| 41 |
|
|
| 42 |
|
OBFingerprint* OBFingerprint::_pDefault; //static variable |
| 43 |
|
const unsigned int OBFingerprint::bitsperint = 8 * sizeof(unsigned int); |
| 44 |
+ |
int OBFingerprint::rubbish = 666; |
| 45 |
|
|
| 46 |
|
void OBFingerprint::SetBit(vector<unsigned int>& vec, unsigned int n) |
| 47 |
|
{ |
| 48 |
< |
vec[n/bitsperint] |= (1 << (n % bitsperint)); |
| 48 |
> |
vec[n/Getbitsperint()] |= (1 << (n % Getbitsperint())); |
| 49 |
|
} |
| 50 |
|
|
| 51 |
|
//////////////////////////////////////// |
| 52 |
|
void OBFingerprint::Fold(vector<unsigned int>& vec, unsigned int nbits) |
| 53 |
|
{ |
| 54 |
< |
while(vec.size()*bitsperint/2 >= nbits) |
| 54 |
> |
while(vec.size()*Getbitsperint()/2 >= nbits) |
| 55 |
|
vec.erase(transform(vec.begin(),vec.begin()+vec.size()/2, |
| 56 |
|
vec.begin()+vec.size()/2, vec.begin(), bit_or()), vec.end()); |
| 57 |
|
} |
| 120 |
|
///The positions of the candidate matching molecules in the original datafile are returned. |
| 121 |
|
|
| 122 |
|
vector<unsigned int> vecwords; |
| 123 |
< |
_pFP->GetFingerprint(pOb,vecwords, _index.header.words * OBFingerprint::bitsperint); |
| 123 |
> |
_pFP->GetFingerprint(pOb,vecwords, _index.header.words * OBFingerprint::Getbitsperint()); |
| 124 |
|
|
| 125 |
|
vector<unsigned int>candidates; //indices of matches from fingerprint screen |
| 126 |
|
candidates.reserve(MaxCandidates); |
| 161 |
|
strstream errorMsg; |
| 162 |
|
#endif |
| 163 |
|
errorMsg << "Stopped looking after " << i << " molecules." << endl; |
| 164 |
< |
obErrorLog.ThrowError(__FUNCTION__, errorMsg.str(), obInfo); |
| 164 |
> |
obErrorLog.ThrowError(__func__, errorMsg.str(), obWarning); |
| 165 |
|
} |
| 166 |
|
|
| 167 |
|
vector<unsigned int>::iterator itr; |
| 177 |
|
double MinTani) |
| 178 |
|
{ |
| 179 |
|
vector<unsigned int> targetfp; |
| 180 |
< |
_pFP->GetFingerprint(pOb,targetfp, _index.header.words * OBFingerprint::bitsperint); |
| 180 |
> |
_pFP->GetFingerprint(pOb,targetfp, _index.header.words * OBFingerprint::Getbitsperint()); |
| 181 |
|
|
| 182 |
|
unsigned int words = _index.header.words; |
| 183 |
|
unsigned int dataSize = _index.header.nEntries; |
| 212 |
|
return false; |
| 213 |
|
|
| 214 |
|
vector<unsigned int> targetfp; |
| 215 |
< |
_pFP->GetFingerprint(pOb,targetfp, _index.header.words * OBFingerprint::bitsperint); |
| 215 |
> |
_pFP->GetFingerprint(pOb,targetfp, _index.header.words * OBFingerprint::Getbitsperint()); |
| 216 |
|
|
| 217 |
|
unsigned int words = _index.header.words; |
| 218 |
|
unsigned int dataSize = _index.header.nEntries; |
| 231 |
|
} |
| 232 |
|
} |
| 233 |
|
return true; |
| 234 |
< |
}///////////////////////////////////////////////////////// |
| 234 |
> |
} |
| 235 |
> |
|
| 236 |
> |
///////////////////////////////////////////////////////// |
| 237 |
|
string FastSearch::ReadIndex(istream* pIndexstream) |
| 238 |
|
{ |
| 239 |
|
//Reads fs index from istream into member variables |
| 240 |
< |
// but first checks whether it is already loaded |
