def test_can_get_domains_details(self):
self.db.query(Domain).delete()
details = Domain.get_domains_details(self.db)
expect(details).to_length(0)
domain = DomainFactory.create(name='domain-1.com',
url='http://domain-1.com/')
domain2 = DomainFactory.create(name='domain-2.com',
url='http://domain-2.com/')
DomainFactory.create()
page = PageFactory.create(domain=domain)
page2 = PageFactory.create(domain=domain)
page3 = PageFactory.create(domain=domain2)
ReviewFactory.create(domain=domain,
page=page,
is_active=True,
number_of_violations=20)
ReviewFactory.create(domain=domain,
page=page2,
is_active=True,
number_of_violations=10)
ReviewFactory.create(domain=domain2,
page=page3,
is_active=True,
number_of_violations=30)
RequestFactory.create(status_code=200,
domain_name=domain.name,
response_time=0.25)
RequestFactory.create(status_code=304,
domain_name=domain.name,
response_time=0.35)
RequestFactory.create(status_code=400,
domain_name=domain.name,
response_time=0.25)
RequestFactory.create(status_code=403,
domain_name=domain.name,
response_time=0.35)
RequestFactory.create(status_code=404,
domain_name=domain.name,
response_time=0.25)
details = Domain.get_domains_details(self.db)
expect(details).to_length(3)
expect(details[0]).to_length(10)
expect(details[0]['url']).to_equal('http://domain-1.com/')
expect(details[0]['name']).to_equal('domain-1.com')
expect(details[0]['violationCount']).to_equal(30)
expect(details[0]['pageCount']).to_equal(2)
expect(details[0]['reviewCount']).to_equal(2)
expect(details[0]['reviewPercentage']).to_equal(100.0)
expect(details[0]['errorPercentage']).to_equal(60.0)
expect(details[0]['is_active']).to_be_true()
expect(details[0]['averageResponseTime']).to_equal(0.3)
def save_request(self, url, response):
if not response:
return
request_time = response.request_time
effective_url = response.effective_url
status_code = response.status_code
domain_name, domain_url = get_domain_from_url(url)
if domain_name not in Domain.get_domain_names(self.db):
return
req = Request(
domain_name=domain_name,
url=url,
effective_url=effective_url,
status_code=int(status_code),
response_time=request_time,
completed_date=datetime.now().date(),
review_url=self.page_url
)
self.db.add(req)
self.cache.increment_requests_count()
self.publish(dumps({
'type': 'new-request',
'url': str(url)
}))
def get(self, domain_name):
domain = Domain.get_domain_by_name(domain_name, self.db)
if not domain:
self.set_status(404, 'Domain %s not found' % domain_name)
return
violation_defs = self.application.violation_definitions
grouped_violations = self.girl.get('violation_count_by_category_for_domains')
total = 0
violations = []
for item in grouped_violations.get(domain.id, []):
key_name, key_category_id, count = item['key_name'], item['category_id'], item['violation_count']
violations.append({
'categoryId': key_category_id,
'categoryName': violation_defs[key_name]['category'],
'count': count
})
total += count
result = {
"domainId": domain.id,
'domainName': domain.name,
'domainURL': domain.url,
'total': total,
'violations': violations
}
self.write_json(result)
def get(self, domain_name):
term = self.get_argument('term', None)
current_page = int(self.get_argument('current_page', 1))
page_size = int(self.get_argument('page_size', 10))
domain = Domain.get_domain_by_name(domain_name, self.db)
if not domain:
self.set_status(404, self._('Domain %s not found') % domain_name)
return
reviews = yield self.application.search_provider.get_domain_active_reviews(
domain=domain,
current_page=current_page,
page_size=page_size,
page_filter=term,
)
if 'error' in reviews:
self.set_status(reviews['error']['status_code'],
reviews['error']['reason'])
self.finish()
return
if 'reviewsCount' not in reviews:
if not term:
reviews[
'reviewsCount'] = yield self.cache.get_active_review_count(
domain)
else:
reviews['reviewsCount'] = None
self.write_json(reviews)
self.finish()
def save_request(self, url, response):
if not response:
return
request_time = response.request_time
effective_url = response.effective_url
status_code = response.status_code
if self.domain_name not in Domain.get_domain_names(self.db):
return
req = Request(
domain_name=self.domain_name,
url=url,
effective_url=effective_url,
status_code=int(status_code),
response_time=request_time,
completed_date=datetime.now().date(),
review_url=self.page_url
)
self.db.add(req)
url = url.encode('utf-8')
self.publish(dumps({
'type': 'new-request',
'url': str(url)
}))
def get(self, domain_name):
domain = Domain.get_domain_by_name(domain_name, self.db)
