C++ OBSmartsPattern::Init方法代码示例

    double Predict(OBBase* pOb, string* param=NULL)
    {
      OBMol* pmol = dynamic_cast<OBMol*> (pOb);
      if(!pmol)
        return 0;

      OBSmartsPattern sp;
      if (sp.Init(_smarts) && sp.Match(*pmol))
        return sp.GetUMapList().size();
      else
        return 0.0;
    }

/** The descriptor name can be s or smarts and is case independent
    The operator to return true for a match can be:
    one or more spaces, =, ==,  or nothing if the SMARTS string
    starts with a letter.
    To return true for a mismatch the operator is !=
    A space or tab should follow the SMARTS string.
 **/      
bool SmartsFilter::Compare(OBBase* pOb, istream& optionText, bool noEval)
{
  OBMol* pmol = dynamic_cast<OBMol*> (pOb);
  if(!pmol)
    return false;
  
  string smarts;
  bool matchornegate = ReadStringFromFilter(optionText, smarts);
  if(noEval)
    return false;
  OBSmartsPattern sp;
  sp.Init(smarts);
  bool ret = sp.Match(*pmol,true);//single match  
  if(!matchornegate)
    ret = !ret;
  return ret;
}

  bool GetFingerprint(OBBase* pOb, vector<unsigned int>&fp, int nbits) 
  {
    OBMol* pmol = dynamic_cast<OBMol*>(pOb);
    
   	unsigned int o=0;
   	unsigned int m=0;
   	unsigned int i=0;
   	unsigned int n=0;
    
    if(!pmol)
      return false;
    
    //Read patterns file if it has not been done already
    if(smartsStrings.empty())
      ReadPatternFile(_patternsfile, smartsStrings);

    //Make fp size the smallest power of two to contain the patterns
    //unsigned int n=Getbitsperint();
    //while(n<smartsStrings.size())n*=2;
    //fp.resize(n/Getbitsperint());
    
    fp.resize(16);

    for(n=0;n<smartsStrings.size();++n)
		{
			OBSmartsPattern sp;
			sp.Init(smartsStrings[n]);
			
			if(sp.Match(*pmol)) {
			    m=sp.GetUMapList().size();
			    //m=sp.NumMatches();
			    o=n*8;
			    for(i=0;i<8;++i) {
			           if(i<m) {SetBit(fp, o+i);
			           //cout << "1";
			       }  
          //cout << endl;  
				}				
				}				
		}

    if(nbits)
      Fold(fp, nbits);
    return true;
  };

void CheckSmarts::checkSmarts() {

	OBSmartsPattern smartsPattern;

	for (unsigned int i = 0; i < data->num_smarts(); i++) {

		cout << *data->get_smarts(i) << "\t[ ";

		for (unsigned int j = 0; j < data->num_smiles(); j++) {

			smartsPattern.Init(*data->get_smarts(i));

			if (smartsPattern.Match(*data->get_mol(j), true))
				cout << data->get_id(j) << " ";
		}

		cout << "]\n";
	}
}

int smartsparse(int argc, char* argv[])
{
  int defaultchoice = 1;
  
  int choice = defaultchoice;

  if (argc > 1) {
    if(sscanf(argv[1], "%d", &choice) != 1) {
      printf("Couldn't parse that input as a number\n");
      return -1;
    }
  }
  
  cout << "# Testing SMARTS Parsing...  \n";

  std::ifstream ifs;
  if (!SafeOpen(ifs, nsmarts_file.c_str()))
    {
      cout << "Bail out! Cannot read " << nsmarts_file << endl;
      return -1; // test failed
    }

  //read in the SMARTS test patterns
  char buffer[BUFF_SIZE];
  OBSmartsPattern sp;
  unsigned int patterns = 0;
  for (;ifs.getline(buffer,BUFF_SIZE);)
    {
      if (buffer[0] == '#') // skip comment line
        continue;
      
      if (sp.Init(buffer))
        cout << "ok " << ++patterns << endl;
      else
        cout << "not ok " << ++patterns << " failed on " << buffer << endl;
    }
  ifs.close();
  // output the number of tests run
  cout << "1.." << patterns << endl;

