medlinecitation.java

一个自然语言处理的Java开源工具包。LingPipe目前已有很丰富的功能

            sb.append("\n");
        }
        if (mOtherAbstracts.length > 0) {
            sb.append("     OTHER ABSTRACTS="
                      + Arrays.asList(mOtherAbstracts));
            sb.append("\n");
        }
        if (mCommentOrCorrections.length > 0) {
            sb.append("     COMMENTS OR CORRECTIONS="
                      + Arrays.asList(mCommentOrCorrections));
            sb.append("\n");
        }
        // sb.append(" XML=");
        // sb.append(xmlString());
        return sb.toString();
    }

    void setXMLString(String xmlString) {
        mXMLString = xmlString;
    }


    public static MedlineCitation parse(InputSource inSource,
                                        XMLReader xmlReader)
        throws IOException, SAXException {

        DelegatingHandler handler = new DelegatingHandler();
        Handler citationHandler = new Handler(handler);
        handler.setDelegate(MedlineCitationSet.MEDLINE_CITATION_ELT,
                            citationHandler);
        xmlReader.setContentHandler(handler);
        xmlReader.parse(inSource);
        return citationHandler.getCitation();
    }

    // <!ELEMENT MedlineCitation (%NlmDcmsID.Ref;, %PMID.Ref;,
    //                            %DateCreated.Ref;, DateCompleted?,
    //                            DateRevised?, Article, MedlineJournalInfo,
    //                            ChemicalList?,
    //                            CitationSubset*, CommentsCorrections?,
    //                            GeneSymbolList?,
    //                            MeshHeadingList?, NumberOfReferences?,
    //                            PersonalNameSubjectList?,
    //                            OtherID*, OtherAbstract*, KeywordList*,
    //                            SpaceFlightMission*,
    //                            InvestigatorList?, GeneralNote*)>
    // <!ATTLIST MedlineCitation
    //           Owner %Owner; "NLM"
    //           Status %Status; >
    // <!ENTITY % Owner "(NLM | NASA | PIP | KIE | HSR | HMD | SIS | NOTNLM)">
    // <!ENTITY % Status "(Completed | In-Process | PubMed-not-MEDLINE |
    //                    In-Data-Review | Publisher) #REQUIRED">
    // <!ENTITY % NlmDcmsID.Ref "NlmDcmsID?">
    // <!ENTITY % PMID.Ref "PMID">
    // <!ENTITY % PubDate.Ref "PubDate">
    // <!ENTITY % DateCreated.Ref "DateCreated">
    // <!ELEMENT PersonalNameSubjectList (PersonalNameSubject+)>
    // returned citation will have null XML bytes, which is OK
    // <!ELEMENT GeneSymbol (#PCDATA)>
    // <!ELEMENT GeneSymbolList (GeneSymbol+)>
    static class Handler extends DelegateHandler {
        private String mOwner;
        private String mStatus;
        private final TextAccumulatorHandler mNlmDcmsIDHandler
            = new TextAccumulatorHandler();
        private final TextAccumulatorHandler mPMIDHandler
            = new TextAccumulatorHandler();
        private final DateHandler mDateCreatedHandler;
        private final DateHandler mDateCompletedHandler;
        private final DateHandler mDateRevisedHandler;
        private final Article.Handler mArticleHandler;
        private final JournalInfo.Handler mJournalInfoHandler;
        private final ArrayList mChemicalList = new ArrayList();
        private final Chemical.Handler mChemicalHandler;
        private final ArrayList mCitationSubsetList = new ArrayList();
        private final TextAccumulatorHandler mCitationSubsetHandler
            = new TextAccumulatorHandler();
        private final ArrayList mCommentOrCorrectionList = new ArrayList();
        private final ArrayList mGeneSymbolList = new ArrayList();
        private final TextAccumulatorHandler mGeneSymbolHandler
            = new TextAccumulatorHandler();
        private final MeshHeading.ListHandler mMeshHeadingListHandler;
        private final TextAccumulatorHandler mNumberOfReferencesHandler
            = new TextAccumulatorHandler();
        private final ArrayList mPersonalNameSubjectList = new ArrayList();
        private final PersonalNameSubject.Handler mPersonalNameSubjectHandler;
        private final ArrayList mOtherIDList = new ArrayList();
        private final OtherID.Handler mOtherIDHandler;
        private final ArrayList mOtherAbstractList = new ArrayList();
        private final OtherAbstract.Handler mOtherAbstractHandler;
        private final ArrayList mKeywordListList = new ArrayList();
        private final KeywordList.Handler mKeywordListHandler;
        private final ArrayList mSpaceFlightMissionList = new ArrayList();
        private final TextAccumulatorHandler mSpaceFlightMissionHandler
            = new TextAccumulatorHandler();
        private final ArrayList mInvestigatorList = new ArrayList();
        private final Investigator.Handler mInvestigatorHandler;
        private final ArrayList mGeneralNoteList = new ArrayList();
        private final GeneralNote.Handler mGeneralNoteHandler;
        public Handler(DelegatingHandler delegator) {
            super(delegator);
            mDateCreatedHandler = new DateHandler(delegator);
            mDateCompletedHandler = new DateHandler(delegator);
            mDateRevisedHandler = new DateHandler(delegator);
            mArticleHandler = new Article.Handler(delegator);
            mJournalInfoHandler = new JournalInfo.Handler(delegator);
            mChemicalHandler = new Chemical.Handler(delegator);
            mMeshHeadingListHandler
                = new MeshHeading.ListHandler(delegator);
            mPersonalNameSubjectHandler
                = new PersonalNameSubject.Handler(delegator);
            mOtherIDHandler = new OtherID.Handler();
            mOtherAbstractHandler = new OtherAbstract.Handler(delegator);
            mKeywordListHandler = new KeywordList.Handler(delegator);
            mInvestigatorHandler = new Investigator.Handler(delegator);
            mGeneralNoteHandler = new GeneralNote.Handler();

