mdlreader.java

化学图形处理软件

            logger.debug(exception);
            throw new CDKException(error, exception);
        }
		try {
			input.close();
		} catch (Exception exc) {
            String error = "Error while closing file: " + exc.getMessage();
            logger.error(error);
			throw new CDKException(error, exc);
		}

        chemFile.addChemSequence(chemSequence);
		return chemFile;
	}



	/**
	 *  Read a Molecule from a file in MDL sd format
	 *
	 *@return    The Molecule that was read from the MDL file.
	 */
	private IMolecule readMolecule(IMolecule molecule) throws CDKException {
        logger.debug("Reading new molecule");
        int linecount = 0;
        int atoms = 0;
        int bonds = 0;
        int atom1 = 0;
        int atom2 = 0;
        int order = 0;
        int stereo = 0;
        int RGroupCounter=1;
        int Rnumber=0;
        String [] rGroup=null;
        double x = 0.0;
        double y = 0.0;
        double z = 0.0;
        double totalZ = 0.0;
        //int[][] conMat = new int[0][0];
        //String help;
        IBond bond;
        IAtom atom;
        String line = "";
        
        try {
            logger.info("Reading header");
            line = input.readLine(); linecount++;
            if (line == null) {
                return null;
            }
            logger.debug("Line " + linecount + ": " + line);

            if (line.startsWith("$$$$")) {
                logger.debug("File is empty, returning empty molecule");
                return molecule;
            }
            if (line.length() > 0) {
                molecule.setProperty(CDKConstants.TITLE, line);
            }
            line = input.readLine(); linecount++;
            logger.debug("Line " + linecount + ": " + line);
            line = input.readLine(); linecount++;
            logger.debug("Line " + linecount + ": " + line);
            if (line.length() > 0) {
                molecule.setProperty(CDKConstants.REMARK, line);
            }
            
            logger.info("Reading rest of file");
            line = input.readLine(); linecount++;
            logger.debug("Line " + linecount + ": " + line);
            atoms = Integer.valueOf(line.substring(0,3).trim()).intValue();
            logger.debug("Atomcount: " + atoms);
            bonds = Integer.valueOf(line.substring(3,6).trim()).intValue();
            logger.debug("Bondcount: " + bonds);
            
            // read ATOM block
            logger.info("Reading atom block");
            for (int f = 0; f < atoms; f++) {
                line = input.readLine(); linecount++;
                x = new Double(line.substring( 0,10).trim()).doubleValue();
                y = new Double(line.substring(10,20).trim()).doubleValue();
                z = new Double(line.substring(20,30).trim()).doubleValue();
                totalZ += Math.abs(z); // *all* values should be zero, not just the sum
                logger.debug("Coordinates: " + x + "; " + y + "; " + z);
                String element = line.substring(31,34).trim();

                logger.debug("Atom type: ", element);
                if (IsotopeFactory.getInstance(molecule.getBuilder()).isElement(element)) {
                    atom = IsotopeFactory.getInstance(molecule.getBuilder()).configure(molecule.getBuilder().newAtom(element));
                } else {
                    logger.debug("Atom ", element, " is not an regular element. Creating a PseudoAtom.");
                    //check if the element is R
                    rGroup=element.split("^R");
                    if (rGroup.length >1){
                    	try{
                    		Rnumber=new Integer(rGroup[(rGroup.length-1)]).intValue();
                    		RGroupCounter=Rnumber;
                    	}catch(Exception ex){
                    		Rnumber=RGroupCounter;
                    		RGroupCounter++;
                    	}
                    	element="R"+Rnumber;
                    }
                    atom = molecule.getBuilder().newPseudoAtom(element);
                }

                // store as 3D for now, convert to 2D (if totalZ == 0.0) later
                atom.setPoint3d(new Point3d(x, y, z));
                
                // parse further fields
                String massDiffString = line.substring(34,36).trim();
                logger.debug("Mass difference: ", massDiffString);
                if (!(atom instanceof IPseudoAtom)) {
                    try {
                        int massDiff = Integer.parseInt(massDiffString);
                        if (massDiff != 0) {
                            IIsotope major = IsotopeFactory.getInstance(molecule.getBuilder()).getMajorIsotope(element);
                            atom.setAtomicNumber(major.getAtomicNumber() + massDiff);
                        }
                    } catch (Exception exception) {
                        logger.error("Could not parse mass difference field");
                    }
                } else {
                    logger.error("Cannot set mass difference for a non-element!");
                }
                
