Example usage for org.apache.poi.openxml4j.opc PackagePart getContentType

Introduction

In this page you can find the example usage for org.apache.poi.openxml4j.opc PackagePart getContentType.

Prototype

public String getContentType() 

Source Link

Usage

From source file:mj.ocraptor.extraction.tika.parser.microsoft.ooxml.AbstractOOXMLExtractor.java

License:Apache License

private void handleEmbeddedParts(ContentHandler handler) throws TikaException, IOException, SAXException {
    try {//  www.  ja  v  a  2s . c o m
        for (PackagePart source : getMainDocumentParts()) {
            for (PackageRelationship rel : source.getRelationships()) {

                URI sourceURI = rel.getSourceURI();
                String sourceDesc;
                if (sourceURI != null) {
                    sourceDesc = getJustFileName(sourceURI.getPath());
                    if (sourceDesc.startsWith("slide")) {
                        sourceDesc += "_";
                    } else {
                        sourceDesc = "";
                    }
                } else {
                    sourceDesc = "";
                }
                if (rel.getTargetMode() == TargetMode.INTERNAL) {
                    PackagePart target;

                    try {
                        target = source.getRelatedPart(rel);
                    } catch (IllegalArgumentException ex) {
                        continue;
                    }

                    String type = rel.getRelationshipType();
                    if (RELATION_OLE_OBJECT.equals(type) && TYPE_OLE_OBJECT.equals(target.getContentType())) {
                        handleEmbeddedOLE(target, handler, sourceDesc + rel.getId());
                    } else if (RELATION_AUDIO.equals(type) || RELATION_IMAGE.equals(type)
                            || RELATION_PACKAGE.equals(type) || RELATION_OLE_OBJECT.equals(type)) {
                        handleEmbeddedFile(target, handler, sourceDesc + rel.getId());
                    }
                }
            }
        }
    } catch (InvalidFormatException e) {
        throw new TikaException("Broken OOXML file", e);
    }
}

---

From source file:mj.ocraptor.extraction.tika.parser.microsoft.ooxml.AbstractOOXMLExtractor.java

License:Apache License

/**
 * Handles an embedded file in the document
 *///from  w  ww.  ja  va  2  s. co m
protected void handleEmbeddedFile(PackagePart part, ContentHandler handler, String rel)
        throws SAXException, IOException {
    Metadata metadata = new Metadata();
    metadata.set(Metadata.EMBEDDED_RELATIONSHIP_ID, rel);

    // Get the name
    String name = part.getPartName().getName();
    metadata.set(Metadata.RESOURCE_NAME_KEY, name.substring(name.lastIndexOf('/') + 1));

    // Get the content type
    metadata.set(Metadata.CONTENT_TYPE, part.getContentType());

    // Call the recursing handler
    if (embeddedExtractor.shouldParseEmbedded(metadata)) {
        embeddedExtractor.parseEmbedded(TikaInputStream.get(part.getInputStream()),
                new EmbeddedContentHandler(handler), metadata, false);
    }
}

---

From source file:mj.ocraptor.extraction.tika.parser.pkg.ZipContainerDetector.java

License:Apache License

/**
 * Detects the type of an OfficeOpenXML (OOXML) file from
 *  opened Package /*  w  w w  .j ava  2  s  .  c o  m*/
 */
public static MediaType detectOfficeOpenXML(OPCPackage pkg) {
    PackageRelationshipCollection core = pkg.getRelationshipsByType(ExtractorFactory.CORE_DOCUMENT_REL);
    if (core.size() != 1) {
        // Invalid OOXML Package received
        return null;
    }

    // Get the type of the core document part
    PackagePart corePart = pkg.getPart(core.getRelationship(0));
    String coreType = corePart.getContentType();

    // Turn that into the type of the overall document
    String docType = coreType.substring(0, coreType.lastIndexOf('.'));

