Example usage for org.dom4j Element add

List of usage examples for org.dom4j Element add

  1. HOME
  2. Java
  3. org.dom4j
  4. org.dom4j.*
  5. Element
  6. add

Introduction

In this page you can find the example usage for org.dom4j Element add.

Prototype

void add(Namespace namespace);

Source Link

Document

Adds the given Namespace to this element.

Usage

From source file:eu.sisob.uma.crawler.ResearcherXMLFileSplitter.java

License:Open Source License

/**
 * Reader folder of one researcher and takes the uri of clean file for to make infoblock. 
 * @param elementResearcher/*www  .  j a  va2s.c om*/
 * @param path
 * @param sInstitutionName
 * @param sWebAddress
 * @param sUnitOfAssessment_Description
 * @param sResearchGroupDescription
 * @param sResearchName
 * @param sResearchInitials
 * @param sStaffIndentifier
 */
@Override
protected boolean actionsInInstitutionNode(org.dom4j.Element elementInstitution, String path,
        String sInstitutionName, String sWebAddress) {
    Document split_doc = DocumentHelper.createDocument();
    Element split_root = split_doc.addElement("root");
    split_root.add((Element) elementInstitution.clone());
    File split_file = new File(work_dir, prefix_file + sInstitutionName.replaceAll("\\W+", "").toLowerCase());
    try {
        FileUtils.write(split_file, split_doc.asXML(), "UTF-8");
    } catch (IOException ex) {
        Logger.getLogger("roor").error("Cannot create xml file for " + split_file.getPath());
    }

    return true;
}

From source file:eu.sisob.uma.NPL.Researchers.GateResearcherAnnCollector.java

License:Open Source License

/**
 *
 * @param doc/*from  ww  w . j  a  va 2s . c om*/
 * @param aoData
 */
@Override
public void collect(Object doc, MiddleData aoData) {
    int n_expressions = 0;
    org.dom4j.Element eOut = org.dom4j.DocumentFactory.getInstance().createElement("blockinfo");
    eOut.addAttribute(DataExchangeLiterals.MIDDLE_ELEMENT_XML_ID_ANNOTATIONRECOLLECTING,
            aoData.getId_annotationrecollecting()); // aoData[MiddleData.I_INDEX_DATA_TYPE].toString());
    eOut.addAttribute(DataExchangeLiterals.MIDDLE_ELEMENT_XML_ID_ENTITY_ATT, aoData.getId_entity()); // aoData[MiddleData.I_INDEX_DATA_ID].toString());

    gate.Document docGate = (gate.Document) doc;

    HashMap<String, String> extra_data = null;
    try {
        extra_data = (HashMap<String, String>) aoData.getData_extra();
    } catch (Exception ex) {
        extra_data = null;
    }
    boolean collect_expressions = false;
    List<String> expression_to_collect = new ArrayList<String>();
    if (extra_data != null) {
        if (extra_data.containsKey(DataExchangeLiterals.MIDDLE_ELEMENT_XML_EXTRADATA_BLOCK_TYPE)) {

            String block_type = extra_data.get(DataExchangeLiterals.MIDDLE_ELEMENT_XML_EXTRADATA_BLOCK_TYPE);
            if (block_type.equals(CVBlocks.CVBLOCK_REST.toString())) {
                expression_to_collect.add("ProfessionalActivityCurrent");
                expression_to_collect.add("ProfessionalActivityNoCurrent");
                expression_to_collect.add("AccreditedUniversityStudiesOtherPostGrade");
                expression_to_collect.add("AccreditedUniversityStudiesDegree");
                expression_to_collect.add("AccreditedUniversityStudiesPhDStudies");
                collect_expressions = true;
            } else if (block_type.equals(CVBlocks.CVBLOCK_PROFESSIONAL_ACTIVITY.toString())) {
                expression_to_collect.add("ProfessionalActivityCurrent");
                expression_to_collect.add("ProfessionalActivityNoCurrent");
                collect_expressions = true;
            } else if (block_type.equals(CVBlocks.CVBLOCK_UNIVERSITY_STUDIES.toString())) {
                expression_to_collect.add("AccreditedUniversityStudiesOtherPostGrade");
                expression_to_collect.add("AccreditedUniversityStudiesDegree");
                expression_to_collect.add("AccreditedUniversityStudiesPhDStudies");
                collect_expressions = true;
            }

        } else {

            expression_to_collect.add("ProfessionalActivityCurrent");
            expression_to_collect.add("ProfessionalActivityNoCurrent");
            expression_to_collect.add("AccreditedUniversityStudiesOtherPostGrade");
            expression_to_collect.add("AccreditedUniversityStudiesDegree");
            expression_to_collect.add("AccreditedUniversityStudiesPhDStudies");
            collect_expressions = true;
        }
    } else {

        expression_to_collect.add("ProfessionalActivityCurrent");
        expression_to_collect.add("ProfessionalActivityNoCurrent");
        expression_to_collect.add("AccreditedUniversityStudiesOtherPostGrade");
        expression_to_collect.add("AccreditedUniversityStudiesDegree");
        expression_to_collect.add("AccreditedUniversityStudiesPhDStudies");
        collect_expressions = true;
    }

    eOut.addAttribute("URL", docGate.getSourceUrl() != null ? docGate.getSourceUrl().toString() : "");

    AnnotationSet annoset = docGate.getAnnotations();

    if (collect_expressions) {

        List<Annotation> anns = new ArrayList<Annotation>();

        for (String exp : expression_to_collect)
            anns.addAll(annoset.get(exp));

        //Collections.sort(anns, new OffsetBeginEndComparator());

