema('hep')
subschema = schema['properties']['languages']
snippet = ('<datafield tag="041" ind1=" " ind2=" ">'
' <subfield code="a">Russian / English</subfield>'
'</datafield>') # record/116959
expected = [
'ru',
'en',
]
result = hep.do(create_record(snippet))
assert validate(result['languages'], subschema) is None
assert expected == result['languages']
expected = [
{
'a': 'russian'
},
{
'a': 'english'
},
]
result = hep2marc.do(result)
assert expected == result['041']
def test_isbns_from_020__a_b_normalizes_print():
schema = load_schema('hep')
subschema = schema['properties']['isbns']
snippet = ('<datafield tag="020" ind1=" " ind2=" ">'
' <subfield code="a">9781786341105</subfield>'
' <subfield code="b">Print</subfield>'
'</datafield>') # record/1509456
expected = [
{
'value': '978-1-78634-110-5',
'medium': 'print',
},
]
result = hep.do(create_record(snippet))
assert validate(result['isbns'], subschema) is None
assert expected == result['isbns']
expected = [
{
'a': '978-1-78634-110-5',
'b': 'print',
},
]
result = hep2marc.do(result)
assert expected == result['020']
def test_report_numbers_from_037__z_9():
schema = load_schema('hep')
subschema = schema['properties']['report_numbers']
snippet = (
'<datafield tag="037" ind1=" " ind2=" ">'
' <subfield code="9">SLAC</subfield>'
' <subfield code="a">SLAC-PUB-16140</subfield>'
'</datafield>'
) # record/1326454
expected = [
{
'source': 'SLAC',
'value': 'SLAC-PUB-16140',
},
]
result = hep.do(create_record(snippet))
assert validate(result['report_numbers'], subschema) is None
assert expected == result['report_numbers']
expected = [
{
'9': 'SLAC',
'a': 'SLAC-PUB-16140',
},
]
result = hep2marc.do(result)
assert expected == result['037']
def test_external_system_numbers_from_035__a_d_h_m_9():
schema = load_schema('hep')
subschema = schema['properties']['external_system_identifiers']
snippet = (
'<datafield tag="035" ind1=" " ind2=" ">'
' <subfield code="9">http://cds.cern.ch/oai2d</subfield>'
' <subfield code="a">oai:cds.cern.ch:325030</subfield>'
' <subfield code="d">2015-06-05T13:24:42Z</subfield>'
' <subfield code="h">2015-11-09T16:22:48Z</subfield>'
' <subfield code="m">marcxml</subfield>'
'</datafield>'
) # record/1403324
expected = [
{
'value': 'oai:cds.cern.ch:325030',
'schema': 'http://cds.cern.ch/oai2d',
}
]
result = hep.do(create_record(snippet))
assert validate(result['external_system_identifiers'], subschema) is None
assert expected == result['external_system_identifiers']
expected = [
{
'9': 'http://cds.cern.ch/oai2d',
'a': 'oai:cds.cern.ch:325030',
},
]
result = hep2marc.do(result)
assert expected == result['035']
def test_dois_from_0247_a_2_double_9_ignores_curator_source():
schema = load_schema('hep')
subschema = schema['properties']['dois']
snippet = (
'<datafield tag="024" ind1="7" ind2=" ">'
' <subfield code="2">DOI</subfield>'
' <subfield code="9">bibcheck</subfield>'
' <subfield code="9">CURATOR</subfield>'
' <subfield code="a">10.1590/S1806-11172008005000006</subfield>'
'</datafield>'
) # record/1117362
expected = [
{
'source': 'bibcheck',
'value': '10.1590/S1806-11172008005000006',
},
]
result = hep.do(create_record(snippet)) # no roundtrip
assert validate(result['dois'], subschema) is None
assert expected == result['dois']
expected = [
{
'a': '10.1590/S1806-11172008005000006',
'9': 'bibcheck',
'2': 'DOI',
},
]
result = hep2marc.do(result)
assert expected == result['0247']
def test_isbns_from_020__a_b_normalizes_print():
schema = load_schema('hep')
subschema = schema['properties']['isbns']
snippet = (
'<datafield tag="020" ind1=" " ind2=" ">'
' <subfield code="a">9781786341105</subfield>'
' <subfield code="b">Print</subfield>'
'</datafield>'
) # record/1509456
expected = [
{
'value': '9781786341105',
'medium': 'print',
},
]
result = hep.do(create_record(snippet))
assert validate(result['isbns'], subschema) is None
assert expected == result['isbns']
expected = [
{
'a': '9781786341105',
'b': 'print',
},
]
result = hep2marc.do(result)
assert expected == result['020']
def test_abstracts_from_520__double_a_ignores_second_a():
schema = load_schema('hep')
subschema = schema['properties']['abstracts']
snippet = (
'<datafield tag="520" ind1=" " ind2=" ">'
' <subfield code="a">$D$ $K$ scattering and the $D_s$ spectrum from lattice QCD 520__</subfield>'
' <subfield code="a">We present results from Lattice QCD calculations of the low-lying charmed-strange meson spectrum using two types of Clover-Wilson lattices. In addition to quark-antiquark interpolating fields we also consider meson-meson interpolators corresponding to D-meson kaon scattering states. To calculate the all-to-all propagation necessary for the backtracking loops we use the (stochastic) distillation technique. For the charm quark we use the Fermilab method. Results for the $J^P=0^+$ $D_{s0}^*(2317)$ charmed-strange meson are presented.</subfield>'
