指向 Facebook 实体的 OGP 端点被 FB 爬虫错误解析？

发布于 2024-12-19 10:34:11 字数 1742 浏览 2 评论 0原文

我们的应用程序呈现指向实际 Facebook 页面的 Like 按钮。然而，我们不是将 Like 按钮的 href 直接指向 FB url，而是通过 opengraph 端点通过我们的服务器代理它。这很有用，因为它允许我们对这些端点的使用时间（除其他外）进行更详细的分析。当这些端点被像“facebookexternalhit”这样的用户代理击中时，它们会呈现如下所示的内容：

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
<head>
    <meta property="fb:app_id" content="158708354177049" />
    <meta property="og:site_name" content="BandPage" />
    <meta property="og:url" content="http://www.facebook.com/stormystrongmusic" />
    <meta property="og:type" content="band" />
    <meta property="og:title" content="Stormy Strong" />
    <meta property="og:description" content="..." />
    <meta property="og:image" content="https://graph.facebook.com/20501829906/picture" />

    <title>Stormy Strong</title>

    <link rel="canonical" href="http://www.facebook.com/stormystrongmusic"/>
</head>
<body>
</body>
</html>

请注意，两个端点： url 和链接 rel="canonical" 指向实际的 Facebook 页面。实际上，当我们尝试通过 Facebook 的调试器运行链接时，这看起来表现良好： https://developers.facebook.com/tools/debug/og/object?q=http%3A%2F%2Fwww.bandpage-s.com%2Fogp%2F11601543111380992

但是，有是一个问题，在一段时间后，“赞”按钮似乎不再解析回页面。这会导致“喜欢”计数关闭，如果您已经喜欢该页面，则该按钮不会被“喜欢”等。通过转到上面的 OG 调试器 URL 手动触发重新抓取可以暂时解决该问题。

显然，手动重新触发这些刮擦并不是一个可行的解决方案。这里是否发生了多种抓取行为？似乎由调试器触发的抓取工具与定期自动重新抓取的抓取工具不同。

为什么会出现这种情况？除了在这里摆脱我们的 ogp 端点“中间人”之外，我们还能做些什么来解决这个问题吗？这似乎是 og:url 的完美利用。

原文

Our app renders Like buttons that point back to an actual Facebook page. However, instead of pointing the Like button's href directly to the FB url, we proxy it through our servers through an opengraph endpoint. This is helpful because it allows us to have more detailed analytics on when these endpoints are used (amongst other things.) When these endpoints are hit with a user agent like 'facebookexternalhit', they render something like the following:

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
<head>
    <meta property="fb:app_id" content="158708354177049" />
    <meta property="og:site_name" content="BandPage" />
    <meta property="og:url" content="http://www.facebook.com/stormystrongmusic" />
    <meta property="og:type" content="band" />
    <meta property="og:title" content="Stormy Strong" />
    <meta property="og:description" content="..." />
    <meta property="og:image" content="https://graph.facebook.com/20501829906/picture" />

    <title>Stormy Strong</title>

    <link rel="canonical" href="http://www.facebook.com/stormystrongmusic"/>
</head>
<body>
</body>
</html>

Note that both og:url and the link rel="canonical" point to the actual Facebook Page. In practice, this looks like it behaves fine when we try running the link through Facebook's debugger: https://developers.facebook.com/tools/debug/og/object?q=http%3A%2F%2Fwww.bandpage-s.com%2Fogp%2F11601543111380992

However, there is a problem where, after an interval, the Like buttons appear to not resolve back to the page anymore. This results in the Like count being off, the button not being 'Liked' if you already like the page, etc. Manually triggering a re-scrape by going to the OG debugger url above solves the problem, for a time.

Obviously manually re-triggering these scrapes is not a tenable solution. Are there multiple scraper behaviors happening here? It seems like the scraper that is triggered w/ the debugger is different than the one that periodically re-scrapes automatically.

Why does this happen? Is there anything we can do to get around this, except for getting rid of our ogp endpoint 'middle-man' here? This seems like a perfect utilization of og:url.

分享到QQ

分享到微博