public string GetMeta(string name)
{
var cc = hostBrowser.GetElementsByTagName(true, "META");
mshtml.IHTMLElement el = cc.item(name, 0) as IHTMLElement;
return(el.getAttribute("content") as string);
}
private bool GotoNextPage(WebBrowser wb, bool isLastPage)
{
mshtml.IHTMLDocument2 doc2 = (mshtml.IHTMLDocument2)wb.Document;
// string html = doc2.body.innerHTML;
var currentUrl = wb.Source.AbsoluteUri;
currentUrl = currentUrl.Replace("#", "");
if (currentUrl == this.TargetUri)
{
mshtml.IHTMLElement example1_next = (mshtml.IHTMLElement)doc2.all.item("example1_next", 0);
example1_next.click();
int pageCounter = 0;
var pagenumStr = example1_next.getAttribute("data-dt-idx");
if (!int.TryParse(pagenumStr.ToString(), out pageCounter))
{
MessageBox.Show(string.Format("翻页失败，请手工执行"));
}
//边界条件
var classstr = example1_next.getAttribute("class");
if (classstr == null)
{
MessageBox.Show(string.Format("example1_next miss {0} attr,请手工点击[下一页]", "class"));
}
else
{
if (classstr.ToString().Contains("disabled"))
{
isLastPage = true;
}
}
// class = paginate_button current 定
}
return(isLastPage);
}
//空间动态里的好友
private void button9_Click(object sender, EventArgs e)
{
string info = "";
mshtml.HTMLDocumentClass htmlDoc = webBrowser1.Document.DomDocument as mshtml.HTMLDocumentClass;
for (int i = 0; i < htmlDoc.frames.length; ++i)
{
object index = i;
mshtml.IHTMLWindow2 frameWindow = htmlDoc.frames.item(ref index) as mshtml.IHTMLWindow2;
IHTMLDocument2 frame = CodecentrixSample.CrossFrameIE.GetDocumentFromWindow(frameWindow);
for (int j = 0; j < frame.frames.length; ++j)
{
try
{
object rj = j;
mshtml.IHTMLWindow2 frameWindow2 = frame.frames.item(ref rj) as mshtml.IHTMLWindow2;
IHTMLDocument2 frame2 = CodecentrixSample.CrossFrameIE.GetDocumentFromWindow(frameWindow2);
mshtml.IHTMLElementCollection userlist = (mshtml.IHTMLElementCollection)frame2.all.tags("div");
IEnumerator users = userlist.GetEnumerator();
while (users.MoveNext())
{
mshtml.IHTMLElement element = (mshtml.IHTMLElement)users.Current;
if (element.className == "user-list" || element.className == "ui-avatar")
{
mshtml.IHTMLElementCollection childrenCol = (mshtml.IHTMLElementCollection)element.children;
IEnumerator cusers = childrenCol.GetEnumerator();
while (cusers.MoveNext())
{
mshtml.IHTMLElement celement = (mshtml.IHTMLElement)cusers.Current;
string href = (string)celement.getAttribute("href", 0);
if (href.Contains("http:"))
{
info += href + "\n";
}
}
}
}
}
catch (Exception)
{
}
}
}
MessageBox.Show(info);
}
private FindAttribute GetCssPathFinder(IHTMLElement element)
{
FindAttribute Finder = Finder = new FindAttribute();
Finder.FindMethod = FindMethods.CssSelector;
string selector = element.getAttribute("__selector", 0) as string;
if (!string.IsNullOrEmpty(selector))
{
Finder.FindValue = selector;
}
else
{
Finder.FindValue = this.AppContext.Browser.Selector.GetCssPath(element, false);
}
return(Finder);
}
//检测是否已评论
private mshtml.IHTMLElement gerCommented()
{
String myqq = getMyQQ();
mshtml.HTMLDocumentClass htmlDoc = webBrowser1.Document.DomDocument as mshtml.HTMLDocumentClass;
for (int i = 0; i < htmlDoc.frames.length; ++i)
{
object index = i;
mshtml.IHTMLWindow2 frameWindow = htmlDoc.frames.item(ref index) as mshtml.IHTMLWindow2;
IHTMLDocument2 frame = CodecentrixSample.CrossFrameIE.GetDocumentFromWindow(frameWindow);
mshtml.IHTMLElementCollection tagls = (mshtml.IHTMLElementCollection)frame.all.tags("div");
