def bring_orgs_to_memory(self):
"""
Make a db call to get all organizations and trim it down to a smaller model to conserve space.
Returns:
small_orgs ([SmallOrganization]): A list of all organizations from the database, converted to
SmallOrganization objects to save some space.
"""
# grab all organizations
tmp_org_dtos = self.org_dao.all(
"id",
"page_rank_info.total",
"organization_url",
"page_rank_info.total_with_self",
"page_rank_info.references.count",
"page_rank_info.references.org_domain",
"page_rank_info.references.pages.url",
"page_rank_info.references.pages.count",
)
# convert the dtos to this smaller model
small_orgs = [
SmallOrganization(DTOConverter.from_dto(Organization, o), o.id) for o in tmp_org_dtos if o.organization_url
]
small_orgs = [o for o in small_orgs if o.org_domain]
# At this point, tmp_org_dtos gets gc'd and we're left with the minimum data needed for calculation
return small_orgs
def test_org_dto_converter(self):
my_org = Organization(
name='Yoyodyne, Inc.',
address="1234 Yoyodyne Way, San Narciso, CA",
types=[OrgTypesEnum.GOVERNMENT, OrgTypesEnum.RESEARCH],
phone_numbers=["4026170423"],
emails=["*****@*****.**"],
contacts=[],
organization_url="www.yoyodyne.com",
page_rank_info=PageRankInfo(
total_with_self=10,
total=8,
references=[
PageRankVector(org_domain='yoyodyne.com',
count=2,
pages=[
UrlCountPair(
url='http://www.yoyodyne.com/',
count=2)
]),
PageRankVector(
org_domain='trystero.org',
count=4,
pages=[
UrlCountPair(url='http://www.yoyodyne.com/',
count=3),
UrlCountPair(
url='http://www.yoyodyne.com/contacts.php',
count=1)
]),
PageRankVector(org_domain='thurnandtaxis.info',
count=4,
pages=[
UrlCountPair(
url='http://www.yoyodyne.com/',
count=4)
])
]))
print 'Converting an organization model to a DTO.'
org_dto = DTOConverter.to_dto(OrganizationDTO, my_org)
print 'Testing equality...'
for attr, value in my_org.__dict__.iteritems():
if attr == 'page_rank_info':
self._compare_page_rank_info(my_org, org_dto)
else:
self.assertEqual(getattr(my_org, attr), getattr(org_dto, attr),
"{0} attribute not equal".format(attr))
print 'Converting a DTO to an organization.'
my_org = DTOConverter.from_dto(Organization, org_dto)
print 'Testing equality...'
for attr, value in my_org.__dict__.iteritems():
if attr == 'page_rank_info':
self._compare_page_rank_info(my_org, org_dto)
else:
self.assertEqual(getattr(my_org, attr), getattr(org_dto, attr),
"{0} attribute not equal".format(attr))
def test_contact_dto_converter(self):
my_contact = Contact(
first_name="Jordan",
last_name="Degner",
phones=['4029813230'],
email="*****@*****.**",
position="Software Engineer",
)
print 'Converting a contact to a DTO.'
contact_dto = DTOConverter.to_dto(ContactDTO, my_contact)
print 'Testing equality...'
for attr, value in my_contact.__dict__.iteritems():
self.assertEqual(getattr(my_contact, attr),
getattr(contact_dto, attr),
"{0} attribute not equal".format(attr))
print 'Converting a DTO to a contact.'
my_contact = DTOConverter.from_dto(Contact, contact_dto)
print 'Testing equality...'
for attr, value in my_contact.__dict__.iteritems():
self.assertEqual(getattr(my_contact, attr),
getattr(contact_dto, attr),
"{0} attribute not equal".format(attr))
def test_item_converter(self):
ctx = ApplicationContext(TestableDAOContext())
print 'Creating organization and contact item.'
org = ctx.get_object('OrganizationDAO')
org_dto = OrganizationDTO(name="Univerisityee of Nyeebraska-Lincoln")
org.create_update(org_dto)
org_model = DTOConverter.from_dto(Organization, org_dto)
contact_item = ScrapedContact(
first_name='Bee',
last_name='Yee',
organization={'name': "Univerisityee of Nyeebraska-Lincoln"})
print 'Converting contact to model.'
converter = ctx.get_object('ModelConverter')
model_contact = converter.to_model(Contact, contact_item)
self.assertEqual(model_contact.organization.name, org_model.name)
def test_item_converter(self):
ctx = ApplicationContext(TestableDAOContext())
print 'Creating organization and contact item.'
org = ctx.get_object('OrganizationDAO')
org_dto = OrganizationDTO(name="Univerisityee of Nyeebraska-Lincoln")
org.create_update(org_dto)
org_model = DTOConverter.from_dto(Organization, org_dto)
contact_item = ScrapedContact(first_name='Bee',
last_name='Yee',
organization={'name': "Univerisityee of Nyeebraska-Lincoln"}
)
print 'Converting contact to model.'
