administrator/ordinaceprojekt
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

z230

Browse Source
This commit is contained in:
vladimir.buzalka
2026-04-24 09:51:45 +02:00
parent 7a7c35f778
commit 365fcd16ba
1 changed files with 45 additions and 29 deletions
Download Patch File Download Diff File Expand all files Collapse all files
SběrDatRůzné/DailyStr8ts/stahni_str8ts.py
+45 -29
View File
@@ -47,46 +47,62 @@ async def main():
context = await browser.new_context( context = await browser.new_context(
viewport={"width": 1280, "height": 900}, viewport={"width": 1280, "height": 900},
) )
page = await context.new_page()
# Krok 1: načti hlavní stránku a zjisti URL iframu s hrou
page = await context.new_page()
print(f"Načítám {URL} ...") print(f"Načítám {URL} ...")
await page.goto(URL, wait_until="networkidle", timeout=60_000) await page.goto(URL, wait_until="networkidle", timeout=60_000)
# Najdi iframe s hrou (solitaire.org vkládá hru do iframe) game_url = None
game_frame = None
for frame in page.frames: for frame in page.frames:
if frame.url != page.url and frame.url.strip() not in ("", "about:blank"): if frame.url != page.url and frame.url.strip() not in ("", "about:blank"):
game_frame = frame game_url = frame.url
print(f" Nalezen iframe: {frame.url}") print(f" Nalezen iframe: {game_url}")
break break
target = game_frame if game_frame else page if not game_url:
# Zkus najít iframe src přímo v DOM
iframe_src = await page.get_attribute("iframe", "src")
if iframe_src:
game_url = iframe_src if iframe_src.startswith("http") else f"https://www.solitaire.org{iframe_src}"
print(f" Iframe src z DOM: {game_url}")
# Zkus kliknout na Print tlačítko (různé možné selektory) await page.close()
print_selectors = [
"text=Print",
"button:has-text('Print')",
"[title*='print' i]",
"[aria-label*='print' i]",
".print-button",
"#print",
]
clicked = False
for sel in print_selectors:
try:
await target.click(sel, timeout=3_000)
clicked = True
print(f" Kliknuto na Print ({sel})")
await page.wait_for_timeout(1_500)
break
except Exception:
pass
if not clicked: # Krok 2: otevři URL hry
print(" Tlačítko Print nenalezeno — ukládám celou stránku jako PDF.") game_page = await context.new_page()
target_url = game_url if game_url else URL
print(f"Načítám hru přímo: {target_url} ...")
await game_page.goto(target_url, wait_until="networkidle", timeout=60_000)
# Uložit jako PDF # Game.printDaily() vytvoří iframe #printFrame a zapíše do něj puzzle HTML
await page.pdf( await game_page.evaluate("Game.printDaily()")
await game_page.wait_for_timeout(1_000)
# Přečti HTML přímo z iframu
iframe_html = await game_page.evaluate("""() => {
const iframe = document.getElementById('printFrame');
if (!iframe) return null;
try {
return iframe.contentDocument.documentElement.outerHTML;
} catch(e) { return null; }
}""")
if not iframe_html:
raise RuntimeError("Nepodařilo se získat HTML z #printFrame")
print(f" iframe HTML zachycen ({len(iframe_html)} znaků), ukládám PDF...")
# Přidej base URL aby se načetly obrázky (img/black.png, img/title.jpg)
base_url = "https://www.solitaire.org/daily-str8ts/"
iframe_html = iframe_html.replace("<head>", f'<head><base href="{base_url}">', 1)
# Načti HTML do nové stránky a ulož jako PDF
pdf_page = await context.new_page()
await pdf_page.set_content(iframe_html, wait_until="networkidle")
await pdf_page.wait_for_timeout(500)
await pdf_page.pdf(
path=str(output_path), path=str(output_path),
format="A4", format="A4",
print_background=True, print_background=True,