void ParseNumberList(std::string inLine,
std::vector<T> &res,
std::istream *inStream=0){
bool foundEnd=false;
while(!foundEnd) {
CommaTokenizer commaTok(inLine);
for(CommaTokenizer::const_iterator commaTokIt=commaTok.begin();
commaTokIt!=commaTok.end();
commaTokIt++){
std::string number=*commaTokIt;
bool atEnd= number.find(";>")!=std::string::npos;
boost::trim_if(number,boost::is_any_of(" \r\n\t;>"));
if(number!="" && !atEnd ){
res.push_back(boost::lexical_cast<T>(number));
} else if(atEnd){
// that's it, we're done:
foundEnd=true;
break;
}
}
if(foundEnd || !inStream || inStream->eof()){
break;
} else {
std::getline(*inStream,inLine);
}
}
if(!foundEnd){
throw FileParseException("no end tag found for numeric list");
}
}
ROMol *SDMolSupplier::next() {
PRECONDITION(dp_inStream, "no stream");
if (df_end && d_last >= d_len) {
throw FileParseException("EOF hit.");
}
// set the stream to the current position
dp_inStream->seekg(d_molpos[d_last]);
std::string tempStr;
ROMol *res = NULL;
// finally if we reached the end of the file set end to be true
if (dp_inStream->eof()) {
// FIX: we should probably be throwing an exception here
df_end = true;
d_len = d_molpos.size();
return res;
}
res = _next();
++d_last;
std::streampos posHold = dp_inStream->tellg();
this->checkForEnd();
if (!this->df_end && d_last >= static_cast<int>(d_molpos.size())) {
d_molpos.push_back(posHold);
}
return res;
}
void TDTMolSupplier::moveTo(unsigned int idx) {
PRECONDITION(dp_inStream,"no stream");
CHECK_INVARIANT(idx >= 0, "");
// dp_inStream->seekg() is called for all idx values
// and earlier calls to next() may have put the stream into a bad state
dp_inStream->clear();
// move until we hit the desired idx
if (idx < d_molpos.size() ) {
dp_inStream->seekg(d_molpos[idx]);
d_last = idx;
}
else {
std::string tempStr;
d_last = d_molpos.size() - 1;
dp_inStream->seekg(d_molpos.back());
while ((d_last < static_cast<int>(idx)) && (!dp_inStream->eof()) ) {
d_line++;
std::getline(*dp_inStream,tempStr);
if (tempStr.find("|") == 0) {
d_molpos.push_back(dp_inStream->tellg());
d_last++;
}
}
// if we reached end of file without reaching "idx" we have an index error
if (dp_inStream->eof()) {
d_len = d_molpos.size();
std::ostringstream errout;
errout << "ERROR: Index error (idx = " << idx << ") : " << " we do no have enough molecule blocks";
throw FileParseException(errout.str());
}
}
}
SDWriter::SDWriter(std::ostream *outStream,bool takeOwnership) {
PRECONDITION(outStream,"null stream");
if (outStream->bad()){
throw FileParseException("Bad output stream");
}
dp_ostream = outStream;
df_owner = takeOwnership;
d_molid = 0;
}
TDTWriter::TDTWriter(std::ostream *outStream, bool takeOwnership) {
PRECONDITION(outStream,"null stream");
if (outStream->bad()) {
throw FileParseException("Bad output stream");
}
dp_ostream = outStream;
df_owner = takeOwnership;
d_molid = 0;
d_numDigits=4;
df_write2D=false;
df_writeNames=true;
}
void SDMolSupplier::moveTo(unsigned int idx) {
PRECONDITION(dp_inStream, "no stream");
// dp_inStream->seekg() is called for all idx values
// and earlier calls to next() may have put the stream into a bad state
