feat: initial douyin crawler

login_douyin.py

@@ -0,0 +1,122 @@
+from __future__ import annotations
+
+import argparse
+import socket
+import subprocess
+import sys
+import time
+from pathlib import Path
+
+from Douyin import DEFAULT_USER_URL
+
+DEFAULT_CHROME_PATH = "/Applications/Google Chrome.app/Contents/MacOS/Google Chrome"
+DEFAULT_BROWSER_PORT = 9223
+DEFAULT_PROFILE_DIR = Path(".douyin-chrome-profile")
+
+
+def derive_chrome_app_path(chrome_path: str) -> str:
+    marker = ".app/"
+    if marker not in chrome_path:
+        return chrome_path
+    prefix, _ = chrome_path.split(marker, 1)
+    return f"{prefix}.app"
+
+
+def build_login_command(
+    chrome_path: str,
+    profile_dir: Path,
+    browser_port: int,
+    user_url: str,
+) -> list[str]:
+    app_path = derive_chrome_app_path(chrome_path)
+    return [
+        "open",
+        "-na",
+        app_path,
+        "--args",
+        f"--user-data-dir={profile_dir}",
+        f"--remote-debugging-port={browser_port}",
+        user_url,
+    ]
+
+
+def build_parser() -> argparse.ArgumentParser:
+    parser = argparse.ArgumentParser(description="启动可见 Chrome，供抖音手动登录后附着抓取")
+    parser.add_argument("--chrome-path", default=DEFAULT_CHROME_PATH, help="Chrome 可执行文件路径")
+    parser.add_argument(
+        "--profile-dir",
+        default=str(DEFAULT_PROFILE_DIR),
+        help="Chrome 用户数据目录，默认复用项目内固定目录",
+    )
+    parser.add_argument(
+        "--browser-port",
+        type=int,
+        default=DEFAULT_BROWSER_PORT,
+        help="Chrome 调试端口，默认 9223",
+    )
+    parser.add_argument("--user-url", default=DEFAULT_USER_URL, help="启动后打开的抖音主页 URL")
+    return parser
+
+
+def launch_browser(command: list[str]) -> subprocess.Popen[str]:
+    return subprocess.Popen(command)
+
+
+def wait_for_browser_debug_port(
+    browser_port: int,
+    timeout_seconds: float = 15.0,
+    interval_seconds: float = 0.25,
+) -> None:
+    deadline = time.monotonic() + timeout_seconds
+    while time.monotonic() < deadline:
+        try:
+            with socket.create_connection(("127.0.0.1", browser_port), timeout=1):
+                return
+        except OSError:
+            time.sleep(interval_seconds)
+
+    raise RuntimeError(
+        f"Chrome 已启动命令，但调试端口 {browser_port} 在限定时间内未就绪。"
+    )
+
+
+def main(argv: list[str] | None = None) -> int:
+    parser = build_parser()
+    args = parser.parse_args(argv)
+
+    if args.browser_port <= 0:
+        parser.error("--browser-port 必须大于 0")
+
+    chrome_path = Path(args.chrome_path)
+    if not chrome_path.exists():
+        print(f"[ERROR] Chrome 可执行文件不存在: {chrome_path}")
+        return 1
+
+    profile_dir = Path(args.profile_dir).resolve()
+    profile_dir.mkdir(parents=True, exist_ok=True)
+    command = build_login_command(
+        chrome_path=str(chrome_path),
+        profile_dir=profile_dir,
+        browser_port=args.browser_port,
+        user_url=args.user_url,
+    )
+
+    try:
+        launch_browser(command)
+    except OSError as exc:
+        print(f"[ERROR] 启动 Chrome 失败: {exc}")
+        return 1
+
+    try:
+        wait_for_browser_debug_port(args.browser_port)
+    except RuntimeError as exc:
+        print(f"[ERROR] {exc}")
+        return 1
+
+    print("[INFO] Chrome 已启动。请在打开的浏览器中完成抖音登录和验证码。")
+    print(f"[INFO] 登录完成后执行: ./.venv/bin/python Douyin.py --browser-port {args.browser_port}")
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())