if not domain:
self.set_status(404, self._('Domain %s not found') % domain_name)
return
prefs = DomainsViolationsPrefs.get_domains_violations_prefs_by_domain(
self.db, domain.name
)
violation_defs = self.application.violation_definitions
result = []
for pref in prefs:
key = violation_defs.get(pref.get('key'))
if key is None:
continue
result.append({
'key': pref.get('key'),
'title': key.get('default_value_description', None),
'category': key.get('category', None),
'value': pref.get('value'),
'default_value': key.get('default_value', None),
'unit': key.get('unit', None)
})
self.write_json(result)
def get_count(self, key, domain_name, expiration, get_count_method):
cache_key = '%s-%s' % (self.get_domain_name(domain_name), key)
count = self.redis.get(cache_key)
if count is not None:
return int(count)
domain = domain_name
if domain and not isinstance(domain, Domain):
domain = Domain.get_domain_by_name(domain_name, self.db)
if domain is None:
count = Page.get_page_count(self.db)
else:
count = get_count_method(domain)
cache_key = '%s-%s' % (self.get_domain_name(domain), key)
self.redis.setex(
cache_key,
expiration,
value=int(count)
)
return int(count)
def test_can_insert_default_violations_values_for_all_domains(self):
DomainsViolationsPrefsFactory.create(domain=Domain(name='globo.com'),
key=Key(name='some.random.fact'),
value='whatever')
for x in range(3):
DomainFactory.create(name='g%d.com' % x)
domains_violations_prefs = \
DomainsViolationsPrefs.get_domains_violations_prefs(self.db)
expect(domains_violations_prefs).to_length(1)
default_violations_values = {
'page.title.size': 100,
'total.requests.img': 5,
}
page_title_size = KeyFactory.create(name='page.title.size')
total_requests_img = KeyFactory.create(name='total.requests.img')
violation_definitions = {
'page.title.size': {
'key': page_title_size,
'default_value': 100
},
'total.requests.img': {
'key': total_requests_img,
'default_value': 5
}
}
DomainsViolationsPrefs.insert_default_violations_values_for_all_domains(
self.db, default_violations_values, violation_definitions,
self.cache)
domains_violations_prefs = \
DomainsViolationsPrefs.get_domains_violations_prefs(self.db)
expect(domains_violations_prefs).to_length(4)
expect(domains_violations_prefs).to_be_like({
'globo.com': {
'some.random.fact': 'whatever',
'total.requests.img': 5,
'page.title.size': 100
},
'g0.com': {
'page.title.size': 100,
'total.requests.img': 5
},
'g1.com': {
'page.title.size': 100,
'total.requests.img': 5
},
'g2.com': {
'page.title.size': 100,
'total.requests.img': 5
},
})
def get(self, domain_name, key_category_id):
domain = Domain.get_domain_by_name(domain_name, self.db)
if not domain:
self.set_status(404, 'Domain %s not found' % domain_name)
return
violation_defs = self.application.violation_definitions
top_violations = yield self.cache.get_top_in_category_for_domain(
domain,
key_category_id,
self.application.config.get('TOP_CATEGORY_VIOLATIONS_LIMIT')
)
violations = []
for key_name, count in top_violations:
violations.append({
'title': violation_defs[key_name]['title'],
'count': count
})
result = {
"domainId": domain.id,
'domainName': domain.name,
'domainURL': domain.url,
'categoryId': key_category_id,
'violations': violations
}
self.write_json(result)
def post(self, domain_name):
domain = Domain.get_domain_by_name(domain_name, self.db)
if not domain:
self.set_status(404, 'Domain %s not found' % domain_name)
return
domain.is_active = not domain.is_active
def test_domains_violations_prefs_str(self):
data = DomainsViolationsPrefsFactory.create(
domain=Domain(name='globo.com'),
key=Key(name='some.random.fact'),
value='whatever')
loaded = self.db.query(DomainsViolationsPrefs).get(data.id)
expect(
str(loaded)).to_be_like('some.random.fact (globo.com): whatever')
def test_can_get_active_domains(self):
self.db.query(Domain).delete()
domain = DomainFactory(is_active=True)
DomainFactory(is_active=False)
domains = Domain.get_active_domains(self.db)
expect(domains).to_length(1)
expect(domains[0].id).to_equal(domain.id)
def test_get_domain_names(self):
self.db.query(Domain).delete()
DomainFactory.create(name="g1.globo.com")
DomainFactory.create(name="globoesporte.globo.com")
domain_names = Domain.get_domain_names(self.db)
expect(domain_names).to_be_like(
['g1.globo.com', 'globoesporte.globo.com'])
def test_can_get_active_domains(self):
self.db.query(Domain).delete()
domain = DomainFactory(is_active=True)
DomainFactory(is_active=False)
domains = Domain.get_active_domains(self.db)
expect(domains).to_length(1)