dp_inStream->clear();
// move until we hit the desired idx
if (idx < d_molpos.size()) {
dp_inStream->seekg(d_molpos[idx]);
d_last = idx;
} else {
std::string tempStr;
dp_inStream->seekg(d_molpos.back());
d_last = d_molpos.size() - 1;
while ((d_last < static_cast<int>(idx)) && (!dp_inStream->eof())) {
d_line++;
tempStr = getLine(dp_inStream);
if (tempStr[0] == '$' && tempStr.substr(0, 4) == "$$$$") {
std::streampos posHold = dp_inStream->tellg();
this->checkForEnd();
if (!this->df_end) {
d_molpos.push_back(posHold);
d_last++;
}
}
}
// if we reached end of file without reaching "idx" we have an index error
if (dp_inStream->eof()) {
d_len = d_molpos.size();
std::ostringstream errout;
errout << "ERROR: Index error (idx = " << idx << ") : "
<< " we do no have enough mol blocks";
throw FileParseException(errout.str());
}
}
}
ROMol *TDTMolSupplier::parseMol(std::string inLine){
PRECONDITION(dp_inStream,"no stream");
Utils::LocaleSwitcher ls;
std::size_t startP=inLine.find("<");
std::size_t endP=inLine.find_last_of(">");
std::string smiles = inLine.substr(startP+1,endP-startP-1);
ROMol *res = SmilesToMol(smiles,0,df_sanitize);
if(res && res->getNumAtoms()>0){
// -----------
// Process the properties:
d_line++;
std::getline(*dp_inStream,inLine);
while(!dp_inStream->eof() && inLine.find("|")!=0){
endP=inLine.find("<");
std::string propName = inLine.substr(0,endP);
boost::trim_if(propName,boost::is_any_of(" \t"));
startP = endP+1;
if(propName=="2D" && d_confId2D>=0){
std::string rest=inLine.substr(startP,inLine.size()-startP);
std::vector<double> coords;
TDTParseUtils::ParseNumberList(rest,coords,dp_inStream);
Conformer *conf=new Conformer(res->getNumAtoms());
conf->setId(d_confId2D);
conf->set3D(false);
for(unsigned int atIdx=0;atIdx<res->getNumAtoms();atIdx++){
if(2*atIdx+1 < coords.size()){
conf->setAtomPos(atIdx,RDGeom::Point3D(coords[2*atIdx],coords[2*atIdx+1],0.0));
} else {
// we're going to let this slide... but maybe we should do something else?
}
}
res->addConformer(conf,false);
} else if(propName=="3D" && d_confId3D>=0){
std::string rest=inLine.substr(startP,inLine.size()-startP);
std::vector<double> coords;
TDTParseUtils::ParseNumberList(rest,coords,dp_inStream);
Conformer *conf=new Conformer(res->getNumAtoms());
conf->setId(d_confId3D);
conf->set3D(true);
for(unsigned int atIdx=0;atIdx<res->getNumAtoms();atIdx++){
if(3*atIdx+2 < coords.size()){
conf->setAtomPos(atIdx,RDGeom::Point3D(coords[3*atIdx],
coords[3*atIdx+1],
coords[3*atIdx+2]));
} else {
// we're going to let this slide... but maybe we should do something else?
}
}
res->addConformer(conf,false);
} else {
endP=inLine.find_last_of(">");
if(endP==std::string::npos){
std::ostringstream errout;
errout << "no end tag found for property" << propName;
throw FileParseException(errout.str());
} else {
std::string propVal = inLine.substr(startP,endP-startP);
res->setProp(propName,propVal);
if(propName==d_nameProp) res->setProp("_Name",propVal);
}
}
std::getline(*dp_inStream,inLine);
}
}
return res;
}
// --------------------------------------------------
//
// Moves to the position of a particular entry in the
// stream.