        //need to bee order
        if (anns.size() > 0) {
            for (Annotation an : anns) {
                String cvnItemName = an.getType();
                org.dom4j.Element eAux = new org.dom4j.DocumentFactory().createElement(cvnItemName);
                //                        eAux.addElement("Domain").addText(gate.Utils.stringFor(docGate,
                //                                                          an.getStartNode().getOffset() > 100 ? an.getStartNode().getOffset() - 100 : an.getStartNode().getOffset(),
                //                                                          an.getEndNode().getOffset() + 100 < docGate.getContent().size() ? an.getEndNode().getOffset() + 100 :  an.getEndNode().getOffset()));
                eAux.addAttribute("action_mode", "add");
                eAux.addElement("Content").addText(gate.Utils.stringFor(docGate, an));
                FeatureMap fmap = an.getFeatures();
                for (Object key : fmap.keySet()) {
                    String fieldName = key.toString();
                    eAux.addElement(fieldName).addText(fmap.get(key).toString());
                }
                eOut.add(eAux);
            }
        }

        n_expressions += eOut.elements().size();
    }

    collect_expressions = false;
    if (extra_data != null) {
        if (extra_data.containsKey(DataExchangeLiterals.MIDDLE_ELEMENT_XML_EXTRADATA_BLOCK_TYPE)) {

            String block_type = extra_data.get(DataExchangeLiterals.MIDDLE_ELEMENT_XML_EXTRADATA_BLOCK_TYPE);
            if (block_type.equals(CVBlocks.CVBLOCK_REST.toString())) {
                collect_expressions = true;
            } else if (block_type.equals(CVBlocks.CVBLOCK_PERSONAL.toString())) {
                collect_expressions = true;
            }

        } else {
            collect_expressions = true;
        }
    } else {

        collect_expressions = true;
    }

    if (collect_expressions) {

        ArrayList<Annotation> anns = new ArrayList<Annotation>();
        anns.addAll(annoset.get("AgentIdentification"));

        if (anns.size() > 0) {
            String lastname = "";
            String initials = "";
            String name = "";
            String firstname = "";

            if (extra_data != null) {
                lastname = extra_data.get(DataExchangeLiterals.MIDDLE_ELEMENT_XML_EXTRADATA_LASTNAME);
                initials = extra_data.get(DataExchangeLiterals.MIDDLE_ELEMENT_XML_EXTRADATA_INITIALS);
                name = extra_data.get(DataExchangeLiterals.MIDDLE_ELEMENT_XML_EXTRADATA_NAME);
                firstname = extra_data.get(DataExchangeLiterals.MIDDLE_ELEMENT_XML_EXTRADATA_FIRSTNAME);
            }

            if (firstname.equals("")) {
                firstname = initials;
            }

            for (Annotation an : anns) {
                String cvnItemName = an.getType();
                org.dom4j.Element eAux = new org.dom4j.DocumentFactory().createElement(cvnItemName);
                //                        eAux.addElement("Domain").addText(gate.Utils.stringFor(docGate,
                //                                                          an.getStartNode().getOffset() > 100 ? an.getStartNode().getOffset() - 100 : an.getStartNode().getOffset(),
                //                                                          an.getEndNode().getOffset() + 100 < docGate.getContent().size() ? an.getEndNode().getOffset() + 100 :  an.getEndNode().getOffset()));
                eAux.addAttribute("action_mode", "overwrite");
                eAux.addAttribute("extra_gets", "getInformation");
                eAux.addElement("Content").addText(gate.Utils.stringFor(docGate, an));
                FeatureMap fmap = an.getFeatures();
                for (Object key : fmap.keySet()) {
                    String fieldName = key.toString();
                    eAux.addElement(fieldName).addText(fmap.get(key).toString());
                }
                eAux.addElement(CVItemExtracted.AgentIdentification.GivenName).addText(firstname);
                eAux.addElement(CVItemExtracted.AgentIdentification.FirstFamilyName).addText(lastname);

                eOut.add(eAux);
            }
        }

        n_expressions += eOut.elements().size();
    }

    ProjectLogger.LOGGER.info(String.format("%3d expressions in %s : ", n_expressions, docGate.getName())); // + docXML.asXML()
    if (eOut == null)
        ProjectLogger.LOGGER.info("Output is null"); // + docXML.asXML()

    aoData.setData_out(eOut);

    if (aoData.getVerbose()) {
        File dest_dir = aoData.getVerboseDir();
        File path = null;
        String fileName = "";

        if (extra_data != null
                && extra_data.containsKey(DataExchangeLiterals.MIDDLE_ELEMENT_XML_EXTRADATA_DOCUMENT_NAME)) {

            fileName = extra_data.get(DataExchangeLiterals.MIDDLE_ELEMENT_XML_EXTRADATA_DOCUMENT_NAME);

        } else {

            URL url = docGate.getSourceUrl();
            try {
                path = new File(url.toURI());
                fileName = path.getName();
            } catch (Exception e) {
                String filename;
                try {
                    filename = URLEncoder.encode(url.toString(), "UTF-8") + ".html";
                } catch (Exception ex) {
                    filename = docGate.getName() + ".html";
                }
                path = new File(filename);
                fileName = path.getName();
            }

            if (!fileName.equals("")) {
                fileName = fileName.substring(0, fileName.lastIndexOf("."));
            }

        }

        File file_result = new File(dest_dir, fileName + "_verbose.html");

        try {
            writeResultsInHTMLFile(docGate, file_result);
        } catch (Exception ex) {
            Logger.getRootLogger().error("Error writing verbose results. " + ex.toString());
        }