expect(domains[0].id).to_equal(domain.id)
def test_can_set_domain_to_inactive(self):
domain = DomainFactory.create(url="http://www.domain.com",
name="domain.com",
is_active=True)
response = yield self.authenticated_fetch(
'/domains/%s/change-status/' % domain.name, method='POST', body='')
expect(response.code).to_equal(200)
domain_from_db = Domain.get_domain_by_name(domain.name, self.db)
expect(domain_from_db.is_active).to_be_false()
def post(self, domain_name):
domain = Domain.get_domain_by_name(domain_name, self.db)
if not domain:
self.set_status(404, self._('Domain %s not found') % domain_name)
return
domain.is_active = not domain.is_active
if not domain.is_active:
yield self.cache.delete_limit_usage_by_domain(domain.url)
def test_can_set_domain_to_active(self):
domain = DomainFactory.create(url="http://www.domain.com", name="domain.com", is_active=False)
response = yield self.http_client.fetch(
self.get_url(r'/domains/%s/change-status/' % domain.name),
method='POST',
body=''
)
expect(response.code).to_equal(200)
domain_from_db = Domain.get_domain_by_name(domain.name, self.db)
expect(domain_from_db.is_active).to_be_true()
def test_can_convert_to_dict(self):
data = DomainsViolationsPrefsFactory.create(
domain=Domain(name='globo.com'),
key=Key(name='some.random.fact'),
value='whatever')
expect(data.to_dict()).to_be_like({
'domain': 'globo.com',
'key': 'some.random.fact',
'value': 'whatever',
})
def test_can_create_domains_violations_prefs(self):
data = DomainsViolationsPrefsFactory.create(
domain=Domain(name='globo.com'),
key=Key(name='some.random.fact'),
value='whatever')
loaded = self.db.query(DomainsViolationsPrefs).get(data.id)
expect(loaded.domain.name).to_equal('globo.com')
expect(loaded.key.name).to_equal('some.random.fact')
expect(loaded.value).to_equal('whatever')
def test_can_set_domain_to_inactive(self):
domain = DomainFactory.create(url="http://www.domain.com", name="domain.com", is_active=True)
response = yield self.authenticated_fetch(
'/domains/%s/change-status/' % domain.name,
method='POST',
body=''
)
expect(response.code).to_equal(200)
domain_from_db = Domain.get_domain_by_name(domain.name, self.db)
expect(domain_from_db.is_active).to_be_false()
def test_get_domain_names(self):
self.db.query(Domain).delete()
DomainFactory.create(name="g1.globo.com")
DomainFactory.create(name="globoesporte.globo.com")
domain_names = Domain.get_domain_names(self.db)
expect(domain_names).to_be_like([
'g1.globo.com',
'globoesporte.globo.com'
])
def test_can_get_domains_violations_prefs(self):
data = DomainsViolationsPrefsFactory.create(
domain=Domain(name='globo.com'),
key=Key(name='some.random.fact'),
value='whatever')
data = DomainsViolationsPrefs.get_domains_violations_prefs(self.db)
expect(data).to_be_like(
{'globo.com': {
'some.random.fact': 'whatever'
}})
def get_next_jobs_count(cls, db, config):
from holmes.models import Domain
active_domains = Domain.get_active_domains(db)
active_domains_ids = [item.id for item in active_domains]
return db \
.query(
sa.func.count(Page.id)
) \
.filter(Page.domain_id.in_(active_domains_ids)) \
.scalar()
def test_can_get_domains_details(self):
self.db.query(Domain).delete()
details = Domain.get_domains_details(self.db)
expect(details).to_length(0)
domain = DomainFactory.create(name='domain-1.com', url='http://domain-1.com/')
domain2 = DomainFactory.create(name='domain-2.com', url='http://domain-2.com/')
DomainFactory.create()
page = PageFactory.create(domain=domain)
page2 = PageFactory.create(domain=domain)
page3 = PageFactory.create(domain=domain2)
ReviewFactory.create(domain=domain, page=page, is_active=True, number_of_violations=20)
ReviewFactory.create(domain=domain, page=page2, is_active=True, number_of_violations=10)
ReviewFactory.create(domain=domain2, page=page3, is_active=True, number_of_violations=30)
RequestFactory.create(status_code=200, domain_name=domain.name, response_time=0.25)
RequestFactory.create(status_code=304, domain_name=domain.name, response_time=0.35)
RequestFactory.create(status_code=400, domain_name=domain.name, response_time=0.25)
RequestFactory.create(status_code=403, domain_name=domain.name, response_time=0.35)
RequestFactory.create(status_code=404, domain_name=domain.name, response_time=0.25)
details = Domain.get_domains_details(self.db)
expect(details).to_length(3)
expect(details[0]).to_length(10)
expect(details[0]['url']).to_equal('http://domain-1.com/')
expect(details[0]['name']).to_equal('domain-1.com')
expect(details[0]['violationCount']).to_equal(30)