            setDelegate(MedlineCitationSet.NLM_DCMS_ID_ELT,
                        mNlmDcmsIDHandler);
            setDelegate(MedlineCitationSet.PMID_ELT,mPMIDHandler);
            setDelegate(MedlineCitationSet.DATE_CREATED_ELT,
                        mDateCreatedHandler);
            setDelegate(MedlineCitationSet.DATE_COMPLETED_ELT,
                        mDateCompletedHandler);
            setDelegate(MedlineCitationSet.DATE_REVISED_ELT,
                        mDateRevisedHandler);
            setDelegate(MedlineCitationSet.CHEMICAL_ELT,
                        mChemicalHandler);
            setDelegate(MedlineCitationSet.CITATION_SUBSET_ELT,
                        mCitationSubsetHandler);
            setDelegate(MedlineCitationSet.MESH_HEADING_LIST_ELT,
                        mMeshHeadingListHandler);
            setDelegate(MedlineCitationSet.NUMBER_OF_REFERENCES_ELT,
                        mNumberOfReferencesHandler);
            setDelegate(MedlineCitationSet.PERSONAL_NAME_SUBJECT_ELT,
                        mPersonalNameSubjectHandler);
            setDelegate(MedlineCitationSet.KEYWORD_LIST_ELT,
                        mKeywordListHandler);
            setDelegate(MedlineCitationSet.GENERAL_NOTE_ELT,
                        mGeneralNoteHandler);
            setDelegate(MedlineCitationSet.GENE_SYMBOL_ELT,
                        mGeneSymbolHandler);
            setDelegate(MedlineCitationSet.SPACE_FLIGHT_MISSION_ELT,
                        mSpaceFlightMissionHandler);
            setDelegate(MedlineCitationSet.OTHER_ID_ELT,
                        mOtherIDHandler);
            setDelegate(MedlineCitationSet.INVESTIGATOR_ELT,
                        mInvestigatorHandler);
            setDelegate(MedlineCitationSet.OTHER_ABSTRACT_ELT,
                        mOtherAbstractHandler);
            for (int i = 0;
                 i < MedlineCitationSet.COMMENT_OR_CORRECTIONS.length; ++i) {
                String elt = MedlineCitationSet.COMMENT_OR_CORRECTIONS[i];
                setDelegate(elt,new CommentOrCorrection.Handler(elt,
                                                                delegator));
            }
            setDelegate(MedlineCitationSet.MEDLINE_JOURNAL_INFO_ELT,
                        mJournalInfoHandler);
            setDelegate(MedlineCitationSet.ARTICLE_ELT,mArticleHandler);
        }
        public void finishDelegate(String qName, DefaultHandler handler) {
            if (qName.equals(MedlineCitationSet.CITATION_SUBSET_ELT)) {
                mCitationSubsetList.add(mCitationSubsetHandler.getText());
            } else if (qName.equals(MedlineCitationSet.CHEMICAL_ELT)) {
                mChemicalList.add(mChemicalHandler.getChemical());
            } else if (qName.equals(MedlineCitationSet.KEYWORD_LIST_ELT)) {
                mKeywordListList.add(mKeywordListHandler.getKeywordList());
            } else if (qName.equals(MedlineCitationSet.GENERAL_NOTE_ELT)) {
                mGeneralNoteList.add(mGeneralNoteHandler.getNote());
            } else if (qName.equals(MedlineCitationSet.GENE_SYMBOL_ELT)) {
                mGeneSymbolList.add(mGeneSymbolHandler.getText());
            } else if (qName.equals(MedlineCitationSet
                                    .SPACE_FLIGHT_MISSION_ELT)) {
                mSpaceFlightMissionList.add(mSpaceFlightMissionHandler
                                            .getText());
            } else if (qName.equals(MedlineCitationSet.OTHER_ID_ELT)) {
                mOtherIDList.add(mOtherIDHandler.getOtherID());
            } else if (qName.equals(MedlineCitationSet.INVESTIGATOR_ELT)) {
                mInvestigatorList.add(mInvestigatorHandler.getInvestigator());
            } else if (qName.equals(MedlineCitationSet.OTHER_ABSTRACT_ELT)) {
                mOtherAbstractList.add(mOtherAbstractHandler
                                       .getOtherAbstract());
            } else if (MedlineCitationSet.COMMENT_OR_CORRECTION_SET
                       .contains(qName)) {
                CommentOrCorrection.Handler ccHandler
                    = (CommentOrCorrection.Handler) handler;
                mCommentOrCorrectionList.add(ccHandler
                                             .getCommentOrCorrection());
            } else if (qName.equals(MedlineCitationSet
                                    .PERSONAL_NAME_SUBJECT_ELT)) {
                mPersonalNameSubjectList
                    .add(mPersonalNameSubjectHandler
                         .getPersonalNameSubject());
            }
        }
        public void startDocument() throws SAXException {
            super.startDocument();
            mNlmDcmsIDHandler.reset();
            mPMIDHandler.reset();
            mDateCreatedHandler.reset();
            mDateCompletedHandler.reset();
            mDateRevisedHandler.reset();
            mCitationSubsetList.clear();
            mChemicalList.clear();
            mKeywordListList.clear();
            mKeywordListHandler.reset();
            mGeneralNoteList.clear();
            mGeneSymbolList.clear();
            mSpaceFlightMissionList.clear();
            mOtherIDList.clear();
            mInvestigatorList.clear();
            mOtherAbstractList.clear();
            mCommentOrCorrectionList.clear();
            mMeshHeadingListHandler.reset();
            mNumberOfReferencesHandler.reset();
            mPersonalNameSubjectList.clear();
        }
        public void startElement(String namespaceURI, String localName,
                                 String qName, Attributes atts)
            throws SAXException {