'</datafield>'
) # record/1297699
expected = [
{'value': '$D$ $K$ scattering and the $D_s$ spectrum from lattice QCD 520__'},
]
result = hep.do(create_record(snippet)) # no roundtrip
assert validate(result['abstracts'], subschema) is None
assert expected == result['abstracts']
expected = [
{
'a': '$D$ $K$ scattering and the $D_s$ spectrum from lattice QCD 520__',
},
]
result = hep2marc.do(result)
assert expected == result['520']
def test_abstracts_from_520__a_9():
schema = load_schema('hep')
subschema = schema['properties']['abstracts']
snippet = (
'<datafield tag="520" ind1=" " ind2=" ">'
' <subfield code="9">Springer</subfield>'
' <subfield code="a">We study a notion of non-commutative integration, in the spirit of modular spectral triples, for the quantum group SU$_{q}$ (2). In particular we define the non-commutative integral as the residue at the spectral dimension of a zeta function, which is constructed using a Dirac operator and a weight. We consider the Dirac operator introduced by Kaad and Senior and a family of weights depending on two parameters, which are related to the diagonal automorphisms of SU$_{q}$ (2). We show that, after fixing one of the parameters, the non-commutative integral coincides with the Haar state of SU$_{q}$ (2). Moreover we can impose an additional condition on the zeta function, which also fixes the second parameter. For this unique choice the spectral dimension coincides with the classical dimension.</subfield>'
'</datafield>'
) # record/1346798
expected = [
{
'source': 'Springer',
'value': 'We study a notion of non-commutative integration, in the spirit of modular spectral triples, for the quantum group SU$_{q}$ (2). In particular we define the non-commutative integral as the residue at the spectral dimension of a zeta function, which is constructed using a Dirac operator and a weight. We consider the Dirac operator introduced by Kaad and Senior and a family of weights depending on two parameters, which are related to the diagonal automorphisms of SU$_{q}$ (2). We show that, after fixing one of the parameters, the non-commutative integral coincides with the Haar state of SU$_{q}$ (2). Moreover we can impose an additional condition on the zeta function, which also fixes the second parameter. For this unique choice the spectral dimension coincides with the classical dimension.',
},
]
result = hep.do(create_record(snippet))
assert validate(result['abstracts'], subschema) is None
assert expected == result['abstracts']
expected = [
{
'9': 'Springer',
'a': 'We study a notion of non-commutative integration, in the spirit of modular spectral triples, for the quantum group SU$_{q}$ (2). In particular we define the non-commutative integral as the residue at the spectral dimension of a zeta function, which is constructed using a Dirac operator and a weight. We consider the Dirac operator introduced by Kaad and Senior and a family of weights depending on two parameters, which are related to the diagonal automorphisms of SU$_{q}$ (2). We show that, after fixing one of the parameters, the non-commutative integral coincides with the Haar state of SU$_{q}$ (2). Moreover we can impose an additional condition on the zeta function, which also fixes the second parameter. For this unique choice the spectral dimension coincides with the classical dimension.',
},
]
result = hep2marc.do(result)
assert expected == result['520']
def test_titles_from_246__a_9():
schema = load_schema('hep')
subschema = schema['properties']['titles']
snippet = ('<datafield tag="246" ind1=" " ind2=" ">'
' <subfield code="a">Superintegrable relativistic systems in'
' spacetime-dependent background fields</subfield>'
' <subfield code="9">arXiv</subfield>'
'</datafield>') # record/1511471
expected = [
{
'source':
'arXiv',
'title':
'Superintegrable relativistic systems in '
'spacetime-dependent background fields',
},
]
result = hep.do(create_record(snippet))
assert validate(result['titles'], subschema) is None
assert expected == result['titles']
expected = [
{
'a':
'Superintegrable relativistic systems in spacetime-dependent background fields',
'9': 'arXiv',
},
]
result = hep2marc.do(result)
assert expected == result['245']
def test_titles_from_245__a_9():
schema = load_schema('hep')
subschema = schema['properties']['titles']
snippet = (
'<datafield tag="245" ind1=" " ind2=" ">'
' <subfield code="a">Exact Form of Boundary Operators Dual to '
'Interacting Bulk Scalar Fields in the AdS/CFT Correspondence</subfield>'
' <subfield code="9">arXiv</subfield>'
'</datafield>') # record/001511698
expected = [
{
'title': 'Exact Form of Boundary Operators Dual to Interacting '
'Bulk Scalar Fields in the AdS/CFT Correspondence',
'source': 'arXiv',
},
]
result = hep.do(create_record(snippet))
