def test_already_harvested_returns_true_when_there_is_one_core_category():
schema = load_schema('hep')
subschema = schema['properties']['arxiv_eprints']
data = {
'arxiv_eprints': [
{
'categories': [
'hep-ph',
'astro-ph.CO',
'gr-qc',
],
'value': '1609.03939',
},
],
}
extra_data = {}
assert validate(data['arxiv_eprints'], subschema) is None
obj = MockObj(data, extra_data)
eng = MockEng()
assert already_harvested(obj, eng)
expected = ('Record with arXiv id 1609.03939 is'
' already being harvested on Legacy.')
result = obj.log._info.getvalue()
assert expected == result
def test_already_harvested_returns_false_otherwise():
schema = load_schema('hep')
subschema = schema['properties']['arxiv_eprints']
data = {
'arxiv_eprints': [
{
'categories': [
'math.CO',
],
'value': '1705.01122',
},
],
}
extra_data = {}
assert validate(data['arxiv_eprints'], subschema) is None
obj = MockObj(data, extra_data)
eng = MockEng()
assert not already_harvested(obj, eng)
expected = ''
result = obj.log._info.getvalue()
assert expected == result
def test_already_harvested_returns_true_when_there_is_one_core_category():
schema = load_schema('hep')
subschema = schema['properties']['arxiv_eprints']
data = {
'arxiv_eprints': [
{
'categories': [
'hep-ph',
'astro-ph.CO',
'gr-qc',
],
'value': '1609.03939',
},
],
}
extra_data = {}
assert validate(data['arxiv_eprints'], subschema) is None
obj = MockObj(data, extra_data)
eng = MockEng()
assert already_harvested(obj, eng)
expected = (
'Record with arXiv id 1609.03939 is'
' already being harvested on Legacy.'
)
result = obj.log._info.getvalue()
assert expected == result
def test_already_harvested_returns_false_otherwise(app):
schema = load_schema('hep')
subschema = schema['properties']['arxiv_eprints']
data = {
'arxiv_eprints': [
{
'categories': [
'math.CO',
],
'value': '1705.01122',
},
],
}
extra_data = {}
assert validate(data['arxiv_eprints'], subschema) is None
obj = MockObj(data, extra_data)
eng = MockEng()
extra_config = {
'ARXIV_CATEGORIES_ALREADY_HARVESTED_ON_LEGACY': [
'astro-ph.CO'
]
}
with patch.dict(app.config, extra_config):
assert not already_harvested(obj, eng)
expected = ''
result = obj.log._info.getvalue()
assert expected == result
def test_already_harvested_returns_false_otherwise():
schema = load_schema('hep')
subschema = schema['properties']['arxiv_eprints']
data = {
'arxiv_eprints': [
{
'categories': [
'astro-ph.IM',
'astro-ph.CO',
'cs.DS',
'stat.ML',
],
'value': '1609.03932',
},
],
}
extra_data = {}
assert validate(data['arxiv_eprints'], subschema) is None
obj = MockObj(data, extra_data)
eng = MockEng()
assert not already_harvested(obj, eng)
expected = ''
result = obj.log._info.getvalue()
assert expected == result