void vmsHttpFlvTrafficAnalyzer::ExtractTitleFromHtml(const vmsHttpTrafficCollector::HttpDialog *pHtmlDlg, wstring &wstrTitle)
{
wstrTitle = L"";
UINT codePage = 0;
ExtractCodePageFromHtml (pHtmlDlg, codePage);
if (!codePage)
codePage = pHtmlDlg->codePage;
assert (pHtmlDlg != NULL);
if (!pHtmlDlg)
return;
if (pHtmlDlg->vbResponseBody.empty ())
return;
LPCSTR psz = strstrni ((LPCSTR)&pHtmlDlg->vbResponseBody [0], "<title>", pHtmlDlg->vbResponseBody.size ());
if (psz)
{
psz += 7;
LPCSTR pszEnd = (LPCSTR)&pHtmlDlg->vbResponseBody [0] + pHtmlDlg->vbResponseBody.size ();
string strTitle;
while (psz < pszEnd)
{
if (*psz != '<')
strTitle += *psz++;
else
break;
}
if (*psz == '<' && !strTitle.empty ())
{
vmsCharsets::DecodeString (strTitle.c_str (),
codePage ? codePage : CP_UTF8, wstrTitle);
while (wstrTitle.empty () == false && wstrTitle [0] <= ' ')
wstrTitle.erase (wstrTitle.begin ());
while (wstrTitle.empty () == false && wstrTitle [wstrTitle.length ()-1] <= ' ')
wstrTitle.erase (wstrTitle.end ()-1);
for (int i = 0; i < wstrTitle.length (); i++)
{
if (wstrTitle [i] < ' ')
wstrTitle [i] = ' ';
if (i && wstrTitle [i] == ' ' && wstrTitle [i-1] == ' ')
wstrTitle.erase (wstrTitle.begin () + i--);
}
}
}
}
void vmsHttpFlvTrafficAnalyzer::ExtractTitleFromXWwwFormUrlEncoded(const vmsHttpTrafficCollector::HttpDialog *pDlg, wstring &wstrTitle)
{
wstrTitle = L"";
assert (pDlg != NULL);
if (!pDlg)
return;
assert (pDlg->enCT == vmsHttpTrafficCollector::HttpDialog::X_WWW_FORM_URL_ENCODED);
if (pDlg->vbResponseBody.empty ())
return;
extern LPCSTR strstrni (LPCSTR pszSrc, LPCSTR pszSrch, int lenSrc);
LPCSTR psz = strstrni ((LPCSTR)&pDlg->vbResponseBody [0], "&title=", pDlg->vbResponseBody.size ());
if (!psz)
return;
psz += 7;
int iLenAvail = (int)pDlg->vbResponseBody.size () - (psz - (LPCSTR)&pDlg->vbResponseBody [0]);
string strTitle;
while (iLenAvail > 0 && *psz != '&')
{
strTitle += *psz++;
iLenAvail--;
}
if (strTitle.empty ())
return;
for (int i = 0; i < strTitle.length (); i++)
{
if (strTitle [i] == '+')
strTitle [i] = ' ';
}
string strDecodedTitle;
vmsHttpHelper::DecodeUrl (strTitle.c_str (), strDecodedTitle);
int n = MultiByteToWideChar (CP_UTF8, 0, strDecodedTitle.c_str (), -1, NULL, 0);
assert (n != 0);
LPWSTR pwsz = new WCHAR [n+1]; *pwsz = 0;
MultiByteToWideChar (CP_UTF8, 0, strDecodedTitle.c_str (), -1, pwsz, n);
wstrTitle = pwsz;
delete [] pwsz;
}
void vmsHttpFlvTrafficAnalyzer::ExtractTitleFromHtml(const vmsHttpTrafficCollector::HttpDialog *pHtmlDlg, wstring &wstrTitle)
{
wstrTitle = L"";
assert (pHtmlDlg != NULL);
if (!pHtmlDlg)
return;
if (pHtmlDlg->vbResponseBody.empty ())
return;
extern LPCSTR strstrni (LPCSTR pszSrc, LPCSTR pszSrch, int lenSrc);
LPCSTR psz = strstrni ((LPCSTR)&pHtmlDlg->vbResponseBody [0], "<title>", pHtmlDlg->vbResponseBody.size ());
if (psz)
{
psz += 7;
LPCSTR pszEnd = (LPCSTR)&pHtmlDlg->vbResponseBody [0] + pHtmlDlg->vbResponseBody.size ();
string strTitle;
while (psz < pszEnd)
{
if (*psz != '<')
strTitle += *psz++;
else
break;
}
if (*psz == '<')
{
const vmsHttpParser::HdrField *pFld = pHtmlDlg->pHttpResponse->FieldByName ("Content-Type");
LPCSTR pszCharset = pFld ? strstr (pFld->strValue.c_str (), "charset") : NULL;
if (pszCharset)
{
pszCharset += 7;
while (*pszCharset == ' ')
pszCharset++;
if (*pszCharset == '=')
{
pszCharset++;
while (*pszCharset == ' ')
pszCharset++;
string strCharset;
while (*pszCharset != ' ' && *pszCharset != 0)
strCharset += *pszCharset++;
UINT nCP = vmsCharsets::GetCpIdFromName (strCharset.c_str ());
if (nCP)
{
int n = MultiByteToWideChar (nCP, 0, strTitle.c_str (), -1, NULL, 0);
