def is_the_link_already_searched_for(starting_url):
with session_scope() as session:
list_of_tuple_urls = session.query(Url).with_entities(
Url.url_string).all()
urls = list(map(lambda url: url[0], list_of_tuple_urls))
if starting_url in urls:
return urls[-1]
return starting_url
def create_plot(choose_option, time_chosen):
with session_scope() as session:
domain_names_raw = session.query(Domain) \
.filter(Domain.visited_at >= time_chosen) \
.with_entities(Domain.domain_name) \
.all()
domain_names = list(map(lambda x: x[0], domain_names_raw))
unique_domain_names = list(set(domain_names))
count_of_domains = ['Unique domains']
plt.rcdefaults()
fig, ax = plt.subplots()
y_pos = [len(unique_domain_names)]
width = [len(unique_domain_names)]
error = [len(unique_domain_names)]
ax.barh(y_pos, width, xerr=error, align='center')
ax.set_yticks(y_pos)
ax.set_yticklabels(count_of_domains)
ax.invert_yaxis() # labels read top-to-bottom
ax.set_xlabel('User statistics')
ax.set_title(f'Domains found in the last {choose_option}')
plt.show()
def is_url_visited(url_string):
with session_scope() as session:
does_url_exist = bool(
session.query(Url).filter(Url.url_string == url_string).first())
return does_url_exist
def get_labels():
with session_scope() as session:
labels_raw = session.query(Server).group_by(
Server.server_name).with_entities(Server.server_name).all()
labels = list(map(lambda x: x[0], labels_raw))
return labels
def add_entry_to_db(entry):
with session_scope() as session:
session.add(entry)