cmlcoremodule.java

化学图形处理软件

                logger.debug("New coord3 xyz3 found: ", currentChars);
                
                try {
                    
                    StringTokenizer st = new StringTokenizer(currentChars);
                    x3.add(st.nextToken());
                    y3.add(st.nextToken());
                    z3.add(st.nextToken());
                    logger.debug("coord3 x3.length: ", x3.size());
                    logger.debug("coord3 y3.length: ", y3.size());
                    logger.debug("coord3 z3.length: ", z3.size());
                } catch (Exception exception) {
                    logger.error(
                    "CMLParsing error while setting coordinate3!");
                    logger.debug(exception);
                }
            } else {
                logger.warn("Unknown coordinate3 BUILTIN: " + BUILTIN);
            }
        } else if ("string".equals(name)) {
            if (BUILTIN.equals("elementType")) {
                logger.debug("Element: ", cData.trim());
                elsym.add(cData);
            } else if (BUILTIN.equals("atomRef")) {
                curRef++;
                logger.debug("Bond: ref #", curRef);
                
                if (curRef == 1) {
                    bondARef1.add(cData.trim());
                } else if (curRef == 2) {
                    bondARef2.add(cData.trim());
                }
            } else if (BUILTIN.equals("order")) {
                logger.debug("Bond: order ", cData.trim());
                order.add(cData.trim());
            } else if (BUILTIN.equals("formalCharge")) {
                // NOTE: this combination is in violation of the CML DTD!!!
                logger.warn("formalCharge BUILTIN accepted but violating CML DTD");
                logger.debug("Charge: ", cData.trim());
                String charge = cData.trim();
                if (charge.startsWith("+") && charge.length() > 1) {
                    charge = charge.substring(1);
                }
                formalCharges.add(charge);
            }
        } else if ("float".equals(name)) {
            if (BUILTIN.equals("x3")) {
                x3.add(cData.trim());
            } else if (BUILTIN.equals("y3")) {
                y3.add(cData.trim());
            } else if (BUILTIN.equals("z3")) {
                z3.add(cData.trim());
            } else if (BUILTIN.equals("x2")) {
                x2.add(cData.trim());
            } else if (BUILTIN.equals("y2")) {
                y2.add(cData.trim());
            } else if (BUILTIN.equals("order")) {
                // NOTE: this combination is in violation of the CML DTD!!!
                order.add(cData.trim());
            } else if (BUILTIN.equals("charge") || BUILTIN.equals("partialCharge")) {
                partialCharges.add(cData.trim());
            }
        } else if ("integer".equals(name)) {
            if (BUILTIN.equals("formalCharge")) {
                formalCharges.add(cData.trim());
            }
        } else if ("coordinate2".equals(name)) {
            if (BUILTIN.equals("xy2")) {
                logger.debug("New coord2 xy2 found.", cData);
                
                try {
                    
                    StringTokenizer st = new StringTokenizer(cData);
                    x2.add(st.nextToken());
                    y2.add(st.nextToken());
                } catch (Exception e) {
                    notify("CMLParsing error: " + e, SYSTEMID, 175, 1);
                }
            }
        } else if ("stringArray".equals(name)) {
            if (BUILTIN.equals("id") || BUILTIN.equals("atomId")
                || BUILTIN.equals("atomID")) { // invalid according to CML1 DTD but found in OpenBabel 1.x output
                
                try {
                    boolean countAtoms = (atomCounter == 0) ? true : false;
                    StringTokenizer st = new StringTokenizer(cData);
                    
                    while (st.hasMoreTokens()) {
                        if (countAtoms) { atomCounter++; }
                        String token = st.nextToken();
                        logger.debug("StringArray (Token): ", token);
                        elid.add(token);
                    }
                } catch (Exception e) {
                    notify("CMLParsing error: " + e, SYSTEMID, 186, 1);
                }
            } else if (BUILTIN.equals("elementType")) {
                
                try {
                    boolean countAtoms = (atomCounter == 0) ? true : false;
                    StringTokenizer st = new StringTokenizer(cData);
                    
                    while (st.hasMoreTokens()) {
                        if (countAtoms) { atomCounter++; }
                        elsym.add(st.nextToken());
                    }
                } catch (Exception e) {
                    notify("CMLParsing error: " + e, SYSTEMID, 194, 1);
                }
            } else if (BUILTIN.equals("atomRefs")) {
                curRef++;
                logger.debug("New atomRefs found: ", curRef);
                
                try {
                    boolean countBonds = (bondCounter == 0) ? true : false;
                    StringTokenizer st = new StringTokenizer(cData);
                    
                    while (st.hasMoreTokens()) {
                        if (countBonds) { bondCounter++; }
                        String token = st.nextToken();
                        logger.debug("Token: ", token);
                        