    }
}

From source file:eu.sisob.uma.NPL.Researchers.GateResearcherAnnCollector_Deprecated.java

License:Open Source License

/**
 *
 * @param doc//from w  ww . ja va 2s .co m
 * @param aoData
 */
@Override
public void collect(Object doc, MiddleData aoData) {
    int n_expressions = 0;
    org.dom4j.Element eOut = org.dom4j.DocumentFactory.getInstance().createElement("blockinfo");
    eOut.addAttribute(DataExchangeLiterals.MIDDLE_ELEMENT_XML_ID_ANNOTATIONRECOLLECTING,
            aoData.getId_annotationrecollecting()); // aoData[MiddleData.I_INDEX_DATA_TYPE].toString());
    eOut.addAttribute(DataExchangeLiterals.MIDDLE_ELEMENT_XML_ID_ENTITY_ATT, aoData.getId_entity()); // aoData[MiddleData.I_INDEX_DATA_ID].toString());

    gate.Document docGate = (gate.Document) doc;

    HashMap<String, String> extra_data = null;
    try {
        extra_data = (HashMap<String, String>) aoData.getData_extra();
    } catch (Exception ex) {
        extra_data = null;
    }
    boolean collect_expressions = true;
    if (extra_data != null) {
        if (extra_data.containsKey(DataExchangeLiterals.MIDDLE_ELEMENT_XML_EXTRADATA_BLOCK_TYPE)) {
            String block_type = extra_data.get(DataExchangeLiterals.MIDDLE_ELEMENT_XML_EXTRADATA_BLOCK_TYPE);
            if (block_type.equals(CVBlocks.CVBLOCK_OTHERS)) {
                collect_expressions = false;
            }
        } else {

        }
    }

    eOut.addAttribute("URL", docGate.getSourceUrl() != null ? docGate.getSourceUrl().toString() : "");

    if (collect_expressions) {
        AnnotationSet annoset = docGate.getAnnotations();
        List<Annotation> anns = new ArrayList<Annotation>();

        //Expressions
        anns.addAll(annoset.get("ProfessionalActivityCurrent"));
        anns.addAll(annoset.get("ProfessionalActivityNoCurrent"));
        anns.addAll(annoset.get("AccreditedUniversityStudiesOtherPostGrade"));
        anns.addAll(annoset.get("AccreditedUniversityStudiesDegree"));
        anns.addAll(annoset.get("AccreditedUniversityStudiesPhDStudies"));

        //Collections.sort(anns, new OffsetBeginEndComparator());

        //need to bee order
        if (anns.size() > 0) {
            for (Annotation an : anns) {
                String cvnItemName = an.getType();
                org.dom4j.Element eAux = new org.dom4j.DocumentFactory().createElement(cvnItemName);
                //                        eAux.addElement("Domain").addText(gate.Utils.stringFor(docGate,
                //                                                          an.getStartNode().getOffset() > 100 ? an.getStartNode().getOffset() - 100 : an.getStartNode().getOffset(),
                //                                                          an.getEndNode().getOffset() + 100 < docGate.getContent().size() ? an.getEndNode().getOffset() + 100 :  an.getEndNode().getOffset()));
                eAux.addAttribute("action_mode", "add");
                eAux.addElement("Content").addText(gate.Utils.stringFor(docGate, an));
                FeatureMap fmap = an.getFeatures();
                for (Object key : fmap.keySet()) {
                    String fieldName = key.toString();
                    eAux.addElement(fieldName).addText(fmap.get(key).toString());
                }
                eOut.add(eAux);
            }
        }

        n_expressions += eOut.elements().size();

        anns = new ArrayList<Annotation>();
        anns.addAll(annoset.get("AgentIdentification"));

        if (anns.size() > 0) {
            String lastname = "";
            String initials = "";
            String name = "";
            String firstname = "";

            if (extra_data != null) {
                lastname = extra_data.get(DataExchangeLiterals.MIDDLE_ELEMENT_XML_EXTRADATA_LASTNAME);
                initials = extra_data.get(DataExchangeLiterals.MIDDLE_ELEMENT_XML_EXTRADATA_INITIALS);
                name = extra_data.get(DataExchangeLiterals.MIDDLE_ELEMENT_XML_EXTRADATA_NAME);
                firstname = extra_data.get(DataExchangeLiterals.MIDDLE_ELEMENT_XML_EXTRADATA_FIRSTNAME);
            }

            if (firstname.equals("")) {
                firstname = initials;
            }

            for (Annotation an : anns) {
                String cvnItemName = an.getType();
                org.dom4j.Element eAux = new org.dom4j.DocumentFactory().createElement(cvnItemName);
                //                        eAux.addElement("Domain").addText(gate.Utils.stringFor(docGate,
                //                                                          an.getStartNode().getOffset() > 100 ? an.getStartNode().getOffset() - 100 : an.getStartNode().getOffset(),
                //                                                          an.getEndNode().getOffset() + 100 < docGate.getContent().size() ? an.getEndNode().getOffset() + 100 :  an.getEndNode().getOffset()));
                eAux.addAttribute("action_mode", "overwrite");
                eAux.addAttribute("extra_gets", "getInformation");
                eAux.addElement("Content").addText(gate.Utils.stringFor(docGate, an));
                FeatureMap fmap = an.getFeatures();
                for (Object key : fmap.keySet()) {
                    String fieldName = key.toString();
                    eAux.addElement(fieldName).addText(fmap.get(key).toString());
                }
                eAux.addElement(CVItemExtracted.AgentIdentification.GivenName).addText(firstname);
                eAux.addElement(CVItemExtracted.AgentIdentification.FirstFamilyName).addText(lastname);

                eOut.add(eAux);
            }
        }

        n_expressions += eOut.elements().size();

