article.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

    /**
     * Returns an array of three-letter language abbreviations
     * indicating the language(s) in which the article was published.
     * A complete list of language abbreviations cna be found at:
     *
     * <blockquote>
     * <a href="http://www.nlm.nih.gov/bsd/language_table.html"
     >MEDLINE Language Table</a>
     * </blockquote>
     *
     * @return The abbreviations of languages in which the article appeared.
     */
    public String[] languages() {
        return mLanguages;
    }

    /**
     * Returns an object representing a list of linkages of
     * molecular sequences mentioned in the paper to their
     * accession numbers in a given data bank.  This method
     * may return <code>null</code> if there are no linkages.
     *
     * @return The data bank linkage list for this article.
     */
    public DataBankList dataBankList() {
        return mDataBankList;
    }

    /**
     * Returns the list of agencies and grant identifiers sponsoring
     * the work reporting in this article.  This method may return
     * <code>null</code> if no grants are provided.
     *
     * @return The grant list for this
     */
    public GrantList grantList() {
        return mGrantList;
    }

    /**
     * Returns a complete list of publication types for this article
     * in alphabetical order.  A full list of publication types
     * may be found at:
     *
     * <blockquote>
     * <a href="http://www.ncbi.nlm.nih.gov/entrez/query/static/help/pmhelp.html#PublicationTypes">PubMed Help: Publication Types</a>
     * </blockquote>
     *
     * @return The complete list of publication types for this article.
     */
    public String[] publicationTypes() {
        return mPublicationTypes;
    }

    /**
     * Returns the title of the article if it was originally published
     * in a language other than English.  All vernacular titles are
     * transliterated into Roman characters.  May be the empty string
     * if there is no vernacular title.
     *
     * @return The original language title for this article.
     */
    public String vernacularTitle() {
        return mVernacularTitle;
    }

    /**
     * Returns the date on which the publisher released an electronic
     * version of this article.  This method may return <code>null</code>
     * if there was no electronic publication date supplied by the
     * publisher or if the article was never published electronically.
     *
     * @return The date on which the publisher produced an electronic
     * version of this article.
     */
    public ArticleDate articleDate() {
        return mArticleDate;
    }

    /**
     * Returns a string representation of this article.
     *
     * @return A string representation of this article.
     */
    public String toString() {
        StringBuffer sb = new StringBuffer();
        sb.append('{');
        sb.append("Title=");
        sb.append(articleTitle());
        sb.append(" Authors=");
        sb.append(authorList());
        sb.append(" Publication Model=");
        sb.append(publicationModel());
        sb.append(" Journal=" + journal());
        if (pagination().length() > 0) {
            sb.append(" Pagination=" + pagination());
        }
        if (abstrct() != null && abstrct().text().length() > 0) {
            sb.append(" Abstract=");
            sb.append(abstrct());
        }
        if (affiliation().length() > 0) {
            sb.append(" Affiliation=");
            sb.append(affiliation());
        }
        sb.append(" Languages=");
        sb.append(Arrays.asList(languages()));
        if (dataBankList() != null) {
            sb.append(" Data Banks=");
            sb.append(dataBankList());
        }
        if (grantList() != null) {
            sb.append(" Grant List=");
            sb.append(grantList());
        }
        if (publicationTypes().length > 0) {
            sb.append(" Publication Types=");
            sb.append(Arrays.asList(publicationTypes()));
        }
        if (vernacularTitle() != null && vernacularTitle().length() > 0) {
            sb.append(" Vernacular Title=");
            sb.append(vernacularTitle());
        }
        if (articleDate() != null) {
            sb.append(" Electronic Article Pub Date=");
            sb.append(articleDate());
        }
        sb.append('}');
        return sb.toString();
    }

    private static final String AUTHORS_TRANS_MARKER = " (author's trans)";