converter = ctx.get_object('ModelConverter')
model_contact = converter.to_model(Contact, contact_item)
self.assertEqual(model_contact.organization.name, org_model.name)
def _monitor_cache(dao, max_size, cache, job_queue, job_cond, fill_cond, empty_cond,
req_doms, blk_doms, srt_list, logger_lock):
while True:
try:
with job_cond:
next_job = job_queue.get(block=False)
except Empty:
with job_cond:
job_cond.wait(1)
try:
next_job = job_queue.get(block=False)
except Empty:
continue
if next_job == CacheJobs.Fill:
with logger_lock:
logger.info('Filling the cache')
with fill_cond:
urls = dao().findmany_by_domains(max_size - cache.qsize(),
req_doms, blk_doms, srt_list)
for u in urls:
url_obj = DTOConverter.from_dto(URLMetadata, u)
try:
cache.put(url_obj)
except Full:
break
fill_cond.notify_all()
elif next_job == CacheJobs.Empty:
with logger_lock:
logger.info('Emptying the cache')
with empty_cond:
while True:
try:
cache.get(block=False)
except Empty:
break
empty_cond.notify()
def test_contact_dto_converter(self):
my_contact = Contact(first_name="Jordan",
last_name="Degner",
phones=['4029813230'],
email="*****@*****.**",
position="Software Engineer",
)
print 'Converting a contact to a DTO.'
contact_dto = DTOConverter.to_dto(ContactDTO, my_contact)
print 'Testing equality...'
for attr, value in my_contact.__dict__.iteritems():
self.assertEqual(getattr(my_contact, attr), getattr(contact_dto, attr),
"{0} attribute not equal".format(attr))
print 'Converting a DTO to a contact.'
my_contact = DTOConverter.from_dto(Contact, contact_dto)
print 'Testing equality...'
for attr, value in my_contact.__dict__.iteritems():
self.assertEqual(getattr(my_contact, attr), getattr(contact_dto, attr),
"{0} attribute not equal".format(attr))
def test_page_rank(self):
print 'Creating PageRankPreprocessor'
prp = self.ctx.get_object('PageRankPreprocessor')
print 'Bring organizations to memory'
orgs = prp.bring_orgs_to_memory()
print 'Cleaning organizations'
orgs = prp.cleanup_data(orgs)
print 'Creating dat matrix'
matrix = prp.create_matrix(orgs)
self.assertIsNotNone(matrix)
print 'Creating the dampened google matrix'
matrix = google_matrix(matrix)
print 'Generating eigenvector'
vector = left_eigenvector(matrix)
print 'Creating PageRankPostprocessor'
post = self.ctx.get_object('PageRankPostprocessor')
print 'Assigning ranks to organizations'
orgs = post.give_orgs_ranks(orgs, vector)
print 'Storing organizations'
post.store_organizations(orgs)
dao = self.ctx.get_object('OrganizationDAO')
new_org_models = [DTOConverter.from_dto(Organization, o) for o in dao.all()]
self.assertEqual(len(new_org_models), len(self.assert_models),
"Error: returned model number different than expected")
for model in self.assert_models:
self._compare_assert_against_test(model, new_org_models)
def test_org_dto_converter(self):
my_org = Organization(name='Yoyodyne, Inc.',
address="1234 Yoyodyne Way, San Narciso, CA",
types=[OrgTypesEnum.GOVERNMENT, OrgTypesEnum.RESEARCH],
phone_numbers=["4026170423"],
emails=["*****@*****.**"],
contacts=[],
organization_url="www.yoyodyne.com",
page_rank_info=PageRankInfo(
total_with_self=10,
total=8,
references=[
PageRankVector(
org_domain='yoyodyne.com',
count=2,
pages=[
UrlCountPair(
url='http://www.yoyodyne.com/',
count=2
)
]
),
PageRankVector(
org_domain='trystero.org',
count=4,
pages=[
UrlCountPair(
url='http://www.yoyodyne.com/',
count=3
),
UrlCountPair(
url='http://www.yoyodyne.com/contacts.php',
count=1
)
]
),
PageRankVector(
org_domain='thurnandtaxis.info',
count=4,
pages=[
UrlCountPair(
url='http://www.yoyodyne.com/',
count=4
)
]
)
]
)
)
print 'Converting an organization model to a DTO.'
org_dto = DTOConverter.to_dto(OrganizationDTO, my_org)
print 'Testing equality...'
for attr, value in my_org.__dict__.iteritems():
if attr == 'page_rank_info':
self._compare_page_rank_info(my_org, org_dto)
else:
self.assertEqual(getattr(my_org, attr), getattr(org_dto, attr),
"{0} attribute not equal".format(attr))
print 'Converting a DTO to an organization.'
my_org = DTOConverter.from_dto(Organization, org_dto)
print 'Testing equality...'
for attr, value in my_org.__dict__.iteritems():
if attr == 'page_rank_info':
self._compare_page_rank_info(my_org, org_dto)
else:
self.assertEqual(getattr(my_org, attr), getattr(org_dto, attr),
"{0} attribute not equal".format(attr))