void vmsHttpFlvTrafficAnalyzer::FindDialogsByUniqueUrlWordsMatch(LPCSTR pszSwfContext, const HTTPDLGLIST &vDlgs, HTTPDLGLIST &vResult)
{
LOGFN ("vmsHttpFlvTrafficAnalyzer::FindDialogsByUniqueUrlWordsMatch");
vResult.clear ();
vector < vector <string> > vvUniqueWords;
vmsWinSockHttpTrafficAnalyzer::FindUniqueUrlWords (vDlgs, vvUniqueWords);
assert (vvUniqueWords.size () == vDlgs.size ());
for (size_t iDlg = 0; iDlg < vvUniqueWords.size (); iDlg++)
{
vector <string> &vWords = vvUniqueWords [iDlg];
if (vWords.empty ())
continue;
for (size_t iW = 0; iW < vWords.size (); iW++)
{
if (strstr (pszSwfContext, vWords [iW].c_str ()) != NULL)
{
vResult.push_back (vDlgs [iDlg]);
LOG ("Unique url word \"%s\" is containing in source text", vWords [iW].c_str ());
LOG ("|-dialog url-%s", vDlgs [iDlg]->strRequestUrl.c_str ());
break;
}
}
}
}
void vmsHttpFlvTrafficAnalyzer::FindDialogsByFileUrlMatch(LPCSTR pszSrcText, const HTTPDLGLIST &vDlgs, HTTPDLGLIST &vResult)
{
assert (pszSrcText != NULL && *pszSrcText != 0);
vResult.clear ();
int iDlgFound = -1;
for (size_t iDlg = 0; iDlg < vDlgs.size (); iDlg++)
{
string strFileUrl; vmsHttpHelper::ExtractFileUrl (vDlgs [iDlg]->strRequestUrl.c_str (),
strFileUrl);
if (strFileUrl.empty ())
continue;
if (strstr (pszSrcText, strFileUrl.c_str ()) != NULL)
{
if (iDlgFound == -1)
{
iDlgFound = iDlg;
}
else if (vDlgs [iDlgFound]->strRequestUrl != vDlgs [iDlg]->strRequestUrl)
{
iDlgFound = -1;
break;
}
}
}
if (iDlgFound != -1)
vResult.push_back (vDlgs [iDlgFound]);
}
void vmsLogHelper::LogHttpDlgList(LPCSTR pszListName, const HTTPDLGLIST &vDlgs)
{
LOG ("%s:", pszListName);
if (vDlgs.empty ())
{
LOGsnl ("|-(empty)");
return;
}
for (size_t i = 0; i < vDlgs.size (); i++)
LOG ("|-(%I64d bytes)-%s", (__int64)vDlgs [i]->nContentLength, vDlgs [i]->strRequestUrl.c_str ());
}
void vmsHttpFlvTrafficAnalyzer::FindDialogsByUniqueUrlWordsMatch2(LPCSTR pszSomeUrl, const HTTPDLGLIST &vDlgs, HTTPDLGLIST &vResult)
{
LOGFN ("vmsHttpFlvTrafficAnalyzer::FindDialogsByUniqueUrlWordsMatch2");
assert (pszSomeUrl != NULL);
vResult.clear ();
if (!pszSomeUrl)
return;
vector <string> vWords;
vmsUrlWords::ExtractWords2 (pszSomeUrl, vWords);
if (vWords.empty ())
return;
string strWords;
for (size_t i = 0; i < vWords.size (); i++)
{
strWords += vWords [i];
strWords += ' ';
}
FindDialogsByUniqueUrlWordsMatch (strWords.c_str (), vDlgs, vResult);
}
void vmsHttpFlvTrafficAnalyzer::FindDialogsByUrlExactMatch(LPCSTR pszSrcText, const HTTPDLGLIST &vDlgs, HTTPDLGLIST &vResult)
{
assert (pszSrcText != NULL && *pszSrcText != 0);
vResult.clear ();
string strTmp;
for (size_t i = 0; i < vDlgs.size (); i++)
{
HTTPDLG pDlg = vDlgs [i];
LPCSTR pszFlvUrl = pDlg->strRequestUrl.c_str ();
if (pszFlvUrl [pDlg->strRequestUrl.length ()-1] == '&')
{
strTmp.assign (pszFlvUrl, pDlg->strRequestUrl.length ()-1);
pszFlvUrl = strTmp.c_str ();
}
if (strstr (pszSrcText, pszFlvUrl))
vResult.push_back (pDlg);
}
}
void vmsHttpFlvTrafficAnalyzer::FindFlvDownloadInInBetweenDlgs(const vmsHttpTrafficCollector *pTraffic, HTTPDLG pWebPage, LPCSTR pszSwfUrl, LPCSTR pszFlashVars, LPCSTR pszSwfContext, HTTPDLGLIST vFlvs, FindFlvDownloadInInBetweenDlgs_Type enFindType)
{
LOGFN ("vmsHttpFlvTrafficAnalyzer::FindFlvDownloadInInBetweenDlgs");
if (!pszSwfUrl)
{
assert (pWebPage != NULL);
if (!pWebPage)
return;
pszSwfUrl = pWebPage->strRequestUrl.c_str ();
}
string strBody;
if (!pszSwfContext)
{
assert (pWebPage != NULL);
if (!pWebPage)
return;
strBody.assign ((LPSTR)&pWebPage->vbResponseBody [0], pWebPage->vbResponseBody.size ());
pszSwfContext = strBody.c_str ();
}
int nStartIndex = pWebPage ? pTraffic->getHttpDialogIndex (pWebPage) + 1 : 0;
HTTPDLGLIST vFlvs2;
DWORD adwCTs [] = {
