bool FastSearchFormat::WriteChemObject(OBConversion* pConv)
{
//Prepares or updates an index file. Called for each molecule indexed
bool update = pConv->IsOption("u")!=NULL;
static ostream* pOs;
static bool NewOstreamUsed;
if(fsi==NULL)
{
//First pass sets up FastSearchIndexer object
pOs = pConv->GetOutStream();// with named index it is already open
NewOstreamUsed=false;
string mes("prepare an");
if(update)
mes = "update the";
clog << "This will " << mes << " index of " << pConv->GetInFilename()
<< " and may take some time..." << flush;
if(!pConv->IsLastFile())
{
obErrorLog.ThrowError(__FUNCTION__,
"There should not be multiple input files. A .fs file is an index of a single datafile.",
obError);
return false;
}
std::string auditMsg = "OpenBabel::Write fastsearch index ";
std::string description(Description());
auditMsg += description.substr( 0, description.find('\n') );
obErrorLog.ThrowError(__FUNCTION__,auditMsg,obAuditMsg);
FptIndex* pidx=NULL; //used with update
//if(pOs==&cout) did not work with GUI
if(!dynamic_cast<ofstream*>(pOs))
{
//No index filename specified
//Derive index name from datafile name
string indexname=pConv->GetInFilename();
string::size_type pos=indexname.find_last_of('.');
if(pos!=string::npos)
indexname.erase(pos);
indexname += ".fs";
bool idxok=true;
if(update)
{
LastSeekpos = 0;
//Read in existing index
idxok=false;
ifstream ifs(indexname.c_str(),ifstream::binary);
if(ifs.good())
{
pidx = new FptIndex;
idxok = pidx->Read(&ifs);
}
}//ifs closed here
pOs = new ofstream(indexname.c_str(),ofstream::binary);
if(!pOs->good() || !idxok)
{
stringstream errorMsg;
errorMsg << "Trouble opening or reading " << indexname << endl;
obErrorLog.ThrowError(__FUNCTION__, errorMsg.str(), obError);
static_cast<ofstream *>(pOs)->close(); // close the file before quitting
delete pOs;
delete pidx; // remove possible memory leak
return false;
}
NewOstreamUsed=true;
}
else // not cout
{
if(update)
{
obErrorLog.ThrowError(__FUNCTION__,
"Currently, updating is only done on index files that"
"have the same name as the datafile.\n"
"Do not specify an output file; use the form:\n"
" babel datafile.xxx -ofs -xu", obError);
return false;
}
}
int nbits = 0;
const char* p = pConv->IsOption("N");
if(p)
nbits = atoi(p);
string fpid; //fingerprint type
p=pConv->IsOption("f");
if(p)
fpid=p;
//Prepare name without path
string datafilename = pConv->GetInFilename();
if(datafilename.empty())
{
obErrorLog.ThrowError(__FUNCTION__, "No datafile!", obError);
return false;
}
string::size_type pos = datafilename.find_last_of("/\\");
if(pos!=string::npos)
datafilename=datafilename.substr(pos+1);
nmols = pConv->NumInputObjects();
if(nmols>0)
clog << "\nIt contains " << nmols << " molecules" << flush;
if(nmols>500000)
{
istream* is = pConv->GetInStream();
streampos origpos = is->tellg();
is->seekg(0,ios_base::end);
long long filesize = is->tellg();
if(filesize > 4294967295u)
{
obErrorLog.ThrowError(__FUNCTION__, "The datafile must not be larger than 4GB", obError);
return false;
}
is->seekg(origpos);
}
sw.Start();
if(update)
{
fsi = new FastSearchIndexer(pidx, pOs, nmols);//using existing index
//Seek to position in datafile of last of old objects
LastSeekpos = *(pidx->seekdata.end()-1);
pConv->GetInStream()->seekg(LastSeekpos);
}
else
fsi = new FastSearchIndexer(datafilename, pOs, fpid, nbits, nmols);
obErrorLog.StopLogging();
}
//All passes provide an object for indexing
OBBase* pOb = pConv->GetChemObject();
OBMol* pmol = dynamic_cast<OBMol*> (pOb);
if(pmol)
pmol->ConvertDativeBonds();//use standard form for dative bonds
streampos seekpos = pConv->GetInPos();
if(!update || seekpos>LastSeekpos)
{
fsi->Add(pOb, seekpos );
if(pConv->GetOutputIndex()==400 && nmols>1000)
{
clog << " Estimated completion time ";
double secs = sw.Elapsed() * nmols / 400; //
streamsize op = clog.precision(0);
if(secs>150)
clog << secs/60 << " minutes" << endl;
else
clog << secs << " seconds" << endl;
clog.precision(op);
}
}
else
//Don't index old objects during update. Don't increment pConv->Index.
