|
| 1 | +import argparse |
| 2 | +import os |
| 3 | +import sys |
| 4 | +import time |
| 5 | +from datetime import datetime |
| 6 | + |
| 7 | +from selenium import webdriver |
| 8 | +from selenium.webdriver.chrome.options import Options |
| 9 | +from selenium.webdriver.support.ui import WebDriverWait |
| 10 | + |
| 11 | + |
| 12 | +def wait_for_ready(driver, timeout: int = 20): |
| 13 | + WebDriverWait(driver, timeout).until( |
| 14 | + lambda d: d.execute_script("return document.readyState") == "complete" |
| 15 | + ) |
| 16 | + |
| 17 | + |
| 18 | +def auto_scroll(driver, pause: float = 0.3, max_steps: int = 50): |
| 19 | + last_height = driver.execute_script("return document.body.scrollHeight") |
| 20 | + steps = 0 |
| 21 | + while steps < max_steps: |
| 22 | + driver.execute_script("window.scrollTo(0, document.body.scrollHeight)") |
| 23 | + time.sleep(pause) |
| 24 | + new_height = driver.execute_script("return document.body.scrollHeight") |
| 25 | + if new_height == last_height: |
| 26 | + break |
| 27 | + last_height = new_height |
| 28 | + steps += 1 |
| 29 | + driver.execute_script("window.scrollTo(0, 0)") |
| 30 | + |
| 31 | + |
| 32 | +def save_as_mhtml(driver, out_path: str): |
| 33 | + # 使用 Chrome DevTools Protocol 保存为单文件 MHTML |
| 34 | + driver.execute_cdp_cmd("Page.enable", {}) |
| 35 | + result = driver.execute_cdp_cmd("Page.captureSnapshot", {"format": "mhtml"}) |
| 36 | + data = result.get("data") if isinstance(result, dict) else result |
| 37 | + if not data: |
| 38 | + raise RuntimeError("Failed to capture MHTML snapshot") |
| 39 | + |
| 40 | + os.makedirs(os.path.dirname(out_path), exist_ok=True) |
| 41 | + with open(out_path, "w", encoding="utf-8") as f: |
| 42 | + f.write(data) |
| 43 | + |
| 44 | + |
| 45 | +def build_driver(headless: bool = True) -> webdriver.Chrome: |
| 46 | + options = Options() |
| 47 | + if headless: |
| 48 | + # 新版 headless,渲染效果更接近非无头 |
| 49 | + options.add_argument("--headless=new") |
| 50 | + options.add_argument("--disable-gpu") |
| 51 | + options.add_argument("--window-size=1920,1080") |
| 52 | + options.add_argument("--no-sandbox") |
| 53 | + options.add_argument("--disable-dev-shm-usage") |
| 54 | + # 让 Selenium 4+ 自动用 Selenium Manager 管理 ChromeDriver(无需 webdriver_manager) |
| 55 | + return webdriver.Chrome(options=options) |
| 56 | + |
| 57 | + |
| 58 | +def default_out_path(url: str) -> str: |
| 59 | + ts = datetime.now().strftime("%Y%m%d-%H%M%S") |
| 60 | + safe_title = url.rstrip("/").split("/")[-1] or "page" |
| 61 | + return os.path.join( |
| 62 | + os.path.dirname(__file__), |
| 63 | + "downloads", |
| 64 | + f"{safe_title}-{ts}.mhtml", |
| 65 | + ) |
| 66 | + |
| 67 | + |
| 68 | +def main(argv=None): |
| 69 | + parser = argparse.ArgumentParser( |
| 70 | + description="Download a web page with Selenium and save as a complete MHTML file." |
| 71 | + ) |
| 72 | + parser.add_argument( |
| 73 | + "--url", |
| 74 | + default="https://en.macromicro.me/series/22748/cnn-fear-and-greed", |
| 75 | + help="Target page URL", |
| 76 | + ) |
| 77 | + parser.add_argument( |
| 78 | + "--out", |
| 79 | + default=None, |
| 80 | + help="Output .mhtml file path (default: selenium/downloads/...)", |
| 81 | + ) |
| 82 | + parser.add_argument( |
| 83 | + "--delay", |
| 84 | + type=float, |
| 85 | + default=0.0, |
| 86 | + help="Extra seconds to wait before capture (for manual actions like cookie consent)", |
| 87 | + ) |
| 88 | + parser.add_argument( |
| 89 | + "--no-headless", |
| 90 | + action="store_true", |
| 91 | + help="Run browser in non-headless mode", |
| 92 | + ) |
| 93 | + |
| 94 | + args = parser.parse_args(argv) |
| 95 | + out_path = args.out or default_out_path(args.url) |
| 96 | + |
| 97 | + driver = None |
| 98 | + try: |
| 99 | + driver = build_driver(headless=not args.no_headless) |
| 100 | + driver.get(args.url) |
| 101 | + |
| 102 | + # 页面就绪 |
| 103 | + wait_for_ready(driver, timeout=25) |
| 104 | + |
| 105 | + # 可选:滚动触发懒加载,以便资源内嵌到 MHTML |
| 106 | + time.sleep(2) |
| 107 | + auto_scroll(driver, pause=0.4, max_steps=60) |
| 108 | + |
| 109 | + # 再等待一小会儿,给异步资源加载时间;可叠加 --delay |
| 110 | + time.sleep(2 + max(0.0, args.delay)) |
| 111 | + |
| 112 | + save_as_mhtml(driver, out_path) |
| 113 | + print(out_path) |
| 114 | + finally: |
| 115 | + if driver is not None: |
| 116 | + driver.quit() |
| 117 | + |
| 118 | + |
| 119 | +if __name__ == "__main__": |
| 120 | + sys.exit(main()) |
| 121 | + |
0 commit comments