            if (qName.equals(MedlineCitationSet.MEDLINE_CITATION_ELT)) {
                mOwner = atts.getValue(MedlineCitationSet.OWNER_ATT);
                mStatus = atts.getValue(MedlineCitationSet.STATUS_ATT);
            } else {
                super.startElement(namespaceURI,localName,qName,atts);
            }
        }
        public MedlineCitation getCitation() {
            String[] citationSubsets = new String[mCitationSubsetList.size()];
            mCitationSubsetList.toArray(citationSubsets);
            Chemical[] chemicals = new Chemical[mChemicalList.size()];
            mChemicalList.toArray(chemicals);
            KeywordList[] keywordLists
                = new KeywordList[mKeywordListList.size()];
            mKeywordListList.toArray(keywordLists);
            GeneralNote[] generalNotes
                = new GeneralNote[mGeneralNoteList.size()];
            mGeneralNoteList.toArray(generalNotes);
            String[] geneSymbols = new String[mGeneSymbolList.size()];
            mGeneSymbolList.toArray(geneSymbols);
            String[] spaceFlightMissions
                = new String[mSpaceFlightMissionList.size()];
            mSpaceFlightMissionList.toArray(spaceFlightMissions);
            OtherID[] otherIDs = new OtherID[mOtherIDList.size()];
            mOtherIDList.toArray(otherIDs);
            Investigator[] investigators
                = new Investigator[mInvestigatorList.size()];
            mInvestigatorList.toArray(investigators);
            OtherAbstract[] otherAbstracts
                = new OtherAbstract[mOtherAbstractList.size()];
            mOtherAbstractList.toArray(otherAbstracts);
            CommentOrCorrection[] commentOrCorrections
                = new CommentOrCorrection[mCommentOrCorrectionList.size()];
            mCommentOrCorrectionList.toArray(commentOrCorrections);
            PersonalNameSubject[] personalNameSubjects
                = new PersonalNameSubject[mPersonalNameSubjectList.size()];
            mPersonalNameSubjectList.toArray(personalNameSubjects);
            return new MedlineCitation(mOwner,
                                       mStatus,
                                       mNlmDcmsIDHandler.getText(),
                                       mPMIDHandler.getText(),
                                       mDateCreatedHandler.getDate(),
                                       mDateCompletedHandler.getDate(),
                                       mDateRevisedHandler.getDate(),
                                       chemicals,
                                       citationSubsets,
                                       mMeshHeadingListHandler
                                       .getMeshHeadings(),
                                       mNumberOfReferencesHandler.getText(),
                                       personalNameSubjects,
                                       keywordLists,
                                       generalNotes,
                                       geneSymbols,
                                       spaceFlightMissions,
                                       otherIDs,
                                       investigators,
                                       otherAbstracts,
                                       commentOrCorrections,
                                       mJournalInfoHandler.getJournalInfo(),
                                       mArticleHandler.getArticle());
        }
    }


