def player_extract_inforamtion(url): import re """ extract name of player and original club and hyper link """ headers={'user-agent':'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/66.0.3359.139 Safari/537.36'} browse=requests_html.HTMLSession()
page=browse.get(url,headers=headers)
values=page.html.find("#yw0 > table > tbody > tr")
players,clubs,links=[],[],[]
for i in values:
values=i.text.split("\n")
players.append(re.sub("[^A-Za-z]"," ",values[0]).strip())
clubs.append(values[1])
link_1=0
for index,j in enumerate(list(i.absolute_links)):
if re.search("spieler",j) != None:
link_1=index
links.append(list(i.absolute_links)[link_1])
return players,clubs,links
def player_club(name,club,players,clubs): res=0 for index,player in enumerate(players): if player==name and clubs[index]==club: res=index return res
def transfer_val(link): browse=requests_html.HTMLSession() page=browse.get(link)
value=page.html.find("div.tm-player-transfer-history-grid__fee")
return value[1].text
for index,row in df.iloc[200:400,:].iterrows(): url=f'https://www.transfermarkt.com/schnellsuche/ergebnis/schnellsuche?query={"+".join(row["name"].split(" "))}' players,clubs,links=extract_links.player_extract_inforamtion(url) res=extract_links.player_club(row["name"],row["new_club"],players,clubs) transfer=extract_links.transfer_val(links[res]) print(transfer) enter image description here