    // The Macro Enabled formats are a little special
    if (docType.toLowerCase().endsWith("macroenabled")) {
        docType = docType.toLowerCase() + ".12";
    }

    if (docType.toLowerCase().endsWith("macroenabledtemplate")) {
        docType = MACRO_TEMPLATE_PATTERN.matcher(docType).replaceAll("macroenabled.12");
    }

    // Build the MediaType object and return
    return MediaType.parse(docType);
}

---

From source file:org.apache.tika.parser.microsoft.ooxml.AbstractOOXMLExtractor.java

License:Apache License

private void handleThumbnail(ContentHandler handler) {
    try {// w ww. j a va  2 s.c  o m
        OPCPackage opcPackage = extractor.getPackage();
        for (PackageRelationship rel : opcPackage.getRelationshipsByType(PackageRelationshipTypes.THUMBNAIL)) {
            PackagePart tPart = opcPackage.getPart(rel);
            InputStream tStream = tPart.getInputStream();
            Metadata thumbnailMetadata = new Metadata();
            String thumbName = tPart.getPartName().getName();
            thumbnailMetadata.set(Metadata.RESOURCE_NAME_KEY, thumbName);

            AttributesImpl attributes = new AttributesImpl();
            attributes.addAttribute(XHTML, "class", "class", "CDATA", "embedded");
            attributes.addAttribute(XHTML, "id", "id", "CDATA", thumbName);
            handler.startElement(XHTML, "div", "div", attributes);
            handler.endElement(XHTML, "div", "div");

            thumbnailMetadata.set(Metadata.EMBEDDED_RELATIONSHIP_ID, thumbName);
            thumbnailMetadata.set(Metadata.CONTENT_TYPE, tPart.getContentType());
            thumbnailMetadata.set(TikaCoreProperties.TITLE, tPart.getPartName().getName());

            if (embeddedExtractor.shouldParseEmbedded(thumbnailMetadata)) {
                embeddedExtractor.parseEmbedded(TikaInputStream.get(tStream),
                        new EmbeddedContentHandler(handler), thumbnailMetadata, false);
            }

            tStream.close();
        }
    } catch (Exception ex) {

    }
}

---

From source file:org.apache.tika.parser.pkg.ZipContainerDetector.java

License:Apache License

/**
 * Detects the type of an OfficeOpenXML (OOXML) file from
 *  opened Package /* ww w.  j a v  a  2 s  .  c o  m*/
 */
public static MediaType detectOfficeOpenXML(OPCPackage pkg) {
    // Check for the normal Office core document
    PackageRelationshipCollection core = pkg.getRelationshipsByType(PackageRelationshipTypes.CORE_DOCUMENT);
    // Otherwise check for some other Office core document types
    if (core.size() == 0) {
        core = pkg.getRelationshipsByType(STRICT_CORE_DOCUMENT);
    }
    if (core.size() == 0) {
        core = pkg.getRelationshipsByType(VISIO_DOCUMENT);
    }

    // If we didn't find a single core document of any type, skip detection
    if (core.size() != 1) {
        // Invalid OOXML Package received
        return null;
    }

    // Get the type of the core document part
    PackagePart corePart = pkg.getPart(core.getRelationship(0));
    String coreType = corePart.getContentType();

    // Turn that into the type of the overall document
    String docType = coreType.substring(0, coreType.lastIndexOf('.'));

    // The Macro Enabled formats are a little special
    if (docType.toLowerCase(Locale.ROOT).endsWith("macroenabled")) {
        docType = docType.toLowerCase(Locale.ROOT) + ".12";
    }

    if (docType.toLowerCase(Locale.ROOT).endsWith("macroenabledtemplate")) {
        docType = MACRO_TEMPLATE_PATTERN.matcher(docType).replaceAll("macroenabled.12");
    }

    // Build the MediaType object and return
    return MediaType.parse(docType);
}

---

From source file:poi.xslf.usermodel.DataExtraction.java

License:Apache License

public static void main(String args[]) throws Exception {

    if (args.length == 0) {
        System.out.println("Input file is required");
        return;/*from w  w  w .  j  a v a 2  s  .  c  om*/
    }