    // <!ENTITY % normal.date "(Year,Month,Day,(Hour,(Minute,Second?)?)?)">
    static class DateHandler extends DelegateHandler {
        private final DateFormat mDateFormat
            = new SimpleDateFormat("yyyy.MM.dd",Locale.ENGLISH);
        private final TextAccumulatorHandler mYearHandler
            = new TextAccumulatorHandler();
        private final TextAccumulatorHandler mMonthHandler
            = new TextAccumulatorHandler();
        private final TextAccumulatorHandler mDayHandler
            = new TextAccumulatorHandler();
        public DateHandler(DelegatingHandler delegator) {
            super(delegator);
            setDelegate(MedlineCitationSet.YEAR_ELT,mYearHandler);
            setDelegate(MedlineCitationSet.MONTH_ELT,mMonthHandler);
            setDelegate(MedlineCitationSet.DAY_ELT,mDayHandler);
        }
        public void startDocument() throws SAXException {
            super.startDocument();
            reset();
        }
        public void reset() {
            mYearHandler.reset();
            mMonthHandler.reset();
            mDayHandler.reset();
        }
        public Date getDate() {
            String yyyy = mYearHandler.getText();
            String mm = mMonthHandler.getText();
            String dd = mDayHandler.getText();
            String date = yyyy + '.' + mm + '.' + dd;
            if (date.length() < 3) return null;
            try {
                return mDateFormat.parse(date);
            } catch (ParseException e) {
                // log
                return null;
            }
        }
    }


}