IEnumerator tagie = tagls.GetEnumerator();
while (tagie.MoveNext())
{
mshtml.IHTMLElement element = (mshtml.IHTMLElement)tagie.Current;
if (element.className == "mod-comments")
{
mshtml.IHTMLElementCollection childrenCol = (mshtml.IHTMLElementCollection)element.all;
IEnumerator cusers = childrenCol.GetEnumerator();
bool cangonext = true;
while (cusers.MoveNext())
{
mshtml.IHTMLElement celement = (mshtml.IHTMLElement)cusers.Current;
Console.WriteLine(celement.className);
if (celement.className == "comments-item bor3")
{
if ((String)celement.getAttribute("data-uin", 0) == myqq)
{
cangonext = false;
break;
}
}
}
//1.打开编辑框
if (cangonext)//自己没有评论
{
return(element);
}
}
}
}
return(null);
}
public static bool IsElementMatch(mshtml.IHTMLElement ele, ElementTag tag, string itemName, string keyword = "")
{
itemName = itemName.Trim();
if (tag == ElementTag.ID)
{
return(!string.IsNullOrEmpty(ele.id) && ele.id.Trim().Equals(itemName));
}
if (tag == ElementTag.name)
{
return((ele.getAttribute("name", 0) != null) && ele.getAttribute("name", 0).ToString().Trim().Equals(itemName));
}
if (tag == ElementTag.outerText)
{
return(!string.IsNullOrEmpty(ele.outerText) && (ele.outerText.Trim().IndexOf(itemName) != -1));
}
if (tag == ElementTag.className)
{
return(!string.IsNullOrEmpty(ele.className) && ele.className.Trim().Equals(itemName));
}
if (tag == ElementTag.outerHTML)
{
return(!string.IsNullOrEmpty(ele.outerHTML) && ele.outerHTML.Trim().Equals(itemName));
}
if (tag == ElementTag.value)
{
return((ele.getAttribute("value", 0) != null) && ele.getAttribute("value", 0).ToString().Trim().Equals(itemName));
}
if (tag == ElementTag.href)
{
return((ele.getAttribute("href", 0) != null) && ele.getAttribute("href", 0).ToString().Trim().Equals(itemName));
}
if (tag != ElementTag.src)
{
return(false);
}
if (ele.getAttribute("src", 0) == null)
{
return(false);
}
return(ele.getAttribute("src", 0).ToString().Trim().Equals(itemName) || (!string.IsNullOrEmpty(keyword) && (ele.getAttribute("src", 0).ToString().Trim().IndexOf(keyword) != -1)));
}
//获取【说说】的好友信息
private void button7_Click(object sender, EventArgs e)
{
string info = null;
mshtml.HTMLDocumentClass htmlDoc = webBrowser1.Document.DomDocument as mshtml.HTMLDocumentClass;
for (int i = 0; i < htmlDoc.frames.length; ++i)
{
object index = i;
mshtml.IHTMLWindow2 frameWindow = htmlDoc.frames.item(ref index) as mshtml.IHTMLWindow2;
IHTMLDocument2 frame = CodecentrixSample.CrossFrameIE.GetDocumentFromWindow(frameWindow);
mshtml.IHTMLElementCollection divs = (mshtml.IHTMLElementCollection)frame.all.tags("div");
IEnumerator div = divs.GetEnumerator();
while (div.MoveNext())
{
mshtml.IHTMLElement item = (mshtml.IHTMLElement)div.Current;
if (item.className == "ui_avatar" || item.className == "feed_like")
{
mshtml.IHTMLElementCollection childrenCol = (mshtml.IHTMLElementCollection)item.children;
IEnumerator cusers = childrenCol.GetEnumerator();
while (cusers.MoveNext())
{
mshtml.IHTMLElement celement = (mshtml.IHTMLElement)cusers.Current;
if (celement.tagName == "a" || celement.tagName == "A")
{
string href = (string)celement.getAttribute("href", 0);
if (href != null && href.Contains("http:"))
{
info += href + "\n";
}
}
}
}
}
}
MessageBox.Show(info);
}
static public string ActiveElementAttribute(IHTMLElement element, string AttributeName)
{
if (element == null)
{
return("");