pConv->SetOutputIndex(pConv->GetOutputIndex()-1);
if(pConv->IsLast())
{
//Last pass
delete fsi; //saves index file
if(NewOstreamUsed)
delete pOs;
//return to starting conditions
fsi=NULL;
obErrorLog.StartLogging();
double secs = sw.Elapsed();
if(secs>150)
clog << "\n It took " << secs/60 << " minutes" << endl;
else
clog << "\n It took " << secs << " seconds" << endl;
}
delete pOb;
return true;
}
int main(int argc,char **argv)
{
char *program_name= argv[0];
int c;
int steps = 2500;
double crit = 1e-6;
bool sd = false;
bool cut = false;
bool newton = false;
bool hydrogens = false;
double rvdw = 6.0;
double rele = 10.0;
int freq = 10;
string basename, filename = "", option, option2, ff = "MMFF94";
char *oext;
OBConversion conv;
OBFormat *format_out = conv.FindFormat("pdb"); // default output format
if (argc < 2) {
cout << "Usage: obminimize [options] <filename>" << endl;
cout << endl;
cout << "options: description:" << endl;
cout << endl;
cout << " -c crit set convergence criteria (default=1e-6)" << endl;
cout << endl;
cout << " -cg use conjugate gradients algorithm (default)" << endl;
cout << endl;
cout << " -sd use steepest descent algorithm" << endl;
cout << endl;
cout << " -newton use Newton2Num linesearch (default=Simple)" << endl;
cout << endl;
cout << " -ff ffid select a forcefield:" << endl;
cout << endl;
cout << " -h add hydrogen atoms" << endl;
cout << endl;
cout << " -n steps specify the maximum numer of steps (default=2500)" << endl;
cout << endl;
cout << " -cut use cut-off (default=don't use cut-off)" << endl;
cout << endl;
cout << " -rvdw rvdw specify the VDW cut-off distance (default=6.0)" << endl;
cout << endl;
cout << " -rele rele specify the Electrostatic cut-off distance (default=10.0)" << endl;
cout << endl;
cout << " -pf freq specify the frequency to update the non-bonded pairs (default=10)" << endl;
cout << endl;
OBPlugin::List("forcefields", "verbose");
exit(-1);
} else {
int ifile = 1;
for (int i = 1; i < argc; i++) {
option = argv[i];
// steps
if ((option == "-n") && (argc > (i+1))) {
steps = atoi(argv[i+1]);
ifile += 2;
}
// vdw cut-off
if ((option == "-rvdw") && (argc > (i+1))) {
rvdw = atof(argv[i+1]);
ifile += 2;
}
// ele cut-off
if ((option == "-rele") && (argc > (i+1))) {
rele = atof(argv[i+1]);
ifile += 2;
}
// pair update frequency
if ((option == "-pf") && (argc > (i+1))) {
freq = atoi(argv[i+1]);
ifile += 2;
}
// steepest descent
if (option == "-sd") {
sd = true;
ifile++;
}
// enable cut-off
if (option == "-cut") {
cut = true;
ifile++;
}
// enable Newton2Num
if (option == "-newton") {
newton = true;
ifile++;
}
if (strncmp(option.c_str(), "-o", 2) == 0) {
oext = argv[i] + 2;
if(!*oext) {
oext = argv[++i]; //space left after -o: use next argument
ifile++;
}
format_out = conv.FindFormat(oext);
ifile++;
}
if (option == "-h") {
hydrogens = true;
ifile++;
}
if (option == "-cg") {
sd = false;
ifile++;
}
if ((option == "-c") && (argc > (i+1))) {
crit = atof(argv[i+1]);
ifile += 2;
}
if ((option == "-ff") && (argc > (i+1))) {
ff = argv[i+1];
ifile += 2;
}
}
basename = filename = argv[ifile];
size_t extPos = filename.rfind('.');
if (extPos!= string::npos) {
basename = filename.substr(0, extPos);
}
}
// Find Input filetype
OBFormat *format_in = conv.FormatFromExt(filename.c_str());
if (!format_in || !format_out || !conv.SetInAndOutFormats(format_in, format_out)) {
cerr << program_name << ": cannot read input/output format!" << endl;
exit (-1);
}
ifstream ifs;
ofstream ofs;
// Read the file
ifs.open(filename.c_str());
if (!ifs) {
cerr << program_name << ": cannot read input file!" << endl;
exit (-1);
}
OBForceField* pFF = OBForceField::FindForceField(ff);
if (!pFF) {
cerr << program_name << ": could not find forcefield '" << ff << "'." <<endl;
exit (-1);
}
// set some force field variables
pFF->SetLogFile(&cerr);
pFF->SetLogLevel(OBFF_LOGLVL_LOW);
pFF->SetVDWCutOff(rvdw);
pFF->SetElectrostaticCutOff(rele);
pFF->SetUpdateFrequency(freq);
pFF->EnableCutOff(cut);
if (newton)
pFF->SetLineSearchType(LineSearchType::Newton2Num);
OBMol mol;
for (c=1;;c++) {
mol.Clear();
if (!conv.Read(&mol, &ifs))
break;
if (mol.Empty())
break;
if (hydrogens)
mol.AddHydrogens();
if (!pFF->Setup(mol)) {
cerr << program_name << ": could not setup force field." << endl;
exit (-1);
}
bool done = true;
OBStopwatch timer;
timer.Start();
if (sd) {
pFF->SteepestDescentInitialize(steps, crit);
} else {
pFF->ConjugateGradientsInitialize(steps, crit);
}
unsigned int totalSteps = 1;
while (done) {
if (sd)
done = pFF->SteepestDescentTakeNSteps(1);
else
done = pFF->ConjugateGradientsTakeNSteps(1);
totalSteps++;
if (pFF->DetectExplosion()) {
cerr << "explosion has occured!" << endl;
conv.Write(&mol, &cout);
return(1);
} else
pFF->GetCoordinates(mol);
}
double timeElapsed = timer.Elapsed();
pFF->GetCoordinates(mol);
conv.Write(&mol, &cout);
cerr << "Time: " << timeElapsed << "seconds. Iterations per second: " << double(totalSteps) / timeElapsed << endl;
} // end for loop
return(0);
}