    // NOTE: Element Book NOT USED IN MEDLINE!
    // <!ELEMENT Article ((Journal | Book), %ArticleTitle.Ref;,
    //                   Pagination, Abstract?,
    //                   Affiliation?, AuthorList?, Language+, DataBankList?,
    //                   GrantList?, PublicationTypeList, VernacularTitle?,
    //                   ArticleDate?)>
    // <!ENTITY % ArticleTitle.Ref "ArticleTitle">
    // <!ELEMENT PublicationType (#PCDATA)>
    // <!ELEMENT PublicationTypeList (PublicationType+)>
    // <!ELEMENT VernacularTitle (#PCDATA)>
    static class Handler extends DelegateHandler {
        private String mPublicationModel;
        private final Journal.Handler mJournalHandler;
        private final TextAccumulatorHandler mArticleTitleHandler
            = new TextAccumulatorHandler();
        private final TextAccumulatorHandler mMedlinePaginationHandler
            = new TextAccumulatorHandler();
        private final Abstract.Handler mAbstractHandler;
        private final TextAccumulatorHandler mAffiliationHandler
            = new TextAccumulatorHandler();
        String mPagination;
        private final ELocationId.Handler mELocationIdHandler
            =  new ELocationId.Handler();
        private final List<ELocationId> mELocationIdList
            = new ArrayList<ELocationId>();
        private final TextAccumulatorHandler mStartPageHandler
            = new TextAccumulatorHandler();
        private final TextAccumulatorHandler mEndPageHandler
            = new TextAccumulatorHandler();
        private final AuthorList.Handler mAuthorListHandler;
        private final ArrayList mLanguageList = new ArrayList();
        private final TextAccumulatorHandler mLanguageHandler
            = new TextAccumulatorHandler();
        private final DataBankList.Handler mDataBankListHandler;
        private final GrantList.Handler mGrantListHandler;
        private final ArrayList mPublicationTypeList = new ArrayList();
        private final TextAccumulatorHandler mPublicationTypeHandler
            = new TextAccumulatorHandler();
        private final TextAccumulatorHandler mVernacularTitleHandler
            = new TextAccumulatorHandler();
        private final ArticleDate.Handler mArticleDateHandler;
        public Handler(DelegatingHandler delegator) {
            super(delegator);
            mJournalHandler = new Journal.Handler(delegator);
            mAbstractHandler = new Abstract.Handler(delegator);
            mAuthorListHandler = new AuthorList.Handler(delegator);
            mDataBankListHandler = new DataBankList.Handler(delegator);
            mGrantListHandler = new GrantList.Handler(delegator);
            mArticleDateHandler = new ArticleDate.Handler(delegator);
            setDelegate(MedlineCitationSet.JOURNAL_ELT,
                        mJournalHandler);
            setDelegate(MedlineCitationSet.ARTICLE_TITLE_ELT,
                        mArticleTitleHandler);
            setDelegate(MedlineCitationSet.MEDLINE_PAGINATION_ELT,
                        mMedlinePaginationHandler);
            setDelegate(MedlineCitationSet.E_LOCATION_ID_ELT,
                        mELocationIdHandler);
            setDelegate(MedlineCitationSet.START_PAGE_ELT,
                        mStartPageHandler);
            setDelegate(MedlineCitationSet.END_PAGE_ELT,
                        mEndPageHandler);
            setDelegate(MedlineCitationSet.ABSTRACT_ELT,
                        mAbstractHandler);
            setDelegate(MedlineCitationSet.AFFILIATION_ELT,
                        mAffiliationHandler);
            setDelegate(MedlineCitationSet.AUTHOR_LIST_ELT,
                        mAuthorListHandler);
            setDelegate(MedlineCitationSet.LANGUAGE_ELT,
                        mLanguageHandler);
            setDelegate(MedlineCitationSet.DATA_BANK_LIST_ELT,
                        mDataBankListHandler);
            setDelegate(MedlineCitationSet.GRANT_LIST_ELT,
                        mGrantListHandler);
            setDelegate(MedlineCitationSet.PUBLICATION_TYPE_ELT,
                        mPublicationTypeHandler);
            setDelegate(MedlineCitationSet.VERNACULAR_TITLE_ELT,
                        mVernacularTitleHandler);
            setDelegate(MedlineCitationSet.ELECTRONIC_PUB_DATE_ELT,
                        mArticleDateHandler);
        }
        public void startDocument() throws SAXException {
            super.startDocument();
            mJournalHandler.reset();
            mArticleTitleHandler.reset();
            mMedlinePaginationHandler.reset();
            mPagination = null;
            mStartPageHandler.reset();
            mEndPageHandler.reset();
            mAbstractHandler.reset();
            mAffiliationHandler.reset();
            mAuthorListHandler.reset();
            mLanguageList.clear();
            mDataBankListHandler.reset();
            mGrantListHandler.reset();
            mPublicationTypeList.clear();
            mVernacularTitleHandler.reset();
            mArticleDateHandler.reset();
            mELocationIdList.clear();
            super.startDocument();
        }
        public void startElement(String x, String y, String qName,
                                 Attributes atts)
            throws SAXException {
            if (qName.equals(MedlineCitationSet.ARTICLE_ELT)) {
                mPublicationModel
                    = atts.getValue(MedlineCitationSet.PUB_MODEL_ATT);
            }
            super.startElement(x,y,qName,atts);
        }
        public Article getArticle() {
            String[] languages = new String[mLanguageList.size()];
            mLanguageList.toArray(languages);
            String[] publicationTypes
                = new String[mPublicationTypeList.size()];
            mPublicationTypeList.toArray(publicationTypes);
            ELocationId[] eLocationIds = new ELocationId[mELocationIdList.size()];
            mELocationIdList.<ELocationId>toArray(eLocationIds);
            return new Article(mPublicationModel,
                               mJournalHandler.getJournal(),
                               mArticleTitleHandler.getText(),
                               mPagination,
                               eLocationIds,
                               mAbstractHandler.getAbstract(),
                               mAffiliationHandler.getText(),
                               mAuthorListHandler.getAuthorList(),
                               languages,
                               mDataBankListHandler.getDataBankList(),
                               mGrantListHandler.getGrantList(),
                               publicationTypes,
                               mVernacularTitleHandler.getText(),
                               mArticleDateHandler.getArticleDate());
        }
        public void finishDelegate(String qName, DefaultHandler handler) {
            if (qName.equals(MedlineCitationSet.MEDLINE_PAGINATION_ELT)) {
                mPagination = mMedlinePaginationHandler.getText();
            } else if (qName.equals(MedlineCitationSet.START_PAGE_ELT)) {
                mPagination = mStartPageHandler.getText();
            } else if (qName.equals(MedlineCitationSet.END_PAGE_ELT)) {
                if (mPagination.length() > 0)
                    mPagination = mPagination + "-" + mEndPageHandler.getText();
                else
                    mPagination = mEndPageHandler.getText();
            } else if (qName.equals(MedlineCitationSet.LANGUAGE_ELT)) {
                mLanguageList.add(mLanguageHandler.getText());
            } else if (qName.equals(MedlineCitationSet.PUBLICATION_TYPE_ELT)) {
                mPublicationTypeList.add(mPublicationTypeHandler.getText());
            } else if (qName.equals(MedlineCitationSet.E_LOCATION_ID_ELT)) {
                mELocationIdList.add(mELocationIdHandler.getELocationId());
            }
        }
    }

}