| 238 |
< |
FptIndexHeader headercopy = _index.header; |
| 239 |
< |
pIndexstream->read((char*)&(_index.header), sizeof(FptIndexHeader)); |
| 240 |
> |
_index.Read(pIndexstream); |
| 241 |
|
|
| 242 |
< |
if(memcmp(&headercopy,&(_index.header),sizeof(FptIndexHeader))) |
| 243 |
< |
{ |
| 244 |
< |
pIndexstream->seekg(_index.header.headerlength);//allows header length to be changed |
| 242 |
> |
_pFP = _index.CheckFP(); |
| 243 |
> |
if(!_pFP) |
| 244 |
> |
*(_index.header.datafilename) = '\0'; |
| 245 |
|
|
| 246 |
< |
unsigned int nwords = _index.header.nEntries * _index.header.words; |
| 247 |
< |
_index.fptdata.resize(nwords); |
| 248 |
< |
_index.seekdata.resize(_index.header.nEntries); |
| 246 |
> |
return _index.header.datafilename; //will be empty on error |
| 247 |
> |
} |
| 248 |
> |
|
| 249 |
> |
////////////////////////////////////////////////////////// |
| 250 |
> |
bool FptIndex::Read(istream* pIndexstream) |
| 251 |
> |
{ |
| 252 |
> |
pIndexstream->read((char*)&(header), sizeof(FptIndexHeader)); |
| 253 |
> |
pIndexstream->seekg(header.headerlength);//allows header length to be changed |
| 254 |
> |
if(pIndexstream->fail() || header.headerlength != sizeof(FptIndexHeader)) |
| 255 |
> |
{ |
| 256 |
> |
*(header.datafilename) = '\0'; |
| 257 |
> |
return false; |
| 258 |
> |
} |
| 259 |
|
|
| 260 |
< |
pIndexstream->read((char*)&(_index.fptdata[0]), sizeof(unsigned int) * nwords); |
| 261 |
< |
pIndexstream->read((char*)&(_index.seekdata[0]), sizeof(unsigned int) * _index.header.nEntries); |
| 262 |
< |
|
| 263 |
< |
if(pIndexstream->fail()) |
| 264 |
< |
*(_index.header.datafilename) = '\0'; |
| 265 |
< |
|
| 266 |
< |
string tempFP(_index.header.fpid); |
| 267 |
< |
_pFP = OBFingerprint::FindFingerprint(tempFP); |
| 268 |
< |
if(!_pFP) |
| 269 |
< |
{ |
| 270 |
< |
#ifdef HAVE_SSTREAM |
| 271 |
< |
stringstream errorMsg; |
| 272 |
< |
#else |
| 273 |
< |
strstream errorMsg; |
| 263 |
< |
#endif |
| 264 |
< |
errorMsg << "Index has Fingerprints of type '" << _index.header.fpid |
| 265 |
< |
<< " which is not currently loaded." << endl; |
| 266 |
< |
obErrorLog.ThrowError(__FUNCTION__, errorMsg.str(), obWarning); |
| 267 |
< |
*(_index.header.datafilename) = '\0'; |
| 268 |
< |
} |
| 260 |
> |
unsigned int nwords = header.nEntries * header.words; |
| 261 |
> |
fptdata.resize(nwords); |
| 262 |
> |
seekdata.resize(header.nEntries); |
| 263 |
> |
|
| 264 |
> |
pIndexstream->read((char*)&(fptdata[0]), sizeof(unsigned int) * nwords); |
| 265 |
> |
pIndexstream->read((char*)&(seekdata[0]), sizeof(unsigned int) * header.nEntries); |
| 266 |
> |
|
| 267 |
> |
if(pIndexstream->fail()) |
| 268 |
> |
{ |
| 269 |
> |
*(header.datafilename) = '\0'; |
| 270 |
> |
return false; |
| 271 |
> |
} |
| 272 |
> |
return true; |
| 273 |
> |
} |
| 274 |
|
|
| 275 |
+ |
////////////////////////////////////////////////////////// |
| 276 |
+ |
OBFingerprint* FptIndex::CheckFP() |
| 277 |
+ |
{ |
| 278 |
+ |
//check that fingerprint type is available |
| 279 |
+ |