assert validate(result['titles'], subschema) is None
assert expected == result['titles']
expected = [
{
'a': 'Exact Form of Boundary Operators Dual to Interacting '
'Bulk Scalar Fields in the AdS/CFT Correspondence',
'9': 'arXiv',
},
]
result = hep2marc.do(result)
assert expected == result['245']
def test_title_translations_from_242__a_b():
schema = load_schema('hep')
subschema = schema['properties']['title_translations']
snippet = (
'<datafield tag="242" ind1=" " ind2=" ">'
' <subfield code="a">Generalized Hamilton-Jacobi Formalism</subfield>'
' <subfield code="b">Field Theories with Upper-Order Derivatives</subfield>'
'</datafield>') # record/1501064
expected = [
{
'language': 'en',
'title': 'Generalized Hamilton-Jacobi Formalism',
'subtitle': 'Field Theories with Upper-Order Derivatives',
},
]
result = hep.do(create_record(snippet))
assert validate(result['title_translations'], subschema) is None
assert expected == result['title_translations']
expected = [
{
'a': 'Generalized Hamilton-Jacobi Formalism',
'b': 'Field Theories with Upper-Order Derivatives',
},
]
result = hep2marc.do(result)
assert expected == result['242']
def test_title_translations_from_242__a():
schema = load_schema('hep')
subschema = schema['properties']['title_translations']
snippet = (
'<datafield tag="242" ind1=" " ind2=" ">'
' <subfield code="a">The redshift of extragalactic nebulae</subfield>'
'</datafield>') # record/8352
expected = [
{
'language': 'en',
'title': 'The redshift of extragalactic nebulae',
},
]
result = hep.do(create_record(snippet))
assert validate(result['title_translations'], subschema) is None
assert expected == result['title_translations']
expected = [
{
'a': 'The redshift of extragalactic nebulae',
},
]
result = hep2marc.do(result)
assert expected == result['242']
def test_titles_from_245__a_b():
schema = load_schema('hep')
subschema = schema['properties']['titles']
snippet = (
'<datafield tag="245" ind1=" " ind2=" ">'
' <subfield code="a">Proceedings, New Observables in Quarkonium Production</subfield>'
' <subfield code="b">Trento, Italy</subfield>'
'</datafield>') # record/1510141
expected = [
{
'title': 'Proceedings, New Observables in Quarkonium Production',
'subtitle': 'Trento, Italy',
},
]
result = hep.do(create_record(snippet))
assert validate(result['titles'], subschema) is None
assert expected == result['titles']
expected = [
{
'a': 'Proceedings, New Observables in Quarkonium Production',
'b': 'Trento, Italy',
},
]
result = hep2marc.do(result)
assert expected == result['245']
def test_ids_from_035__a_9_with_scopus():
schema = load_schema('authors')
subschema = schema['properties']['ids']
snippet = (
'<datafield tag="035" ind1=" " ind2=" ">'
' <subfield code="9">SCOPUS</subfield>'
' <subfield code="a">7103280792</subfield>'
'</datafield>'
) # record/1017182/export/xme
expected = [
{
'schema': 'SCOPUS',
'value': '7103280792',
},
]
result = hepnames.do(create_record(snippet))
assert validate(result['ids'], subschema) is None
assert expected == result['ids']
expected = [
{
'9': 'SCOPUS',
'a': '7103280792',
},
]
result = hepnames2marc.do(result)
assert expected == result['035']
def test_other_names_from_400__triple_a():
schema = load_schema('authors')
subschema = schema['properties']['other_names']
snippet = (
'<datafield tag="400" ind1=" " ind2=" ">'
' <subfield code="a">Yosef Cohen, Hadar</subfield>'
' <subfield code="a">Josef Cohen, Hadar</subfield>'
' <subfield code="a">Cohen, Hadar Josef</subfield>'
'</datafield>'
) # record/1292399/export/xme
expected = [
'Yosef Cohen, Hadar',
'Josef Cohen, Hadar',
'Cohen, Hadar Josef',
]
result = hepnames.do(create_record(snippet))
assert validate(result['other_names'], subschema) is None
assert expected == result['other_names']
expected = [
{'a': 'Yosef Cohen, Hadar'},
{'a': 'Josef Cohen, Hadar'},
{'a': 'Cohen, Hadar Josef'},
]
result = hepnames2marc.do(result)
assert expected == result['400']
def test_copyright_from_542__d_e_g():
schema = load_schema('hep')
subschema = schema['properties']['copyright']
snippet = (
'<datafield tag="542" ind1=" " ind2=" ">'
' <subfield code="d">American Physical Society</subfield>'
' <subfield code="g">2017</subfield>'
' <subfield code="e">Article</subfield>'
'</datafield>'
) # record/1511489
expected = [
{'holder': 'American Physical Society'},
]
result = hep.do(create_record(snippet)) # no roundtrip
assert validate(result['copyright'], subschema) is None
assert expected == result['copyright']
expected = [
{'d': 'American Physical Society'},
]
result = hep2marc.do(result)
assert expected == result['542']
def test_isbns_from_020__a_b_normalizes_ebook():
schema = load_schema('hep')
subschema = schema['properties']['isbns']
snippet = (
'<datafield tag="020" ind1=" " ind2=" ">'
' <subfield code="a">9783319259017</subfield>'
' <subfield code="b">eBook</subfield>'