    FileInputStream is = new FileInputStream(args[0]);
    XMLSlideShow ppt = new XMLSlideShow(is);
    is.close();

    // Get the document's embedded files.
    List<PackagePart> embeds = ppt.getAllEmbedds();
    for (PackagePart p : embeds) {
        String type = p.getContentType();
        String name = p.getPartName().getName(); //typically file name

        InputStream pIs = p.getInputStream();
        // make sense of the part data
        pIs.close();

    }

    // Get the document's embedded files.
    List<XSLFPictureData> images = ppt.getAllPictures();
    for (XSLFPictureData data : images) {
        PackagePart p = data.getPackagePart();

        String type = p.getContentType();
        String name = data.getFileName();

        InputStream pIs = p.getInputStream();
        // make sense of the image data
        pIs.close();

    }

    Dimension pageSize = ppt.getPageSize(); // size of the canvas in points
    for (XSLFSlide slide : ppt.getSlides()) {
        for (XSLFShape shape : slide) {
            Rectangle2D anchor = shape.getAnchor(); // position on the canvas
            if (shape instanceof XSLFTextShape) {
                XSLFTextShape txShape = (XSLFTextShape) shape;
                System.out.println(txShape.getText());
            } else if (shape instanceof XSLFPictureShape) {
                XSLFPictureShape pShape = (XSLFPictureShape) shape;
                XSLFPictureData pData = pShape.getPictureData();
                System.out.println(pData.getFileName());
            } else {
                System.out.println("Process me: " + shape.getClass());
            }
        }
    }
}

---

From source file:poi.xssf.usermodel.examples.EmbeddedObjects.java

License:Apache License

public static void main(String[] args) throws Exception {
    OPCPackage pkg = OPCPackage.open(args[0]);
    XSSFWorkbook workbook = new XSSFWorkbook(pkg);
    for (PackagePart pPart : workbook.getAllEmbedds()) {
        String contentType = pPart.getContentType();
        // Excel Workbook - either binary or OpenXML
        if (contentType.equals("application/vnd.ms-excel")) {
            HSSFWorkbook embeddedWorkbook = new HSSFWorkbook(pPart.getInputStream());
        }// w ww .ja  v  a 2s  .  c om
        // Excel Workbook - OpenXML file format
        else if (contentType.equals("application/vnd.openxmlformats-officedocument.spreadsheetml.sheet")) {
            XSSFWorkbook embeddedWorkbook = new XSSFWorkbook(pPart.getInputStream());
        }
        // Word Document - binary (OLE2CDF) file format
        else if (contentType.equals("application/msword")) {
            HWPFDocument document = new HWPFDocument(pPart.getInputStream());
        }
        // Word Document - OpenXML file format
        else if (contentType
                .equals("application/vnd.openxmlformats-officedocument.wordprocessingml.document")) {
            XWPFDocument document = new XWPFDocument(pPart.getInputStream());
        }
        // PowerPoint Document - binary file format
        else if (contentType.equals("application/vnd.ms-powerpoint")) {
            HSLFSlideShow slideShow = new HSLFSlideShow(pPart.getInputStream());
        }
        // PowerPoint Document - OpenXML file format
        else if (contentType
                .equals("application/vnd.openxmlformats-officedocument.presentationml.presentation")) {
            OPCPackage docPackage = OPCPackage.open(pPart.getInputStream());
            XSLFSlideShow slideShow = new XSLFSlideShow(docPackage);
        }
        // Any other type of embedded object.
        else {
            System.out.println("Unknown Embedded Document: " + contentType);
            InputStream inputStream = pPart.getInputStream();
        }
    }
    pkg.close();
}

---

From source file:test.unit.be.fedict.eid.applet.service.signer.OOXMLSignatureVerifierTest.java

License:Open Source License

@Test
public void testOPC() throws Exception {
    // setup//from   w ww. ja va2 s . c om
    InputStream inputStream = OOXMLSignatureVerifierTest.class.getResourceAsStream("/hello-world-signed.docx");

    // operate
    OPCPackage opcPackage = OPCPackage.open(inputStream);