        ProjectLogger.LOGGER
                .info(String.format("%3d expressions in %s : ", n_expressions, docGate.getSourceUrl())); // + docXML.asXML()
        if (eOut == null)
            ProjectLogger.LOGGER.info("Output is null"); // + docXML.asXML()

        aoData.setData_out(eOut);
    }

    if (aoData.getVerbose()) {
        File dest_dir = aoData.getVerboseDir();
        File path = null;
        String fileName = "";

        if (extra_data != null
                && extra_data.containsKey(DataExchangeLiterals.MIDDLE_ELEMENT_XML_EXTRADATA_DOCUMENT_NAME)) {

            fileName = extra_data.get(DataExchangeLiterals.MIDDLE_ELEMENT_XML_EXTRADATA_DOCUMENT_NAME);

        } else {

            URL url = docGate.getSourceUrl();
            try {
                path = new File(url.toURI());
                fileName = path.getName();
            } catch (Exception e) {
                String filename;
                try {
                    filename = URLEncoder.encode(url.toString(), "UTF-8") + ".html";
                } catch (Exception ex) {
                    filename = docGate.getName() + ".html";
                }
                path = new File(filename);
                fileName = path.getName();
            }

            if (!fileName.equals("")) {
                fileName = fileName.substring(0, fileName.lastIndexOf("."));
            }

        }

        File file_result = new File(dest_dir, fileName + "_verbose.html");

        try {
            writeResultsInHTMLFile(docGate, file_result);
        } catch (Exception ex) {
            Logger.getRootLogger().error("Error writing verbose results. " + ex.toString());
        }

    }
}

From source file:eu.sisob.uma.NPL.Researchers.TextMiningParserGateDetector.java

License:Open Source License

/**
 * Define annotator collector acoording index
 * I_TYPE_CONTENT_ENTIRE_WEB_PAGE => Extract info from CV and personal web page of researchers
 * @param lstAnnColl_ list of annotator collector
 *///from   ww w  . ja  v a2s.c  o  m
@Override
protected void iniAnnotatorCollectors(TreeMap lstAnnColl_) {
    AnnotatorCollector a = null;

    a = new AnnotatorCollector(DataExchangeLiterals.ID_TEXTMININGPARSER_GATERESEARCHER_DEFAULTANNREC) {
        @Override
        public void collect(Object doc, MiddleData aoData) {
            org.dom4j.Element eOut = org.dom4j.DocumentFactory.getInstance().createElement("blockinfo");
            eOut.addAttribute(DataExchangeLiterals.MIDDLE_ELEMENT_XML_ID_ANNOTATIONRECOLLECTING,
                    aoData.getId_annotationrecollecting()); // aoData[MiddleData.I_INDEX_DATA_TYPE].toString());
            eOut.addAttribute(DataExchangeLiterals.MIDDLE_ELEMENT_XML_ID_ENTITY_ATT, aoData.getId_entity()); // aoData[MiddleData.I_INDEX_DATA_ID].toString());

            gate.Document docGate = (gate.Document) doc;
            AnnotationSet annoset = docGate.getAnnotations();
            List<Annotation> anns = new ArrayList<Annotation>();

            //Expressions
            //anns.addAll(annoset.get("JobTitleTest"));
            //anns.addAll(annoset.get("DegreeTest"));
            anns.addAll(annoset.get("OrgTest"));

            //Collections.sort(anns, new OffsetBeginEndComparator());

            //need to bee order
            if (anns.size() > 0) {
                for (Annotation an : anns) {
                    String cvnItemName = an.getType();
                    org.dom4j.Element eAux = new org.dom4j.DocumentFactory().createElement(cvnItemName);
                    //eAux.addElement("Domain").addText(gate.Utils.stringFor(docGate,
                    //                                  an.getStartNode().getOffset() > 100 ? an.getStartNode().getOffset() - 100 : an.getStartNode().getOffset(),
                    //                                  an.getEndNode().getOffset() + 100 < docGate.getContent().size() ? an.getEndNode().getOffset() + 100 :  an.getEndNode().getOffset()));
                    eAux.addText(gate.Utils.stringFor(docGate, an));
                    eOut.add(eAux);
                }
            }

            Logger.getLogger("MyLog").info(String.format("%3d expressions in %s : ",
                    eOut != null ? eOut.elements().size() : 0, docGate.getSourceUrl())); // + docXML.asXML()

