signed char GZipEndAnalyzer::analyze(AnalysisResult& idx, InputStream* in) { if(!in) return -1; GZipInputStream stream(in); // since this is gzip file, its likely that it contains a tar file const char* start = 0; int32_t nread = stream.read(start, 1024, 0); if (nread < -1) { printf("Error reading gzip: %s\n", stream.error()); return -2; } idx.addValue(factory->typeField, "http://www.semanticdesktop.org/ontologies/2007/03/22/nfo#Archive"); stream.reset(0); if (TarInputStream::checkHeader(start, nread)) { return TarEndAnalyzer::staticAnalyze(idx, &stream); } else { std::string file = idx.fileName(); size_t len = file.length(); if (len > 3 && file.substr(len-3) == ".gz") { file = file.substr(0, len-3); } signed char r = idx.indexChild(file, idx.mTime(), &stream); idx.finishIndexChild(); return r; } }
signed char LzmaEndAnalyzer::analyze(AnalysisResult& idx, InputStream* in) { if(!in) return -1; LZMAInputStream stream(in); // since this is lzma file, its likely that it contains a tar file const char* start = 0; int32_t nread = stream.read(start, 1024, 0); if (nread < -1) { fprintf(stderr, "Error reading lzma: %s\n", stream.error()); return -2; } idx.addValue(factory->typeField, "http://www.semanticdesktop.org/ontologies/2007/03/22/nfo#Archive"); stream.reset(0); if (TarInputStream::checkHeader(start, nread)) { return TarEndAnalyzer::staticAnalyze(idx, &stream); } else { std::string name = idx.fileName(); string::size_type len = name.length(); if (len > 5 && name.substr(len-5)==".lzma") { name = name.substr(0, len-5); } return idx.indexChild(name, idx.mTime(), &stream); } }
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; }