scossa/scrapeje
1
1
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Update 'ScrapeJE.py'

Browse Source
This commit is contained in:
scossa
2022-10-01 03:09:17 +02:00
parent 96b7a02aa4
commit 59850d9868
1 changed files with 13 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
ScrapeJE.py
View File

@@ -8,11 +8,13 @@ npezzi=[]
prezzo=[] prezzo=[]
n=0 n=0
linkJE = input('Incolla il link del ristorante: ')
linkJE
#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(linkJE).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()
@@ -22,7 +24,7 @@ menu = soup.find(attrs={"data-test-id": "menu-item"})
#Stora nome ristorante #Stora nome ristorante
nrist=soup.title.text[8:24] nrist=soup.title.text[8:-32]
menu.find(attrs={"allergenPhoneNumber": "menu-item-name"}) menu.find(attrs={"allergenPhoneNumber": "menu-item-name"})
# alla riga 870 dell html c'è "allergenPhoneNumber":"3389529446" # alla riga 870 dell html c'è "allergenPhoneNumber":"3389529446"
@@ -69,4 +71,9 @@ print(nrist)
print("lista nome",len(nome),) print("lista nome",len(nome),)
print("lista desc:",len(desc)) print("lista desc:",len(desc))
print("lista npezzi:",len(npezzi)) print("lista npezzi:",len(npezzi))
print("lista prezzi:",len(prezzo)) print("lista prezzi:",len(prezzo)) #da trattare come stringa perchè ci sono anche i es. "da 1,00 €"
#Creare un file json formattato per jawanndenn
#automatizzare l inserimento in jawandenn
#jawanndenn include la funzione per argomento
#"jawanndenn --loaddata FILE.json" --> Load a JSON export of the database from FILE.json, then quit.