C++ OBMol::GetFormula方法代码示例

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

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

  OBMol mol;
  OBConversion conv(&ifs, &cout);

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

  for (;ifs;)
    {
      mol.Clear();
      conv.Read(&mol);
      if (mol.Empty())
        continue;

      //write out formula, molecular weight and exact mass
      ofs << mol.GetFormula() << " " << mol.GetMolWt() << " " 
          << mol.GetExactMass() << endl;
    }

	cerr << " Molecular formula results written successfully" << endl;
  return;
}

extern "C" char *
ob_hillformula (char *smiles)
{

  string tmpStr (smiles);
  istringstream molstream (tmpStr);
  string molfmla;
  OBMol mol;
  OBConversion conv;
  char *tmpFormula;

  conv.SetInAndOutFormats ("SMI", "SMI");
  conv.Read (&mol, &molstream);

  molfmla = mol.GetFormula ();

  tmpFormula = strdup (molfmla.c_str ());

  return (tmpFormula);
}

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);

  if (argc != 1)
    {
      if (strncmp(argv[1], "-g", 2))
        {
          cout << "Usage: formula" << endl;
          cout << "   Tests Open Babel molecular formula, weight, and exact mass." << endl;
          return 0;
        }
      else
        {
          GenerateFormulaReference();
          return 0;
        }
    }

  cout << "# Testing molecular formulas..." << endl;

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

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

  char buffer[BUFF_SIZE];
  vector<string> vs;
  OBMol mol;
  OBConversion conv(&mifs, &cout);
  unsigned int currentTest = 0;
  // double mass;

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

  for (;mifs;)
    {
      mol.Clear();
      conv.Read(&mol);
      if (mol.Empty())
        continue;
      if (!rifs.getline(buffer,BUFF_SIZE))
        {
          cout << "Bail out! error reading reference data" << endl;
          return -1; // test failed
        }

      tokenize(vs,buffer);
      if (vs.size() != 3)
        {
          cout << "Bail out! Reference data has incorrect format" << endl;
          return -1; // test failed
        }

      if (vs[0] != mol.GetFormula())
        {
          cout << "not ok " << ++currentTest << " # molecular formula incorrect"
               << " for molecule " << mol.GetTitle() << "\n";
        }
      else
        cout << "ok " << ++currentTest << " # molecular formula\n";

      if ( fabs(atof(vs[1].c_str()) - mol.GetMolWt() ) > 1.0e-3)
        {
          cout << "not ok " << ++currentTest << " # molecular weight incorrect"
               << " for molecule " << mol.GetTitle() << "\n";
          cout << "# Expected " << atof(vs[1].c_str()) << " found " <<
            mol.GetMolWt() << "\n";
        }
      else
        cout << "ok " << ++currentTest << " # molecular weight\n";

      if ( fabs(atof(vs[2].c_str()) - mol.GetExactMass() ) > 1.0e-3)
        {
          cout << "not ok " << ++currentTest << " # exact mass incorrect"
               << " for molecule " << mol.GetTitle() << "\n";
          cout << "# Expected " << atof(vs[2].c_str()) << " found " <<
            mol.GetExactMass() << "\n";
        }
      else
        cout << "ok " << ++currentTest << " # molecular exact mass\n";


      // now after adding explict hydrogens -- should be identical
      //  since we'll add hydrogens that were implicit before

//.........这里部分代码省略.........

void formula_test()
{
#ifdef TESTDATADIR
  string testdatadir = TESTDATADIR;
  string results_file = testdatadir + "formularesults.txt";
  string smilestypes_file = testdatadir + "attype.00.smi";
#else
  string results_file = "files/formularesults.txt";
  string smilestypes_file = "files/attype.00.smi";
#endif

  cout << "# Testing molecular formulas..." << endl;

  std::ifstream mifs;
  BOOST_REQUIRE_MESSAGE( SafeOpen(mifs, smilestypes_file.c_str()), "Bail out! Cannot read file " );

  std::ifstream rifs;
  BOOST_REQUIRE_MESSAGE( SafeOpen(rifs, results_file.c_str()), "Bail out! Cannot read file " );

  char buffer[BUFF_SIZE];
  char message[BUFF_SIZE];
  vector<string> vs;
  OBMol mol;
  OBConversion conv(&mifs, &cout);
  unsigned int currentTest = 0;
  // double mass;

