/// <summary>
/// 初始化抓取程序类型设定
/// </summary>
public static void InitCrawller(TreeView tree)
{
if (tree.Nodes.Count > 0)
{
SaveConfigs(tree);
}
tree.Nodes.Clear();
Dictionary <string, Type> dictCrawllers = ToolComm.DictTypeCrawllers;
DateTime dt = DateTime.Now;
foreach (string typeFullName in dictCrawllers.Keys)
{
Type crawllerType = dictCrawllers[typeFullName];
WebSiteCrawller crawller = Assembly.GetAssembly(crawllerType).CreateInstance(typeFullName) as WebSiteCrawller;
if (!string.IsNullOrEmpty(crawller.Title))
{
string groupName = string.IsNullOrEmpty(crawller.Group) ? "其他" : crawller.Group;
TreeNode[] tn_arr = tree.Nodes.Find(groupName, true);
TreeNode group = null;
if (tn_arr != null && tn_arr.Length > 0)
{
group = tn_arr[0];
}
else
{
group = new TreeNode();
group.Text = groupName;
group.Name = groupName;
group.Checked = true;
tree.Nodes.Add(group);
}
if (group != null)
{
TreeNode node = new TreeNode();
node.Text = crawller.Title;
node.Name = typeFullName;
node.ToolTipText = crawller.Description;
node.Checked = crawller.Enabled;
node.ForeColor = crawller.Enabled ? Color.Blue : Color.Gray;
node.Tag = crawller;
group.Nodes.Add(node);
}
}
}
LoadConfigs(tree);
tree.ExpandAll();
}
protected static void GetNodeConfigs(TreeNode node, List <WebSiteCrawlConfig> configs)
{
WebSiteCrawller crawller = node.Tag as WebSiteCrawller;
if (crawller != null)
{
configs.Add(crawller.Config);
}
if (node.Nodes.Count > 0)
{
foreach (TreeNode child in node.Nodes)
{
GetNodeConfigs(child, configs);
}
}
}
protected static void GetNodeCrawllers(TreeNode node, Dictionary <string, WebSiteCrawller> crawllers)
{
if (node.Checked)
{
WebSiteCrawller crawller = node.Tag as WebSiteCrawller;
if (crawller != null)
{
crawllers.Add(crawller.Key, crawller);
}
if (node.Nodes.Count > 0)
{
foreach (TreeNode child in node.Nodes)
{
GetNodeCrawllers(child, crawllers);
}
}
}
}
public static void UpdateNodeConfigs(TreeNode node, bool updateChilds)
{
WebSiteCrawller crawller = node.Tag as WebSiteCrawller;
if (crawller != null)
{
node.Text = crawller.Title;
node.Name = crawller.Key;
node.ToolTipText = crawller.Description;
node.Checked = crawller.Enabled;
node.ForeColor = crawller.Enabled ? Color.Blue : Color.Gray;
}
if (node.Nodes.Count > 0)
{
foreach (TreeNode child in node.Nodes)
{
UpdateNodeConfigs(child, updateChilds);
}
}
}
private void ClearSelect()
{
foreach (TreeNode tn in treCrawlers.Nodes)
{
if (tn.Text.Contains(txtSertch.Text.Trim()))
{
tn.ForeColor = Color.Blue;
}
if (tn.Nodes.Count > 1)
{
foreach (TreeNode node in tn.Nodes)
{
WebSiteCrawller crawller = node.Tag as WebSiteCrawller;
if (crawller != null)
{
node.ForeColor = crawller.Enabled ? Color.Blue : Color.Gray;
}
}
}
}
}
private void treCrawlers_AfterCheck(object sender, TreeViewEventArgs e)
{
if (e.Node != null)
{
WebSiteCrawller crawller = e.Node.Tag as WebSiteCrawller;
if (crawller != null)
{
crawller.Enabled = e.Node.Checked;
}
foreach (TreeNode child in e.Node.Nodes)
{
child.Checked = e.Node.Checked;
WebSiteCrawller childCrawller = child.Tag as WebSiteCrawller;
if (childCrawller != null)
{
childCrawller.Enabled = child.Checked;
}
}
}
}
/// <summary>
/// 启动定时抓取程序
/// </summary>
/// <param name="sender"></param>
/// <param name="e"></param>
private void BtnBegin_Click(object sender, EventArgs e)
{
dictTimerCrawllers.Clear();
int count = 0;
Dictionary <string, WebSiteCrawller> crawllers = ToolComm.GetEnabledCrawllers(treCrawlers);
foreach (string key in crawllers.Keys)
{
count++;
WebSiteCrawller crawller = crawllers[key];
if (dictTimerCrawllers.ContainsKey(crawller.PlanTime))
{
dictTimerCrawllers[crawller.PlanTime].Add(crawller);
}
else
{
List <WebSiteCrawller> list = new List <WebSiteCrawller>();
list.Add(crawller);
dictTimerCrawllers.Add(crawller.PlanTime, list);
}
}
if (count == 0)
{
MessageBox.Show("请选择要抓取的选项！");
return;
}
ToolComm.SaveConfigs(treCrawlers);