'</datafield>'
) # record/1430829
expected = [
{
'value': '9783319259017',
'medium': 'online',
},
]
result = hep.do(create_record(snippet))
assert validate(result['isbns'], subschema) is None
assert expected == result['isbns']
expected = [
{
'a': '9783319259017',
'b': 'online',
},
]
result = hep2marc.do(result)
assert expected == result['020']
def test_public_notes_from_500__a_9():
schema = load_schema('hep')
subschema = schema['properties']['public_notes']
snippet = (
'<datafield tag="500" ind1=" " ind2=" ">'
' <subfield code="9">arXiv</subfield>'
' <subfield code="a">5 pages</subfield>'
'</datafield>'
) # record/1450044
expected = [
{
'source': 'arXiv',
'value': '5 pages',
},
]
result = hep.do(create_record(snippet))
assert validate(result['public_notes'], subschema) is None
assert expected == result['public_notes']
expected = [
{
'9': 'arXiv',
'a': '5 pages',
},
]
result = hep2marc.do(result)
assert expected == result['500']
def test_texkeys_from_035__a_9():
schema = load_schema('hep')
subschema = schema['properties']['texkeys']
snippet = (
'<datafield tag="035" ind1=" " ind2=" ">'
' <subfield code="9">INSPIRETeX</subfield>'
' <subfield code="a">Hagedorn:1963hdh</subfield>'
'</datafield>'
) # record/1403324
expected = [
'Hagedorn:1963hdh',
]
result = hep.do(create_record(snippet))
assert validate(result['texkeys'], subschema) is None
assert expected == result['texkeys']
expected = [
{
'9': 'INSPIRETeX',
'a': 'Hagedorn:1963hdh',
}
]
result = hep2marc.do(result)
assert expected == result['035']
def test_hidden_notes_from_595__a_9():
schema = load_schema('hep')
subschema = schema['properties']['hidden_notes']
snippet = (
'<datafield tag="595" ind1=" " ind2=" ">'
' <subfield code="9">SPIRES-HIDDEN</subfield>'
' <subfield code="a">Title changed from ALLCAPS</subfield>'
'</datafield>'
) # record/109310
expected = [
{
'source': 'SPIRES-HIDDEN',
'value': 'Title changed from ALLCAPS',
},
]
result = hep.do(create_record(snippet))
assert validate(result['hidden_notes'], subschema) is None
assert expected == result['hidden_notes']
expected = [
{
'9': 'SPIRES-HIDDEN',
'a': 'Title changed from ALLCAPS',
},
]
result = hep2marc.do(result)
assert expected == result['595']
def test_report_numbers_from_two_037__a():
schema = load_schema('hep')
subschema = schema['properties']['report_numbers']
snippet = (
'<record>'
' <datafield tag="037" ind1=" " ind2=" ">'
' <subfield code="a">UTPT-89-27</subfield>'
' </datafield>'
' <datafield tag="037" ind1=" " ind2=" ">'
' <subfield code="a">CALT-68-1585</subfield>'
' </datafield>'
'</record>'
) # record/26564
expected = [
{
'value': 'UTPT-89-27',
},
{
'value': 'CALT-68-1585',
},
]
result = hep.do(create_record(snippet))
assert validate(result['report_numbers'], subschema) is None
assert expected == result['report_numbers']
expected = [
{'a': 'UTPT-89-27'},
{'a': 'CALT-68-1585'},
]
result = hep2marc.do(result)
assert expected == result['037']
def test_urls_from_8564_u_w_y_ignores_w():
schema = load_schema('hep')
subschema = schema['properties']['urls']
snippet = (
'<datafield tag="856" ind1="4" ind2=" ">'
' <subfield code="w">12-316</subfield>'
' <subfield code="y">FERMILABPUB</subfield>'
' <subfield code="u">http://lss.fnal.gov/cgi-bin/find_paper.pl?pub-12-316</subfield>'
'</datafield>') # record/1120360
expected = [
{
'description': 'FERMILABPUB',
'value': 'http://lss.fnal.gov/cgi-bin/find_paper.pl?pub-12-316',
},
]
result = hep.do(create_record(snippet)) # no roundtrip
assert validate(result['urls'], subschema) is None
assert expected == result['urls']
expected = [
{
'u': 'http://lss.fnal.gov/cgi-bin/find_paper.pl?pub-12-316',
'y': 'FERMILABPUB',
},
]
result = hep2marc.do(result)
assert expected == result['8564']
def test_languages_from_041__a_handles_multiple_languages_in_one_a():
schema = load_schema('hep')
subschema = schema['properties']['languages']
snippet = (
'<datafield tag="041" ind1=" " ind2=" ">'
' <subfield code="a">Russian / English</subfield>'
'</datafield>'
) # record/116959
expected = [
'ru',
'en',
]
result = hep.do(create_record(snippet))
assert validate(result['languages'], subschema) is None
assert expected == result['languages']
expected = [
{'a': 'russian'},
{'a': 'english'},
]
result = hep2marc.do(result)
assert expected == result['041']
def test_acquisition_source_from_541__double_a_b_c_e():
schema = load_schema('elements/acquisition_source')
snippet = ('<datafield tag="541" ind1=" " ind2=" ">'
' <subfield code="a">inspire:uid:52524</subfield>'
' <subfield code="a">orcid:0000-0002-1048-661X</subfield>'
' <subfield code="b">[email protected]</subfield>'
' <subfield code="c">submission</subfield>'
' <subfield code="e">504296</subfield>'
'</datafield>') # record/1416571/export/xme
expected = {
'source': 'orcid:0000-0002-1048-661X',
'email': '*****@*****.**',
'method': 'submission',