            aoData.setData_out(eOut);
        }
    };
    lstAnnColl_.put(a.type, a);
}

From source file:fr.gouv.culture.vitam.command.VitamCommand.java

License:Open Source License

public static void computeDigest() {
    XMLWriter writer = null;/*from   ww w .  j  av a  2  s  . c  o m*/
    try {
        writer = new XMLWriter(outputStream, StaticValues.defaultOutputFormat);
    } catch (UnsupportedEncodingException e1) {
        System.err.println(StaticValues.LBL.error_writer.get() + ": " + e1.toString());
        return;
    }
    File basedir = new File(checkDigest);
    List<File> files;
    try {
        files = DroidHandler.matchedFiled(new File[] { basedir }, extensions,
                StaticValues.config.argument.recursive);
    } catch (CommandExecutionException e1) {
        System.err.println(StaticValues.LBL.error_error.get() + e1.toString());
        return;
    }
    System.out.println("Digest...");
    Element root = null;
    VitamResult vitamResult = new VitamResult();
    if (basedir.isFile()) {
        basedir = basedir.getParentFile();
    }
    if (StaticValues.config.argument.outputModel == VitamOutputModel.OneXML) {
        root = XmlDom.factory.createElement("digests");
        root.addAttribute("source", basedir.getAbsolutePath());
        vitamResult.unique = XmlDom.factory.createDocument(root);
    }
    int currank = 0;
    int error = 0;
    for (File file : files) {
        currank++;
        String shortname;
        shortname = StaticValues.getSubPath(file, basedir);
        FileInputStream inputstream;
        try {
            inputstream = new FileInputStream(file);
        } catch (FileNotFoundException e) {
            System.err.println(StaticValues.LBL.error_computedigest.get() + ": " + shortname);
            continue;
        }
        String[] shas = DigestCompute.computeDigest(inputstream, StaticValues.config.argument);
        //SEDA type since already configured
        Element result = XmlDom.factory.createElement(StaticValues.config.DOCUMENT_FIELD);
        Element attachment = XmlDom.factory.createElement(StaticValues.config.ATTACHMENT_FIELD);
        attachment.addAttribute(StaticValues.config.FILENAME_ATTRIBUTE.substring(1), shortname);
        result.add(attachment);
        if (shas[0] != null) {
            Element integrity = XmlDom.factory.createElement(StaticValues.config.INTEGRITY_FIELD);
            integrity.addAttribute(StaticValues.config.ALGORITHME_ATTRIBUTE.substring(1),
                    StaticValues.XML_SHA1);
            integrity.setText(shas[0]);
            result.add(integrity);
        }
        if (shas[1] != null) {
            Element integrity = XmlDom.factory.createElement(StaticValues.config.INTEGRITY_FIELD);
            integrity.addAttribute(StaticValues.config.ALGORITHME_ATTRIBUTE.substring(1),
                    StaticValues.XML_SHA256);
            integrity.setText(shas[1]);
            result.add(integrity);
        }
        if (shas[2] != null) {
            Element integrity = XmlDom.factory.createElement(StaticValues.config.INTEGRITY_FIELD);
            integrity.addAttribute(StaticValues.config.ALGORITHME_ATTRIBUTE.substring(1),
                    StaticValues.XML_SHA512);
            integrity.setText(shas[2]);
            result.add(integrity);
        }
        if ((shas[0] == null && StaticValues.config.argument.sha1)
                || (shas[1] == null && StaticValues.config.argument.sha256)
                || (shas[2] == null && StaticValues.config.argument.sha512)) {
            result.addAttribute("status", "error");
            error++;
        } else {
            result.addAttribute("status", "ok");
        }
        XmlDom.addDate(StaticValues.config.argument, StaticValues.config, result);
        if (root != null) {
            root.add(result);
        } else {
            // multiple
            root = XmlDom.factory.createElement("digests");
            root.addAttribute("source", basedir.getAbsolutePath());
            root.add(result);
            try {
                writer.write(root);
            } catch (IOException e) {
                System.err.println(StaticValues.LBL.error_error.get() + e.toString());
            }
            root = null;
        }
    }
    if (root != null) {
        if (error == 0) {
            root.addAttribute("status", "ok");
        } else {
            root.addAttribute("status", "error on " + error + " / " + currank + " file checks");
        }
        XmlDom.addDate(StaticValues.config.argument, StaticValues.config, root);
        try {
            writer.write(vitamResult.unique);
        } catch (IOException e) {
            System.err.println(StaticValues.LBL.error_analysis.get() + e);
        }
    }
    System.out.println(StaticValues.LBL.action_digest.get() + " [ " + currank
            + (error > 0 ? " (" + StaticValues.LBL.error_error.get() + error + " ) " : "") + " ]");
}

From source file:fr.gouv.culture.vitam.command.VitamCommand.java

License:Open Source License

private static Element addPdfaElement(Element root, Element pdfa, File basedir, File baseoutdir, boolean error,
        String serror, String puid, VitamResult vitamResult) {
    XmlDom.addDate(StaticValues.config.argument, StaticValues.config, pdfa);
    if (puid != null) {
        pdfa.addAttribute("puid", puid);
    }/*  w w  w.j a  va2  s .c  o  m*/
    if (root != null) {
        root.add(pdfa);
    } else {
        // multiple
        root = XmlDom.factory.createElement("transform");
        root.addAttribute("source", basedir.getAbsolutePath());
        root.addAttribute("target", baseoutdir.getAbsolutePath());
        Document document = XmlDom.factory.createDocument(root);
        root.add(pdfa);
        if (error) {
            root.addAttribute("status", serror);
        } else {
            root.addAttribute("status", "ok");
        }
        vitamResult.multiples.add(document);
    }
    return root;
}