  BOOST_REQUIRE_MESSAGE( conv.SetInAndOutFormats("SMI","SMI"), "Bail out! SMILES format is not loaded" );

  for (;mifs;)
    {
      mol.Clear();
      conv.Read(&mol);
      if (mol.Empty())
        continue;
      BOOST_REQUIRE_MESSAGE( rifs.getline(buffer,BUFF_SIZE), "Bail out! error reading reference data" );

      tokenize(vs,buffer);
      BOOST_REQUIRE_MESSAGE( vs.size() == 3, "Bail out! Reference data has incorrect format" );

      currentTest++;
      snprintf(message, BUFF_SIZE, "not ok %d # molecular formula incorrect for molecule %s"
		                   " # Expected %s, found %s", currentTest, mol.GetTitle(),
				   vs[0].c_str(), mol.GetFormula().c_str());
      BOOST_CHECK_MESSAGE(vs[0] == mol.GetFormula(), message );

      currentTest++;
      snprintf(message, BUFF_SIZE, "not ok %d # molecular weight incorrect for molecule %s"
		                   " # Expected %f, found %f ", currentTest, mol.GetTitle(),
				   atof(vs[1].c_str()), mol.GetMolWt());
      BOOST_CHECK_MESSAGE( fabs(atof(vs[1].c_str()) - mol.GetMolWt() ) < 1.0e-3, message );

      currentTest++;
      snprintf(message, BUFF_SIZE, "not ok %d # exact mass incorrect for molecule %s"
		                   " # Expected %f, found %f ", currentTest, mol.GetTitle(),
				   atof(vs[2].c_str()), mol.GetExactMass());
      BOOST_CHECK_MESSAGE( fabs(atof(vs[2].c_str()) - mol.GetExactMass() ) < 1.0e-3, message );

      // now after adding explict hydrogens -- should be identical
      //  since we'll add hydrogens that were implicit before

      // PR#1485580
      BOOST_CHECK( mol.AddHydrogens() );

      currentTest++;
      snprintf(message, BUFF_SIZE, "not ok %d # molecular formula incorrect for "
		                   "hydrogen-added molecule %s", currentTest, mol.GetTitle());
      BOOST_CHECK_MESSAGE(vs[0] == mol.GetFormula(), message);

      currentTest++;
      snprintf(message, BUFF_SIZE, "not ok %d # molecular weight incorrect for hydrogen-added "
		                   "molecule %s # Expected %f, found %f ", currentTest, mol.GetTitle(),
				   atof(vs[1].c_str()), mol.GetMolWt());
      BOOST_CHECK_MESSAGE( fabs(atof(vs[1].c_str()) - mol.GetMolWt() ) < 1.0e-3, message);

      currentTest++;
      snprintf(message, BUFF_SIZE, "not ok %d # exact mass  incorrect for hydrogen-added "
		                   "molecule %s # Expected %f, found %f ", currentTest, mol.GetTitle(),
				   atof(vs[2].c_str()), mol.GetExactMass());
      BOOST_CHECK_MESSAGE( fabs(atof(vs[2].c_str()) - mol.GetExactMass() ) < 1.0e-3, message);

    }

}

int main(int argc,char **argv)
{
  char *program_name= argv[0];
  int c;
  char *FileIn = NULL;

  if (argc != 2)
    {
      string err = "Usage: ";
      err += program_name;
      err += " <filename>\n"
      "Output format:\n"
        "name NAME\n"
        "formula  FORMULA\n"
        "mol_weight MOLECULAR_WEIGHT\n"
        "exact_mass ISOTOPIC MASS\n"
        "canonical_SMILES STRING\n"
        "InChI  STRING\n"
        "num_atoms  NUM\n"
        "num_bonds  NUM\n"
        "num_residues  NUM\n"
	"num_rotors NUM\n"
        "sequence RESIDUE_SEQUENCE\n"
        "num_rings NUMBER_OF_RING_(SSSR)\n"
        "logP   NUM\n"
        "PSA    POLAR_SURFACE_AREA\n"
        "MR     MOLAR REFRACTIVITY";
      err += "$$$$";
//      ThrowError(err); wasn't being output because error level too low
      cerr << err; //Why not do directly
      exit(-1);
    }
  else
    {
      FileIn  = argv[1];
    }

  // Find Input filetype
  OBConversion conv;
  OBFormat *format = conv.FormatFromExt(FileIn);
    
  if (!format || !conv.SetInFormat(format))
    {
      cerr << program_name << ": cannot read input format!" << endl;
      exit (-1);
    }

  ifstream ifs;

  // Read the file
  ifs.open(FileIn);
  if (!ifs)
    {
      cerr << program_name << ": cannot read input file!" << endl;
      exit (-1);
    }
  
  OBMol mol;
  OBFormat *canSMIFormat = conv.FindFormat("can");
  OBFormat *inchiFormat = conv.FindFormat("inchi");


  ////////////////////////////////////////////////////////////////////////////
  // List of properties
  // Name
  // Molecular weight (Standard molar mass given by IUPAC atomic masses)
  // Number of rings : the size of the smallest set of smallest rings (SSSR)
  
  //.....ADD YOURS HERE.....
  