'submission_number': '504296',
}
result = hep.do(create_record(snippet)) # no roundtrip
assert validate(result['acquisition_source'], schema) is None
assert expected == result['acquisition_source']
expected = {
'a': 'orcid:0000-0002-1048-661X',
'b': '*****@*****.**',
'c': 'submission',
'e': '504296',
}
result = hep2marc.do(result)
assert expected == result['541']
def test_urls_from_8564_s_u_ignores_s():
schema = load_schema('hep')
subschema = schema['properties']['urls']
snippet = (
'<datafield tag="856" ind1="4" ind2=" ">'
' <subfield code="s">443981</subfield>'
' <subfield code="u">http://inspirehep.net/record/1511347/files/HIG-16-034-pas.pdf</subfield>'
'</datafield>') # record/1511347
expected = [
{
'value':
'http://inspirehep.net/record/1511347/files/HIG-16-034-pas.pdf'
},
]
result = hep.do(create_record(snippet)) # no roundtrip
assert validate(result['urls'], subschema) is None
assert expected == result['urls']
expected = [
{
'u':
'http://inspirehep.net/record/1511347/files/HIG-16-034-pas.pdf'
},
]
result = hep2marc.do(result)
assert expected == result['8564']
def test_languages_from_double_041__a():
schema = load_schema('hep')
subschema = schema['properties']['languages']
snippet = ('<record>'
' <datafield tag="041" ind1=" " ind2=" ">'
' <subfield code="a">French</subfield>'
' </datafield>'
' <datafield tag="041" ind1=" " ind2=" ">'
' <subfield code="a">German</subfield>'
' </datafield>'
'</record>') # record/1231408
expected = [
'fr',
'de',
]
result = hep.do(create_record(snippet))
assert validate(result['languages'], subschema) is None
assert expected == result['languages']
expected = [
{
'a': 'french'
},
{
'a': 'german'
},
]
result = hep2marc.do(result)
assert expected == result['041']
def test_urls_from_8564_u_double_y_selects_the_first_y():
schema = load_schema('hep')
subschema = schema['properties']['urls']
snippet = (
'<datafield tag="856" ind1="4" ind2=" ">'
' <subfield code="u">http://link.springer.com/journal/10909/176/5/page/1</subfield>'
' <subfield code="y">Part II</subfield>'
' <subfield code="y">Springer</subfield>'
'</datafield>') # record/1312672
expected = [
{
'description': 'Part II',
'value': 'http://link.springer.com/journal/10909/176/5/page/1',
},
]
result = hep.do(create_record(snippet)) # no roundtrip
assert validate(result['urls'], subschema) is None
assert expected == result['urls']
expected = [
{
'u': 'http://link.springer.com/journal/10909/176/5/page/1',
'y': 'Part II',
},
]
result = hep2marc.do(result)
assert expected == result['8564']
def test_ids_from_035__a_with_bai():
schema = load_schema('authors')
subschema = schema['properties']['ids']
snippet = (
'<datafield tag="035" ind1=" " ind2=" ">'
' <subfield code="a">Jian.Long.Han.1</subfield>'
'</datafield>'
) # record/1464894/export/xme
expected = [
{
'schema': 'INSPIRE BAI',
'value': 'Jian.Long.Han.1',
},
]
result = hepnames.do(create_record(snippet))
assert validate(result['ids'], subschema) is None
assert expected == result['ids']
expected = [
{
'9': 'BAI',
'a': 'Jian.Long.Han.1',
},
]
result = hepnames2marc.do(result)
assert expected == result['035']
def test_ids_from_035__a_9_with_viaf():
schema = load_schema('authors')
subschema = schema['properties']['ids']
snippet = (
'<datafield tag="035" ind1=" " ind2=" ">'
' <subfield code="9">VIAF</subfield>'
' <subfield code="a">34517183</subfield>'
'</datafield>'
) # record/1008109/export/xme
expected = [
{
'schema': 'VIAF',
'value': '34517183',
},
]
result = hepnames.do(create_record(snippet))
assert validate(result['ids'], subschema) is None
assert expected == result['ids']
expected = [
{
'9': 'VIAF',
'a': '34517183',
},
]
result = hepnames2marc.do(result)
assert expected == result['035']
def test_isbns_from_020__a_b_normalizes_online():
schema = load_schema('hep')
subschema = schema['properties']['isbns']
snippet = ('<datafield tag="020" ind1=" " ind2=" ">'
' <subfield code="a">978-94-024-0999-4</subfield>'
' <subfield code="b">Online</subfield>'
'</datafield>') # record/1504286
expected = [
{
'value': '978-94-024-0999-4',
'medium': 'online',
},
]
result = hep.do(create_record(snippet))
assert validate(result['isbns'], subschema) is None
assert expected == result['isbns']
expected = [
{
'a': '978-94-024-0999-4',
'b': 'online',
},
]
result = hep2marc.do(result)
assert expected == result['020']
def test_ids_from_035__a_9_with_googlescholar():
schema = load_schema('authors')
subschema = schema['properties']['ids']
snippet = (
'<datafield tag="035" ind1=" " ind2=" ">'
' <subfield code="9">GoogleScholar</subfield>'
' <subfield code="a">Tnl-9KoAAAAJ</subfield>'
'</datafield>'
) # record/1467553/export/xme
expected = [
{
'schema': 'GOOGLESCHOLAR',
'value': 'Tnl-9KoAAAAJ',
},
]
result = hepnames.do(create_record(snippet))
assert validate(result['ids'], subschema) is None
assert expected == result['ids']