From source file:fr.gouv.culture.vitam.command.VitamCommand.java

License:Open Source License

public static void convertPdfa() {
    XMLWriter writer = null;/*from  ww  w .  j ava  2  s  .  com*/
    try {
        writer = new XMLWriter(outputStream, StaticValues.defaultOutputFormat);
    } catch (UnsupportedEncodingException e1) {
        System.err.println(StaticValues.LBL.error_writer.get() + ": " + e1.toString());
        return;
    }
    File basedir = new File(fromPdfA);
    List<File> files;
    try {
        files = DroidHandler.matchedFiled(new File[] { basedir }, extensions,
                StaticValues.config.argument.recursive);
    } catch (CommandExecutionException e1) {
        System.err.println(StaticValues.LBL.error_error.get() + e1.toString());
        return;
    }
    if (basedir.isFile()) {
        basedir = basedir.getParentFile();
    }
    File baseoutdir = new File(toPdfA);
    if (!baseoutdir.exists()) {
        baseoutdir.mkdirs();
    }
    if (baseoutdir.isFile()) {
        baseoutdir = baseoutdir.getParentFile();
    }
    int errorcpt = 0;
    boolean checkDroid = false;
    try {
        StaticValues.config.initDroid();
        checkDroid = true;
    } catch (CommandExecutionException e) {
        System.err.println(StaticValues.LBL.error_initdroid.get() + e.toString());
    }
    System.out.println("\nTransform PDF/A-1B\n");
    Element root = null;
    Element temp = null;
    VitamResult vitamResult = new VitamResult();
    if (StaticValues.config.argument.outputModel == VitamOutputModel.OneXML) {
        root = XmlDom.factory.createElement("transform");
        root.addAttribute("source", basedir.getAbsolutePath());
        root.addAttribute("target", baseoutdir.getAbsolutePath());
        vitamResult.unique = XmlDom.factory.createDocument(root);
    } else {
        // force multiple
        vitamResult.multiples = new ArrayList<Document>();
    }
    for (File file : files) {
        String basename = file.getName();
        File rootdir;
        String subpath = null;
        if (file.getParentFile().equals(basedir)) {
            rootdir = basedir;
            subpath = File.separator;
        } else {
            rootdir = file.getParentFile();
            subpath = rootdir.getAbsolutePath().replace(basedir.getAbsolutePath(), "") + File.separator;
        }
        String fullname = subpath + basename;
        String puid = null;
        if (checkDroid) {
            try {
                List<DroidFileFormat> list = StaticValues.config.droidHandler.checkFileFormat(file,
                        StaticValues.config.argument);
                if (list == null || list.isEmpty()) {
                    System.err.println("Ignore: " + fullname);
                    Element pdfa = XmlDom.factory.createElement("convert");
                    Element newElt = XmlDom.factory.createElement("file");
                    newElt.addAttribute("filename", fullname);
                    pdfa.add(newElt);
                    addPdfaElement(root, pdfa, basedir, baseoutdir, true, "Error: filetype not found", null,
                            vitamResult);
                    errorcpt++;
                    continue;
                }
                DroidFileFormat type = list.get(0);
                puid = type.getPUID();
                if (puid.startsWith(StaticValues.FORMAT_XFMT) || puid.equals("fmt/411")) { // x-fmt or RAR
                    System.err.println("Ignore: " + fullname + " " + puid);
                    Element pdfa = XmlDom.factory.createElement("convert");
                    Element newElt = XmlDom.factory.createElement("file");
                    newElt.addAttribute("filename", fullname);
                    pdfa.add(newElt);
                    addPdfaElement(root, pdfa, basedir, baseoutdir, true, "Error: filetype not allowed", puid,
                            vitamResult);
                    errorcpt++;
                    continue;
                }
            } catch (CommandExecutionException e) {
                // ignore
            }
        }
        System.out.println("PDF/A-1B convertion... " + fullname);
        long start = System.currentTimeMillis();
        Element pdfa = PdfaConverter.convertPdfA(subpath, basename, basedir, baseoutdir, StaticValues.config);
        long end = System.currentTimeMillis();
        boolean error = false;
        if (pdfa.selectSingleNode(".[@status='ok']") == null) {
            error = true;
            errorcpt++;
        }
        if (error) {
            System.err.println(StaticValues.LBL.error_pdfa.get() + " PDF/A-1B KO: " + fullname + " "
                    + ((end - start) * 1024 / file.length()) + " ms/KB " + (end - start) + " ms " + "\n");
        } else {
            System.out.println("PDF/A-1B OK: " + fullname + " " + ((end - start) * 1024 / file.length())
                    + " ms/KB " + (end - start) + " ms " + "\n");
        }
        temp = addPdfaElement(root, pdfa, basedir, baseoutdir, error, "error", puid, vitamResult);
        try {
            Thread.sleep(500);
        } catch (InterruptedException e) {
        }
    }
    if (root != null) {
        XmlDom.addDate(StaticValues.config.argument, StaticValues.config, root);
        if (errorcpt > 0) {
            root.addAttribute("status", "error found");
        } else {
            root.addAttribute("status", "ok");
        }
        try {
            writer.write(vitamResult.unique);
        } catch (IOException e) {
            System.err.println(StaticValues.LBL.error_analysis.get() + e);
        }
    } else {
        XmlDom.addDate(StaticValues.config.argument, StaticValues.config, temp);
        try {
            writer.write(temp);
        } catch (IOException e) {
        }
    }
    if (errorcpt < files.size()) {
        System.out.println(StaticValues.LBL.action_pdfa.get() + " [ " + files.size()
                + (errorcpt > 0 ? " (" + StaticValues.LBL.error_error.get() + errorcpt + " )" : "") + " ]");
    } else {
        System.err.println(StaticValues.LBL.error_pdfa.get() + " [ " + StaticValues.LBL.error_error.get()
                + errorcpt + " ]");
    }
}

