def getOriginFromWikidata(name: str):
toReturn = Optional.empty()
placeOfBirth = None
page = wptools.page(name)
try:
page.get_wikidata()
except:
print("Couldn't get wikidata from name.")
try:
placeOfBirth = page.data['wikidata']['place of birth (P19)']
except:
None
try:
placeOfBirth = page.data['wikidata']['location of formation (P740)']
except:
None
try:
index = placeOfBirth.index("(")
placeOfBirth = placeOfBirth[0:index]
except:
None
if placeOfBirth != None:
toReturn = Optional.of(placeOfBirth)
return toReturn
def test_get_or_raise_on_an_empty_optional_throws_wrapped_exception(self):
optional = Optional.empty()
class RandomDomainException(Exception):
pass
with pytest.raises(RandomDomainException):
optional.get_or_raise(RandomDomainException())
def test_will_raise_on_or_else_raise_from_if_present_when_not_present(
self):
class TestException(Exception):
pass
optional = Optional.empty()
with pytest.raises(TestException):
optional.if_present(lambda x: x).or_else_raise(
TestException("Something"))
def test_will_not_run_consumer_if_not_present(self):
scope = {'seen': False}
def some_thing_consumer(thing):
scope['seen'] = True
optional = Optional.empty()
optional.if_present(some_thing_consumer)
assert not scope['seen']
def test_will_run_or_else_from_if_present_when_not_present(self):
scope = {
'if_seen': False,
'else_seen': False,
}
def some_thing_consumer(thing):
scope['if_seen'] = True
def or_else_procedure():
scope['else_seen'] = True
optional = Optional.empty()
optional.if_present(some_thing_consumer).or_else(or_else_procedure)
assert not scope['if_seen']
assert scope['else_seen']
def getArtistOriginFromScraping(name: str):
result = Optional.empty()
if result.is_empty():
result = getOriginFromMusicbrainz(name)
if result.is_empty():
result = getOriginFromWikipedia(name + " Musician")
if result.is_empty():
result = getOriginFromWikipedia(name + " Band")
if result.is_empty():
result = getOriginFromWikipedia(name)
if result.is_empty():
result = getOriginFromWikidata(name)
return result
def test_empty_optionals_are_falsy(self):
assert not Optional.empty()
def test_get_or_default_on_an_empty_optional_returns_default_value(self):
optional = Optional.empty()
assert optional.get_or_default("pants") == "pants"
def test_can_instantiate_an_empty_optional_via_the_zero_arity_of(self):
assert Optional.of() == Optional.empty()
def test_can_eval_the_representation_of_an_empty_optional(self):
optional = Optional.empty()
assert eval(repr(optional)) == optional
def test_flat_map_returns_empty_if_value_is_empty(self):
def does_stuff(thing):
return Optional.of("PANTS")
optional = Optional.empty()
assert optional.flat_map(does_stuff).is_empty()
def does_nothing(thing):
return Optional.empty()
def test_instantiate_empty(self):
optional = Optional.empty()
assert optional.is_empty()
def test_will_return_optional_of_return_val_when_not_present(self):
def or_else_supplier():
return "pants"
optional = Optional.empty()
assert optional.or_else(or_else_supplier) == Optional.of("pants")
def getOriginFromMusicbrainz(name: str):
toReturn = Optional.empty()
wiki_location = Optional.empty()
location = ""
sort_name = ""
url = "https://musicbrainz.org/search?query=" + urllib.parse.quote_plus(
name) + "&type=artist&method=indexed"
page = requests.get(url)
print(page)
bs = BeautifulSoup(page.content, features="lxml")
table = bs.find_all("table")
try:
titles = [title.text.strip() for title in table[0].find_all("th")]
except:
return toReturn
artist_data = []
for data in table[0].find_all("td"):
if len(artist_data) < len(titles):
artist_data.append(data.text.strip())
else:
break
found_category = 0
for key, value in zip(titles, artist_data):
if (key == "Area" or key == "Begin Area"):
if value != "":
found_category += 1
if key == "Area":
location += value
if key == "Begin Area":
if value != "":
location = value + ", " + location
if key == "Sort Name":
sort_name = value
# attempt to verify this is the right artist using sort name, if not stop the function
name_list = name.lower().split()
sort_name = sort_name.lower()
dont_match_score = 0
for word in name_list:
try:
try:
index = sort_name.index(word)
except:
# removes a potential 's from the end of the name
index = sort_name.index(word[0:len(word) - 2])
except:
dont_match_score += 1
dont_match_index = dont_match_score / len(name_list)
print("Dissimilarity rating is", dont_match_index)
if dont_match_index > 0.4:
return toReturn
# if we didnt find precise city, search wikipedia with location name
if found_category < 2:
try:
wiki_location = getOriginFromWikipedia(name + " " + location +
" Musician")
print(wiki_location)
except:
None
if wiki_location.is_present():
toReturn = wiki_location
elif location != "":
toReturn = Optional.of(location)
return toReturn
def test_cannot_get_from_empty_even_after_checking(self):
optional = Optional.empty()
assert optional.is_empty()
with pytest.raises(OptionalAccessOfEmptyException):
optional.get()
def test_empty_optionals_are_equal(self):
assert Optional.empty() == Optional.empty()
def test_empty_optional_not_equal_non_empty_optional(self):
assert Optional.empty() != Optional.of("thing")
def get_open_class(self, period: int, subject: Subject) -> Optional.of(Class):
newest_class = self.matrix[period][subject.id][-1]
if newest_class.is_open():
return Optional.of(newest_class)
return Optional.empty()
def getOriginFromWikipedia(name: str):
toReturn = Optional.empty()
try:
searches = wiki.search(name)
while len(searches) == 0:
if len(name) <= 2:
return Optional.empty()
name = ' '.join(name.split(' ')[:-1])
print(name)
searches = wiki.search(name)
artistPage = wiki.page(searches[0], auto_suggest=False).html()
except:
return Optional.empty()
bs = BeautifulSoup(artistPage, features="lxml")
tables = bs.find_all("table")
tab = None
for tab in tables:
if "infobox" in tab["class"]:
break
if tab != None:
node_list = [node for node in tab.find_all('tr')]
zipped_key_value = []
for node in node_list:
try:
entry = (node.find_all('th')[0].text.strip(),
node.find_all('td')[0].text.strip())
zipped_key_value.append(entry)
except:
None
# search for 'origin' to return
for key, value in zipped_key_value:
if key == "Origin":
toReturn = Optional.of(value)
# edge case for some solo artists, find birthplace
if toReturn == Optional.empty():
for div in tab.find_all("div", {"class": "birthplace"}):
if "birthplace" in div["class"]:
toReturn = Optional.of(div.text.strip())
# worse case, birthplace not tagged
if toReturn == Optional.empty():
for key, value in zipped_key_value:
if key == "Born":
try:
index = value.index("age")
value = value[index + 7:]
toReturn = Optional.of(value)
except:
index = 0
if key == "Died":
try:
index = value.index("age")
value = value[index + 7:]
toReturn = Optional.of(value)
except:
index = 0
return toReturn