expected = [
{
'9': 'GOOGLESCHOLAR',
'a': 'Tnl-9KoAAAAJ',
},
]
result = hepnames2marc.do(result)
assert expected == result['035']
def test_collaboration_from_710__g():
schema = load_schema('hep')
subschema = schema['properties']['collaboration']
snippet = ('<datafield tag="710" ind1=" " ind2=" ">'
' <subfield code="g">Pierre Auger</subfield>'
'</datafield>') # record/1510404
expected = [
{
'value': 'Pierre Auger'
},
]
result = hep.do(create_record(snippet))
assert validate(result['collaboration'], subschema) is None
assert expected == result['collaboration']
expected = [
{
'g': 'Pierre Auger'
},
]
result = hep2marc.do(result)
assert expected == result['710']
def test_ids_from_035__a_9_with_researcherid():
schema = load_schema('authors')
subschema = schema['properties']['ids']
snippet = (
'<datafield tag="035" ind1=" " ind2=" ">'
' <subfield code="9">RESEARCHERID</subfield>'
' <subfield code="a">B-4717-2008</subfield>'
'</datafield>'
) # record/1051026/export/xme
expected = [
{
'schema': 'RESEARCHERID',
'value': 'B-4717-2008',
},
]
result = hepnames.do(create_record(snippet))
assert validate(result['ids'], subschema) is None
assert expected == result['ids']
expected = [
{
'9': 'RESEARCHERID',
'a': 'B-4717-2008',
},
]
result = hepnames2marc.do(result)
assert expected == result['035']
def test_collaboration_from_710__g_0():
schema = load_schema('hep')
subschema = schema['properties']['collaboration']
snippet = ('<datafield tag="710" ind1=" " ind2=" ">'
' <subfield code="g">ANTARES</subfield>'
' <subfield code="0">1110619</subfield>'
'</datafield>') # record/1422032/export/xme
expected = [
{
'record': {
'$ref': 'http://localhost:5000/api/experiments/1110619',
},
'value': 'ANTARES',
},
]
result = hep.do(create_record(snippet))
assert validate(result['collaboration'], subschema) is None
assert expected == result['collaboration']
expected = [
{
'g': 'ANTARES'
},
]
result = hep2marc.do(result)
assert expected == result['710']
def test_status_from_100__a_g_q():
schema = load_schema('authors')
subschema = schema['properties']['status']
snippet = (
'<datafield tag="100" ind1=" " ind2=" ">'
' <subfield code="a">Abarbanel, Henry D.I.</subfield>'
' <subfield code="q">Henry D.I. Abarbanel</subfield>'
' <subfield code="g">ACTIVE</subfield>'
'</datafield>'
) # record/1019100
expected = 'active'
result = hepnames.do(create_record(snippet))
assert validate(result['status'], subschema) is None
assert expected == result['status']
expected = {
'a': 'Abarbanel, Henry D.I.',
'q': 'Henry D.I. Abarbanel',
'g': 'active',
}
result = hepnames2marc.do(result)
assert expected == result['100']
def test_keywords_from_695__a_2():
schema = load_schema('hep')
subschema = schema['properties']['keywords']
snippet = (
'<datafield tag="695" ind1=" " ind2=" ">'
' <subfield code="a">REVIEW</subfield>'
' <subfield code="2">INSPIRE</subfield>'
'</datafield>'
) # record/200123
expected = [
{
'keyword': 'REVIEW',
'classification_scheme': 'INSPIRE',
},
]
result = hep.do(create_record(snippet))
assert validate(result['keywords'], subschema) is None
assert expected == result['keywords']
expected = [
{
'a': 'REVIEW',
'2': 'INSPIRE',
},
]
result = hep2marc.do(result)
assert expected == result['695']
def test_positions_from_371__a():
schema = load_schema('authors')
subschema = schema['properties']['positions']
snippet = (
'<datafield tag="371" ind1=" " ind2=" ">'
' <subfield code="a">Aachen, Tech. Hochsch.</subfield>'
'</datafield>'
) # record/997958
expected = [
{
'current': False,
'institution': {
'curated_relation': False,
'name': 'Aachen, Tech. Hochsch.',
},
},
]
result = hepnames.do(create_record(snippet))
assert validate(result['positions'], subschema) is None
assert expected == result['positions']
expected = [
{'a': 'Aachen, Tech. Hochsch.'}
]
result = hepnames2marc.do(result)
assert expected == result['371']
def test_energy_ranges_from_695__2_e():
schema = load_schema('hep')
subschema = schema['properties']['energy_ranges']
snippet = (
'<datafield tag="695" ind1=" " ind2=" ">'
' <subfield code="2">INSPIRE</subfield>'
' <subfield code="e">7</subfield>'
'</datafield>'
) # record/1124337
expected = [7]
result = hep.do(create_record(snippet))
assert validate(result['energy_ranges'], subschema) is None
assert expected == result['energy_ranges']
expected = [
{
'2': 'INSPIRE',
'e': 7,
},
]
result = hep2marc.do(result)
assert expected == result['695']
def test_isbns_from_020__a_b_normalizes_electronic():
schema = load_schema('hep')
subschema = schema['properties']['isbns']
snippet = (
'<datafield tag="020" ind1=" " ind2=" ">'
' <subfield code="a">9783319006260</subfield>'
' <subfield code="b">electronic version</subfield>'
'</datafield>'
) # record/1292006
expected = [
{
'value': '9783319006260',
},
]
result = hep.do(create_record(snippet))
