Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -26,50 +26,13 @@ def visualize(url):
|
|
| 26 |
# Fetch and display the website content
|
| 27 |
with st.spinner("loading website data ..."):
|
| 28 |
# innerHTML = get_innerHTML(url)
|
| 29 |
-
html_content,
|
| 30 |
#st.subheader("Website title:")
|
| 31 |
-
if
|
| 32 |
-
st.code(
|
| 33 |
-
|
| 34 |
-
for row in baris:
|
| 35 |
-
rank = row.select_one(".tableCellRank")
|
| 36 |
-
team = row.select_one(".tableCellParticipant__name")
|
| 37 |
-
mp = row.select_one("span.table__cell:nth-of-type(3)")
|
| 38 |
-
w = row.select_one("span.table__cell:nth-of-type(4)")
|
| 39 |
-
d = row.select_one("span.table__cell:nth-of-type(5)")
|
| 40 |
-
l = row.select_one("span.table__cell:nth-of-type(6)")
|
| 41 |
-
g = row.select_one(".table__cell--score")
|
| 42 |
-
sg = row.select_one(".table__cell--goalsForAgainstDiff")
|
| 43 |
-
pts = row.select_one(".table__cell--points")
|
| 44 |
-
|
| 45 |
-
data.append({
|
| 46 |
-
"Peringkat": rank.text.strip() if rank else "",
|
| 47 |
-
"Tim": team.text.strip() if team else "",
|
| 48 |
-
"Main": mp.text.strip() if mp else "",
|
| 49 |
-
"Menang": w.text.strip() if w else "",
|
| 50 |
-
"Seri": d.text.strip() if d else "",
|
| 51 |
-
"Kalah": l.text.strip() if l else "",
|
| 52 |
-
"Gol": g.text.strip() if g else "",
|
| 53 |
-
"Selisih Gol": sg.text.strip() if sg else "",
|
| 54 |
-
"Poin": pts.text.strip() if pts else ""
|
| 55 |
-
})
|
| 56 |
-
|
| 57 |
-
# === 5. Buat DataFrame ===
|
| 58 |
-
df = pd.DataFrame(data)
|
| 59 |
-
st.dataframe(df)
|
| 60 |
else:
|
| 61 |
-
st.warning("
|
| 62 |
-
if xdataframe:
|
| 63 |
-
#st.info(html_content)
|
| 64 |
-
|
| 65 |
-
# Konversi tabel HTML ke DataFrame Pandas
|
| 66 |
-
#df = pd.read_html(str(parse_element))[0]
|
| 67 |
-
#st.dataframe(df) # tampilkan tabel dengan format rapi Streamlit
|
| 68 |
-
st.components.v1.html(str(xdataframe), height=600, scrolling=True)
|
| 69 |
-
# === 6. Tampilkan di Streamlit ===
|
| 70 |
-
#st.dataframe(xdataframe)
|
| 71 |
-
else:
|
| 72 |
-
st.warning("Tabel tidak ditemukan.")
|
| 73 |
|
| 74 |
except Exception as e:
|
| 75 |
st.error(f"Error: {e}")
|
|
@@ -84,15 +47,30 @@ def take_webdata(url):
|
|
| 84 |
wd = webdriver.Chrome(options=options)
|
| 85 |
wd.set_window_size(1080, 720) # Adjust the window size here
|
| 86 |
wd.get(url)
|
| 87 |
-
wd.implicitly_wait(
|
| 88 |
# Get the page title
|
| 89 |
page_title = wd.title
|
| 90 |
#screenshot = wd.get_screenshot_as_png()
|
| 91 |
-
WebDriverWait(wd, 20).until(EC.presence_of_element_located((By.ID, "tournament-table")))
|
| 92 |
html = wd.execute_script("return document.documentElement.outerHTML;")
|
| 93 |
soup = BeautifulSoup(html, "html.parser")
|
| 94 |
-
|
| 95 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 96 |
|
| 97 |
#rows = div_find.find("ui-table__row ") if div_find else None
|
| 98 |
#rows = soup.find("div", class_="ui-table__row ")
|
|
@@ -104,7 +82,7 @@ def take_webdata(url):
|
|
| 104 |
if wd:
|
| 105 |
wd.quit()
|
| 106 |
|
| 107 |
-
return html ,
|
| 108 |
|
| 109 |
if __name__ == "__main__":
|
| 110 |
main()
|
|
|
|
| 26 |
# Fetch and display the website content
|
| 27 |
with st.spinner("loading website data ..."):
|
| 28 |
# innerHTML = get_innerHTML(url)
|
| 29 |
+
html_content, xtarget_dropdown, xurl = take_webdata(url)
|
| 30 |
#st.subheader("Website title:")
|
| 31 |
+
if xtarget_dropdown:
|
| 32 |
+
st.code(xtarget_dropdown, language='html')
|
| 33 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 34 |
else:
|
| 35 |
+
st.warning("tidak ditemukan.")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 36 |
|
| 37 |
except Exception as e:
|
| 38 |
st.error(f"Error: {e}")
|
|
|
|
| 47 |
wd = webdriver.Chrome(options=options)
|
| 48 |
wd.set_window_size(1080, 720) # Adjust the window size here
|
| 49 |
wd.get(url)
|
| 50 |
+
wd.implicitly_wait(15)
|
| 51 |
# Get the page title
|
| 52 |
page_title = wd.title
|
| 53 |
#screenshot = wd.get_screenshot_as_png()
|
| 54 |
+
#WebDriverWait(wd, 20).until(EC.presence_of_element_located((By.ID, "tournament-table")))
|
| 55 |
html = wd.execute_script("return document.documentElement.outerHTML;")
|
| 56 |
soup = BeautifulSoup(html, "html.parser")
|
| 57 |
+
|
| 58 |
+
target_dropdown = soup.find('div', class_='dropdown-menu', attrs={'aria-labelledby': 'navbar-match'})
|
| 59 |
+
|
| 60 |
+
if target_dropdown:
|
| 61 |
+
links = target_dropdown.find_all('a', class_='dropdown-item')
|
| 62 |
+
|
| 63 |
+
for link in links:
|
| 64 |
+
urlx = link.get('href')
|
| 65 |
+
text = link.get_text(strip=True)
|
| 66 |
+
#print(f"Text: {text}")
|
| 67 |
+
#print(f"URL: {url}")
|
| 68 |
+
#print("---")
|
| 69 |
+
else:
|
| 70 |
+
print("Dropdown menu tidak ditemukan")
|
| 71 |
+
|
| 72 |
+
#div_find = soup.find("div", id="tournament-table", class_="tournament-table-standings")
|
| 73 |
+
#rows = div_find.select("div[class*=ui-table__row]")
|
| 74 |
|
| 75 |
#rows = div_find.find("ui-table__row ") if div_find else None
|
| 76 |
#rows = soup.find("div", class_="ui-table__row ")
|
|
|
|
| 82 |
if wd:
|
| 83 |
wd.quit()
|
| 84 |
|
| 85 |
+
return html ,target_dropdown, urlx
|
| 86 |
|
| 87 |
if __name__ == "__main__":
|
| 88 |
main()
|