  // Passed Test
  return 0;
}

/** The descriptor name can be s or smarts and is case independent
    The operator to return true for a match can be:
    one or more spaces, =, ==,  or nothing if the SMARTS string
    starts with a letter.
    To return true for a mismatch the operator is !=
    A space or tab should follow the SMARTS string.
 **/
bool SmartsFilter::Compare(OBBase *pOb, istream &optionText, bool noEval,
                           std::string *) {
  OBMol *pmol = dynamic_cast<OBMol *>(pOb);
  if (!pmol)
    return false;

  string smarts;
  bool matchornegate = ReadStringFromFilter(optionText, smarts);
  if (noEval)
    return false;
  OBSmartsPattern sp;
  if (!sp.Init(smarts))
    return false; // can't initialize the SMARTS, so fail gracefully

  bool ret = sp.Match(*pmol, true); // single match
  if (!matchornegate)
    ret = !ret;
  return ret;
}

bool OBGroupContrib::ParseFile(const char *filename)
{
    OBSmartsPattern *sp;

    // open data file
    ifstream ifs;

    if (OpenDatafile(ifs, filename).length() == 0) {
        obErrorLog.ThrowError(__FUNCTION__, " Could not find contribution data file.", obError);
        return false;
    }

    vector<string> vs;
    bool heavy = false;

    char buffer[80];
    while (ifs.getline(buffer, 80)) {
        if (EQn(buffer, "#", 1)) continue;
        if (EQn(buffer, ";heavy", 6))
            heavy = true;
        else if (EQn(buffer, ";", 1)) continue;


        tokenize(vs, buffer);
        if (vs.size() < 2)
            continue;

        sp = new OBSmartsPattern;
        if (sp->Init(vs[0])) {
            if (heavy)
                _contribsHeavy.push_back(pair<OBSmartsPattern*, double> (sp, atof(vs[1].c_str())));
            else
                _contribsHydrogen.push_back(pair<OBSmartsPattern*, double> (sp, atof(vs[1].c_str())));
        } else {
            delete sp;
            sp = NULL;
            obErrorLog.ThrowError(__FUNCTION__, " Could not parse SMARTS from contribution data file", obInfo);
            return false;
        }
    }

    return true;
}

  void OBBondTyper::ParseLine(const char *buffer)
  {
    vector<string> vs;
    vector<int>    bovector;
    OBSmartsPattern *sp;

    if (buffer[0] != '#')
      {
        tokenize(vs,buffer);
        // Make sure we actually have a SMARTS pattern plus at least one triple
        // and make sure we have the correct number of integers
        if (vs.empty() || vs.size() < 4)
          return; // just ignore empty (or short lines)
        else if (!vs.empty() && vs.size() >= 4 && (vs.size() % 3 != 1))
          {
            stringstream errorMsg;
            errorMsg << " Error in OBBondTyper. Pattern is incorrect, found "
                     << vs.size() << " tokens." << endl;
            errorMsg << " Buffer is: " << buffer << endl;
            obErrorLog.ThrowError(__FUNCTION__, errorMsg.str(), obInfo);
            return;
          }

        sp = new OBSmartsPattern;
        if (sp->Init(vs[0]))
          {
            for (unsigned int i = 1; i < vs.size() ; ++i)
              {
                bovector.push_back( atoi((char *)vs[i].c_str()) );
              }
            _fgbonds.push_back(pair<OBSmartsPattern*,vector<int> >
                               (sp, bovector));
          }
        else
          {
            delete sp;
            sp = NULL;
          }
      }
  }

  void OBRingTyper::ParseLine(const char *buffer)
  {
    vector<string> vs;
    OBSmartsPattern *sp;

    if (EQn(buffer,"RINGTYP",7)) {
      tokenize(vs,buffer);
      if (vs.empty() || vs.size() < 3) {
        obErrorLog.ThrowError(__FUNCTION__, " Could not parse RING line in ring type table from ringtyp.txt", obInfo);
        return;
      }
      sp = new OBSmartsPattern;
      if (sp->Init(vs[2]))
        _ringtyp.push_back(pair<OBSmartsPattern*,string> (sp,vs[1]));
      else {
        delete sp;
        sp = NULL;
        obErrorLog.ThrowError(__FUNCTION__, " Could not parse RING line in ring type table from ringtyp.txt", obInfo);
        return;
      }
    }
  }

int main(int argc,char *argv[])
{
  // turn off slow sync with C-style output (we don't use it anyway).
  std::ios::sync_with_stdio(false);