}
string strValue = "";
try
{
strValue = element.getAttribute(AttributeName, 0) as string ?? "";
if (string.IsNullOrEmpty(strValue))
{
strValue = element.innerText;
}
}
catch (Exception)
{
}
return(strValue);
}
public static bool IsElementMatchType(mshtml.IHTMLElement ele, string typeName)
{
return((ele.getAttribute("type", 0) != null) && ele.getAttribute("type", 0).ToString().ToLower().Trim().Equals(typeName));
}
public void StartCrawl(object view, Button btn)// private void BtnDownload_Click(object sender, RoutedEventArgs e)
{
bool IsUnConfirmChecked = true;
WebBrowser wb = null;
var vw = view as WebBoxView;
if (vw != null)
{
wb = vw.FindName("webBox") as WebBrowser;
Uri uri = new Uri(this.TargetUri);
if (wb == null || wb.Source != uri)
{
MessageBox.Show("没有适合的数据");
return;
}
}
#region 确认标志 IsUnConfirmChecked (qrzt0~未确认 / qrzt1~已确认)
var currentUrl = wb.Source.AbsoluteUri;
if (currentUrl.Contains("#"))
{
currentUrl = currentUrl.Replace("#", "");
}
if (currentUrl == this.TargetUri)
{
mshtml.IHTMLDocument2 doc2 = (mshtml.IHTMLDocument2)wb.Document;
mshtml.IHTMLElement qrzt0 = (mshtml.IHTMLElement)doc2.all.item("qrzt0", 0);
if (qrzt0 != null)
{
IsUnConfirmChecked = (bool)qrzt0.getAttribute("CHECKED");
// login_header.setAttribute("style", "display: none;");
Debug.WriteLine(IsUnConfirmChecked);
}
}
#endregion
bool isOffline = string.Compare(this.IsOffline, "1", StringComparison.InvariantCultureIgnoreCase) == 0 ? true : false;
#region Auto pagenate ： id = example1_next
this.IsBusy = true;
this.BusyText = "正在同步数据，请稍后......";
wb.Visibility = Visibility.Hidden;
ProgressBar pBar = vw.FindName("ProgBar") as ProgressBar;
if (pBar != null)
{
ConfigProgressBar(pBar);
}
//下一页定总页数和边界条件
var isLastPage = false;
var pNum = 1;
bool hasValidData = true;
do
{
#region Crawl Current Page
//法1：<N>基于Httphelper，这样下载会要求程序自己实现验证授权
//HttpHelper httpHelper = new HttpHelper();
//HttpItem rq = new HttpItem();
//rq.URL = uri.AbsoluteUri;
//HttpResult html = httpHelper.GetHtml(rq);
//Debug.WriteLine(html.Html);
//法2：直接基于WebBrowser，授权是由用户手动实现的
CrawlCurrentPage(wb, isOffline, IsUnConfirmChecked, ref hasValidData);
#endregion
if (isOffline)
{
MessageBox.Show("离线测试结束");
return;
}
if (hasValidData)
{
this.IsBusy = true;
this.BusyText = string.Format("已同步第{0}页数据", (pNum++).ToString());
isLastPage = GotoNextPage(wb, isLastPage);
}
else
{
this.IsBusy = false;
wb.Visibility = Visibility.Visible;
isLastPage = true;
}
}while (!isLastPage);
#endregion
if (isLastPage && hasValidData)
{
this.IsBusy = false;
wb.Visibility = Visibility.Visible;
MessageBox.Show("所有数据已同步完成");
}
}
public override void Tick()
{
if (Enable)
{
try
{
string info = null;
mshtml.HTMLDocumentClass htmlDoc = webBrowser1.Document.DomDocument as mshtml.HTMLDocumentClass;
for (int i = 0; i < htmlDoc.frames.length; ++i)
{
object index = i;
mshtml.IHTMLWindow2 frameWindow = htmlDoc.frames.item(ref index) as mshtml.IHTMLWindow2;
IHTMLDocument2 frame = CodecentrixSample.CrossFrameIE.GetDocumentFromWindow(frameWindow);
mshtml.IHTMLElementCollection divs = (mshtml.IHTMLElementCollection)frame.all.tags("div");
IEnumerator div = divs.GetEnumerator();
while (div.MoveNext())
{
mshtml.IHTMLElement item = (mshtml.IHTMLElement)div.Current;
if (item.className == "ui_avatar" || item.className == "feed_like")
{
mshtml.IHTMLElementCollection childrenCol = (mshtml.IHTMLElementCollection)item.children;
