/// <summary>
/// 解析导航规则
/// </summary>
/// <param name="startingUrl">起始地址</param>
/// <returns>导航地址</returns>
private StringCollection ParseNavigationRuleItem(StringCollection startingUrl)
{
StringCollection urls = new StringCollection();
foreach (string u in startingUrl)
{
foreach (NavigationRule rule in _urlItem.NavigationRules)
{
/*
* 描述:
* 加入最终页面地址
*
* 步骤:
* 1.判断是否终端页面地址，如果是则直接加入并引发事件.
* 2.否则,请求web服务器并返回html文本,根据导航规则解析出终端页面地址.
*
* 修改标志:王亚 201204244
*/
if (rule.Terminal)
{
urls.Add(u); //最终页面直接加入导航地址
if (onSingleComplete != null)
{
this.onSingleComplete(this, u); //引发增加一条网址事件
}
}
else
{
try
{
HttpHelper http = new HttpHelper();
/*
*修改标志 20120601 王亚 解析导航地址时增加Http请求编码
*/
http._encoding = Encoding.GetEncoding(_urlItem.UrlEncoding);
string htmlText = http.RequestResult(u); //发送Http请求获取导航地址
StringCollection navUrlItem = ParseNavigationRuleHtmlText(rule, htmlText);
foreach (string r in navUrlItem)
{
/*
* 处理相对路径网址问题如：/html/gndy/jddy/20120425/37418.html
* 如果不包含http://选项，则在相对路径前边加上主机地址。
*/
string path = r;
if (!r.Contains("http://") && r.Length > 0)
{
path = r.Insert(0, "http://" + http.WebResponse.ResponseUri.Authority);
}
urls.Add(path);
if (onSingleComplete != null)
{
this.onSingleComplete(this, path); //引发增加一条网址事件
}
}
}
catch (Exception ex)
{
if (OnAppendSingileLog != null)
{
OnAppendSingileLog(this, new LogEventArgs(ex.Message));
}
}
}
}
}
return urls;
}
/// <summary>
/// 销毁资源
/// </summary>
public void Dispose() {
//保存采集结果
SaveResult();
this._HttpHelper = null;
this._Results = null;
this._TaskConfig = null;
}
/// <summary>
/// 构造函数
/// </summary>
public TaskUnit() {
this._HttpHelper = new HttpHelper(Encoding.GetEncoding(this._TaskConfig.UrlListManager.UrlEncoding));
time = new Timer(new TimerCallback(Start), "", Timeout.Infinite, Timeout.Infinite);
}
/// <summary>
/// 解析导航规则
/// </summary>
/// <param name="startingUrl">起始地址</param>
/// <returns>导航地址</returns>
private StringCollection ParseNavigationRuleItem(StringCollection startingUrl)
{
StringCollection urls = new StringCollection();
foreach (string u in startingUrl)
{
foreach (NavigationRule rule in _urlItem.NavigationRules)
{
/*
* 描述:
* 加入最终页面地址
*
* 步骤:
* 1.判断是否终端页面地址，如果是则直接加入并引发事件.
* 2.否则,请求web服务器并返回html文本,根据导航规则解析出终端页面地址.
*
* 修改标志:王亚 201204244
*/
if (rule.Terminal)
{
urls.Add(u); //最终页面直接加入导航地址
if (onSingleComplete != null)
{
this.onSingleComplete(this, u); //引发增加一条网址事件
}
}
else
{
try
{
HttpHelper http = new HttpHelper();
/*
*修改标志 20120601 王亚 解析导航地址时增加Http请求编码
*/
http._encoding = Encoding.GetEncoding(_urlItem.UrlEncoding);
string htmlText = http.RequestResult(u); //发送Http请求获取导航地址
StringCollection navUrlItem = ParseNavigationRuleHtmlText(rule, htmlText);
foreach (string r in navUrlItem)
{
/*
* 处理相对路径网址问题如：/html/gndy/jddy/20120425/37418.html
* 如果不包含http://选项，则在相对路径前边加上主机地址。
*/
string path = r;
if (!r.Contains("http://") && r.Length > 0)
{
path = r.Insert(0, "http://" + http.WebResponse.ResponseUri.Authority);
}
urls.Add(path);
if (onSingleComplete != null)
{
this.onSingleComplete(this, path); //引发增加一条网址事件
}
}
}
catch (Exception ex)
{
if (OnAppendSingileLog != null)
{
OnAppendSingileLog(this, new LogEventArgs(ex.Message));
}
}
}
}
}
return(urls);
}