//
// If insufficient entries are present, a FileParseException
// will be thrown
//
void SmilesMolSupplier::moveTo(unsigned int idx) {
PRECONDITION(dp_inStream, "bad instream");
// get the easy situations (boundary conditions) out of the
// way first:
if (d_len > -1 && idx >= static_cast<unsigned int>(d_len)) {
df_end = true;
std::ostringstream errout;
errout << "ERROR: Index error (idx = " << idx << "): "
<< "ran out of lines\n";
throw FileParseException(errout.str());
}
// dp_inStream->seekg() is called for all idx values
// and earlier calls to next() may have put the stream into a bad state
dp_inStream->clear();
// -----------
// Case 1: we have already read the particular entry:
//
// Set the stream position and return
// -----------
if (!d_molpos.empty() && d_molpos.size() > idx) {
dp_inStream->clear(); // clear the EOF tag if it has been set
df_end = false;
dp_inStream->seekg(d_molpos[idx]);
d_next = idx;
d_line = d_lineNums[idx];
return;
}
// -----------
// Case 2: we haven't read the entry, so move forward until
// we've gone far enough.
// -----------
if (d_molpos.empty()) {
// if we are just starting out, process the title line
dp_inStream->seekg(0);
if (df_title) this->processTitleLine();
} else {
// move to the last position we've seen:
dp_inStream->seekg(d_molpos.back());
// read that line:
std::string tmp = getLine(dp_inStream);
}
// the stream pointer is now at the last thing we read in
while (d_molpos.size() <= idx) {
int nextP = this->skipComments();
if (nextP < 0) {
std::ostringstream errout;
errout << "ERROR: Index error (idx = " << idx << "): "
<< "ran out of lines\n";
throw FileParseException(errout.str());
} else {
d_molpos.push_back(nextP);
d_lineNums.push_back(d_line);
if (d_molpos.size() == idx + 1 && df_end) {
// boundary condition: we could read the point we were looking for
// but not the next one.
// indicate that we've reached EOF:
dp_inStream->clear();
dp_inStream->seekg(0, std::ios_base::end);
d_len = d_molpos.size();
break;
}
}
}
POSTCONDITION(d_molpos.size() > idx, "not enough lines");
dp_inStream->seekg(d_molpos[idx]);
d_next = idx;
return;
}
ROMol *SmilesMolSupplier::processLine(std::string inLine) {
ROMol *res = NULL;
try {
// -----------
// tokenize the input line:
// -----------
boost::char_separator<char> sep(d_delim.c_str(), "",
boost::keep_empty_tokens);
tokenizer tokens(inLine, sep);
STR_VECT recs;
for (tokenizer::iterator tokIter = tokens.begin(); tokIter != tokens.end();
++tokIter) {
std::string rec = strip(*tokIter);
recs.push_back(rec);
}
if (recs.size() <= static_cast<unsigned int>(d_smi)) {
std::ostringstream errout;
errout << "ERROR: line #" << d_line << "does not contain enough tokens\n";
throw FileParseException(errout.str());
}
// -----------
// get the smiles and create a molecule
// -----------
res = SmilesToMol(recs[d_smi], 0, df_sanitize);
if (!res) {
std::stringstream errout;
errout << "Cannot create molecule from : '" << recs[d_smi] << "'";
throw SmilesParseException(errout.str());
}
// -----------
// get the name (if there's a name column)
// -----------
if (d_name == -1) {
// if no name defaults it to the line number we read it from string
std::ostringstream tstr;
tstr << d_line;
std::string mname = tstr.str();
res->setProp(common_properties::_Name, mname);
} else {
if (d_name >= static_cast<int>(recs.size())) {
BOOST_LOG(rdWarningLog) << "WARNING: no name column found on line "
<< d_line << std::endl;
} else {
res->setProp(common_properties::_Name, recs[d_name]);
}
}
// -----------
// read in the properties
// -----------
unsigned int iprop = 0;
for (unsigned int col = 0; col < recs.size(); col++) {