From source file:fr.gouv.culture.vitam.command.VitamCommand.java

License:Open Source License

public static void checkFilesType() {
    File fic = new File(FILEarg);
    if (!fic.exists()) {
        System.err.println(StaticValues.LBL.error_filenotfile.get() + ": " + FILEarg);
        return;//from  w w w  .j  a  v a  2s.c o  m
    } else {
        System.out.println("\n" + StaticValues.LBL.tools_dir_format_output.get() + "\n");
        Document global = null;
        Element root = null;
        XMLWriter writer = null;
        try {
            writer = new XMLWriter(outputStream, StaticValues.defaultOutputFormat);
        } catch (UnsupportedEncodingException e1) {
            System.err.println(StaticValues.LBL.error_writer.get() + ": " + e1.toString());
            return;
        }
        if (StaticValues.config.argument.outputModel == VitamOutputModel.OneXML) {
            root = XmlDom.factory.createElement("checkfiles");
            root.addAttribute("source", FILEarg);
            global = XmlDom.factory.createDocument(root);
            EmlExtract.filEmls.clear();
        }
        if (showFormat) {
            if (StaticValues.config.droidHandler == null && StaticValues.config.exif == null
                    && StaticValues.config.jhove == null) {
                System.err.println(StaticValues.LBL.error_initfits.get());
                return;
            }
            try {
                List<File> files = DroidHandler.matchedFiled(new File[] { fic }, extensions,
                        StaticValues.config.argument.recursive);
                for (File file : files) {
                    String shortname;
                    if (fic.isDirectory()) {
                        shortname = StaticValues.getSubPath(file, fic);
                    } else {
                        shortname = FILEarg;
                    }
                    Element result = Commands.showFormat(shortname, null, null, file, StaticValues.config,
                            StaticValues.config.argument);
                    XmlDom.addDate(StaticValues.config.argument, StaticValues.config, result);
                    if (root != null) {
                        root.add(result);
                    } else {
                        writer.write(result);
                        System.out.println("\n========================================================");
                    }
                }
            } catch (CommandExecutionException e) {
                System.err.println(StaticValues.LBL.error_analysis.get() + e);
                e.printStackTrace();
            } catch (IOException e) {
                System.err.println(StaticValues.LBL.error_analysis.get() + e);
            }
        } else {
            if (StaticValues.config.droidHandler == null) {
                System.err.println(StaticValues.LBL.error_initdroid.get());
                return;
            }
            if (root != null) {
                Element newElt = XmlDom.factory.createElement("toolsversion");
                if (StaticValues.config.droidHandler != null) {
                    newElt.addAttribute("pronom", StaticValues.config.droidHandler.getVersionSignature());
                }
                if (StaticValues.config.droidHandler != null) {
                    newElt.addAttribute("droid", "6.1");
                }
                root.add(newElt);
            }
            List<DroidFileFormat> list;
            try {
                VitamArgument argument = new VitamArgument(StaticValues.config.argument.archive,
                        StaticValues.config.argument.recursive, true, true, true,
                        StaticValues.config.argument.outputModel, StaticValues.config.argument.checkSubFormat,
                        StaticValues.config.argument.extractKeyword);
                List<File> files = DroidHandler.matchedFiled(new File[] { fic }, extensions,
                        argument.recursive);
                list = StaticValues.config.droidHandler.checkFilesFormat(files, argument, null);
                String pathBeforeArg = fic.getCanonicalPath();
                pathBeforeArg = pathBeforeArg.substring(0, pathBeforeArg.indexOf(FILEarg));
                for (DroidFileFormat droidFileFormat : list) {
                    Element fileformat = droidFileFormat.toElement(true);
                    Attribute filename = fileformat.attribute("filename");
                    if (filename != null) {
                        String value = filename.getText();
                        filename.setText(value.replace(pathBeforeArg, ""));
                    }
                    XmlDom.addDate(StaticValues.config.argument, StaticValues.config, fileformat);
                    if (root != null) {
                        root.add(fileformat);
                    } else {
                        writer.write(fileformat);
                        System.out.println("\n========================================================");
                    }
                }
            } catch (CommandExecutionException e) {
                System.err.println(StaticValues.LBL.error_analysis.get() + e);
                e.printStackTrace();
            } catch (UnsupportedEncodingException e) {
                System.err.println(StaticValues.LBL.error_analysis.get() + e);
            } catch (IOException e) {
                System.err.println(StaticValues.LBL.error_analysis.get() + e);
            }
        }
        if (global != null) {
            XmlDom.addDate(StaticValues.config.argument, StaticValues.config, root);
            if (!EmlExtract.filEmls.isEmpty()) {
                Element sortEml = XmlDom.factory.createElement("emlsort");
                for (String parent : EmlExtract.filEmls.keySet()) {
                    Element eparent = XmlDom.factory.createElement("parent");
                    String fil = EmlExtract.filEmls.get(parent);
                    eparent.addAttribute("messageId", parent);
                    String[] fils = fil.split(",");
                    for (String mesg : fils) {
                        if (mesg != null && mesg.length() > 1) {
                            Element elt = XmlDom.factory.createElement("descendant");
                            elt.addAttribute("messageId", mesg);
                            eparent.add(elt);
                        }
                    }
                    sortEml.add(eparent);
                }
                root.add(sortEml);
            }
            try {
                writer.write(global);
            } catch (IOException e) {
                System.err.println(StaticValues.LBL.error_analysis.get() + e);
            }
        }
    }
}