expect(details[0]['pageCount']).to_equal(2)
expect(details[0]['reviewCount']).to_equal(2)
expect(details[0]['reviewPercentage']).to_equal(100.0)
expect(details[0]['errorPercentage']).to_equal(60.0)
expect(details[0]['is_active']).to_be_true()
expect(details[0]['averageResponseTime']).to_equal(0.3)
def handle(has_key):
domain = domain_name
if domain and not isinstance(domain, Domain):
domain = Domain.get_domain_by_name(domain_name, self.db)
if has_key:
self.redis.incrby(key, increment, callback=callback)
else:
if domain is None:
value = Page.get_page_count(self.db) + increment - 1
else:
value = get_default_method(domain) + increment - 1
self.redis.set(key, value, callback=callback)
def test_can_get_pages_per_domain(self):
domain = DomainFactory.create()
domain2 = DomainFactory.create()
DomainFactory.create()
PageFactory.create(domain=domain)
PageFactory.create(domain=domain)
PageFactory.create(domain=domain2)
PageFactory.create(domain=domain2)
PageFactory.create(domain=domain2)
pages_per_domain = Domain.get_pages_per_domain(self.db)
expect(pages_per_domain).to_be_like({domain.id: 2, domain2.id: 3})
def handle(has_key):
domain = domain_name
if domain and not isinstance(domain, Domain):
domain = Domain.get_domain_by_name(domain_name, self.db)
if has_key:
self.redis.incrby(key, increment, callback=callback)
else:
if domain is None:
value = Page.get_page_count(self.db) + increment - 1
else:
value = get_default_method(domain) + increment - 1
self.redis.set(key, value, callback=callback)
def get(self, key_name):
current_page = int(self.get_argument('current_page', 1))
page_size = int(self.get_argument('page_size', 10))
domain_filter = self.get_argument('domain_filter', None)
page_filter = self.get_argument('page_filter', None)
domain = None
if domain_filter is not None:
domain = Domain.get_domain_by_name(domain_filter, self.db)
if not domain:
self.set_status(404,
self._('Domain %s not found') % domain_filter)
self.finish()
return
violations = self.application.violation_definitions
if key_name not in violations:
self.set_status(404, self._('Invalid violation key %s') % key_name)
self.finish()
return
violation_title = violations[key_name]['title']
key_id = violations[key_name]['key'].id
violation = yield self.application.search_provider.get_by_violation_key_name(
key_id=key_id,
current_page=current_page,
page_size=page_size,
domain=domain,
page_filter=page_filter,
)
if 'error' in violation:
self.set_status(violation['error']['status_code'],
violation['error']['reason'])
self.finish()
return
if 'reviewsCount' not in violation:
if not domain and not page_filter:
violation['reviewsCount'] = Review.count_by_violation_key_name(
self.db, key_id)
else:
violation['reviewsCount'] = None
violation['title'] = violation_title
self.write_json(violation)
self.finish()
def get(self, domain_name):
domain = Domain.get_domain_by_name(domain_name, self.db)
if not domain:
self.set_status(404, self._('Domain %s not found') % domain_name)
return
result = self.girl.get('domains_details') or []
data = next((l for l in result if l['name'] == domain_name), None)
if not data:
self.set_status(404, self._('Domain %s not found') % domain_name)
return
page_count = data.get('pageCount', 0)
review_count = data.get('reviewCount', 0)
violation_count = data.get('violationCount', 0)
error_percentage = data.get('errorPercentage', 0)
response_time_avg = data.get('averageResponseTime', 0)
review_percentage = data.get('reviewPercentage', 0)
domain_json = {
"id": domain.id,
"name": domain.name,
"url": domain.url,
"pageCount": page_count,
"reviewCount": review_count,
"violationCount": violation_count,
"reviewPercentage": review_percentage,
"is_active": domain.is_active,
"errorPercentage": error_percentage,
"averageResponseTime": response_time_avg,
"homepageId": "",
"homepageReviewId": "",
}
homepage = domain.get_homepage(self.db)
if homepage:
if homepage.uuid:
domain_json["homepageId"] = str(homepage.uuid)
if homepage.last_review_uuid:
domain_json["homepageReviewId"] = str(
homepage.last_review_uuid)
self.write_json(domain_json)
def get(self, domain_name):
domain = Domain.get_domain_by_name(domain_name, self.db)
if not domain:
self.set_status(404, self._('Domain %s not found') % domain_name)
return
violations_per_day = domain.get_violations_per_day(self.db)
domain_json = {
"id": domain.id,
"name": domain.name,
"url": domain.url,
"violations": violations_per_day
}
self.write_json(domain_json)
def test_can_get_pages_per_domain(self):
domain = DomainFactory.create()
