git
This commit is contained in:
@@ -30,10 +30,11 @@ HOW_MANY_TO_CHECK = 0
|
||||
|
||||
COOKIE_FILE = Path("sktorrent_cookies.json")
|
||||
|
||||
BASE_URL = (
|
||||
"https://sktorrent.eu/torrent/torrents.php"
|
||||
"?active=0&category=23&order=data&by=DESC"
|
||||
)
|
||||
CATEGORIES = {
|
||||
24: "Knihy a časopisy",
|
||||
32: "Mluvené slovo"
|
||||
}
|
||||
MAX_PAGES = 10
|
||||
|
||||
SLEEP_BETWEEN_PAGES = 2.0 # pauza mezi stránkami
|
||||
SLEEP_BEFORE_DOWNLOAD = 1.5 # pauza před stažením každého .torrent souboru
|
||||
@@ -221,6 +222,8 @@ def main():
|
||||
else:
|
||||
mode_desc = f"kontrola posledních {HOW_MANY_TO_CHECK} torrentů"
|
||||
print(f"Pořadí: nejnovější → nejstarší | {mode_desc}")
|
||||
print(f"Kategorie: {', '.join(CATEGORIES.values())}")
|
||||
print(f"Stránek na kategorii: {MAX_PAGES}")
|
||||
print("=" * 60)
|
||||
|
||||
session = build_session()
|
||||
@@ -230,79 +233,87 @@ def main():
|
||||
new_count = 0
|
||||
checked_count = 0
|
||||
skipped_count = 0
|
||||
page = 0
|
||||
stop = False
|
||||
|
||||
while not stop:
|
||||
|
||||
url = f"{BASE_URL}&page={page}"
|
||||
r = None
|
||||
for attempt in range(1, 6):
|
||||
try:
|
||||
r = session.get(url, timeout=15)
|
||||
r.raise_for_status()
|
||||
break
|
||||
except Exception as e:
|
||||
if attempt < 5:
|
||||
print(f"⚠️ Stránka {page} — pokus {attempt}/5 selhal: {e} — čekám 10s")
|
||||
time.sleep(10)
|
||||
else:
|
||||
print(f"⚠️ Stránka {page} — všech 5 pokusů selhalo: {e}")
|
||||
if r is None or not r.ok:
|
||||
for category_id, category_name in CATEGORIES.items():
|
||||
if stop:
|
||||
break
|
||||
|
||||
if "login.php" in r.url or "Prihlas sa" in r.text:
|
||||
print("❌ Cookies expiraly — spusť přihlašovací Selenium skript a obnov cookies.")
|
||||
break
|
||||
print(f"\n📚 Kategorie: {category_name} (ID: {category_id})")
|
||||
|
||||
rows = parse_page(r.text)
|
||||
for page in range(MAX_PAGES):
|
||||
|
||||
if not rows:
|
||||
print(f" Stránka {page} — žádné záznamy, konec.")
|
||||
break
|
||||
|
||||
print(f"\n📄 Stránka {page} ({len(rows)} torrentů)")
|
||||
|
||||
for item in rows:
|
||||
|
||||
if HOW_MANY_TO_CHECK > 0 and checked_count >= HOW_MANY_TO_CHECK:
|
||||
print(f" ⏹ Zkontrolováno {checked_count} torrentů — limit dosažen.")
|
||||
stop = True
|
||||
break
|
||||
|
||||
checked_count += 1
|
||||
|
||||
cursor.execute(
|
||||
"SELECT 1 FROM torrents WHERE torrent_hash = %s",
|
||||
(item["torrent_hash"],)
|
||||
url = (
|
||||
"https://sktorrent.eu/torrent/torrents.php"
|
||||
f"?active=0&category={category_id}&order=data&by=DESC&page={page}"
|
||||
)
|
||||
exists = cursor.fetchone()
|
||||
r = None
|
||||
for attempt in range(1, 6):
|
||||
try:
|
||||
r = session.get(url, timeout=15)
|
||||
r.raise_for_status()
|
||||
break
|
||||
except Exception as e:
|
||||
if attempt < 5:
|
||||
print(f"⚠️ Stránka {page} — pokus {attempt}/5 selhal: {e} — čekám 10s")
|
||||
time.sleep(10)
|
||||
else:
|
||||
print(f"⚠️ Stránka {page} — všech 5 pokusů selhalo: {e}")
|
||||
if r is None or not r.ok:
|
||||
break
|
||||
|
||||
if exists:
|
||||
if HOW_MANY_TO_CHECK == -1:
|
||||
print(f" ⏹ Již v DB: {item['title_visible']} → zastavuji import.")
