网页信息抓取

之前做聊天室时，由于在聊天室中提供了新闻阅读的功能，写了一个从网页中抓取信息（如最新的头条新闻，新闻的来源，标题，内容等）的类，本文将介绍如何使用这个类来抓取网页中需要的信息。本文将以抓取博客园首页的博客标题和链接为例：

上图显示的是博客园首页的DOM树，显然只需提取出class为post_item的div，再重中提取出class为titlelnk的a标志即可。这样的功能可以通过以下函数来实现：

/// <summary>
/// 在文本html的文本查找标志名为tagName,并且属性attrName的值为attrValue的所有标志
/// 例如：FindTagByAttr(html, "div", "class", "demo")
/// 返回所有class为demo的div标志
/// </summary>
public static List<HtmlTag> FindTagByAttr(String html, String tagName, String attrName, String attrValue)
{
    String format = String.Format(@"<{0}\s[^<>]*{1}\s*=\s*(\x27|\x22){2}(\x27|\x22)[^<>]*>", tagName, attrName, attrValue);
    return FindTag(html, tagName, format);
}

public static List<HtmlTag> FindTag(String html, String name, String format)
{
    Regex reg = new Regex(format, RegexOptions.IgnoreCase);
    Regex tagReg = new Regex(String.Format(@"<(\/|)({0})(\s[^<>]*|)>", name), RegexOptions.IgnoreCase);

    List<HtmlTag> tags = new List<HtmlTag>();
    int start = 0;

    while (true)
    {
        Match match = reg.Match(html, start);
        if (match.Success)
        {
            start = match.Index + match.Length;
            Match tagMatch = null;
            int beginTagCount = 1;

            while (true)
            {
                tagMatch = tagReg.Match(html, start);
                if (!tagMatch.Success)
                {
                    tagMatch = null;
                    break;
                }
                start = tagMatch.Index + tagMatch.Length;
                if (tagMatch.Groups[1].Value == "/") beginTagCount--;
                else beginTagCount++;
                if (beginTagCount == 0) break;
            }

            if (tagMatch != null)
            {
                HtmlTag tag = new HtmlTag(name, match.Value, html.Substring(match.Index + match.Length, tagMatch.Index - match.Index - match.Length));
                tags.Add(tag);
            }
            else
            {
                break;
            }
        }
        else
        {
            break;
        }
    }

    return tags;
}

有了以上函数，就可以提取需要的HTML标志了，要实现抓取，还需要一个下载网页的函数：

public static String GetHtml(string url)
{
    try
    {
        HttpWebRequest req = HttpWebRequest.Create(url) as HttpWebRequest;
        req.Timeout = 30 * 1000;
        HttpWebResponse response = req.GetResponse() as HttpWebResponse;
        Stream stream = response.GetResponseStream();

        MemoryStream buffer = new MemoryStream();
        Byte[] temp = new Byte[4096];
        int count = 0;
        while ((count = stream.Read(temp, 0, 4096)) > 0)
        {
            buffer.Write(temp, 0, count);
        }

        return Encoding.GetEncoding(response.CharacterSet).GetString(buffer.GetBuffer());
    }
    catch
    {
        return String.Empty;
    }
}

以下以抓取博客园首页的文章标题和链接为例，介绍如何使用HtmlTag类来抓取网页信息：

class Program
{
    static void Main(string[] args)
    {
        String html = HtmlTag.GetHtml("http://www.cnblogs.com");
        List<HtmlTag> tags = HtmlTag.FindTagByAttr(html, "div", "id", "post_list");
        if (tags.Count > 0)
        {
            List<HtmlTag> item_tags = tags[0].FindTagByAttr("div", "class", "post_item");
            foreach (HtmlTag item_tag in item_tags)
            {
                List<HtmlTag> a_tags = item_tag.FindTagByAttr("a", "class", "titlelnk");
                if (a_tags.Count > 0)
                {
                    Console.WriteLine("标题:{0}", a_tags[0].InnerHTML);
                    Console.WriteLine("链接:{0}", a_tags[0].GetAttribute("href"));
                    Console.WriteLine("");
                }
            }
        }
    }
}

运行结果如下：

源代码下载