domain2 = DomainFactory.create()
DomainFactory.create()
PageFactory.create(domain=domain)
PageFactory.create(domain=domain)
PageFactory.create(domain=domain2)
PageFactory.create(domain=domain2)
PageFactory.create(domain=domain2)
pages_per_domain = Domain.get_pages_per_domain(self.db)
expect(pages_per_domain).to_be_like({
domain.id: 2,
domain2.id: 3
})
def get(self, domain_name):
domain = Domain.get_domain_by_name(domain_name, self.db)
if not domain:
self.set_status(404, 'Domain %s not found' % domain_name)
return
violations_per_day = domain.get_violations_per_day(self.db)
domain_json = {
"id": domain.id,
"name": domain.name,
"url": domain.url,
"violations": violations_per_day
}
self.write_json(domain_json)
def increment_count(self, key, domain_name, get_default_method, increment=1):
key = '%s-%s' % (self.get_domain_name(domain_name), key)
has_key = self.has_key(key)
domain = domain_name
if domain and not isinstance(domain, Domain):
domain = Domain.get_domain_by_name(domain_name, self.db)
if has_key:
self.redis.incrby(key, increment)
else:
if domain is None:
value = Page.get_page_count(self.db) + increment - 1
else:
value = get_default_method(domain) + increment - 1
self.redis.set(key, value)
def add_domain(cls, url, db, publish_method, config, girl,
default_violations_values, violation_definitions, cache):
from holmes.models import Domain, DomainsViolationsPrefs
from holmes.material import expire_materials
domain_name, domain_url = get_domain_from_url(url)
domains = db.query(Domain).filter(
or_(Domain.name == domain_name,
Domain.name == domain_name.rstrip('/'),
Domain.name == "%s/" % domain_name)).all()
if not domains:
domain = None
else:
domain = domains[0]
if not domain:
url_hash = hashlib.sha512(domain_url).hexdigest()
domain = Domain(url=domain_url,
url_hash=url_hash,
name=domain_name)
db.add(domain)
db.flush()
expire_materials(girl)
publish_method(
dumps({
'type': 'new-domain',
'domainUrl': str(domain_url)
}))
keys = default_violations_values.keys()
DomainsViolationsPrefs.insert_default_violations_values_for_domain(
db, domain, keys, violation_definitions, cache)
from holmes.models import Limiter
connections = config.DEFAULT_NUMBER_OF_CONCURRENT_CONNECTIONS
Limiter.add_or_update_limiter(db, domain_url, connections)
return domain
def test_can_get_violations_per_domain(self):
domain = DomainFactory.create()
domain2 = DomainFactory.create()
DomainFactory.create()
page = PageFactory.create(domain=domain)
page2 = PageFactory.create(domain=domain)
page3 = PageFactory.create(domain=domain2)
page4 = PageFactory.create(domain=domain2)
page5 = PageFactory.create(domain=domain2)
ReviewFactory.create(domain=domain,
page=page,
number_of_violations=40,
is_active=True,
is_complete=True)
ReviewFactory.create(domain=domain,
page=page2,
number_of_violations=20,
is_active=True,
is_complete=True)
ReviewFactory.create(domain=domain2,
page=page3,
number_of_violations=15,
is_active=True,
is_complete=True)
ReviewFactory.create(domain=domain2,
page=page4,
number_of_violations=25,
is_active=True,
is_complete=True)
ReviewFactory.create(domain=domain2,
page=page5,
number_of_violations=50,
is_active=True,
is_complete=True)
violations_per_domain = Domain.get_violations_per_domain(self.db)
expect(violations_per_domain).to_be_like({
domain.id: 60,
domain2.id: 90
})
def get(self, domain_name):
domain = Domain.get_domain_by_name(domain_name, self.db)
if not domain:
self.set_status(404, 'Domain %s not found' % domain_name)
return
page_count = yield self.cache.get_page_count(domain)
review_count = yield self.cache.get_active_review_count(domain)
violation_count = yield self.cache.get_violation_count(domain)
bad_request_count = yield self.cache.get_bad_request_count(domain)
good_request_count = yield self.cache.get_good_request_count(domain)
total_request_count = good_request_count + bad_request_count
if total_request_count > 0:
error_percentage = round(float(bad_request_count) / total_request_count * 100, 2)
else:
error_percentage = 0
response_time_avg = yield self.cache.get_response_time_avg(domain)
status_code_info = Request.get_status_code_info(domain_name, self.db)
if page_count > 0:
review_percentage = round(float(review_count) / page_count * 100, 2)
else:
review_percentage = 0
domain_json = {
"id": domain.id,
"name": domain.name,
"url": domain.url,
"pageCount": page_count,
"reviewCount": review_count,
"violationCount": violation_count,
"reviewPercentage": review_percentage,
"is_active": domain.is_active,
"statusCodeInfo": status_code_info,
"errorPercentage": error_percentage,