  // Define location of file formats for testing
  #ifdef FORMATDIR
    char env[BUFF_SIZE];
    snprintf(env, BUFF_SIZE, "BABEL_LIBDIR=%s", FORMATDIR);
    putenv(env);
  #endif

  if (argc != 1)
    {
      if (strncmp(argv[1], "-g", 2))
        {
          cout << "Usage: smartstest\n";
          cout << "   Tests Open Babel SMILES/SMARTS pattern matching." << endl;
          return 0;
        }
      else
        {
          GenerateSmartsReference();
          return 0;
        }
    }
  
  cout << endl << "# Testing SMARTS...  \n";

  std::ifstream ifs;
  if (!SafeOpen(ifs, smarts_file.c_str()))
    {
      cout << "Bail out! Cannot read " << smarts_file << endl;
      return -1; // test failed
    }

  //read in the SMARTS test patterns
  char buffer[BUFF_SIZE];
  vector<OBSmartsPattern*> vsp;
  for (;ifs.getline(buffer,BUFF_SIZE);)
    {
      if (buffer[0] == '#') // skip comment line
        continue;

      OBSmartsPattern *sp = new OBSmartsPattern;

      if (sp->Init(buffer))
        vsp.push_back(sp);
      else
        delete sp;
    }
  ifs.close();

  std::ifstream rifs;
  if (!SafeOpen(rifs, results_file.c_str()))
    {
      cout << "Bail out! Cannot read in results file " << results_file << endl;
      return -1; // test failed
    }
  unsigned int npats;
  rifs.getline(buffer,BUFF_SIZE);
  sscanf(buffer,"%d %*s",&npats);

  //make sure the number of SMARTS patterns is the same as in the
  //reference data
  if (npats != vsp.size())
    {
      cout << "Bail out! Correct number of patterns not read in" <<
        "Read in " << vsp.size() << " expected " << npats << endl;
      return -1; // test failed
    }

  std::ifstream mifs;
  if (!SafeOpen(mifs, smilestypes_file.c_str()))
    {
      cout << "Bail out! Cannot read atom types " << smilestypes_file << endl;
      return -1; // test failed
    }

  unsigned int k;
  unsigned int res_line = 0;
  OBMol mol;
  vector<string> vs;
  vector<OBSmartsPattern*>::iterator i;
  vector<vector<int> > mlist;
  unsigned int currentMol = 0; // each molecule is a separate test
  bool molPassed = true;

  OBConversion conv(&mifs, &cout);
  if (! conv.SetInAndOutFormats("SMI","SMI"))
    {
      cout << "Bail out! SMILES format is not loaded" << endl;
      return -1;
    }

  //read in molecules, match SMARTS, and compare results to reference data
  for (;mifs;)
    {
      mol.Clear();
      conv.Read(&mol);
//.........这里部分代码省略.........

void GenerateSmartsReference()
{
  std::ifstream ifs;
  if (!SafeOpen(ifs,smarts_file.c_str()))
    return;

  char buffer[BUFF_SIZE];
  vector<OBSmartsPattern*> vsp;
  for (;ifs.getline(buffer,BUFF_SIZE);)
    {
      if (buffer[0] == '#') // skip comment line
        continue;

