sintamar commited on
Commit
fb70667
·
verified ·
1 Parent(s): ffd88e4

Update extract.py

Browse files
Files changed (1) hide show
  1. extract.py +27 -2
extract.py CHANGED
@@ -23,13 +23,38 @@ def take_webdata(url):
23
  soup = BeautifulSoup(html, "html.parser")
24
  #div_find = soup.find("div", id="tournament-table", class_="tournament-table-standings")
25
  #table_find = div_find.find("table") if div_find else None
26
- tournament_div = soup.find("div", id="tournament-table", class_="tournament-table-standings")
 
 
 
 
 
 
 
 
 
 
 
27
 
 
 
 
 
 
 
 
 
 
 
 
28
 
 
 
 
29
  except WebDriverException as e:
30
  return page_title
31
  finally:
32
  if wd:
33
  wd.quit()
34
 
35
- return html , tournament_div
 
23
  soup = BeautifulSoup(html, "html.parser")
24
  #div_find = soup.find("div", id="tournament-table", class_="tournament-table-standings")
25
  #table_find = div_find.find("table") if div_find else None
26
+ tournament_div = soup.find("div", class_="ui-table__row ")
27
+ data = []
28
+ for row in rows:
29
+ rank = row.select_one(".tableCellRank")
30
+ team = row.select_one(".tableCellParticipant__name")
31
+ mp = row.select_one("span.table__cell:nth-of-type(3)")
32
+ w = row.select_one("span.table__cell:nth-of-type(4)")
33
+ d = row.select_one("span.table__cell:nth-of-type(5)")
34
+ l = row.select_one("span.table__cell:nth-of-type(6)")
35
+ g = row.select_one(".table__cell--score")
36
+ sg = row.select_one(".table__cell--goalsForAgainstDiff")
37
+ pts = row.select_one(".table__cell--points")
38
 
39
+ data.append({
40
+ "Peringkat": rank.text.strip() if rank else "",
41
+ "Tim": team.text.strip() if team else "",
42
+ "Main": mp.text.strip() if mp else "",
43
+ "Menang": w.text.strip() if w else "",
44
+ "Seri": d.text.strip() if d else "",
45
+ "Kalah": l.text.strip() if l else "",
46
+ "Gol": g.text.strip() if g else "",
47
+ "Selisih Gol": sg.text.strip() if sg else "",
48
+ "Poin": pts.text.strip() if pts else ""
49
+ })
50
 
51
+ # === 5. Buat DataFrame ===
52
+ df = pd.DataFrame(data)
53
+
54
  except WebDriverException as e:
55
  return page_title
56
  finally:
57
  if wd:
58
  wd.quit()
59
 
60
+ return html , df