assert validate(result['isbns'], subschema) is None
assert expected == result['isbns']
expected = [
{
'a': '9783319006260',
},
]
result = hep2marc.do(result)
assert expected == result['020']
def test_keywords_from_653__a_9():
schema = load_schema('hep')
subschema = schema['properties']['keywords']
snippet = (
'<datafield tag="653" ind1=" " ind2=" ">'
' <subfield code="9">author</subfield>'
' <subfield code="a">Data</subfield>'
'</datafield>'
) # record/1260876
expected = [
{
'source': 'author',
'keyword': 'Data',
},
]
result = hep.do(create_record(snippet))
assert validate(result['keywords'], subschema) is None
assert expected == result['keywords']
expected = [
{
'9': 'author',
'a': 'Data',
},
]
result = hep2marc.do(result)
assert expected == result['653']
def test_isbns_from_020__a_b_normalizes_hardcover():
schema = load_schema('hep')
subschema = schema['properties']['isbns']
snippet = (
'<datafield tag="020" ind1=" " ind2=" ">'
' <subfield code="a">978-981-4571-66-1</subfield>'
' <subfield code="b">hardcover</subfield>'
'</datafield>'
) # record/1351311
expected = [
{
'value': '9789814571661',
'medium': 'hardcover',
},
]
result = hep.do(create_record(snippet))
assert validate(result['isbns'], subschema) is None
assert expected == result['isbns']
expected = [
{
'a': '9789814571661',
'b': 'hardcover',
},
]
result = hep2marc.do(result)
assert expected == result['020']
def test_external_system_numbers_from_970__a():
schema = load_schema('hep')
subschema = schema['properties']['external_system_numbers']
snippet = ('<datafield tag="970" ind1=" " ind2=" ">'
' <subfield code="a">SPIRES-10325093</subfield>'
'</datafield>') # record/1297176
expected = [
{
'institute': 'SPIRES',
'obsolete': True,
'value': 'SPIRES-10325093',
},
]
result = hep.do(create_record(snippet))
assert validate(result['external_system_numbers'], subschema) is None
assert expected == result['external_system_numbers']
expected = [
{
'a': 'SPIRES-10325093'
},
]
result = hep2marc.do(result)
assert expected == result['970']
def test_dois_from_0247_a_2():
schema = load_schema('hep')
subschema = schema['properties']['dois']
snippet = (
'<datafield tag="024" ind1="7" ind2=" ">'
' <subfield code="2">DOI</subfield>'
' <subfield code="a">10.1088/0264-9381/31/24/245004</subfield>'
'</datafield>'
) # record/1302395
expected = [
{'value': '10.1088/0264-9381/31/24/245004'},
]
result = hep.do(create_record(snippet))
assert validate(result['dois'], subschema) is None
assert expected == result['dois']
expected = [
{
'a': '10.1088/0264-9381/31/24/245004',
'2': 'DOI',
},
]
result = hep2marc.do(result)
assert expected == result['0247']
def test_collections_from_980__a():
schema = load_schema('hep')
subschema = schema['properties']['collections']
snippet = ('<datafield tag="980" ind1=" " ind2=" ">'
' <subfield code="a">HEP</subfield>'
'</datafield>') # record/1508886
expected = [
{
'primary': 'HEP'
},
]
result = hep.do(create_record(snippet))
assert validate(result['collections'], subschema) is None
assert expected == result['collections']
expected = [
{
'a': 'HEP'
},
]
result = hep2marc.do(result)
assert expected == result['980']
def test_external_system_identifiers_from_035__z_9_handles_cernkey():
schema = load_schema('hep')
subschema = schema['properties']['external_system_identifiers']
snippet = (
'<datafield tag="035" ind1=" " ind2=" ">'
' <subfield code="9">CERNKEY</subfield>'
' <subfield code="z">0263439</subfield>'
'</datafield>'
) # record/451647
expected = [
{
'schema': 'CERNKEY',
'value': '0263439',
},
]
result = hep.do(create_record(snippet))
assert validate(result['external_system_identifiers'], subschema) is None
assert expected == result['external_system_identifiers']
expected = [
{
'9': 'CERNKEY',
'z': '0263439',
},
]
result = hep2marc.do(result)
assert expected == result['035']
def test_inspire_categories_from_65017a_2():
schema = load_schema('elements/inspire_field')
snippet = ('<datafield tag="650" ind1="1" ind2="7">'
' <subfield code="2">Inspire</subfield>'
' <subfield code="a">Experiment-HEP</subfield>'
'</datafield>') # record/1426196
expected = [
{
'source': 'undefined',
'term': 'Experiment-HEP',
},
]
result = hep.do(create_record(snippet)) # no roundtrip
assert validate(result, schema) is None
assert expected == result['inspire_categories']
expected = [
{
'9': 'undefined',
'a': 'Experiment-HEP',
},
]
result = hep2marc.do(result)
assert expected == result['65017']
def test_arxiv_eprints_from_037__a_c_9():
schema = load_schema('hep')
subschema = schema['properties']['arxiv_eprints']
snippet = (
'<datafield tag="037" ind1=" " ind2=" ">'
' <subfield code="9">arXiv</subfield>'
' <subfield code="a">arXiv:1505.01843</subfield>'
' <subfield code="c">hep-ph</subfield>'
'</datafield>'
) # record/1368891
expected = [
{
'categories': [
'hep-ph',
],
'value': '1505.01843',
},
]