                        if (curRef == 1) {
                            bondARef1.add(token);
                        } else if (curRef == 2) {
                            bondARef2.add(token);
                        }
                    }
                } catch (Exception e) {
                    notify("CMLParsing error: " + e, SYSTEMID, 194, 1);
                }
            } else if (BUILTIN.equals("atomRef")) {
                curRef++;
                logger.debug("New atomRef found: ", curRef); // this is CML1 stuff, we get things like:
                /*
                  <bondArray>
                  <stringArray builtin="atomRef">a2 a2 a2 a2 a3 a3 a4 a4 a5 a6 a7 a9</stringArray>
                  <stringArray builtin="atomRef">a9 a11 a12 a13 a5 a4 a6 a9 a7 a8 a8 a10</stringArray>
                  <stringArray builtin="order">1 1 1 1 2 1 2 1 1 1 2 2</stringArray>
                  </bondArray>
                */
                
                try {
                    boolean countBonds = (bondCounter == 0) ? true : false;
                    StringTokenizer st = new StringTokenizer(cData);
                    
                    while (st.hasMoreTokens()) {
                        if (countBonds) { bondCounter++; }
                        String token = st.nextToken();
                        logger.debug("Token: ", token);
                        
                        if (curRef == 1) {
                            bondARef1.add(token);
                        } else if (curRef == 2) {
                            bondARef2.add(token);
                        }
                    }
                } catch (Exception e) {
                    notify("CMLParsing error: " + e, SYSTEMID, 194, 1);
                }
            } else if (BUILTIN.equals("order")) {
                logger.debug("New bond order found.");
                
                try {
                    
                    StringTokenizer st = new StringTokenizer(cData);
                    
                    while (st.hasMoreTokens()) {
                        
                        String token = st.nextToken();
                        logger.debug("Token: ", token);
                        order.add(token);
                    }
                } catch (Exception e) {
                    notify("CMLParsing error: " + e, SYSTEMID, 194, 1);
                }
            }
        } else if ("integerArray".equals(name)) {
            logger.debug("IntegerArray: builtin = ", BUILTIN);
            
            if (BUILTIN.equals("formalCharge")) {
                
                try {
                    
                    StringTokenizer st = new StringTokenizer(cData);
                    
                    while (st.hasMoreTokens()) {
                        
                        String token = st.nextToken();
                        logger.debug("Charge added: ", token);
                        formalCharges.add(token);
                    }
                } catch (Exception e) {
                    notify("CMLParsing error: " + e, SYSTEMID, 205, 1);
                }
            }
        } else if ("scalar".equals(name)) {
            if (xpath.endsWith("crystal", "scalar")) {
                logger.debug("Going to set a crystal parameter: " + crystalScalar, 
                    " to ", cData);
                try {
                    unitcellparams[crystalScalar-1] = Double.parseDouble(cData.trim());
                } catch (NumberFormatException exception) {
                    logger.error("Content must a float: " + cData);
                }
            } else if (xpath.endsWith("bond", "scalar")) {
                if (DICTREF.equals("mdl:stereo")) {
                	bondStereo.add(cData.trim());
                    stereoGiven=true;
                }
            } else if (xpath.endsWith("atom", "scalar")) {
                if (DICTREF.equals("cdk:partialCharge")) {
                    partialCharges.add(cData.trim());
                }
            } else if (xpath.endsWith("molecule", "scalar")) {
                if (DICTREF.equals("pdb:id")) {
//                	cdo.setObjectProperty("Molecule", DICTREF, cData);
                	currentMolecule.setProperty(new DictRef(DICTREF, cData), cData);
                } else if (DICTREF.equals("cdk:molecularProperty")) {
                	currentMolecule.setProperty(elementTitle, cData);
                }
            } else {
                logger.warn("Ignoring scalar: " + xpath);
            }
        } else if ("floatArray".equals(name)) {
            if (BUILTIN.equals("x3")) {
                
