yongge/tvbox
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

12

Browse Source
This commit is contained in:
qist
2026-05-21 19:03:54 +08:00
parent d38766eb4a
commit 26a4584c62
1 changed files with 54 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
py/czzyv_v2.py
+54 -2
View File
@@ -17,9 +17,12 @@ class Spider(Spider):
self.host = "https://czzyv.com" self.host = "https://czzyv.com"
self.timeout = 20 self.timeout = 20
self._proxy = "" self._proxy = ""
self._debug = True
self._hosts = [ self._hosts = [
"https://czzyv.com", "https://czzyv.com",
"https://www.cz4k.com", "https://www.cz4k.com",
"https://cz01.vip",
"https://cz01.tv",
] ]
self.headers = { self.headers = {
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/136.0.0.0 Safari/537.36", "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/136.0.0.0 Safari/537.36",
@@ -30,7 +33,7 @@ class Spider(Spider):
self.session = None self.session = None
self._text_cache = {} self._text_cache = {}
self._text_cache_ttl = 300 self._text_cache_ttl = 300
self._last = {"url": "", "status": 0, "len": 0, "host": "", "ua": "", "proxy": "", "err": ""} self._last = {"url": "", "final_url": "", "status": 0, "len": 0, "host": "", "ua": "", "proxy": "", "err": ""}
self._api_base = "" self._api_base = ""
self._use_api = False self._use_api = False
self._ua_fallback = "Dalvik/2.1.0 (Linux; U; Android 10)" self._ua_fallback = "Dalvik/2.1.0 (Linux; U; Android 10)"
@@ -56,13 +59,30 @@ class Spider(Spider):
def getName(self): def getName(self):
return "厂长资源" return "厂长资源"
def _log(self, data):
if not self._debug:
return
try:
fn = getattr(self, "log", None)
if callable(fn):
fn(data)
else:
if isinstance(data, (dict, list)):
print(json.dumps(data, ensure_ascii=False))
else:
print(str(data))
except Exception:
pass
def init(self, extend=""): def init(self, extend=""):
proxy = "" proxy = ""
debug = None
if isinstance(extend, dict): if isinstance(extend, dict):
host = (extend.get("host") or extend.get("site") or "").strip() host = (extend.get("host") or extend.get("site") or "").strip()
if host: if host:
self.host = host.rstrip("/") self.host = host.rstrip("/")
proxy = (extend.get("proxy") or extend.get("http_proxy") or extend.get("https_proxy") or "").strip() proxy = (extend.get("proxy") or extend.get("http_proxy") or extend.get("https_proxy") or "").strip()
debug = extend.get("debug")
elif isinstance(extend, str) and extend.strip(): elif isinstance(extend, str) and extend.strip():
ext_str = extend.strip() ext_str = extend.strip()
if (ext_str.startswith("{") and ext_str.endswith("}")) or (ext_str.startswith("[") and ext_str.endswith("]")): if (ext_str.startswith("{") and ext_str.endswith("}")) or (ext_str.startswith("[") and ext_str.endswith("]")):
@@ -73,11 +93,15 @@ class Spider(Spider):
if host: if host:
self.host = host.rstrip("/") self.host = host.rstrip("/")
proxy = (ext_obj.get("proxy") or ext_obj.get("http_proxy") or ext_obj.get("https_proxy") or "").strip() proxy = (ext_obj.get("proxy") or ext_obj.get("http_proxy") or ext_obj.get("https_proxy") or "").strip()
debug = ext_obj.get("debug")
except Exception: except Exception:
pass pass
elif ext_str.startswith("http"): elif ext_str.startswith("http"):
self.host = ext_str.rstrip("/") self.host = ext_str.rstrip("/")
if debug is not None:
self._debug = bool(debug)
self.headers["Referer"] = self.host + "/" self.headers["Referer"] = self.host + "/"
self.headers["Origin"] = self.host self.headers["Origin"] = self.host
self.session = requests.Session() self.session = requests.Session()
@@ -85,6 +109,7 @@ class Spider(Spider):
self._proxy = proxy self._proxy = proxy
if proxy: if proxy:
self.session.proxies.update({"http": proxy, "https": proxy}) self.session.proxies.update({"http": proxy, "https": proxy})
self._log({"event": "init", "host": self.host, "proxy": self._proxy, "ua": self.session.headers.get("User-Agent", "")})
self._choose_host() self._choose_host()
self._detect_api() self._detect_api()
self._warmup() self._warmup()
@@ -114,26 +139,33 @@ class Spider(Spider):
try: try:
r = self.session.get(h + "/", timeout=self.timeout, allow_redirects=True, verify=False) r = self.session.get(h + "/", timeout=self.timeout, allow_redirects=True, verify=False)
if not r or r.status_code != 200: if not r or r.status_code != 200:
