first draft (proof of concept) for getting data from single row/company

This commit is contained in:
Daniil Fajnberg 2021-11-08 18:40:34 +01:00
parent 5274008ab0
commit 29b37c756d

View File

@ -0,0 +1,30 @@
import re
from requests import get
from bs4 import BeautifulSoup
from bs4.element import Tag
row_type = tuple[str, str, str, str, str]
def get_row_data(table_row: Tag) -> row_type:
tds = table_row.find_all('td')
company_name = str(tds[0].a.contents[0]).strip()
stock_symbol = str(tds[0].a.contents[1].contents[0]).strip()
stock_symbol_pattern = re.compile(r'\((\w+)\)')
stock_symbol = re.search(stock_symbol_pattern, stock_symbol).group(1)
country = str(tds[1].contents[0])
exchange = str(tds[2].contents[0])
sector = str(tds[3].contents[0])
return company_name, stock_symbol, country, exchange, sector
if __name__ == '__main__':
response = get('https://www.marketwatch.com/tools/markets/stocks/a-z/0-9')
soup = BeautifulSoup(response.text, 'html.parser')
trs = soup.find('div', {'id': 'marketsindex'}).table.tbody.find_all('tr')
print(get_row_data(trs[0]))
print(get_row_data(trs[1]))
print(get_row_data(trs[149]))
data: list[row_type] = []