"averageResponseTime": response_time_avg,
}
self.write_json(domain_json)
def get(self, key_name):
current_page = int(self.get_argument("current_page", 1))
page_size = int(self.get_argument("page_size", 10))
domain_filter = self.get_argument("domain_filter", None)
page_filter = self.get_argument("page_filter", None)
domain = None
if domain_filter is not None:
domain = Domain.get_domain_by_name(domain_filter, self.db)
if not domain:
self.set_status(404, self._("Domain %s not found") % domain_filter)
self.finish()
return
violations = self.application.violation_definitions
if key_name not in violations:
self.set_status(404, self._("Invalid violation key %s") % key_name)
self.finish()
return
violation_title = violations[key_name]["title"]
key_id = violations[key_name]["key"].id
violation = yield self.application.search_provider.get_by_violation_key_name(
key_id=key_id, current_page=current_page, page_size=page_size, domain=domain, page_filter=page_filter
)
if "error" in violation:
self.set_status(violation["error"]["status_code"], violation["error"]["reason"])
self.finish()
return
if "reviewsCount" not in violation:
if not domain and not page_filter:
violation["reviewsCount"] = Review.count_by_violation_key_name(self.db, key_id)
else:
violation["reviewsCount"] = None
violation["title"] = violation_title
self.write_json(violation)
self.finish()
def get_next_job_list(cls, db, expiration, current_page=1, page_size=200):
from holmes.models import Domain
lower_bound = (current_page - 1) * page_size
upper_bound = lower_bound + page_size
active_domains = Domain.get_active_domains(db)
active_domains_ids = [item.id for item in active_domains]
pages_query = db \
.query(
Page.uuid,
Page.url,
Page.score,
Page.last_review_date
) \
.filter(Page.domain_id.in_(active_domains_ids)) \
.order_by(Page.score.desc())
return pages_query[lower_bound:upper_bound]
def post(self, domain_name):
if not self.validate_superuser():
return
domain = Domain.get_domain_by_name(domain_name, self.db)
if not domain:
self.set_status(404, self._('Domain %s not found') % domain_name)
return
post_data = loads(self.request.body)
DomainsViolationsPrefs.update_by_domain(
self.db, self.cache, domain, post_data
)
self.write_json({
'reason': self._('Preferences successfully saved!'),
'description': self._('Preferences successfully saved!')
})
def get(self, domain_name, key_category_id):
domain = Domain.get_domain_by_name(domain_name, self.db)
if not domain:
self.set_status(404, self._('Domain %s not found') % domain_name)
return
key_category = KeysCategory.get_by_id(self.db, key_category_id)
if not key_category:
self.set_status(
404,
self._('Key category %s not found') % key_category_id)
return
violation_defs = self.application.violation_definitions
top_violations = self.girl.get(
'top_violations_in_category_for_domains') or {}
violations = []
for top_violation in top_violations.get(domain_name,
{}).get(key_category.id, []):
violations.append({
'title':
self._(violation_defs[top_violation['key_name']]['title']),
'count':
top_violation['count'],
'key':
top_violation['key_name'],
})
result = {
"domainId": domain.id,
'domainName': domain.name,
'domainURL': domain.url,
'categoryId': key_category_id,
'violations': violations
}
self.write_json(result)
def insert_default_violations_values_for_all_domains(
cls, db, default_violations_values, violation_definitions, cache):
from holmes.models import Domain
domains_violations_prefs = DomainsViolationsPrefs.get_domains_violations_prefs(
db)
domains = Domain.get_all_domains(db)
for domain in domains:
domain_data = domains_violations_prefs.get(domain.name, None)
if domain_data:
keys = set(default_violations_values.keys()) - set(
domain_data.keys())
else:
keys = default_violations_values.keys()
DomainsViolationsPrefs.insert_default_violations_values_for_domain(
db, domain, keys, violation_definitions, cache)
def get_count(self, key, domain_name, expiration, get_count_method):
cache_key = '%s-%s' % (self.get_domain_name(domain_name), key)
count = self.redis.get(cache_key)
if count is not None:
return int(count)
domain = domain_name
if domain and not isinstance(domain, Domain):
domain = Domain.get_domain_by_name(domain_name, self.db)
if domain is None:
count = Page.get_page_count(self.db)
else:
count = get_count_method(domain)
cache_key = '%s-%s' % (self.get_domain_name(domain), key)
self.redis.setex(cache_key, expiration, value=int(count))
return int(count)
def increment_count(self,
key,
domain_name,
get_default_method,
increment=1):