|
||||
if "login.php" in r.url or "Prihlas sa" in r.text:
|
||||
print("❌ Cookies expiraly — spusť přihlašovací Selenium skript a obnov cookies.")
|
||||
break
|
||||
|
||||
rows = parse_page(r.text)
|
||||
|
||||
if not rows:
|
||||
print(f" Stránka {page} — žádné záznamy, konec.")
|
||||
break
|
||||
|
||||
print(f"\n📄 Stránka {page} ({len(rows)} torrentů)")
|
||||
|
||||
for item in rows:
|
||||
|
||||
if HOW_MANY_TO_CHECK > 0 and checked_count >= HOW_MANY_TO_CHECK:
|
||||
print(f" ⏹ Zkontrolováno {checked_count} torrentů — limit dosažen.")
|
||||
stop = True
|
||||
break
|
||||
|
||||
checked_count += 1
|
||||
|
||||
cursor.execute(
|
||||
"SELECT 1 FROM torrents WHERE torrent_hash = %s",
|
||||
(item["torrent_hash"],)
|
||||
)
|
||||
exists = cursor.fetchone()
|
||||
|
||||
if exists:
|
||||
if HOW_MANY_TO_CHECK == -1:
|
||||
print(f" ⏹ Již v DB: {item['title_visible']} → zastavuji import.")
|
||||
stop = True
|
||||
break
|
||||
else:
|
||||
skipped_count += 1
|
||||
print(f" ⏭ Již v DB: {item['title_visible']} — přeskakuji")
|
||||
continue
|
||||
|
||||
print(f" ⬇️ Nový: {item['title_visible']}")
|
||||
time.sleep(SLEEP_BEFORE_DOWNLOAD)
|
||||
|
||||
content = download_torrent(session, item["download_url"])
|
||||
if content:
|
||||
print(f" ✔ Staženo ({len(content):,} B)")
|
||||
else:
|
||||
skipped_count += 1
|
||||
print(f" ⏭ Již v DB: {item['title_visible']} — přeskakuji")
|
||||
continue
|
||||
print(f" ✖ Nepodařilo se stáhnout, ukládám bez obsahu")
|
||||
|
||||
print(f" ⬇️ Nový: {item['title_visible']}")
|
||||
time.sleep(SLEEP_BEFORE_DOWNLOAD)
|
||||
item["torrent_content"] = content
|
||||
cursor.execute(INSERT_SQL, item)
|
||||
new_count += 1
|
||||
|
||||
content = download_torrent(session, item["download_url"])
|
||||
if content:
|
||||
print(f" ✔ Staženo ({len(content):,} B)")
|
||||
else:
|
||||
print(f" ✖ Nepodařilo se stáhnout, ukládám bez obsahu")
|
||||
|
||||
item["torrent_content"] = content
|
||||
cursor.execute(INSERT_SQL, item)
|
||||
new_count += 1
|
||||
|
||||
if not stop:
|
||||
page += 1
|
||||
if stop:
|
||||
break
|
||||
time.sleep(SLEEP_BETWEEN_PAGES)
|
||||
|
||||
# ============================================================
|
||||
@@ -314,7 +325,6 @@ def main():
|
||||
print(f"Nových torrentů uloženo : {new_count}")
|
||||
print(f"Zkontrolováno celkem : {checked_count}")
|
||||
print(f"Přeskočeno (v DB) : {skipped_count}")
|
||||
print(f"Stránek prošlo : {page}")
|
||||
print("=" * 60)
|
||||
|
||||
db.close()
|
||||
|
||||
Reference in New Issue
Block a user