IEnumerator cusers = childrenCol.GetEnumerator();
while (cusers.MoveNext())
{
mshtml.IHTMLElement celement = (mshtml.IHTMLElement)cusers.Current;
if (celement.tagName == "a" || celement.tagName == "A")
{
string href = (string)celement.getAttribute("href", 0);
if (href != null && href.Contains("http:"))
{
info += href + "\n";
spider.AddQQ(href);
}
}
}
}
}
}
//string info = "";
//for (int i = 0; i < webBrowser1.Document.Window.Frames.Count; ++i)
//{
// HtmlDocument frame = webBrowser1.Document.Window.Frames[i].Document;
// HtmlElementCollection divs = frame.GetElementsByTagName("div");
// for (int j = 0; j < divs.Count; ++j)
// {
// string className = divs[j].GetAttribute("className");
// if (className == "ui_avatar")
// {
// HtmlElementCollection a = divs[j].GetElementsByTagName("a");
// if (a != null && a.Count > 0)
// {
// string url = a[0].GetAttribute("href");
// info += url + "\n";
// spider.AddQQ(url);
// }
// }
// else if (className == "feed_like")
// {
// HtmlElementCollection a = divs[j].GetElementsByTagName("a");
// if (a != null && a.Count > 0)
// {
// for (int k = 0; k < a.Count; ++k)
// {
// if (a[k].GetAttribute("className") == "c_tx")
// {
// string url = a[0].GetAttribute("href");
// info += url + "\n";
// spider.AddQQ(url);
// }
// }
// }
// }
// }
//}
spider.UserGetOver();
Log(info);
ChangeCommon("NewUser");
}
catch (Exception e)
{
Log(e.Message);
ChangeCommon("NewUser");
}
}
}
void IDocHostUIHandler.TranslateAccelerator(ref tagMSG lpmsg, ref Guid pguidCmdGroup, uint nCmdID)
{
const int Ok = 0;
const int Error = 1;
const int VK_BACK = 0x08;
const int WM_KEYDOWN = 0x0100;
const int WM_KEYUP = 0x0101;
const int VK_CONTROL = 0x11;
const int VK_ALT = 0x12;
if (lpmsg.message != WM_KEYUP)
{
if (GetAsyncKeyState(VK_BACK) < 0)
{
mshtml.IHTMLDocument2 htmlDoc = axWebBrowser1.Document as mshtml.IHTMLDocument2;
mshtml.IHTMLElement htmlElement = htmlDoc.activeElement as mshtml.IHTMLElement;
string activeTag = htmlElement.tagName.ToLower();
if (activeTag == "a")
{
throw new COMException("", Ok); // returns HRESULT = S_OK
}
}
}
if (lpmsg.message != WM_KEYDOWN)
{
// allow message
throw new COMException("", Error); // returns HRESULT = S_FALSE
}
if (GetAsyncKeyState(VK_CONTROL) < 0) //CONTROL KEY PRESSED
{
// disable all control keys except Ctrl-A Ctrl-C and Ctrl-V
lpmsg.lParam &= 0xFF; // get the virtual keycode
if ((lpmsg.lParam != 'A') || ((lpmsg.lParam != 'C')) || ((lpmsg.lParam != 'V')))
{
throw new COMException("", Ok); // returns HRESULT = S_OK
}
}
else if (GetAsyncKeyState(VK_ALT) < 0) //ALT KEY PRESSED
{
//disable all Alt key actions
throw new COMException("", Ok); // returns HRESULT = S_OK
}
else
{
// Ctrl key not pressed: allow message
mshtml.IHTMLDocument2 htmlDoc = axWebBrowser1.Document as mshtml.IHTMLDocument2;
mshtml.IHTMLElement htmlElement = htmlDoc.activeElement as mshtml.IHTMLElement;
string activeTag = htmlElement.tagName.ToLower();
string activeElementType = htmlElement.getAttribute("type", 0).ToString();
if (GetAsyncKeyState(VK_BACK) < 0)
{
if ((activeTag == "input" && activeElementType == "text") || activeTag == "textarea" || activeTag == "text" || activeTag == "password")
{
throw new COMException("", Error); // returns HRESULT = S_FALSE
}
else
{
throw new COMException("", Ok); // returns HRESULT = S_OK
}
}
}
// allow everything else
throw new COMException("", Error); // returns HRESULT = S_FALSE
}
