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,
}
def insert_institutions():
"""
Insert institution rows.
"""
Institution.insert_institutions()
def cited(out_file):
"""
CSV with institution id, name, and citation count.
"""
# CSV writer.
cols = ["id", "count", "name"]
writer = csv.DictWriter(out_file, cols)
writer.writeheader()
count = fn.Count(HLOM_Citation.id)
cited = (
Institution.select(Institution, count)
.join(Document_Institution)
# Join citations.
.join(HLOM_Citation, on=(Document_Institution.document == HLOM_Citation.document))
.group_by(Institution.id)
.order_by(count.desc())
)
for inst in cited.naive():
writer.writerow({"count": inst.count, "id": inst.id, "name": inst.metadata["Institution_Name"]})
def queue_geocode():
"""
Queue geocoding tasks in the worker.
:param in_file: A handle on the input CSV.
"""
for inst in Institution.select():
config.rq.enqueue(geocode, inst.id)
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 geocode(id):
"""
Geocode an institution.
Args:
id (int): The institution id.
"""
coder = OpenMapQuest(config['mapquest']['api_key'])
inst = Institution.get(Institution.id==id)
# Geocode.
location = coder.geocode(inst.geocoding_query, timeout=10)
if location:
# Write the coordinate.
inst.metadata['Latitude'] = location.latitude
inst.metadata['Longitude'] = location.longitude
inst.save()
def lonlats(out_file):
"""
CSV with institution name and lon/lat, for Fusion Tables.
"""
# CSV writer.
cols = ["name", "longitude", "latitude"]
writer = csv.DictWriter(out_file, cols)
writer.writeheader()
# Select rows with coordinates.
geocoded = Institution.select().where(Institution.metadata.contains("Latitude"))
for inst in geocoded:
writer.writerow(
{
"name": inst.metadata["Institution_Name"],
"longitude": inst.metadata["Longitude"],
"latitude": inst.metadata["Latitude"],
}
)
