/// <summary>
/// 自动翻页
/// </summary>
/// <param name="args">DataReceivedEventArgs</param>
/// <returns>URL</returns>
private static void CustomParseLinkEvent_Next(CustomParseLinkEvent2Args args)
{
#region 20150930之前的代码
/*20150930之前的代码
string url = "";
string html = args.Html;
string strReg = "<a .+ href='(.+)'>下一页</a>";
Regex regex = new Regex(strReg);
Match mat = regex.Match(html);
if (mat.Success)
{
if (IsUrlable(mat.Groups[1].Value))
{
url = mat.Groups[1].Value;
}
else
{
Int32 index = args.Url.LastIndexOf("/");
//url = args.Url.Substring(0, index) + "/" + mat.Groups[1].Value;
url = args.Url.Substring(0, index) + "/list.php" + mat.Groups[1].Value;
Console.WriteLine("************************");
Console.WriteLine(url);
Console.WriteLine("************************");
File.AppendAllText(urlFilePath, "************************" + "\r\n");
File.AppendAllText(urlFilePath, args.Url + "\r\n");
File.AppendAllText(urlFilePath, "************************" + "\r\n");
}
}
return IsUrlable(url) ? new UrlInfo(url) { Depth = args.Depth + 1 } : null;
*/
#endregion 20150930之前的代码
//urlAndHtml.Html = args.Html;
//urlAndHtml.Url = args.Url;
//string url = AutoNextPage(urlAndHtml, "<a .+ href='(.+)'>下一页</a>", 1);
//return IsUrlable(url) ? new UrlInfo(url) { Depth = args.Depth + 1 } : null;
}
private static Dictionary<string, string> CustomParseLinkE_MainList(CustomParseLinkEvent2Args args, string patternStr)
{
Dictionary<string, string> temp = new Dictionary<string, string>();
foreach (var item in args.UrlDictionary)
{
string href = item.Key;
string text = item.Value;
if (!string.IsNullOrEmpty(href))
{
Regex regex = new Regex(patternStr);
Match mat = regex.Match(href);
if (mat.Success)
{
temp.Add(href, text);
}
}
}
return temp;
}
//用ref或out改写该方法
private static Dictionary<string, string> Master_CustomParseLinkEvent2(CustomParseLinkEvent2Args args)
{
args.UrlDictionary = CustomParseLinkE_MainList(args, "(view).+?([0-9]{5})");//去除
return CustomParseLinkE_NextPageSdau(args, "<a .+ href='(.+)'>下一页</a>", 1);//添加
}
private static Dictionary<string, string> CustomParseLinkE_NextPageSdau(CustomParseLinkEvent2Args args, string patternStr, int groupIndex)
{
string url = "";
if (args != null && !string.IsNullOrEmpty(args.Html))
{
Regex regex = new Regex(patternStr);
Match mat = regex.Match(args.Html);
if (mat.Success)
{
url = mat.Groups[groupIndex].Value;
var baseUri = new Uri(args.UrlInfo.UrlString);
Uri currentUri = url.StartsWith("http", StringComparison.OrdinalIgnoreCase)
? new Uri(url)
: new Uri(baseUri, url);//根据指定的基 URI 和相对 URI 字符串，初始化 System.Uri 类的新实例。
//如果不包含http，则认为超链接是相对路径，根据baseUrl建立绝对路径
url = currentUri.AbsoluteUri;
//Console.WriteLine("######" + url + "######");
args.UrlDictionary.Add(url, Guid.NewGuid().ToString());
}
}
return args.UrlDictionary;
}