                
                String chargeCodeString = line.substring(36,39).trim();
                logger.debug("Atom charge code: ", chargeCodeString);
                int chargeCode = Integer.parseInt(chargeCodeString);
                if (chargeCode == 0) {
                    // uncharged species
                } else if (chargeCode == 1) {
                    atom.setFormalCharge(+3);
                } else if (chargeCode == 2) {
                        atom.setFormalCharge(+2);
                } else if (chargeCode == 3) {
                        atom.setFormalCharge(+1);
                } else if (chargeCode == 4) {
                } else if (chargeCode == 5) {
                        atom.setFormalCharge(-1);
                } else if (chargeCode == 6) {
                        atom.setFormalCharge(-2);
                } else if (chargeCode == 7) {
                        atom.setFormalCharge(-3);
                }
                
                try {
                    // read the mmm field as position 61-63
                    String reactionAtomIDString = line.substring(60,63).trim();
                    logger.debug("Parsing mapping id: ", reactionAtomIDString);
                    try {
                        int reactionAtomID = Integer.parseInt(reactionAtomIDString);
                        if (reactionAtomID != 0) {
                            atom.setID(reactionAtomIDString);
                        }
                    } catch (Exception exception) {
                        logger.error("Mapping number ", reactionAtomIDString, " is not an integer.");
                        logger.debug(exception);
                    }
                } catch (Exception exception) {
                    // older mol files don't have all these fields...
                    logger.warn("A few fields are missing. Older MDL MOL file?");
                }
                
                //shk3: This reads shifts from after the molecule. I don't think this is an official format, but I saw it frequently 80=>78 for alk
                if(line.length()>=78){
                	double shift=Double.parseDouble(line.substring(69,80).trim());
                	atom.setProperty("first shift",new Double(shift));
                }
                if(line.length()>=87){
                	double shift=Double.parseDouble(line.substring(79,87).trim());
                	atom.setProperty("second shift",new Double(shift));
                }
                
                molecule.addAtom(atom);
            }
            
            // convert to 2D, if totalZ == 0
            if (totalZ == 0.0 && !forceReadAs3DCoords.isSet()) {
                logger.info("Total 3D Z is 0.0, interpreting it as a 2D structure");
                java.util.Iterator atomsToUpdate = molecule.atoms();
                while (atomsToUpdate.hasNext()) {
                    IAtom atomToUpdate = (IAtom)atomsToUpdate.next();
                    Point3d p3d = atomToUpdate.getPoint3d();
                    atomToUpdate.setPoint2d(new Point2d(p3d.x, p3d.y));
                    atomToUpdate.setPoint3d(null);
                }
            }
            
            // read BOND block
            logger.info("Reading bond block");
            for (int f = 0; f < bonds; f++) {
                line = input.readLine(); linecount++;
                atom1 = java.lang.Integer.valueOf(line.substring(0,3).trim()).intValue();
                atom2 = java.lang.Integer.valueOf(line.substring(3,6).trim()).intValue();
                order = java.lang.Integer.valueOf(line.substring(6,9).trim()).intValue();
                if (line.length() > 12) {
                	stereo = java.lang.Integer.valueOf(line.substring(9,12).trim()).intValue();
                } else {
                	logger.warn("Missing expected stereo field at line: " + line);
                }
                if (logger.isDebugEnabled()) {
                    logger.debug("Bond: " + atom1 + " - " + atom2 + "; order " + order);
                }
                if (stereo == 1) {
                    // MDL up bond
                    stereo = CDKConstants.STEREO_BOND_UP;
                } else if (stereo == 6) {
                    // MDL down bond
                    stereo = CDKConstants.STEREO_BOND_DOWN;
                } else if (stereo == 4) {
                    //MDL bond undefined
                    stereo = CDKConstants.STEREO_BOND_UNDEFINED;
                }
                // interpret CTfile's special bond orders
                IAtom a1 = molecule.getAtom(atom1 - 1);
                IAtom a2 = molecule.getAtom(atom2 - 1);
                if (order == 4) {
                    // aromatic bond
                    bond = molecule.getBuilder().newBond(a1, a2, CDKConstants.BONDORDER_AROMATIC, stereo);
                    // mark both atoms and the bond as aromatic
                    bond.setFlag(CDKConstants.ISAROMATIC, true);
                    a1.setFlag(CDKConstants.ISAROMATIC, true);
                    a2.setFlag(CDKConstants.ISAROMATIC, true);
                    molecule.addBond(bond);
                } else {
                    bond = molecule.getBuilder().newBond(a1, a2, (double) order, stereo);
                    molecule.addBond(bond);
                }
            }
            
		} catch (Exception exception) {
			exception.printStackTrace();
            String error = "Error while parsing line " + linecount + ": " + line + " -> " + exception.getMessage();
            logger.error(error);
            logger.debug(exception);
            throw new CDKException(error, exception);
		}
		return molecule;
	}
    
    public void close() throws IOException {
        input.close();
    }
    
    private void initIOSettings() {
        forceReadAs3DCoords = new BooleanIOSetting("ForceReadAs3DCoordinates", IOSetting.LOW,
          "Should coordinates always be read as 3D?", 
          "false");
    }
    
    public void customizeJob() {
        fireIOSettingQuestion(forceReadAs3DCoords);
    }

    public IOSetting[] getIOSettings() {
        IOSetting[] settings = new IOSetting[1];
        settings[0] = forceReadAs3DCoords;
        return settings;
    }
}