self._log({"event": "choose_host", "try": h, "status": int(r.status_code) if r else 0, "final_url": r.url if r else ""})
continue continue
try: try:
if urlparse(r.url).netloc != urlparse(h).netloc: if urlparse(r.url).netloc != urlparse(h).netloc:
self._log({"event": "choose_host", "try": h, "status": int(r.status_code), "final_url": r.url, "skip": "redirect"})
continue continue
except Exception: except Exception:
pass pass
r.encoding = "utf-8" r.encoding = "utf-8"
text = r.text or "" text = r.text or ""
if "访问已被拦截" in text or "已被拦截" in text: if "访问已被拦截" in text or "已被拦截" in text:
self._log({"event": "choose_host", "try": h, "status": int(r.status_code), "final_url": r.url, "skip": "blocked"})
continue continue
if ("公告" in text and "域名" in text) or ("最新发布" in text) or ("备用网址" in text): if ("公告" in text and "域名" in text) or ("最新发布" in text) or ("备用网址" in text):
self._log({"event": "choose_host", "try": h, "status": int(r.status_code), "final_url": r.url, "skip": "notice"})
continue continue
self.host = h self.host = h
self.headers["Referer"] = self.host + "/" self.headers["Referer"] = self.host + "/"
self.headers["Origin"] = self.host self.headers["Origin"] = self.host
self.session.headers.update(self.headers) self.session.headers.update(self.headers)
self._text_cache.clear() self._text_cache.clear()
self._log({"event": "choose_host", "ok": True, "host": self.host, "final_url": r.url})
return return
except Exception: except Exception:
self._log({"event": "choose_host", "try": h, "ok": False, "err": "exception"})
continue continue
self._log({"event": "choose_host", "ok": False, "host": self.host})
def _detect_api(self): def _detect_api(self):
candidates = [ candidates = [
@@ -177,6 +209,19 @@ class Spider(Spider):
except Exception as e: except Exception as e:
last_exc = f"{type(e).__name__}: {e}" last_exc = f"{type(e).__name__}: {e}"
time.sleep(1) time.sleep(1)
if r is None:
try:
r = requests.get(
url,
headers=(self.session.headers if self.session else self.headers),
timeout=self.timeout,
allow_redirects=True,
verify=False,
proxies=(self.session.proxies if self.session else None),
)
except Exception as e:
if not last_exc:
last_exc = f"{type(e).__name__}: {e}"
if not r or r.status_code != 200: if not r or r.status_code != 200:
if r and r.status_code in (403, 406, 412): if r and r.status_code in (403, 406, 412):
self.session.headers["User-Agent"] = self._ua_fallback self.session.headers["User-Agent"] = self._ua_fallback
@@ -187,13 +232,15 @@ class Spider(Spider):
if not r or r.status_code != 200: if not r or r.status_code != 200:
self._last = { self._last = {
"url": url, "url": url,
"final_url": r.url if r else "",
"status": int(r.status_code) if r else 0, "status": int(r.status_code) if r else 0,
"len": 0, "len": 0,
"host": self.host, "host": self.host,
"ua": (self.session.headers.get("User-Agent") if self.session else ""), "ua": (self.session.headers.get("User-Agent") if self.session else ""),
"proxy": self._proxy, "proxy": self._proxy,
"err": last_exc or "bad_status", "err": last_exc or ("no_response" if r is None else "bad_status"),
} }
self._log({"event": "fetch", **self._last})
return "" return ""
r.encoding = "utf-8" r.encoding = "utf-8"
text = r.text or "" text = r.text or ""
@@ -208,6 +255,7 @@ class Spider(Spider):
pass pass
self._last = { self._last = {
"url": url, "url": url,
"final_url": r.url if r else "",
"status": int(r.status_code) if r else 0, "status": int(r.status_code) if r else 0,
"len": len(text), "len": len(text),
"host": self.host, "host": self.host,
@@ -215,12 +263,15 @@ class Spider(Spider):
"proxy": self._proxy, "proxy": self._proxy,
"err": "", "err": "",
} }
if not text:
self._log({"event": "fetch", **self._last, "warn": "empty_text"})
if text: if text:
self._text_cache[url] = (now + self._text_cache_ttl, text) self._text_cache[url] = (now + self._text_cache_ttl, text)
return text return text
except Exception as e: except Exception as e:
self._last = { self._last = {
"url": url, "url": url,
"final_url": "",
"status": 0, "status": 0,
"len": 0, "len": 0,
"host": self.host, "host": self.host,
@@ -228,6 +279,7 @@ class Spider(Spider):
"proxy": self._proxy, "proxy": self._proxy,
"err": f"{type(e).__name__}: {e}", "err": f"{type(e).__name__}: {e}",
} }
self._log({"event": "fetch", **self._last})
return "" return ""
def _debug_item(self, tag): def _debug_item(self, tag):