string tempFP(header.fpid); |
| 280 |
+ |
OBFingerprint* pFP = OBFingerprint::FindFingerprint(tempFP); |
| 281 |
+ |
if(!pFP) |
| 282 |
+ |
{ |
| 283 |
+ |
#ifdef HAVE_SSTREAM |
| 284 |
+ |
stringstream errorMsg; |
| 285 |
+ |
#else |
| 286 |
+ |
strstream errorMsg; |
| 287 |
+ |
#endif |
| 288 |
+ |
errorMsg << "Index has Fingerprints of type '" << header.fpid |
| 289 |
+ |
<< " which is not currently loaded." << endl; |
| 290 |
+ |
obErrorLog.ThrowError(__func__, errorMsg.str(), obError); |
| 291 |
|
} |
| 292 |
< |
return _index.header.datafilename; |
| 292 |
> |
return pFP; //NULL if not available |
| 293 |
|
} |
| 294 |
|
|
| 295 |
|
//******************************************************* |
| 298 |
|
{ |
| 299 |
|
///Starts indexing process |
| 300 |
|
_indexstream = os; |
| 280 |
– |
_pFP = OBFingerprint::FindFingerprint(fpid); |
| 281 |
– |
if(!_pFP) |
| 282 |
– |
{ |
| 283 |
– |
#ifdef HAVE_SSTREAM |
| 284 |
– |
stringstream errorMsg; |
| 285 |
– |
#else |
| 286 |
– |
strstream errorMsg; |
| 287 |
– |
#endif |
| 288 |
– |
errorMsg << "Fingerprint type '" << fpid << "' not available" << endl; |
| 289 |
– |
obErrorLog.ThrowError(__FUNCTION__, errorMsg.str(), obWarning); |
| 290 |
– |
} |
| 291 |
– |
|
| 301 |
|
_nbits=FptBits; |
| 302 |
|
_pindex= new FptIndex; |
| 303 |
|
_pindex->header.headerlength = sizeof(FptIndexHeader); |
| 304 |
|
strncpy(_pindex->header.fpid,fpid.c_str(),15); |
| 305 |
|
strncpy(_pindex->header.datafilename, datafilename.c_str(), 255); |
| 306 |
+ |
|
| 307 |
+ |
//check that fingerprint type is available |
| 308 |
+ |
_pFP = _pindex->CheckFP(); |
| 309 |
+ |
} |
| 310 |
+ |
|
| 311 |
+ |
///////////////////////////////////////////////////////////// |
| 312 |
+ |
FastSearchIndexer::FastSearchIndexer(FptIndex* pindex, std::ostream* os) |
| 313 |
+ |
{ |
| 314 |
+ |
//Uses existing index |
| 315 |
+ |
_indexstream = os; |
| 316 |
+ |
_pindex = pindex; |
| 317 |
+ |
_nbits = _pindex->header.words * OBFingerprint::Getbitsperint(); |
| 318 |
+ |
|
| 319 |
+ |
//check that fingerprint type is available |
| 320 |
+ |
_pFP = _pindex->CheckFP(); |
| 321 |
|
} |
| 322 |
|
|
| 323 |
|
///////////////////////////////////////////////////////////// |
| 329 |
|
_indexstream->write((const char*)&_pindex->fptdata[0], _pindex->fptdata.size()*sizeof(unsigned int)); |
| 330 |
|
_indexstream->write((const char*)&_pindex->seekdata[0], _pindex->seekdata.size()*sizeof(unsigned int)); |
| 331 |
|
if(!_indexstream) |
| 332 |
< |
obErrorLog.ThrowError(__FUNCTION__, |
| 332 |
> |
obErrorLog.ThrowError(__func__, |
| 333 |
|
"Difficulty writing index", obWarning); |
| 334 |
|
delete _pindex; |
| 335 |
|
} |
| 350 |
|
_pindex->seekdata.push_back(seekpos); |
| 351 |
|
return true; |
| 352 |
|
} |
| 353 |
< |
obErrorLog.ThrowError(__FUNCTION__, "Failed to make a fingerprint", obWarning); |
| 353 |
> |
obErrorLog.ThrowError(__func__, "Failed to make a fingerprint", obWarning); |
| 354 |
|
return false; |
| 331 |
– |
|
| 355 |
|
} |
| 356 |
|
|
| 357 |
|
/*! |