vmsHttpTrafficCollector::HttpDialog::AMF,
vmsHttpTrafficCollector::HttpDialog::X_WWW_FORM_URL_ENCODED,
vmsHttpTrafficCollector::HttpDialog::XML,
vmsHttpTrafficCollector::HttpDialog::HTML,
};
for (int iCT = 0; iCT < sizeof (adwCTs) / sizeof (DWORD); iCT++)
{
HTTPDLGLIST vBetwDlgs;
pTraffic->GetHttpDialogs (nStartIndex, -1, adwCTs [iCT], vBetwDlgs);
HTTPDLGLIST vBetwDlgs2;
vector <HTTPDLGLIST> vBetwToFlvMap;
for (size_t iDlg = 0; iDlg < vBetwDlgs.size (); iDlg++)
{
HTTPDLG pDlg = vBetwDlgs [iDlg];
if (pDlg->vbResponseBody.empty ())
continue;
if (pDlg->enCT == vmsHttpTrafficCollector::HttpDialog::HTML &&
pDlg->vbResponseBody.size () > 1000)
continue;
string strBody;
if (pDlg->enCT == vmsHttpTrafficCollector::HttpDialog::AMF)
{
for (size_t i = 0; i < pDlg->vbResponseBody.size (); i++)
{
if (pDlg->vbResponseBody [i] >= ' ')
strBody += pDlg->vbResponseBody [i];
}
}
else if (pDlg->enCT == vmsHttpTrafficCollector::HttpDialog::X_WWW_FORM_URL_ENCODED)
{
string str; str.assign ((LPSTR)&pDlg->vbResponseBody [0], pDlg->vbResponseBody.size ());
vmsHttpHelper::DecodeUrl (str.c_str (), strBody);
}
else
{
strBody.assign ((LPSTR)&pDlg->vbResponseBody [0], pDlg->vbResponseBody.size ());
}
if (enFindType == FDT_EXACTURLMATCH)
FindDialogsByUrlExactMatch (strBody.c_str (), vFlvs, vFlvs2);
else if (enFindType == FDT_FILEURLMATCH)
FindDialogsByFileUrlMatch (strBody.c_str (), vFlvs, vFlvs2);
else if (enFindType == FDT_BYUNIQUEWORDS)
FindDialogsByUniqueUrlWordsMatch (strBody.c_str (), vFlvs, vFlvs2);
else
return;
if (!vFlvs2.empty ())
{
vBetwDlgs2.push_back (pDlg);
vBetwToFlvMap.push_back (vFlvs2);
}
}
if (vBetwDlgs2.empty ())
continue;
LOGHTTPDLGLIST ("List of dialogs with FLVs", vBetwDlgs2);
LOGsnl ("Found dialogs with FLVs. Find their source.");
HTTPDLGLIST vBetwDlgsFoundInSwfContext;
FindDialogsByUrlExactMatch (pszSwfContext, vBetwDlgs2, vBetwDlgsFoundInSwfContext);
LOGHTTPDLGLIST ("List of these dialogs belonging to source page (exact match)", vBetwDlgsFoundInSwfContext);
if (vBetwDlgsFoundInSwfContext.empty ())
{
FindDialogsByFileUrlMatch (pszSwfContext, vBetwDlgs, vBetwDlgsFoundInSwfContext);
LOGHTTPDLGLIST ("List of these dialogs belonging to source page (file url match)", vBetwDlgsFoundInSwfContext);
}
if (vBetwDlgsFoundInSwfContext.empty ())
{
FindDialogsByUniqueUrlWordsMatch (pszSwfContext, vBetwDlgs, vBetwDlgsFoundInSwfContext);
LOGHTTPDLGLIST ("List of these dialogs belonging to source page (unique url word match)", vBetwDlgsFoundInSwfContext);
}
if (vBetwDlgsFoundInSwfContext.empty ())
{
FindDialogsByUniqueUrlWordsMatch2 (pszSwfUrl, vBetwDlgs, vBetwDlgsFoundInSwfContext);
LOGHTTPDLGLIST ("List of these dialogs belonging to source page (unique url words2 match)", vBetwDlgsFoundInSwfContext);
}
if (!vBetwDlgsFoundInSwfContext.empty ())
{
wstring wstrTitle;
if (pszFlashVars)
ExtractTitleFromFlashVars (pWebPage, pszFlashVars, wstrTitle);
if (wstrTitle.empty () && pWebPage)
ExtractTitleFromHtml (pWebPage, wstrTitle);
for (size_t i = 0; i < vBetwDlgsFoundInSwfContext.size (); i++)
{
for (size_t j = 0; j < vBetwDlgs2.size (); j++)
{
if (vBetwDlgs2 [j] == vBetwDlgsFoundInSwfContext [i])
{
HTTPDLG pBetwDlg = vBetwDlgs2 [j];
for (size_t k = 0; k < vBetwToFlvMap [j].size (); k++)
{
HTTPDLG pFlvDlg = vBetwToFlvMap [j][k];
wstring wstrTitle2;
if (pBetwDlg->enCT == vmsHttpTrafficCollector::HttpDialog::X_WWW_FORM_URL_ENCODED)
ExtractTitleFromXWwwFormUrlEncoded (pBetwDlg, wstrTitle2);
else if (pBetwDlg->enCT == vmsHttpTrafficCollector::HttpDialog::XML)
ExtractTitleFromXml (pBetwDlg, wstrTitle2, pFlvDlg);
if (wstrTitle2.empty ())
wstrTitle2 = wstrTitle;
AddFlvDownload (pWebPage, pFlvDlg, wstrTitle2.c_str ());
}
break;
}
}
}
return;
}
}
}