      OBSmartsPattern *sp = new OBSmartsPattern;

      if (sp->Init(buffer))
        vsp.push_back(sp);
      else
        delete sp;
    }

  std::ofstream ofs;
  if (!SafeOpen(ofs, results_file.c_str()))
    return;

  ofs << vsp.size() << " patterns" << endl;
  std::ifstream mifs;
  if (!SafeOpen(mifs, smilestypes_file.c_str()))
    return;

  vector<int> vm;
  vector<vector<int> > mlist;
  vector<vector<int> >::iterator j;
  vector<OBSmartsPattern*>::iterator i;
  OBMol mol;
  OBConversion conv(&mifs, &cout);

  if(! conv.SetInAndOutFormats("SMI","SMI"))
    {
      cerr << "SMILES format is not loaded" << endl;
      return;
    }

  for (;mifs;)
    {
      mol.Clear();
      conv.Read(&mol);

      if (mol.Empty())
        continue;
      for (i = vsp.begin();i != vsp.end();i++)
        {
          (*i)->Match(mol);
          mlist = (*i)->GetMapList();
          for (j = mlist.begin();j != mlist.end();j++)
            {
              sprintf(buffer,"%3d",*(j->begin()));
              ofs << buffer;
            }
          ofs << endl;
        }
    }


  cerr << " SMARTS test results written successfully" << endl;
  return;
}

int main(int argc, char* argv[])
{
	const array<string, 5> SmartsPatterns =
	{
		"[!#1]", // heavy
		"[#6+0!$(*~[#7,#8,F]),SH0+0v2,s+0,S^3,Cl+0,Br+0,I+0]", // hydrophobic
		"[a]", // aromatic
		"[$([O,S;H1;v2]-[!$(*=[O,N,P,S])]),$([O,S;H0;v2]),$([O,S;-]),$([N&v3;H1,H2]-[!$(*=[O,N,P,S])]),$([N;v3;H0]),$([n,o,s;+0]),F]", // acceptor
		"[N!H0v3,N!H0+v4,OH+0,SH+0,nH+0]", // donor
	};
	OBConversion obConversion;
	obConversion.SetInFormat("pdbqt");
	while (true)
	{
		vector<array<double, 3>> atoms;
		atoms.reserve(80);
		stringstream ss;
		for (string line; getline(cin, line);)
		{
			ss << line << endl;
			const auto record = line.substr(0, 6);
			if (record == "TORSDO") break;
			if (record != "ATOM  " && record != "HETATM") continue;
			atoms.push_back({ stod(line.substr(30, 8)), stod(line.substr(38, 8)), stod(line.substr(46, 8)) });
		}
		if (atoms.empty()) break;
		OBMol obMol;
		obConversion.Read(&obMol, &ss);
		array<vector<int>, 5> subsets;
		for (size_t k = 0; k < 5; ++k)
		{
			auto& subset = subsets[k];
			subset.reserve(atoms.size());
			OBSmartsPattern smarts;
			smarts.Init(SmartsPatterns[k]);
			smarts.Match(obMol);
			for (const auto& map : smarts.GetMapList())
			{
				subset.push_back(map.front() - 1);
			}
		}
		const auto& subset0 = subsets.front();
		const auto n = subset0.size();
		const auto v = 1.0 / n;
		array<double, 3> ctd{};
		array<double, 3> cst{};
		array<double, 3> fct{};
		array<double, 3> ftf{};
		for (size_t k = 0; k < 3; ++k)
		{
			for (const auto i : subset0)
			{
				const auto& a = atoms[i];
				ctd[k] += a[k];
			}
			ctd[k] *= v;
		}
		double cst_dist = numeric_limits<double>::max();
		double fct_dist = numeric_limits<double>::lowest();
		double ftf_dist = numeric_limits<double>::lowest();
		for (const auto i : subset0)
		{
			const auto& a = atoms[i];
			const auto this_dist = dist2(a, ctd);
			if (this_dist < cst_dist)
			{
				cst = a;
				cst_dist = this_dist;
			}
			if (this_dist > fct_dist)
			{
				fct = a;
				fct_dist = this_dist;
			}
		}
		for (const auto i : subset0)
		{
			const auto& a = atoms[i];
			const auto this_dist = dist2(a, fct);
			if (this_dist > ftf_dist)
			{
				ftf = a;
				ftf_dist = this_dist;
			}
		}
		for (const auto& subset : subsets)
		{
			const auto n = subset.size();
			const auto v = 1.0 / n;
			for (const auto& rpt : { ctd, cst, fct, ftf })
			{
				vector<double> dists(n);
				for (size_t i = 0; i < n; ++i)
				{
					dists[i] = sqrt(dist2(atoms[subset[i]], rpt));
				}
				array<double, 3> m{};
				for (size_t i = 0; i < n; ++i)
				{
					const auto d = dists[i];
//.........这里部分代码省略.........