  for (c = 1;; ++c)
    {
      mol.Clear();
      conv.Read(&mol, &ifs);
      if (mol.Empty())
        break;
      
      if (!mol.HasHydrogensAdded())
        mol.AddHydrogens();
      // Print the properties
      if (strlen(mol.GetTitle()) != 0)
        cout << "name             " << mol.GetTitle() << endl;
      else 
        cout << "name             " << FileIn << " " << c << endl;

      cout << "formula          " << mol.GetFormula() << endl;
      cout << "mol_weight       " << mol.GetMolWt() << endl;
      cout << "exact_mass       " << mol.GetExactMass() << endl;

      string smilesString = "-";
      if (canSMIFormat) {
        conv.SetOutFormat(canSMIFormat);
        smilesString = conv.WriteString(&mol);
        if ( smilesString.length() == 0 )
        {
          smilesString = "-";
        }
      }
      cout << "canonical_SMILES " << smilesString << endl;

//.........这里部分代码省略.........

//.........这里部分代码省略.........
  doubleBondMol.BeginModify();
  a1 = doubleBondMol.NewAtom();
  a1->SetVector(0.0, 0.0, 0.0);
  a1->SetAtomicNum(6);
  a2 = doubleBondMol.NewAtom();
  a2->SetVector(1.6, 0.0, 0.0);
  a2->SetAtomicNum(6);
  b = doubleBondMol.NewBond();
  b->SetBegin(a1);
  b->SetEnd(a2);
  a1->AddBond(b);
  a2->AddBond(b);
  doubleBondMol.EndModify();
  cout << "ok 9" << endl;

  // test AddHydrogens
  OBMol testMolH;
  testMolH.BeginModify();
  OBAtom *testAtom = testMolH.NewAtom();
  testAtom->SetVector(0.5f, 0.5f, 0.5f);
  testAtom->SetAtomicNum(6);
  testAtom->SetImplicitHCount(4);
  testMolH.EndModify();
  testMolH.AddHydrogens();
  if (testMolH.NumAtoms() == 5) {
    cout << "ok 10" << endl;
  } else {
    cout << "not ok 10" << endl;
  }

  // test AddHydrogens (pr #1665519)
  OBMol testMolH2;
  OBAtom *testAtom2 = testMolH2.NewAtom();
  testAtom2->SetVector(0.5f, 0.5f, 0.5f);
  testAtom2->SetAtomicNum(6);
  testAtom2->SetImplicitHCount(4);
  testMolH2.AddHydrogens();
  if (testMolH2.NumAtoms() == 5) {
    cout << "ok 11" << endl;
  } else {
    cout << "not ok 11 # hydrogen additions" << endl;
  }
  
  // Attempt to write an empty InChI (PR#2864334)
  pFormat = conv.FindFormat("InChI");
  if ( pFormat != NULL && conv.SetOutFormat(pFormat))
    {
      if (conv.Write(&emptyMol))
        cout << "ok 12" << endl;
      else
        cout << "not ok 12 # failed empty InChI" << endl;
    }

  OBMol testMolFormula;
  string formula("C6");
  testMolFormula.SetFormula(formula);
  if ( testMolFormula.GetFormula() == formula ) {
     cout << "ok 13" << endl;
  } else {
    cout << "not ok 13 # SetFormula "<< endl;
  }
  // Reset the formula to test for a double delete error
  testMolFormula.SetFormula(formula);
  
  // Test molecular formulas with large atomic numbers
  OBMol testLgAtNo;
  testLgAtNo.BeginModify();
  OBAtom *lgAtom = testLgAtNo.NewAtom();
  lgAtom->SetAtomicNum(118);
  // Undefined atomic numbers should be ignored with an obWarning instead of segfault
  lgAtom = testLgAtNo.NewAtom();
  lgAtom->SetAtomicNum(200);
  lgAtom = testLgAtNo.NewAtom();
  lgAtom->SetAtomicNum(1);
  lgAtom->SetIsotope(2);
  testLgAtNo.EndModify();
  if ( testLgAtNo.GetFormula() == "DOg" ) {
    cout << "ok 14" << endl;
  } else {
    cout << "not ok 14" << endl;
  }
  

  double dihedral = CalcTorsionAngle(vector3(-1., -1.,  0.),
                                     vector3(-1.,  0.,  0.),
                                     vector3( 1.,  0.,  0.),
                                     vector3( 1.,  1.,  0.));

  double dihedral_error = fabs(dihedral) - 180.0;

  if (fabs(dihedral_error) < 0.001) {
      std::cout << "ok 15 " << dihedral_error << std::endl;
  } else {

      std::cout << "not ok 15 # CalcTorsionAngle " << dihedral << "!= 180.0" << std::endl;
  }

  cout << "1..15\n"; // total number of tests for Perl's "prove" tool
  return(0);
}