                try {
                    
                    StringTokenizer st = new StringTokenizer(cData);
                    
                    while (st.hasMoreTokens())
                        x3.add(st.nextToken());
                } catch (Exception e) {
                    notify("CMLParsing error: " + e, SYSTEMID, 205, 1);
                }
            } else if (BUILTIN.equals("y3")) {
                
                try {
                    
                    StringTokenizer st = new StringTokenizer(cData);
                    
                    while (st.hasMoreTokens())
                        y3.add(st.nextToken());
                } catch (Exception e) {
                    notify("CMLParsing error: " + e, SYSTEMID, 213, 1);
                }
            } else if (BUILTIN.equals("z3")) {
                
                try {
                    
                    StringTokenizer st = new StringTokenizer(cData);
                    
                    while (st.hasMoreTokens())
                        z3.add(st.nextToken());
                } catch (Exception e) {
                    notify("CMLParsing error: " + e, SYSTEMID, 221, 1);
                }
            } else if (BUILTIN.equals("x2")) {
                logger.debug("New floatArray found.");
                
                try {
                    
                    StringTokenizer st = new StringTokenizer(cData);
                    
                    while (st.hasMoreTokens())
                        x2.add(st.nextToken());
                } catch (Exception e) {
                    notify("CMLParsing error: " + e, SYSTEMID, 205, 1);
                }
            } else if (BUILTIN.equals("y2")) {
                logger.debug("New floatArray found.");
                
                try {
                    
                    StringTokenizer st = new StringTokenizer(cData);
                    
                    while (st.hasMoreTokens())
                        y2.add(st.nextToken());
                } catch (Exception e) {
                    notify("CMLParsing error: " + e, SYSTEMID, 454, 1);
                }
            } else if (BUILTIN.equals("partialCharge")) {
                logger.debug("New floatArray with partial charges found.");
                
                try {
                    
                    StringTokenizer st = new StringTokenizer(cData);
                    
                    while (st.hasMoreTokens())
                        partialCharges.add(st.nextToken());
                } catch (Exception e) {
                    notify("CMLParsing error: " + e, SYSTEMID, 462, 1);
                }
            }
        } else if ("basic".equals(name)) {
            // assuming this is the child element of <identifier>
            this.inchi = cData;
        } else if ("name".equals(name)) {
            if (xpath.endsWith("molecule", "name")) {
            	if (DICTREF.length() > 0) {
//            		cdo.setObjectProperty("Molecule", DICTREF, cData);
            		
            		currentMolecule.setProperty(new DictRef(DICTREF, cData), cData);
            	} else {
//            		cdo.setObjectProperty("Molecule", "Name", cData);
            		currentMolecule.setProperty(CDKConstants.TITLE, cData);
            	}
            }
        } else {
            logger.warn("Skipping element: " + name);
        }

        currentChars = "";
        BUILTIN = "";
        elementTitle = "";
    }

    public void characterData(CMLStack xpath, char[] ch, int start, int length) {
        currentChars = currentChars + new String(ch, start, length);
        logger.debug("CD: ", currentChars);
    }

    protected void notify(String message, String systemId, int line, 
                          int column) {
        logger.debug("Message: ", message);
        logger.debug("SystemId: ", systemId);
        logger.debug("Line: ", line);
        logger.debug("Column: ", column);
    }

    protected void storeData() {
        if (inchi != null) {
//            cdo.setObjectProperty("Molecule", "inchi", inchi);
        	currentMolecule.setProperty(CDKConstants.INCHI, inchi);
        }
        storeAtomData();
        storeBondData();
    }

    protected void storeAtomData() {
        logger.debug("No atoms: ", atomCounter);
        if (atomCounter == 0) {
            return;
        }

        boolean hasID = false;
        boolean has3D = false;
        boolean has3Dfract = false;
        boolean has2D = false;
        boolean hasFormalCharge = false;
        boolean hasPartialCharge = false;
        boolean hasHCounts = false;
        boolean hasSymbols = false;
        boolean hasTitles = false;
        boolean hasIsotopes = false;