def test_generate_category_pistol_returns_correct_category():
url = "https://www.2alfa.no/ammunisjon/ladet-ammo/pistol.hml"
result = Category.extract(url)
assert result != None
assert result == Category.HANDGUN
def test_generate_category_rifle():
url = "/produkter/ammunisjon/rifle"
result = Category.extract(url)
assert result != None
assert result == Category.RIFLE
def test_generate_category_shotgun():
url = "/produkter/ammunisjon/haandvaapen"
result = Category.extract(url)
assert result != None
assert result == Category.HANDGUN
def test_generate_category_shotgun():
url = "/produkter/ammunisjon/hagle"
result = Category.extract(url)
assert result != None
assert result == Category.SHOTGUN
def fetch(self):
elements = []
for url in self.urls:
compounded_url = f"{self.root_url}{url}"
# print(f"Fetching data @ {compounded_url}")
response = requests.get(compounded_url)
# 2Alpha has a bit more cryptioc site setup
soup = BeautifulSoup(response.content, "html.parser")
data = soup.find_all("div", class_="ut2-gl__body")
for container in data:
# Extract image info
image_div = container.find("div", class_="ut2-gl__image")
image_url = image_div.find("img", class_="ty-pict")["src"]
# Get the first anchor which has the details link
details_url = image_div.find("a")["href"]
# Extract name
name_div = container.find("div", class_="ut2-gl__name")
name = name_div.find("a")["title"]
# Extract price
price_div = container.find("div", class_="ut2-gl__price")
price_str = price_div.find("span", class_="ty-price-num").text
product = Product(cat=Category.extract(compounded_url),
img_url=image_url,
price=price_str,
name=name,
details_url=details_url)
elements.append(product)
return elements
def fetch(self):
elements = []
for url in urls:
compounded_url = f"{root_url}{url}"
print(f"Fetching data @ {compounded_url}")
response = requests.get(compounded_url)
soup = BeautifulSoup(response.content, "html.parser")
data = soup.find_all("div", class_="ProdItem")
for container in data:
# print(container)
image_link = container.find("img")["src"]
# print(image_link)
name = container.find("a", class_="ItemTitleLink")["title"]
# print(title)
details_url = container.find("a",
class_="ItemTitleLink")["href"]
# print(title_link)
price_string = container.find("span", class_="Price").text
# print(price_string)
product = Product(cat=Category.extract(compounded_url),
img_url=f"{root_url}{image_link}",
price=price_string,
name=name,
details_url=details_url)
elements.append(product)
return elements
def add_sentence(self, sentence, categoryName):
sentence = self.preprocess_sentence(sentence)
category = self._categories.get(categoryName)
if category is None:
category = Category(categoryName)
for word in get_words(sentence):
category.add_words(word)
if word not in self._words:
self._words.append(word)
self._categories[categoryName] = category
category.setValueN(len(get_words(sentence)))
category.setNumberOfSentence(1)
else:
for word in get_words(sentence):
category.add_words(word)
if word not in self._words:
self._words.append(word)
category.setValueN(len(get_words(sentence)) + category.getValueN())
category.setNumberOfSentence(category.getNumberOfSentence()+1)