//.........这里部分代码省略.........
  ifstream ifs;
  if (useInFile && FileIn != NULL)
    {
      // Read the file
      ifs.open(FileIn);
      if (!ifs)
        {
          cerr << program_name << ": cannot read input file!" << endl;
          exit (-1);
        }
      conv.SetInStream(&ifs);
	
	
      // Find Input filetype
      if (pFormat == NULL) {
          pFormat = conv.FormatFromExt(FileIn);
          if (pFormat == NULL)
            {
              cerr << program_name << ": cannot read input format!" << endl;
              return (-1);
            }
      }
    }

  if (! conv.SetInAndOutFormats(pFormat, pFormat))
    {
      cerr << program_name << ": cannot read or write to this file format" << endl;
      return (-1);
    }

  // Match the SMART
  OBSmartsPattern sp;
  vector< vector <int> > maplist;      // list of matched atoms
  sp.Init(Pattern);

  OBMol mol;

  bool impossible_match;

  // Search for pattern
  for (c=0;;)
    {
      mol.Clear();
      conv.Read(&mol);
      if (mol.Empty())
        break;


      ////////////////////////////////////////////////////////////////
      // Do not loose time trying to match the pattern if the matching
      // is impossible.
      // It is impossible to make a full match if the number of atoms is
      // different
      if (full )
        impossible_match = (sp.NumAtoms() == mol.NumHvyAtoms()) ? false : true;
      else
        impossible_match = false;

      if (impossible_match)
        { // -> avoid useless SMART matching attempt
          if (invert)
            {
              if (!count)
                {
                  if ( name_only )
                    cout << mol.GetTitle() << endl;

//.........这里部分代码省略.........
          {
            string attr(txt.substr(0,pos)), val(txt.substr(pos+1));
            //Update value if it already exists
            OBPairData* dp = dynamic_cast<OBPairData*>(GetData(attr));
            if(dp) {
              dp->SetValue(val);
              dp->SetOrigin(userInput);
            }
            else {
              // Pair did not exist; make new one
              dp = new OBPairData;
              dp->SetAttribute(attr);
              dp->SetValue(val);
              dp->SetOrigin(userInput);
              SetData(dp);
            }
          }
      }

    itr = pOptions->find("add");  //adds new properties from descriptors in list
    if(itr!=pOptions->end())
      OBDescriptor::AddProperties(this, itr->second);

    itr = pOptions->find("delete"); //deletes the specified properties
    if(itr!=pOptions->end())
      OBDescriptor::DeleteProperties(this, itr->second);

    itr = pOptions->find("append"); //Appends values of descriptors or properties to title
    if(itr!=pOptions->end())
      {
        string title(GetTitle());
        title += OBDescriptor::GetValues(this, itr->second);
        if(ispunct(title[0]))
          title[0]=' ';//a leading punct char is used only as a separator, not at start
        SetTitle(Trim(title).c_str());
      }



      //Filter using OBDescriptor comparison and (older) SMARTS tests
    //Continue only if previous test was true.
    bool fmatch = true;
    itr = pOptions->find("filter");
    if(itr!=pOptions->end())
      {
        std::istringstream optionText(itr->second);
        fmatch = OBDescriptor::FilterCompare(this, optionText, false);
      }

    if(fmatch)
      {
        itr = pOptions->find("v");
        if(itr!=pOptions->end() && !itr->second.empty())
          {
            //inverse match quoted SMARTS string which follows
            OBSmartsPattern sp;
            sp.Init(itr->second);
            fmatch = !sp.Match(*this); //(*pmol) ;
          }
      }
    if(fmatch)
    {
      itr = pOptions->find("s");
      if(itr!=pOptions->end() && !itr->second.empty())
        {
          //SMARTS filter
          //If exactmatch option set (probably in fastsearchformat) the
          //number of atoms in the pattern (passed as a string in the option text)
          //has to be the same as in the molecule.
          itr2 = pOptions->find("exactmatch");
          if(itr2!=pOptions->end() && NumHvyAtoms()!=atoi(itr2->second.c_str()))
            fmatch=false;
          else
            {
              //match quoted SMARTS string which follows
              OBSmartsPattern sp;
                sp.Init(itr->second.c_str());
                fmatch = sp.Match(*this);
            }
        }
    }