if (static_cast<int>(col) == d_smi || static_cast<int>(col) == d_name)
continue;
std::string pname, pval;
if (d_props.size() > col) {
pname = d_props[col];
} else {
pname = "Column_";
std::stringstream ss;
ss << col;
pname += ss.str();
}
pval = recs[col];
res->setProp(pname, pval);
iprop++;
}
} catch (const SmilesParseException &pe) {
// Couldn't parse the passed in smiles
// Simply print out a message
BOOST_LOG(rdErrorLog) << "ERROR: Smiles parse error on line " << d_line
<< "\n";
BOOST_LOG(rdErrorLog) << "ERROR: " << pe.message() << "\n";
res = NULL;
} catch (const MolSanitizeException &se) {
// We couldn't sanitize the molecule
// write out an error message
BOOST_LOG(rdErrorLog) << "ERROR: Could not sanitize molecule on line "
<< d_line << std::endl;
BOOST_LOG(rdErrorLog) << "ERROR: " << se.message() << "\n";
res = NULL;
} catch (...) {
// write out an error message
BOOST_LOG(rdErrorLog) << "ERROR: Could not process molecule on line "
<< d_line << std::endl;
res = NULL;
}
return res;
}
void ForwardSDMolSupplier::readMolProps(ROMol *mol){
PRECONDITION(dp_inStream,"no stream");
PRECONDITION(mol,"no molecule");
d_line++;
std::string tempStr;
std::getline(*dp_inStream,tempStr);
// FIX: report files missing the $$$$ marker
while(!(dp_inStream->eof()) && (tempStr[0]!='$'||tempStr.substr(0,4)!="$$$$") ){
tempStr = strip(tempStr);
if(tempStr!=""){
if (tempStr[0] == '>') { // data header line: start of a data item
// ignore all other crap and seek for for a data label enclosed
// by '<' and '>'
// FIX: "CTfile.pdf" (page 51) says that the a data header line does not
// have to contain a data label (instead can have something line field
// id into a MACCS db). But we do not currently know what to do in this
// situation - so ignore such data items for now
tempStr.erase(0,1); // remove the first ">" sign
int sl = tempStr.find("<"); // begin datalabel
int se = tempStr.find(">"); // end datalabel
if ((sl == -1) || (se == -1) || (se == (sl+1)) ) {
// we either do not have a data label or the label is emtpy
// no data label ignore until next data item
// i.e. until we hit a blank line
d_line++;
std::getline(*dp_inStream,tempStr);
std::string stmp = strip(tempStr);
while (stmp.length() != 0) {
d_line++;
std::getline(*dp_inStream,tempStr);
if(dp_inStream->eof()) throw FileParseException("End of data field name not found");
}
} else {
std::string dlabel = tempStr.substr(sl+1, se-sl-1);
// we know the label - now read in the relevant properties
// until we hit a blank line
d_line++;
std::getline(*dp_inStream,tempStr);
std::string prop="";
std::string stmp = strip(tempStr);
int nplines = 0; // number of lines for this property
while (stmp.length() != 0 || tempStr[0]==' ' || tempStr[0]=='\t') {
nplines++;
if (nplines > 1) {
prop += "\n";
}
// take off \r if it's still in the property:
if (tempStr[tempStr.length()-1]=='\r'){
tempStr.erase(tempStr.length()-1);
}
prop += tempStr;
d_line++;
// erase tempStr in case the file does not end with a carrier
// return (we will end up in an infinite loop if we don't do
// this and we do not check for EOF in this while loop body)
tempStr.erase();
std::getline(*dp_inStream,tempStr);
stmp = strip(tempStr);
}
mol->setProp(dlabel, prop);
}
} else {
// at this point we should always be at a line starting with '>'
// following a blank line. If this is not true throw an exception
// FIX: should we be deleting the molecule (which is probably fine)
// because we couldn't read the data ???
throw FileParseException("Problems encountered parsing data fields");
}
}
d_line++;
std::getline(*dp_inStream,tempStr);
}
}
