如果缺少”http://”,使用url.ResolveReference()解析URL会出错。

huangapple go评论85阅读模式
英文:

Wrong URL parsing with url.ResolveReference() if http:// is missing

问题

我已经构建了一个网络爬虫,用于搜索网页上的所有链接,并且会继续搜索这些链接上的链接,直到整个页面都被爬取。在遇到一个特殊的网站时出现了问题。

问题出在他们的链接上:

正常情况1:绝对路径,如'http://www.example.com/test'

正常情况2:相对路径,如'/test'

问题的新情况:没有'http://'的绝对路径,只有'www.example.com'

下面是展示问题的示例代码:

package main

import (
	"fmt"
	"log"
	"net/url"
)

func main() {

	u, err := url.Parse("http://www.example.com")
	if err != nil {
		log.Fatal(err)
	}
	base, err := url.Parse("http://example.com/directory/")
	if err != nil {
		log.Fatal(err)
	}

	u2, err := url.Parse("www.example.com")
	if err != nil {
		log.Fatal(err)
	}
	base2, err := url.Parse("http://example.com/directory/")
	if err != nil {
		log.Fatal(err)
	}

	fmt.Println(base.ResolveReference(u))
	fmt.Println(base2.ResolveReference(u2))
}

输出结果为:

http://www.example.com
http://example.com/test/www.example.com

可以看到第二行返回了一个错误的URL,因为如果缺少'http://',则绝对URL的测试结果为u.IsAbs() = false...

有什么办法可以修复这个问题吗?我每天需要测试10万到100万个链接,甚至更多,所以性能很重要。

英文:

I've build a web crawler that searches a website for all links on that page and take this links and search on them for more links until the whole page is crawled. Worked perfectly until I came across a special site.

Problem with their linking:

Normal case 1: absolute path like 'http://www.example.com/test'

Normal case 2: relative path like '/test'

Problematic new case: absolute path without the http:// - just 'www.example.com'

Example code that shows the problem:

package main

import (
    "fmt"
    "log"
    "net/url"
)

func main() {

    u, err := url.Parse("http://www.example.com")
    if err != nil {
	    log.Fatal(err)
    }
    base, err := url.Parse("http://example.com/directory/")
        if err != nil {
	        log.Fatal(err)
        }

    u2, err := url.Parse("www.example.com")
    if err != nil {
	    log.Fatal(err)
    }
    base2, err := url.Parse("http://example.com/directory/")
        if err != nil {
	        log.Fatal(err)
        }

    fmt.Println(base.ResolveReference(u))
    fmt.Println(base2.ResolveReference(u2))
}

http://www.example.com
http://example.com/test/www.example.com

As you can see the second line gives back a wrong URL because the test for an absolute URL is u.IsAbs() = false if the http:// is missing ...

Any ideas how to fix that? I have to test 100.000 - 1.000.000 links on a daily base, maybe more and it needs to be performant.

答案1

得分: 1

很遗憾,对于这个问题没有真正的“修复”方法,因为如果你得到一个带有如下 href 的链接:

www.example.com

在一般情况下,它可能是模棱两可的:

http://host.tld/path/to/www.example.com
http://www.example.com

事实上,大多数浏览器会将这样的链接:

<a href="www.example.com">

解释为:

<a href="/current/path/www.example.com">

我建议你做同样的处理(因为这是网站的错误),如果你得到一个 404 错误,就像对待其他错误一样处理。

英文:

Unfortunately there's no real "fix" for this, because if you get a link with an href like this:

www.example.com

In the general case it's ambiguous between:

http://host.tld/path/to/www.example.com
http://www.example.com

In fact, most browsers treat a link like this:

&lt;a href=&quot;www.example.com&quot;&gt;

As this:

&lt;a href=&quot;/current/path/www.example.com&quot;&gt;

I'd suggest doing the same (since this is a bug with the person's website), and if you get a 404 just treat it as you would any other.

huangapple
  • 本文由 发表于 2016年3月13日 04:34:37
  • 转载请务必保留本文链接:https://go.coder-hub.com/35962971.html
匿名

发表评论

匿名网友

:?: :razz: :sad: :evil: :!: :smile: :oops: :grin: :eek: :shock: :???: :cool: :lol: :mad: :twisted: :roll: :wink: :idea: :arrow: :neutral: :cry: :mrgreen:

确定