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::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;
}
}
}
HRESULT vmsHttpFlvTrafficAnalyzer4::FindFlvDownloads(LPCSTR pszHtmlPageUrl, const vmsHttpTrafficCollector *pTraffic)
{
LOGFN ("vmsHttpFlvTrafficAnalyzer4::FindFlvDownloads");
m_vDownloads.clear ();
const vmsHttpTrafficCollector::HttpDialog* pWebPage =
pTraffic->FindDialogByRequestUrl (pszHtmlPageUrl, vmsHttpTrafficCollector::HttpDialog::HTML);
if (!pWebPage)
return E_INVALIDARG;
if (pWebPage->vbResponseBody.empty ())
return S_FALSE;
string strPageText;
strPageText.assign ((LPSTR)&pWebPage->vbResponseBody [0], pWebPage->vbResponseBody.size ());
int nStartIndex = pTraffic->getHttpDialogIndex (pWebPage) + 1;
HTTPDLGLIST vFlvs;
pTraffic->GetHttpDialogs (nStartIndex, -1,
vmsHttpTrafficCollector::HttpDialog::FLV, vFlvs);
HTTPDLGLIST vFlvs2;
FindDialogsByUrlExactMatch (strPageText.c_str (), vFlvs, vFlvs2);
if (vFlvs2.empty ())
FindDialogsByFileUrlMatch (strPageText.c_str (), vFlvs, vFlvs2);
if (!vFlvs2.empty ())
{
wstring wstrTitle;
ExtractTitleFromHtml (pWebPage, wstrTitle);
for (size_t i = 0; i < vFlvs2.size (); i++)
AddFlvDownload (pWebPage, vFlvs2 [i], wstrTitle.c_str ());
return S_OK;
}
string strBody;
strBody.assign ((LPSTR)&pWebPage->vbResponseBody [0], pWebPage->vbResponseBody.size ());
FindFlvDownloadInInBetweenDlgs (pTraffic, pWebPage, NULL, NULL, strBody.c_str (), vFlvs, FDT_EXACTURLMATCH);
if (!m_vDownloads.empty ())
return S_OK;
FindFlvDownloadInInBetweenDlgs (pTraffic, pWebPage, NULL, NULL, strBody.c_str (), vFlvs, FDT_FILEURLMATCH);
if (!m_vDownloads.empty ())
return S_OK;
FindDialogsByUniqueUrlWordsMatch (strBody.c_str (), vFlvs, vFlvs2);
if (vFlvs2.empty ())
FindDialogsByUniqueUrlWordsMatch2 (pWebPage->strRequestUrl.c_str (), vFlvs, vFlvs2);
if (!vFlvs2.empty ())
{
wstring wstrTitle;
ExtractTitleFromHtml (pWebPage, wstrTitle);
for (size_t i = 0; i < vFlvs2.size (); i++)
AddFlvDownload (pWebPage, vFlvs2 [i], wstrTitle.c_str ());
return S_OK;
}
FindFlvDownloadInInBetweenDlgs (pTraffic, pWebPage, NULL, NULL, strBody.c_str (), vFlvs, FDT_BYUNIQUEWORDS);
if (!m_vDownloads.empty ())
return S_OK;
return m_vDownloads.empty () ? S_FALSE : S_OK;
}
