def match_doc(id):
"""
Find an institution with the same base URL as a document.
Args:
id (int): A document id.
"""
doc = Document.get(Document.id==id)
# Break if no manifest.
if not doc.syllabus.registered_domain:
return
# Form the domain query.
q = '%'+doc.syllabus.registered_domain+'%'
inst = (
Institution
.select()
.where(Institution.metadata['Institution_Web_Address'] ** (q))
.order_by(Institution.id)
.first()
)
if inst:
Document_Institution.create(
document=doc.id,
institution=inst
)
def test_state_abbreviations(add_hlom, add_doc):
"""
HLOM_Citation.index_state() should denormalize state abbreviations.
"""
t1 = add_hlom()
t2 = add_hlom()
t3 = add_hlom()
s1 = add_doc('syllabus1')
s2 = add_doc('syllabus2')
s3 = add_doc('syllabus3')
c1 = HLOM_Citation.create(document=s1, record=t1)
c2 = HLOM_Citation.create(document=s2, record=t2)
c3 = HLOM_Citation.create(document=s3, record=t3)
# Create institutions with states.
AL = Institution.create(metadata={'Institution_State': 'AL'})
CT = Institution.create(metadata={'Institution_State': 'CT'})
CA = Institution.create(metadata={'Institution_State': 'CA'})
# Link documents -> institutions.
Document_Institution.create(document=t1, institution=AL)
Document_Institution.create(document=t2, institution=CT)
Document_Institution.create(document=t3, institution=CA)
HLOM_Citation.index_institutions()
c1 = HLOM_Citation.reload(c1)
c2 = HLOM_Citation.reload(c2)
c3 = HLOM_Citation.reload(c3)
assert c1.state == 'AL'
assert c2.state == 'CT'
assert c3.state == 'CA'
def test_institution_ids(add_hlom, add_doc):
"""
HLOM_Citation.index_state() should denormalize institution ids.
"""
t1 = add_hlom()
t2 = add_hlom()
t3 = add_hlom()
s1 = add_doc('syllabus1')
s2 = add_doc('syllabus2')
s3 = add_doc('syllabus3')
c1 = HLOM_Citation.create(document=s1, record=t1)
c2 = HLOM_Citation.create(document=s2, record=t2)
c3 = HLOM_Citation.create(document=s3, record=t3)
i1 = Institution.create()
i2 = Institution.create()
i3 = Institution.create()
# Link documents -> institutions.
Document_Institution.create(document=t1, institution=i1)
Document_Institution.create(document=t2, institution=i2)
Document_Institution.create(document=t3, institution=i3)
HLOM_Citation.index_institutions()
c1 = HLOM_Citation.reload(c1)
c2 = HLOM_Citation.reload(c2)
c3 = HLOM_Citation.reload(c3)
assert c1.institution == i1
assert c2.institution == i2
assert c3.institution == i3
def test_institution_counts(models):
"""
Document_Institution.institution_counts() should provide syllabus counts
for each institution id.
"""
i1 = Institution.create()
i2 = Institution.create()
i3 = Institution.create()
d1 = Document.create(path='d1')
d2 = Document.create(path='d2')
d3 = Document.create(path='d3')
d4 = Document.create(path='d4')
d5 = Document.create(path='d5')
d6 = Document.create(path='d6')
# 1 document for institution 1.
Document_Institution.create(institution=i1, document=d1)
# 2 documents for institution 2.
Document_Institution.create(institution=i2, document=d2)
Document_Institution.create(institution=i2, document=d3)
# 3 documents for institution 3.
Document_Institution.create(institution=i3, document=d4)
Document_Institution.create(institution=i3, document=d5)
Document_Institution.create(institution=i3, document=d6)
assert Document_Institution.institution_counts() == {
d1.id: 1,
d2.id: 2,
d3.id: 3,
}
