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

blog 크롤링 안되는 부분 추가

Browse Source 
git-svn-id: svn://192.168.0.12/source@240 8346c931-da38-4b9b-9d4c-e48b93cbd075
This commit is contained in:
admin
2016-01-19 06:48:49 +00:00
parent f25230c652
commit 73ede17add
1 changed files with 6 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
CrawlerProcess/scrawler.cpp
View File

@@ -601,6 +601,10 @@ bool SCrawler::saveFrameUrl(QWebFrame *frame)
{ {
title = Find(frame->documentElement(), "div", "class", "se_textView"); title = Find(frame->documentElement(), "div", "class", "se_textView");
} }
if(title.isNull())
{
title = Find(frame->documentElement(), "h3", "class", "se_textarea");
}
if (title.toPlainText().isEmpty()==false) if (title.toPlainText().isEmpty()==false)
{ {
str[E_DATA_TITLE] = title.toPlainText(); str[E_DATA_TITLE] = title.toPlainText();
@@ -631,6 +635,8 @@ bool SCrawler::saveFrameUrl(QWebFrame *frame)
QWebElement body = post.findFirst("div[class^='post-view pcol2 _param(1)']"); QWebElement body = post.findFirst("div[class^='post-view pcol2 _param(1)']");
if(body.isNull()) if(body.isNull())
body = post.findFirst("div[class*='pcol2 _param(1)']"); body = post.findFirst("div[class*='pcol2 _param(1)']");
if(body.isNull())
body = Find(post, "class", "se_component_wrap sect_dsc __se_component_area");
if (body.toPlainText().isEmpty()==false) if (body.toPlainText().isEmpty()==false)
{ {
str[E_DATA_DATA] = body.toPlainText(); str[E_DATA_DATA] = body.toPlainText();