key = '%s-%s' % (self.get_domain_name(domain_name), key)
has_key = self.has_key(key)
domain = domain_name
if domain and not isinstance(domain, Domain):
domain = Domain.get_domain_by_name(domain_name, self.db)
if has_key:
self.redis.incrby(key, increment)
else:
if domain is None:
value = Page.get_page_count(self.db) + increment - 1
else:
value = get_default_method(domain) + increment - 1
self.redis.set(key, value)
def get(self, domain_name):
term = self.get_argument('term', None)
current_page = int(self.get_argument('current_page', 1))
page_size = int(self.get_argument('page_size', 10))
domain = Domain.get_domain_by_name(domain_name, self.db)
if not domain:
self.set_status(404, 'Domain %s not found' % domain_name)
return
reviews = domain.get_active_reviews(
self.db,
url_starts_with=term,
current_page=current_page,
page_size=page_size
)
if term:
review_count = len(reviews)
else:
review_count = yield self.cache.get_active_review_count(domain)
result = {
'reviewCount': review_count,
'pages': [],
}
for page in reviews:
result['pages'].append({
"url": page.url,
"uuid": str(page.uuid),
"violationCount": page.violations_count,
"completedAt": page.last_review_date,
"reviewId": str(page.last_review_uuid)
})
self.write_json(result)
def test_can_get_violations_per_domain(self):
domain = DomainFactory.create()
domain2 = DomainFactory.create()
DomainFactory.create()
page = PageFactory.create(domain=domain)
page2 = PageFactory.create(domain=domain)
page3 = PageFactory.create(domain=domain2)
page4 = PageFactory.create(domain=domain2)
page5 = PageFactory.create(domain=domain2)
ReviewFactory.create(domain=domain, page=page, number_of_violations=40, is_active=True, is_complete=True)
ReviewFactory.create(domain=domain, page=page2, number_of_violations=20, is_active=True, is_complete=True)
ReviewFactory.create(domain=domain2, page=page3, number_of_violations=15, is_active=True, is_complete=True)
ReviewFactory.create(domain=domain2, page=page4, number_of_violations=25, is_active=True, is_complete=True)
ReviewFactory.create(domain=domain2, page=page5, number_of_violations=50, is_active=True, is_complete=True)
violations_per_domain = Domain.get_violations_per_domain(self.db)
expect(violations_per_domain).to_be_like({
domain.id: 60,
domain2.id: 90
})
def get(self, domain_name):
domain = Domain.get_domain_by_name(domain_name, self.db)
if not domain:
self.set_status(404, self._('Domain %s not found') % domain_name)
return
violation_defs = self.application.violation_definitions
grouped_violations = self.girl.get(
'violation_count_by_category_for_domains') or {}
total = 0
violations = []
for item in grouped_violations.get(domain.id, []):
key_name, key_category_id, count = item['key_name'], item[
'category_id'], item['violation_count']
violations.append({
'categoryId':
key_category_id,
'categoryName':
self._(violation_defs[key_name]['category']),
'count':
count
})
total += count
result = {
"domainId": domain.id,
'domainName': domain.name,
'domainURL': domain.url,
'total': total,
'violations': violations
}
self.write_json(result)
def insert_default_violations_values_for_all_domains(
cls, db, default_violations_values, violation_definitions, cache):
from holmes.models import Domain
domains_violations_prefs = DomainsViolationsPrefs.get_domains_violations_prefs(db)
domains = Domain.get_all_domains(db)
for domain in domains:
domain_data = domains_violations_prefs.get(domain.name, None)
if domain_data:
keys = set(default_violations_values.keys()) - set(domain_data.keys())
else:
keys = default_violations_values.keys()
DomainsViolationsPrefs.insert_default_violations_values_for_domain(
db,
domain,
keys,
violation_definitions,
cache
)
def _verify_workers_limits(self, url, avg_links_per_page=10):
active_domains = Domain.get_active_domains(self.db)
return LimiterModel.has_limit_to_work(self.db, active_domains, url, avg_links_per_page)
def load_all_domains_violations_prefs(self):
from holmes.models import Domain
for domain in Domain.get_all_domains(self.db):
self.cache.get_domain_violations_prefs(domain.name)
def load_all_domains_violations_prefs(self):
from holmes.models import Domain
for domain in Domain.get_all_domains(self.db):
self.cache.get_domain_violations_prefs(domain.name)
def get_domain(self, domain_name):
domain = domain_name
if domain and not isinstance(domain, Domain):
domain = Domain.get_domain_by_name(domain_name, self.db)
return domain
def test_can_get_domain_by_name(self):
domain = DomainFactory.create()
loaded_domain = Domain.get_domain_by_name(domain.name, self.db)