result = hep.do(create_record(snippet))
assert validate(result['arxiv_eprints'], subschema) is None
assert expected == result['arxiv_eprints']
expected = [
{
'9': 'arXiv',
'a': 'arXiv:1505.01843',
'c': 'hep-ph',
},
]
result = hep2marc.do(result)
assert expected == result['037']
def test_inspire_categories_from_65017a_2_9_discards_conference():
schema = load_schema('elements/inspire_field')
snippet = ('<datafield tag="650" ind1="1" ind2="7">'
' <subfield code="2">INSPIRE</subfield>'
' <subfield code="9">conference</subfield>'
' <subfield code="a">Accelerators</subfield>'
'</datafield>') # record/1479228
expected = [
{
'source': 'undefined',
'term': 'Accelerators',
},
]
result = hep.do(create_record(snippet)) # no roundtrip
assert validate(result, schema) is None
assert expected == result['inspire_categories']
expected = [
{
'9': 'undefined',
'a': 'Accelerators',
},
]
result = hep2marc.do(result)
assert expected == result['65017']
def test_report_numbers_hidden_from_037__z():
schema = load_schema('hep')
subschema = schema['properties']['report_numbers']
snippet = (
'<datafield tag="037" ind1=" " ind2=" ">'
' <subfield code="z">FERMILAB-PUB-17-011-CMS</subfield>'
'</datafield>'
) # record/1508174
expected = [
{
'hidden': True,
'value': 'FERMILAB-PUB-17-011-CMS',
},
]
result = hep.do(create_record(snippet))
assert validate(result['report_numbers'], subschema) is None
assert expected == result['report_numbers']
expected = [
{'z': 'FERMILAB-PUB-17-011-CMS'}
]
result = hep2marc.do(result)
assert expected == result['037']
def test_inspire_categories_from_65017a_2_9_converts_automatically_added():
schema = load_schema('elements/inspire_field')
snippet = (
'<datafield tag="650" ind1="1" ind2="7">'
' <subfield code="2">INSPIRE</subfield>'
' <subfield code="a">Instrumentation</subfield>'
' <subfield code="9">automatically added based on DCC, PPF, DK</subfield>'
'</datafield>') # record/669400
expected = [
{
'source': 'curator',
'term': 'Instrumentation',
},
]
result = hep.do(create_record(snippet)) # no roundtrip
assert validate(result, schema) is None
assert expected == result['inspire_categories']
expected = [
{
'9': 'curator',
'a': 'Instrumentation',
},
]
result = hep2marc.do(result)
assert expected == result['65017']
def test_isbns_from_020__a_b_normalizes_online():
schema = load_schema('hep')
subschema = schema['properties']['isbns']
snippet = (
'<datafield tag="020" ind1=" " ind2=" ">'
' <subfield code="a">978-94-024-0999-4</subfield>'
' <subfield code="b">Online</subfield>'
'</datafield>'
) # record/1504286
expected = [
{
'value': '9789402409994',
'medium': 'online',
},
]
result = hep.do(create_record(snippet))
assert validate(result['isbns'], subschema) is None
assert expected == result['isbns']
expected = [
{
'a': '9789402409994',
'b': 'online',
},
]
result = hep2marc.do(result)
assert expected == result['020']
def test_inspire_categories_from_65017a_2_9_converts_submitter():
schema = load_schema('elements/inspire_field')
snippet = ('<datafield tag="650" ind1="1" ind2="7">'
' <subfield code="a">Math and Math Physics</subfield>'
' <subfield code="9">submitter</subfield>'
' <subfield code="2">INSPIRE</subfield>'
'</datafield>') # record/1511089
expected = [
{
'source': 'user',
'term': 'Math and Math Physics',
},
]
result = hep.do(create_record(snippet)) # no roundtrip
assert validate(result, schema) is None
assert expected == result['inspire_categories']
expected = [
{
'9': 'user',
'a': 'Math and Math Physics'
},
]
result = hep2marc.do(result)
assert expected == result['65017']
def test_languages_from_double_041__a():
schema = load_schema('hep')
subschema = schema['properties']['languages']
snippet = (
'<record>'
' <datafield tag="041" ind1=" " ind2=" ">'
' <subfield code="a">French</subfield>'
' </datafield>'
' <datafield tag="041" ind1=" " ind2=" ">'
' <subfield code="a">German</subfield>'
' </datafield>'
'</record>'
) # record/1231408
expected = [
'fr',
'de',
]
result = hep.do(create_record(snippet))
assert validate(result['languages'], subschema) is None
assert expected == result['languages']
expected = [
{'a': 'french'},
{'a': 'german'},
]
result = hep2marc.do(result)
assert expected == result['041']
def test_urls_from_8564_u_y():
schema = load_schema('hep')
subschema = schema['properties']['urls']
snippet = (
'<datafield tag="856" ind1="4" ind2=" ">'
' <subfield code="u">http://www-lib.kek.jp/ar/ar.html</subfield>'
' <subfield code="y">KEK</subfield>'
'</datafield>') # record/1405358
expected = [
{
'description': 'KEK',
'value': 'http://www-lib.kek.jp/ar/ar.html',
},
]
result = hep.do(create_record(snippet))
assert validate(result['urls'], subschema) is None
assert expected == result['urls']
expected = [
{
'u': 'http://www-lib.kek.jp/ar/ar.html',
'y': 'KEK',
},
]
result = hep2marc.do(result)
assert expected == result['8564']