From source file:fr.gouv.culture.vitam.database.DbSchema.java

License:Open Source License

public Element getElement(boolean recursive, boolean data) {
    DocumentFactory factory = DocumentFactory.getInstance();
    Element schema = factory.createElement(SCHEMA_FIELD);
    schema.addAttribute(NAME_ATTRIBUTE, name);
    schema.addAttribute(DESCRIPTION_ATTRIBUTE, description);
    if (identifier != null) {
        Element eltidentifier = factory.createElement(IDENTIFIER_FIELD);
        eltidentifier.addAttribute(TYPE_ATTRIBUTE, identifier.type.name());
        eltidentifier.addAttribute(DATAFILE_ATTRIBUTE, identifier.refDataFile);
        switch (identifier.type) {
        case CSVTYPE:
            eltidentifier.addAttribute(SEPARATOR_ATTRIBUTE, identifier.separator);
            break;
        case MULTIPLETYPE: {
            Element poslistm = factory.createElement(MULTIPLEPOSITIONS_FIELD);
            for (int[] posmul : identifier.positionsMultiple) {
                Element poslist = factory.createElement(POSITIONS_FIELD);
                for (int ipos : posmul) {
                    Element pos = factory.createElement(POSITION_FIELD);
                    pos.setText(Integer.toString(ipos));
                    poslist.add(pos);
                }/*from   w  w w.j  a  v  a2  s. co  m*/
                poslistm.add(poslist);
            }
            eltidentifier.add(poslistm);
        }
            break;
        case UNIQUETYPE: {
            Element poslist = factory.createElement(POSITIONS_FIELD);
            for (int ipos : identifier.positions) {
                Element pos = factory.createElement(POSITION_FIELD);
                pos.setText(Integer.toString(ipos));
                poslist.add(pos);
            }
            eltidentifier.add(poslist);
        }
            break;
        }
        schema.add(eltidentifier);
    }
    if (constructionOrder != null) {
        Element order = factory.createElement(ORDER_FIELD);
        int rank = 0;
        for (String table : constructionOrder) {
            Element etable = factory.createElement(TABLE_FIELD);
            etable.addAttribute(RANK_ATTRIBUTE, Integer.toString(rank));
            etable.setText(table);
            order.add(etable);
            rank++;
        }
        schema.add(order);
    }
    if (recursive) {
        schema.addAttribute(NB_ATTRIBUTE, Integer.toString(tables.size()));
        for (DbTable table : tables) {
            if (data && table.getRows() != null && table.getRows().size() > 0) {
                Element etable = table.getElement(data);
                schema.add(etable);
            } else {
                Element etable = table.getElement(false);
                if (data) {
                    // fill data using file
                    try {
                        XmlWriter.loadDbTableDataFromFile(this, table, etable);
                    } catch (IOException e) {
                        logger.warn(StaticValues.LBL.error_error.get() + e);
                    } catch (WaarpDatabaseSqlException e) {
                        logger.warn(StaticValues.LBL.error_error.get() + e);
                    }
                }
                schema.add(etable);
            }
        }
    }
    return schema;
}

From source file:fr.gouv.culture.vitam.database.DbTable.java

License:Open Source License

public Element getElement(boolean data) {
    DocumentFactory factory = DocumentFactory.getInstance();
    Element table = factory.createElement(DbSchema.TABLE_FIELD);
    table.addAttribute(DbSchema.NAME_ATTRIBUTE, name);
    table.addAttribute(DbSchema.DESCRIPTION_ATTRIBUTE, description);
    table.addAttribute(DbSchema.RANK_ATTRIBUTE, Integer.toString(rank));
    table.addAttribute(DbSchema.TYPE_ATTRIBUTE, type.toString());
    if (datafile != null) {
        table.addAttribute(DbSchema.DATAFILE_ATTRIBUTE, datafile);
    }/* w  w  w. j av a 2 s  .  c om*/
    if (fields != null) {
        Element efields = factory.createElement(DbSchema.FIELDS_FIELD);
        for (DbField field : fields) {
            Element efield = field.getElement();
            efields.add(efield);
        }
        efields.addAttribute(DbSchema.NB_ATTRIBUTE, Integer.toString(fields.size()));
        table.add(efields);
    }
    if (primaryKeys != null) {
        Element efields = factory.createElement(DbSchema.PRIMARY_KEYS_FIELD);
        for (DbField field : primaryKeys) {
            Element pk = factory.createElement(DbSchema.PRIMARY_KEY_FIELD);
            pk.addAttribute(DbSchema.NAME_ATTRIBUTE, field.name);
            efields.add(pk);
        }
        table.add(efields);
    }
    if (indexes != null) {
        Element efields = factory.createElement(DbSchema.INDEXES_FIELD);
        for (String idx : indexes.keySet()) {
            Element index = factory.createElement(DbSchema.INDEX_FIELD);
            index.addAttribute(DbSchema.NAME_ATTRIBUTE, idx);
            for (DbField field : indexes.get(idx)) {
                Element efield = factory.createElement(DbSchema.FIELD_FIELD);
                efield.addAttribute(DbSchema.NAME_ATTRIBUTE, field.name);
                index.add(efield);
            }
            efields.add(index);
        }
        efields.addAttribute(DbSchema.NB_ATTRIBUTE, Integer.toString(indexes.size()));
        table.add(efields);
    }
    if (constraints != null) {
        Element efields = factory.createElement(DbSchema.CONSTRAINTS_FIELD);
        for (String name : constraints.keySet()) {
            Element index = factory.createElement(DbSchema.CONSTRAINT_FIELD);
            index.addAttribute(DbSchema.FIELD_FIELD, name);
            DbField field = constraints.get(name);
            index.addAttribute(DbSchema.TARGETTABLE_ATTRIBUTE, field.table.name);
            index.addAttribute(DbSchema.TARGETFIELD_ATTRIBUTE, field.name);
            efields.add(index);
        }
        efields.addAttribute(DbSchema.NB_ATTRIBUTE, Integer.toString(constraints.size()));
        table.add(efields);
    }
    if (data && rows != null && rows.size() > 0) {
        Element erows = factory.createElement(DbSchema.ROWS_FIELD);
        erows.addAttribute(DbSchema.NB_ATTRIBUTE, Integer.toString(rows.size()));
        for (DbTableRow row : rows) {
            Element erow = row.getElement();
            erows.add(erow);
        }
        table.add(erows);
    }
    return table;
}