timer.AutoReset = true;
timer.Enabled = true;
timer.Interval = 60 * 1000; //1分钟执行一次
timer.Elapsed += new System.Timers.ElapsedEventHandler(timer_Tick);
AppendText("开始抓取数据......");
BtnBegin.Enabled = false;
grpCurr.Enabled = false;
}
/// <summary>
/// 得到某个类的抓取时间
/// </summary>
public static string GetCrawlerInfo(WebSiteCrawller crawller)
{
if (crawller == null)
{
return(string.Empty);
}
StringBuilder sb = new StringBuilder();
sb.AppendFormat("抓取说明：{0}", crawller.Description);
sb.AppendLine();
sb.AppendFormat("抓取程序：{0}", crawller.Key);
sb.AppendLine();
sb.AppendFormat("上次抓取：{0:yyyy-MM-dd HH:mm:ss}", crawller.LastCrawlEnd);
sb.AppendLine();
sb.AppendFormat("计划时间：{0}", crawller.PlanTime.Replace(",", ", "));
sb.AppendLine();
sb.AppendFormat("抓取数量：最近 {0} 条", crawller.MaxCount);
sb.AppendLine();
sb.AppendFormat("抓取地址：{0}", crawller.SiteUrl);
sb.AppendLine();
return(sb.ToString());
}
protected static void SetNodeConfigs(TreeNode node, Dictionary <string, WebSiteCrawlConfig> configs)
{
WebSiteCrawller crawller = node.Tag as WebSiteCrawller;
if (crawller != null && !string.IsNullOrEmpty(crawller.Key))
{
if (configs.ContainsKey(crawller.Key))
{
WebSiteCrawlConfig config = configs[crawller.Key];
if (config != null && !string.IsNullOrEmpty(config.Key))
{
crawller.Config = config;
UpdateNodeConfigs(node, false);
}
}
}
if (node.Nodes.Count > 0)
{
foreach (TreeNode child in node.Nodes)
{
SetNodeConfigs(child, configs);
}
}
}
private void treCrawlers_AfterSelect(object sender, TreeViewEventArgs e)
{
currNode = e.Node;
currCrawller = currNode.Tag as WebSiteCrawller;
SetCurrCrawllerDetails();
}
public WebSiteCollection(WebSiteCrawller crawller)
{
this.WebCrawller = crawller;
}
/// <summary>
/// 根据反射，从实体中调用方法，得到返回结果
/// </summary>
/// <param name="assemblyPath"></param>
/// <param name="entityFullName"></param>
/// <param name="crawlAll">是否抓取所有数据</param>
/// <returns></returns>
public static string DealEntity(WebSiteCrawller crawller)
{
StringBuilder result = new StringBuilder();
try
{
//处理信息
IList infoList = crawller.Crawl(false);
if (infoList != null && infoList.Count > 0)
{
object[] successList;
List <BaseAttach> tattachList = crawller.AttachList;
int count = ToolCoreDb.SaveDatas(infoList, crawller.ExistCompareFields, tattachList, out successList, crawller.ExistsUpdate, crawller.ExistsHtlCtx, crawller.ExistsUpdateAttach);
result.Append("【").Append(crawller.Title).Append("】信息【").Append(count).Append("/").Append(infoList.Count).Append("】条；");
Base.KdService.CrawlerService ser = new Base.KdService.CrawlerService();
int resultCount;
//处理附件
List <BaseAttach> newAttachList = new List <BaseAttach>();
if (crawller.ExistsUpdateAttach)
{
if (infoList != null && infoList.Count > 0)
{
List <BaseAttach> attch = null;
foreach (var item in infoList)
{
Type types = item.GetType();
string id = types.GetProperty("Id").GetValue(item, null).ToString();
attch = tattachList.FindAll(a => a.SourceID == id);
for (int i = 0; i < attch.Count; i++)
{
newAttachList.Add(attch[i]);
}
}
}
}
else
{
if (successList != null && successList.Length > 0)
{
List <BaseAttach> attch = null;
foreach (var item in successList)
{
Type types = item.GetType();
string id = types.GetProperty("Id").GetValue(item, null).ToString();
attch = tattachList.FindAll(a => a.SourceID == id);
for (int i = 0; i < attch.Count; i++)
{
newAttachList.Add(attch[i]);
}
}
}
}
count = ToolCoreDb.SaveDatas(newAttachList, "SourceID,AttachServerPath");
result.Append("附件【").Append(count).Append("/").Append(tattachList.Count).Append("】条；");
crawller.AttachList.Clear();
}
else
{
result.Append("【").Append(crawller.Title).Append("】信息【").Append("0").Append("/").Append("0").Append("】条；");
}
}
catch (Exception ex)
{
Logger.Error(ex);
result.Remove(0, result.Length);
result.Append("抓取【").Append(crawller.Title).Append("】出现异常，详见日志文件！");
}
return(result.ToString());
}
