// parse with info from
// http://www.wotsit.org/getfile.asp?file=wword8&sc=230027800
bool
OleEndAnalyzer::tryFIB(AnalysisResult& ar, InputStream* in) {
const char* d;
int32_t size = 426;
int32_t nread = in->read(d, size, size);
in->reset(0);
if (nread != size
|| (unsigned char)d[0] != 0xec || (unsigned char)d[1] != 0xa5) {
return false;
}
bool complex = (d[10] & 4) == 4;
if (complex) return false;
int32_t fcMin = readLittleEndianInt32(d+24);
int32_t fcMac = readLittleEndianInt32(d+28);
// for some reason we sometimes need to add 512 here. No clue why.
// if the first 512 bytes are 0 we do this
size = fcMin+512;
nread = in->read(d, size, size);
in->reset(0);
if (nread != size) {
return false;
}
int i;
for (i=0; i<512 && d[i+fcMin] == 0; i++) ;
if (i == 512) {
fcMin += 512;
fcMac += 512;
}
size = fcMac;
nread = in->read(d, size, size);
in->reset(0);
if (nread != size) {
return false;
}
wordtext.reset();
for (int32_t dp = fcMin; dp < fcMac; dp += size) {
size = fcMac-dp;
if (size > 512) size = 512;
wordtext.addText(d+dp, size);
}
wordtext.cleanText();
ar.addText(wordtext.text(), (int32_t)wordtext.length());
wordtext.reset();
return true;
}
signed char
IFilterEndAnalyzer::analyze(AnalysisResult& idx, InputStream *in) {
const string& filename = idx.fileName();
int p = filename.find_last_of('.');
if (p < 0 || extensions.find(filename.substr(p)) == extensions.end()) {
return -1;
}
string filepath;
bool fileisondisk = checkForFile(idx.depth(), filename);
if (fileisondisk) {
filepath = filename;
} else {
int p = filename.find_last_of(".");
if ( p > 0 ){
string ext = filename.substr(p).c_str();
strlwr((char*)ext.c_str());
p = ext.find_first_not_of("._abcdefghijklmnopqrstuvwxyz0123456789");
if ( p >= 0 )
filepath = writeToTempFile(in, "");
else
filepath = writeToTempFile(in, ext.c_str());
}else
filepath = writeToTempFile(in, "");
}
if (filepath.length() > 0) {
IFilter* filter = NULL;
void* pvfilter=NULL;
wchar_t tmp[MAX_PATH];
_cpycharToWide(tmp,filepath.c_str(),MAX_PATH);
HRESULT hr = LoadIFilter(tmp,NULL,&pvfilter);
if (hr == S_OK) {
filter = (IFilter*)pvfilter;
ULONG __i=0;
hr = filter->Init(IFILTER_INIT_APPLY_INDEX_ATTRIBUTES,0,NULL,&__i);
if (FAILED( hr )) {
if (!fileisondisk)
unlink(filepath.c_str());
return -1;
}
const int sbBufferLen = 1024;
wchar_t sbBuffer[sbBufferLen];
STAT_CHUNK ps;
hr = filter->GetChunk(&ps);
while ( SUCCEEDED(hr) ) {
if (ps.flags == CHUNK_TEXT) {
int resultText = 0;
while ( resultText >= 0 ) {
ULONG sizeBuffer=sbBufferLen;
resultText = filter->GetText(&sizeBuffer, sbBuffer);
if (sizeBuffer > 0 ) {
string str = wchartoutf8(sbBuffer,sbBuffer+sizeBuffer);
idx.addText(str.c_str(),str.length());
}
}
} else if ( ps.flags == CHUNK_VALUE ) {
PROPVARIANT *pVar;
while ( SUCCEEDED( hr = filter->GetValue( &pVar ) ) ) {
//printf("propid: %d\nkind:%d\n",ps.attribute.psProperty.propid,ps.attribute.psProperty.ulKind);
if ( ps.attribute.psProperty.propid == 2 &&
ps.attribute.psProperty.ulKind == 1 &&
pVar->vt == VT_LPWSTR ) {
string str = wchartoutf8(pVar->pwszVal,pVar->pwszVal+wcslen(pVar->pwszVal));
idx.addValue("title", str );
}
PropVariantClear( pVar );
CoTaskMemFree( pVar );
}
} else {
printf("other flag %d\n",ps.flags);
}
hr = filter->GetChunk(&ps);
}
filter->Release();
if (!fileisondisk)
unlink(filepath.c_str());
return 0;
}
DWORD dw = GetLastError();
if ( dw != 0 ) {
LPVOID lpMsgBuf;
FormatMessage(
FORMAT_MESSAGE_ALLOCATE_BUFFER |
FORMAT_MESSAGE_FROM_SYSTEM,
NULL,
dw,
MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT),
(LPTSTR) &lpMsgBuf,
0, NULL );
wprintf(L"%s\n", lpMsgBuf);
LocalFree(lpMsgBuf);
}
}
if (!fileisondisk && filepath.length()>0) {
unlink(filepath.c_str());
}
return -1;
}