assert (n != 0);
LPWSTR pwsz = new WCHAR [n+1];
MultiByteToWideChar (nCP, 0, strTitle.c_str (), -1, pwsz, n);
wstrTitle = pwsz;
delete [] pwsz;
}
}
}
if (wstrTitle.empty ())
{
USES_CONVERSION;
wstrTitle = A2W (strTitle.c_str ());
}
while (wstrTitle.empty () == false && wstrTitle [0] <= ' ')
wstrTitle.erase (wstrTitle.begin ());
while (wstrTitle.empty () == false && wstrTitle [wstrTitle.length ()-1] <= ' ')
wstrTitle.erase (wstrTitle.end ()-1);
for (int i = 0; i < wstrTitle.length (); i++)
{
if (wstrTitle [i] < ' ')
wstrTitle [i] = ' ';
if (i && wstrTitle [i] == ' ' && wstrTitle [i-1] == ' ')
wstrTitle.erase (wstrTitle.begin () + i--);
}
}
}
}
void vmsHttpFlvTrafficAnalyzer::ExtractTitleFromXml(const vmsHttpTrafficCollector::HttpDialog *pDlg, wstring &wstrTitle, const vmsHttpTrafficCollector::HttpDialog* pFlvDlg)
{
extern LPCSTR strstrni (LPCSTR pszSrc, LPCSTR pszSrch, int lenSrc);
extern LPCSTR strstrn (LPCSTR pszSrc, LPCSTR pszSrch, int lenSrc);
wstrTitle = L"";
assert (pDlg != NULL);
if (!pDlg)
return;
assert (pDlg->enCT == vmsHttpTrafficCollector::HttpDialog::XML);
if (pDlg->vbResponseBody.empty ())
return;
LPCSTR pszXml = (LPCSTR)&pDlg->vbResponseBody [0];
int iXmlLen = pDlg->vbResponseBody.size ();
LPCSTR pszXml2 = strstrni (pszXml, "encoding=\"", iXmlLen);
if (!pszXml2)
return;
pszXml2 += 10;
iXmlLen -= pszXml2 - pszXml;
pszXml = pszXml2;
if (iXmlLen < 5)
return;
if (strnicmp (pszXml, "utf-8", 5))
return;
USES_CONVERSION;
tstring tstrRequestUrl = A2CT (pFlvDlg->strRequestUrl.c_str ());
string strUrl = vmsXmlHelper::toUtf8noEncode (tstrRequestUrl);
pszXml2 = strstrn (pszXml, strUrl.c_str (), iXmlLen);
if (!pszXml2)
{
strUrl = vmsXmlHelper::toUtf8 (tstrRequestUrl);
pszXml2 = strstrn (pszXml, strUrl.c_str (), iXmlLen);
if (!pszXml2)
{
strUrl = pFlvDlg->strRequestUrl;
pszXml2 = strstrn (pszXml, strUrl.c_str (), iXmlLen);
}
}
LPCSTR pszVideoSectionTag = pszXml;
if (pszXml2)
{
bool bAsAttr = pszXml2 [-1] != '>';
if (bAsAttr)
{
while (pszXml2 > pszXml && *pszXml2 != '"')
pszXml2--;
pszXml2--;
if (pszXml2 > pszXml)
{
bool bInQ = false;
while (pszXml2 > pszXml)
{
if (*pszXml2 == '"')
bInQ = !bInQ;
else if (*pszXml2 == '<' && bInQ == false)
break;
pszXml2--;
}
if (pszXml2 > pszXml)
pszVideoSectionTag = pszXml2;
}
}
else
{
while (pszXml2 > pszXml && *pszXml2 != '<')
pszXml2--;
pszXml2--;
if (pszXml2 > pszXml)
{
bool bInT = false;
while (pszXml2 > pszXml)
{
if (*pszXml2 == '<')
{
if (bInT)
{
bInT = false;
}
else
{
if (pszXml2 [1] == '/')
{
bInT = true;
}
else
{
break;
}
}
}
pszXml2--;
}
if (pszXml2 > pszXml)
pszVideoSectionTag = pszXml2;
}
}
}
int iXmlLen2 = iXmlLen - (pszVideoSectionTag - pszXml);
LPCSTR apszTags [] = {"title=\"", "<title>", "name=\"", "<name>"};
int iTag;
LPCSTR pszTitle = NULL;
do
{
for (iTag = 0; iTag < sizeof (apszTags)/sizeof (LPCSTR) && pszTitle == NULL; iTag++)
pszTitle = strstrni (pszVideoSectionTag, apszTags [iTag], iXmlLen2);
if (!pszTitle)
{
if (pszVideoSectionTag != pszXml)
{
pszVideoSectionTag = pszXml;
iXmlLen2 = iXmlLen;
}
else
{
return;
}
}
}
while (pszTitle == NULL);
pszTitle += strlen (apszTags [iTag]);
LPCSTR pszTitleE = pszTitle;
int iXmlLen3 = iXmlLen - (pszTitle - pszXml);
char chEnd = pszTitle [-1] == '>' ? '<' : '"';
while (*pszTitleE != chEnd && iXmlLen3)
{
pszTitleE++;
iXmlLen3--;
}
if (!iXmlLen3)
return;
string strTitle; strTitle.assign (pszTitle, pszTitleE-pszTitle);
int n = MultiByteToWideChar (CP_UTF8, 0, strTitle.c_str (), -1, NULL, 0);
assert (n != 0);
LPWSTR pwsz = new WCHAR [n+1]; *pwsz = 0;
MultiByteToWideChar (CP_UTF8, 0, strTitle.c_str (), -1, pwsz, n);
wstrTitle = pwsz;
delete [] pwsz;
}
