etc/clients
1
0
Fork 0
Code Issues Pull Requests Releases Activity

blog_list 에서 url 가져오는 방식 변경

Browse Source 
git-svn-id: svn://192.168.0.12/source@214 8346c931-da38-4b9b-9d4c-e48b93cbd075
This commit is contained in:
admin
2015-10-22 04:29:21 +00:00
parent 980f597433
commit c42d1c3d20
1 changed files with 7 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
CrawlerProcess/scrawler.cpp
View File

@@ -323,7 +323,7 @@ void SCrawler::saveFrameList(QWebFrame *frame)
{ {
QString str = "sp_blog_"; QString str = "sp_blog_";
QWebElement sub = Find(eleMain,"li","id",str+QString::number(i+1)); QWebElement sub = Find(eleMain,"li","id",str+QString::number(i+1));
QString strUrl = Find(sub,"a","class","url").toPlainText(); QString strUrl = Find(sub,"a","class","url").attribute("href");
if (strUrl.isEmpty()) if (strUrl.isEmpty())
{ {
//m_bLast = true; //m_bLast = true;
@@ -331,6 +331,9 @@ void SCrawler::saveFrameList(QWebFrame *frame)
break; break;
} }
strUrl = strUrl.replace("http://","");
strUrl = strUrl.replace("?Redirect=Log&logNo=","/", Qt::CaseInsensitive);
QStringList strList = strUrl.split('/'); QStringList strList = strUrl.split('/');
QString strBlogMe = "blog.me"; QString strBlogMe = "blog.me";
@@ -385,7 +388,7 @@ void SCrawler::saveFrameList(QWebFrame *frame)
{ {
QString str = "sp_blog_"; QString str = "sp_blog_";
QWebElement sub = Find(eleMain,"li","id",str+QString::number(i+1)); QWebElement sub = Find(eleMain,"li","id",str+QString::number(i+1));
QString strUrl = Find(sub,"a","class","url").toPlainText(); QString strUrl = Find(sub,"a","class","url").attribute("href");
if (strUrl.isEmpty()) if (strUrl.isEmpty())
{ {
@@ -393,7 +396,8 @@ void SCrawler::saveFrameList(QWebFrame *frame)
m_bUse = true; m_bUse = true;
break; break;
} }
strUrl = strUrl.replace("http://","");
strUrl = strUrl.replace("?Redirect=Log&logNo=","/", Qt::CaseInsensitive);
QStringList strList = strUrl.split('/'); QStringList strList = strUrl.split('/');
QString strBlogMe = "blog.me"; QString strBlogMe = "blog.me";