Update 'BozzaScrapeJE.py'

This commit is contained in:
2022-09-29 19:07:26 +02:00
parent e2dd43c884
commit 876a648211

View File

@@ -4,62 +4,57 @@ import cloudscraper
nome=[] nome=[]
desc=[] desc=[]
desc2=[] npezzi=[]
prezzo=[] prezzo=[]
n=0 n=0
#scrape html #scrape html
scraper = cloudscraper.create_scraper(browser={'browser': 'firefox','platform': 'windows','mobile': False}) # scraper = cloudscraper.create_scraper(browser={'browser': 'firefox','platform': 'windows','mobile': False})
page = scraper.get("https://www.justeat.it/restaurants-saporedialeppo/menu").content # page = scraper.get("https://www.justeat.it/restaurants-saporedialeppo/menu").content
with open('aleppo.html', 'wb') as f: # with open('aleppo.html', 'wb') as f:
f.write(page) # f.write(page)
with open('aleppo.html', 'rb') as f: with open('aleppo.html', 'rb') as f:
page = f.read() page = f.read()
soup = BeautifulSoup(page, "html.parser") soup = BeautifulSoup(page, "html.parser")
menu = soup.find(attrs={"data-test-id": "menu-item-name"}) menu = soup.find(attrs={"data-test-id": "menu-item"})
#riempie la lista "nome" con i nomi dei prodotti
for menu in soup.find_all(attrs={"data-test-id": "menu-item-name"}):
nome.append(menu.text.splitlines()[2])
nome[n]=nome[n].lstrip()
n=n+1
n=0
#riempie la lista "desc" con le descrizioni dei prodotti for menu in soup.find_all(attrs={"data-test-id": "menu-item"}):
for menu in soup.find_all(attrs={"data-test-id": "menu-item-description"}): att=menu
if menu.text.find("pezza") > 0: #poichè c'è un piatto con lo sPEZZAtino
desc.append(menu.text.splitlines()[1]) #riempie la lista "nome"
desc[n]=desc[n].lstrip() for att in menu.find(attrs={"data-test-id": "menu-item-name"}):
desc2.append("") if att != " ":
n=n+1 nome.append(att.lstrip().splitlines()[0])
elif menu.text.find("pezz") > 0:
desc2.append(menu.text.splitlines()[1]) #riempie la lista "desc"
desc2[n]=desc2[n].lstrip() att=menu.find("p", class_="c-menuItems-description")
desc.append("") if att != None:
n=n+1 for att in menu.find("p", class_="c-menuItems-description"):
desc.append(att.lstrip().splitlines()[0])
else: else:
desc.append(menu.text.splitlines()[1]) desc.append(None)
desc[n]=desc[n].lstrip()
desc2.append("")
n=n+1
n=0
#riempie la lista "prezzo" con le descrizioni dei prodotti # #riempie la lista "npezzi"
for menu in soup.find_all(class_="c-menuItems-price notranslate"): # for att in menu.find_next(attrs={"data-test-id": "menu-item-description"}):
prezzo.append(menu.text.splitlines()[1]) # npezzi.append(att.lstrip().splitlines()[0])
prezzo[n]=prezzo[n].lstrip()
n=n+1
n=0
#riempie la lista "prezzo"
for att in menu.find("p", class_="c-menuItems-price notranslate"):
prezzo.append(att.lstrip().splitlines()[0])
#stampa liste
for x in range(len(nome)): for x in range(len(nome)):
print("\n")
print(nome[x]) print(nome[x])
print(desc[x]) print(desc[x])
print(desc2[x]) # print(npezzi[x])
print(prezzo[x],"\n") print(prezzo[x])
print(len(nome)) #stampa lunghezza liste
print(len(desc)) print("\nlista nomi:",len(nome))
print(len(desc2)) print("lista desc:",len(desc))
print(len(prezzo)) print("lista npezzi:",len(npezzi))
print("lista prezzi:",len(prezzo))