    if(!fmatch)
      {
        //filter failed: delete OBMol and return NULL
        delete this;
        return NULL;
      }
    else
      {
        if(ret==false)
          {
            obErrorLog.ThrowError(__FUNCTION__, "Error executing an option", obError);
            delete this; //added 9March2006
            return NULL;
          }
        else
          return this;
      }
  }

  void OBBondTyper::AssignFunctionalGroupBonds(OBMol &mol)
  {
    if (!_init)
      Init();

    OBSmartsPattern *currentPattern;
    OBBond *b1, *b2;
    OBAtom *a1,*a2, *a3;
    double angle, dist1, dist2;
    vector<int> assignments;
    vector<vector<int> > mlist;
    vector<vector<int> >::iterator matches, l;
    vector<pair<OBSmartsPattern*, vector<int> > >::iterator i;
    unsigned int j;

    // Loop through for all the functional groups and assign bond orders
    for (i = _fgbonds.begin();i != _fgbonds.end();++i)
      {
        currentPattern = i->first;
        assignments = i->second;

        if (currentPattern && currentPattern->Match(mol))
          {
            mlist = currentPattern->GetUMapList();
            for (matches = mlist.begin(); matches != mlist.end(); ++matches)
              {
                // Now loop through the bonds to assign from _fgbonds
                for (j = 0; j < assignments.size(); j += 3)
                  {
                    // along the assignments vector: atomID1 atomID2 bondOrder
                    a1 = mol.GetAtom((*matches)[ assignments[j] ]);
                    a2 = mol.GetAtom((*matches)[ assignments[j+1 ] ]);
                    if (!a1 || !a2) continue;

                    b1 = a1->GetBond(a2);

                    if (!b1) continue;
                    b1->SetBO(assignments[j+2]);
                  } // bond order assignments
              } // each match
          } // current pattern matches

      } // for(functional groups)

    // FG with distance and/or bond criteria
    // Carbonyl oxygen C=O
    OBSmartsPattern carbo; carbo.Init("[#8D1][#6](*)(*)");

    if (carbo.Match(mol))
      {
        mlist = carbo.GetUMapList();
        for (l = mlist.begin(); l != mlist.end(); ++l)
          {
            a1 = mol.GetAtom((*l)[0]);
            a2 = mol.GetAtom((*l)[1]);

            angle = a2->AverageBondAngle();
            dist1 = a1->GetDistance(a2);

            // carbonyl geometries ?
            if (angle > 115 && angle < 150 && dist1 < 1.28) {

              if ( !a1->HasDoubleBond() ) {// no double bond already assigned
                b1 = a1->GetBond(a2);

                if (!b1 ) continue;
                b1->SetBO(2);
              }
            }
          }
      } // Carbonyl oxygen

    // thione C=S
    OBSmartsPattern thione; thione.Init("[#16D1][#6](*)(*)");

    if (thione.Match(mol))
      {
        mlist = thione.GetUMapList();
        for (l = mlist.begin(); l != mlist.end(); ++l)
          {
            a1 = mol.GetAtom((*l)[0]);
            a2 = mol.GetAtom((*l)[1]);

            angle = a2->AverageBondAngle();
            dist1 = a1->GetDistance(a2);

            // thione geometries ?
            if (angle > 115 && angle < 150 && dist1 < 1.72) {

              if ( !a1->HasDoubleBond() ) {// no double bond already assigned
                b1 = a1->GetBond(a2);

                if (!b1 ) continue;
                b1->SetBO(2);
              }
            }
          }
      } // thione

    // Isocyanate N=C=O or Isothiocyanate
//.........这里部分代码省略.........