Git-Mirrors/RetroShare
1
0
Fork 0
mirror of https://github.com/RetroShare/RetroShare.git synced 2024-12-28 08:59:37 -05:00
Code Issues Actions Packages Projects Releases Wiki Activity

Merge pull request #103 from AsamK/html_parsing

Browse Source 
Improve HTML parsing in libresapi
This commit is contained in:
Cyril Soler 2015-09-09 19:57:07 -04:00
commit 4a0539c035
2 changed files with 73 additions and 56 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

127
libresapi/src/api/ChatHandler.cpp
View File

@ -294,62 +294,7 @@ void ChatHandler::tick()
// remove html tags from chat message // remove html tags from chat message
// extract links form href // extract links form href
const std::string& in = msg.msg; getPlainText(msg.msg, m.msg, m.links);
std::string out;
bool ignore = false;
bool keep_link = false;
std::string last_six_chars;
Triple current_link;
std::vector<Triple> links;
for(unsigned int i = 0; i < in.size(); ++i)
{
if(keep_link && in[i] == '"')
{
keep_link = false;
current_link.second = out.size();
}
if(last_six_chars == "href=\"")
{
keep_link = true;
current_link.first = out.size();
}
// "rising edge" sets mode to ignore
if(in[i] == '<')
{
ignore = true;
}
if(!ignore || keep_link)
out += in[i];
// "falling edge" resets mode to keep
if(in[i] == '>')
ignore = false;
last_six_chars += in[i];
if(last_six_chars.size() > 6)
last_six_chars = last_six_chars.substr(1);
std::string a = "</a>";
if( current_link.first != -1
&& last_six_chars.size() >= a.size()
&& last_six_chars.substr(last_six_chars.size()-a.size()) == a)
{
// only allow these protocols
// we don't want for example javascript:alert(0)
std::string http = "http://";
std::string https = "https://";
std::string retroshare = "retroshare://";
if( out.substr(current_link.first, http.size()) == http
|| out.substr(current_link.first, https.size()) == https
|| out.substr(current_link.first, retroshare.size()) == retroshare)
{
current_link.third = out.size();
links.push_back(current_link);
}
current_link = Triple();
}
}
m.msg = out;
m.links = links;
m.recv_time = msg.recvTime; m.recv_time = msg.recvTime;
m.send_time = msg.sendTime; m.send_time = msg.sendTime;
@ -390,6 +335,76 @@ void ChatHandler::tick()
} }
} }
void ChatHandler::getPlainText(const std::string& in, std::string &out, std::vector<Triple> &links)
{
if (in.size() == 0)
return;
if (in[0] != '<' || in[in.size() - 1] != '>')
{
// It's a plain text message without HTML
out = in;
return;
}
bool ignore = false;
bool keep_link = false;
std::string last_six_chars;
unsigned int tag_start_index = 0;
Triple current_link;
for(unsigned int i = 0; i < in.size(); ++i)
{
if(keep_link && in[i] == '"')
{
keep_link = false;
current_link.second = out.size();
}
if(last_six_chars == "href=\"")
{
keep_link = true;
current_link.first = out.size();
}
// "rising edge" sets mode to ignore
if(in[i] == '<')
{
tag_start_index = i;
ignore = true;
}
if(!ignore || keep_link)
out += in[i];
// "falling edge" resets mode to keep
if(in[i] == '>') {
// leave ignore mode on, if it's a style tag
if (tag_start_index == 0 || tag_start_index + 6 > i || in.substr(tag_start_index, 6) != "<style")
ignore = false;
}
last_six_chars += in[i];
if(last_six_chars.size() > 6)
last_six_chars = last_six_chars.substr(1);
std::string a = "</a>";
if( current_link.first != -1
&& last_six_chars.size() >= a.size()
&& last_six_chars.substr(last_six_chars.size()-a.size()) == a)
{
// only allow these protocols
// we don't want for example javascript:alert(0)
std::string http = "http://";
std::string https = "https://";
std::string retroshare = "retroshare://";
if( out.substr(current_link.first, http.size()) == http
|| out.substr(current_link.first, https.size()) == https
|| out.substr(current_link.first, retroshare.size()) == retroshare)
{
current_link.third = out.size();
links.push_back(current_link);
}
current_link = Triple();
}
}
}
void ChatHandler::handleWildcard(Request &req, Response &resp) void ChatHandler::handleWildcard(Request &req, Response &resp)
{ {
RS_STACK_MUTEX(mMtx); /********** LOCKED **********/ RS_STACK_MUTEX(mMtx); /********** LOCKED **********/

2
libresapi/src/api/ChatHandler.h
View File

@ -102,6 +102,8 @@ private:
void handleSendStatus(Request& req, Response& resp); void handleSendStatus(Request& req, Response& resp);
void handleUnreadMsgs(Request& req, Response& resp); void handleUnreadMsgs(Request& req, Response& resp);
void getPlainText(const std::string& in, std::string &out, std::vector<Triple> &links);
StateTokenServer* mStateTokenServer; StateTokenServer* mStateTokenServer;
RsNotify* mNotify; RsNotify* mNotify;
RsMsgs* mRsMsgs; RsMsgs* mRsMsgs;