expect(loaded_domain.id).to_equal(domain.id)
def get_domain(self, domain_name):
domain = domain_name
if domain and not isinstance(domain, Domain):
domain = Domain.get_domain_by_name(domain_name, self.db)
return domain
def test_invalid_domain_returns_None(self):
domain_name = 'domain-details.com'
domain = Domain.get_domain_by_name(domain_name, self.db)
expect(domain).to_be_null()
def get_next_job(cls, db, expiration, cache, lock_expiration, avg_links_per_page=10):
from holmes.models import Settings, Worker, Domain, Limiter # Avoid circular dependency
page = None
lock = None
settings = Settings.instance(db)
workers = db.query(Worker).all()
number_of_workers = len(workers)
active_domains = Domain.get_active_domains(db)
active_domains_ids = [item.id for item in active_domains]
all_domains_pages_in_need_of_review = {}
for domain_id in active_domains_ids:
pages = db \
.query(
Page.uuid,
Page.url,
Page.score,
Page.last_review_date
) \
.filter(Page.domain_id == domain_id) \
.order_by(Page.score.desc())[:number_of_workers]
if pages:
all_domains_pages_in_need_of_review[domain_id] = pages
pages_in_need_of_review = []
current_domain = 0
while all_domains_pages_in_need_of_review:
domains = all_domains_pages_in_need_of_review.keys()
if current_domain >= len(domains):
current_domain = 0
domain_id = domains[current_domain]
item = all_domains_pages_in_need_of_review[domain_id].pop(0)
pages_in_need_of_review.append(item)
if not all_domains_pages_in_need_of_review[domain_id]:
del all_domains_pages_in_need_of_review[domain_id]
current_domain += 1
if not pages_in_need_of_review:
return None
if settings.lambda_score > 0 and settings.lambda_score > pages_in_need_of_review[0].score:
cls.update_pages_score_by(settings, settings.lambda_score, db)
for i in range(len(pages_in_need_of_review)):
if not Limiter.has_limit_to_work(db, active_domains, pages_in_need_of_review[i].url, avg_links_per_page):
continue
lock = cache.has_next_job_lock(
pages_in_need_of_review[i].url,
lock_expiration
)
if lock is not None:
page = pages_in_need_of_review[i]
break
if page is None:
return None
return {
'page': str(page.uuid),
'url': page.url,
'score': page.score,
'lock': lock
}
def fill_job_bucket(self,
expiration,
look_ahead_pages=1000,
avg_links_per_page=10.0):
try:
with Lock('next-job-fill-bucket-lock', redis=self.redis):
logging.info('Refilling job bucket. Lock acquired...')
expired_time = datetime.utcnow() - timedelta(
seconds=expiration)
active_domains = Domain.get_active_domains(self.db)
if not active_domains:
return
active_domains_ids = [item.id for item in active_domains]
limiter_buckets = self.get_limiter_buckets(
active_domains, avg_links_per_page)
all_domains_pages_in_need_of_review = []
for domain_id in active_domains_ids:
pages = self.db \
.query(
Page.uuid,
Page.url,
Page.score,
Page.last_review_date
) \
.filter(Page.domain_id == domain_id) \
.filter(or_(
Page.last_review_date == None,
Page.last_review_date <= expired_time
))[:look_ahead_pages]
if pages:
all_domains_pages_in_need_of_review.append(pages)
logging.debug(
'Total of %d pages found to add to redis.' % (sum([
len(item)
for item in all_domains_pages_in_need_of_review
])))
item_count = int(self.redis.zcard('next-job-bucket'))
current_domain = 0
while item_count < look_ahead_pages and len(
all_domains_pages_in_need_of_review) > 0:
if current_domain >= len(
all_domains_pages_in_need_of_review):
current_domain = 0
item = all_domains_pages_in_need_of_review[
current_domain].pop(0)
has_limit = True
logging.debug('Available Limit Buckets: %s' %
limiter_buckets)
for index, (limit,
available) in enumerate(limiter_buckets):
if limit.matches(item.url):
if available <= 0:
has_limit = False
break
limiter_buckets[index] = (limit, available - 1)
if has_limit:
self.add_next_job_bucket(item.uuid, item.url)
item_count += 1
# if there are not any more pages in this domain remove it from dictionary
if not all_domains_pages_in_need_of_review[current_domain]:
del all_domains_pages_in_need_of_review[current_domain]
current_domain += 1
logging.debug('ADDED A TOTAL of %d ITEMS TO REDIS...' %
item_count)
except LockTimeout:
logging.info("Can't acquire lock. Moving on...")
def test_invalid_domain_returns_None(self):
domain_name = 'domain-details.com'
domain = Domain.get_domain_by_name(domain_name, self.db)
expect(domain).to_be_null()
def test_can_get_domain_by_name(self):
domain = DomainFactory.create()
loaded_domain = Domain.get_domain_by_name(domain.name, self.db)
expect(loaded_domain.id).to_equal(domain.id)