main.py - master - harrisson55/pythonprojectnba

main.py
87 строк3 KB
notebook
Изменение колонок.
20 июл 2024, 15:43
20 июл 2024, 15:438ecd1ad
import requests
import pymssql
from datetime import  datetime
from bs4 import BeautifulSoup

SERVER = "80.76.42.208:1433"
USER = "user3"
PASSWORD = "Rom@wka55"
DATABASE = "NbaDashoboard"

connection = pymssql.connect(server=SERVER, user=USER,
                                     password=PASSWORD, database=DATABASE)
pymssql.connect()

cursor = connection.cursor()
cursor.execute("SELECT * FROM dbo.TestTable")

name_team = input("Введите название команды: ")
previously_extracted_links_2 = ['https://www.sports.ru/basketball/club/'+name_team+'/calendar/']
links_massive1 = []
links_massive2 = []
links_massive3 = []
links_massive4 = []
symbols_to_remove = ["/n","Дата","'dot',","\xa0"]
today = datetime.today()
print(today)


for url in previously_extracted_links_2:
    response = requests.get(url)
    if response.status_code == 200:
        soup = BeautifulSoup(response.content, 'html.parser')
        symbols_to_remove = ["\n"]
        links = soup.find_all('td', 'name-td alLeft' in 'class')
        links_massive = []
        for link in links:
            if ':' in link.get_text() and '|' not in link.get_text():
                links_massive1.append(link.get_text())
        print(links_massive1)


for url in previously_extracted_links_2:
    response = requests.get(url)
    if response.status_code == 200:
        soup1 = BeautifulSoup(response.content, 'html.parser')
        symbols_to_remove2 = "Дата"
        links2 = soup1.find_all('td', class_='name-td alLeft bordR')
        links_massive2 = []
        for link2 in links2:
            if symbols_to_remove2 not in link2.get_text():
                links_massive2.append(link2.get_text())
        print(links_massive2)

for url in previously_extracted_links_2:
    response = requests.get(url)
    if response.status_code == 200:
        soup3 = BeautifulSoup(response.content, 'html.parser')
        name_tds = soup3.find_all('td', class_='name-td')
        teams = [td.find('a').get_text(strip=True) for td in name_tds if td.find('a') ]
        evert_thrid_team = teams[2::3]
        for team in evert_thrid_team:
            links_massive3.append(team)
        print(links_massive3)


for url in previously_extracted_links_2:
    response = requests.get(url)
    if response.status_code == 200:
        soup = BeautifulSoup(response.content, 'html.parser')
        symbols_to_remove4 = "\xa0"
        links = soup.find_all('td', class_= ['alRight padR20'] )
        for link4 in links:
            if symbols_to_remove4 not in link4.get_text():
                links_massive4.append(link4.get_text())
        print(links_massive4)



for i in range(len(links_massive1)):
    links_massive1[i] = "".join([char for char in links_massive1[i] if char not in symbols_to_remove])
    links_massive2[i] = "".join([char for char in links_massive2[i] if char not in symbols_to_remove])
    links_massive4[i] = "".join([char for char in links_massive4[i] if char not in symbols_to_remove])
    sql_query = f"INSERT INTO dbo.TestTable (calendar_date,enemy,score,team,upload_date,home) VALUES ('{links_massive2[i]}',N'{links_massive3[i]}','{links_massive1[i]}','{name_team}','{today}',N'{links_massive4[i]}')"
    cursor.execute(sql_query)
    connection.commit()

connection.close()