    ArrayList<PackagePart> parts = opcPackage.getParts();
    for (PackagePart part : parts) {
        LOG.debug("part name: " + part.getPartName().getName());
        LOG.debug("part content type: " + part.getContentType());
    }

    ArrayList<PackagePart> signatureParts = opcPackage
            .getPartsByContentType("application/vnd.openxmlformats-package.digital-signature-xmlsignature+xml");
    assertFalse(signatureParts.isEmpty());

    PackagePart signaturePart = signatureParts.get(0);
    LOG.debug("signature part class type: " + signaturePart.getClass().getName());

    PackageDigitalSignatureManager packageDigitalSignatureManager = new PackageDigitalSignatureManager();
    // yeah... POI implementation still missing
}

---

From source file:uk.ac.liverpool.spreadsheet.ExcelFeatureAnalysis.java

License:Apache License

private static void analyseSpreadsheet(Element da, ExcelFeatureAnalysis efa) {

    Element s = new Element("spreadsheets", sn);
    da.addContent(s);/* www . j  a va  2  s.  co m*/
    s.setAttribute("numberOfSheets", "" + efa.wb.getNumberOfSheets());
    // workbook wide features

    List<? extends PictureData> allPictures = efa.wb.getAllPictures();
    if (allPictures != null && allPictures.size() > 0) {
        Element oo = new Element("Pictures", sn);
        s.addContent(oo);
        for (PictureData pd : allPictures) {
            Element ob = new Element("Picture", sn);
            ob.setAttribute("mimeType", pd.getMimeType());
            oo.addContent(ob);
        }
    }

    int numfonts = efa.wb.getNumberOfFonts();
    if (numfonts > 0) {
        Element oo = new Element("Fonts", sn);
        s.addContent(oo);
        for (int i = 0; i < numfonts; i++) {
            Font cs = efa.wb.getFontAt((short) i);
            Element ob = new Element("Font", sn);
            ob.setAttribute("Name", cs.getFontName());

            ob.setAttribute("Charset", "" + cs.getCharSet());
            oo.addContent(ob);
        }
    }

    if (efa.hswb != null) {

        DocumentSummaryInformation dsi = efa.hswb.getDocumentSummaryInformation();
        if (dsi != null)
            s.setAttribute("OSVersion", "" + dsi.getOSVersion());
        // Property[] properties = dsi.getProperties();
        // CustomProperties customProperties = dsi.getCustomProperties();

        List<HSSFObjectData> eo = efa.hswb.getAllEmbeddedObjects();
        if (eo != null && eo.size() > 0) {
            Element oo = new Element("EmbeddedObjects", sn);
            s.addContent(oo);
            for (HSSFObjectData o : eo) {
                Element ob = new Element("EmbeddedObject", sn);
                ob.setAttribute("name", o.getOLE2ClassName());
                oo.addContent(ob);
            }

        }
    } else if (efa.xswb != null) {
        try {
            POIXMLProperties properties = efa.xswb.getProperties();
            List<PackagePart> allEmbedds = efa.xswb.getAllEmbedds();
            if (allEmbedds != null && allEmbedds.size() > 0) {
                Element oo = new Element("EmbeddedObjects", sn);
                s.addContent(oo);

                for (PackagePart p : allEmbedds) {
                    Element ob = new Element("EmbeddedObject", sn);
                    ob.setAttribute("mimeType", p.getContentType());
                    ob.setAttribute("name", p.getPartName().getName());

                    oo.addContent(ob);
                }
            }
        } catch (OpenXML4JException e) {
            // TODO Auto-generated catch block
            e.printStackTrace();
        }

    }
    int nn = efa.wb.getNumberOfNames();
    if (nn > 0) {
        Element oo = new Element("NamedCells", sn);
        s.addContent(oo);
    }

    // sheet specific features
    int total = efa.wb.getNumberOfSheets();
    for (int c = 0; c < total; c++) {
        Sheet sheet = efa.wb.getSheetAt(c);
        Element single = new Element("sheet", sn);
        s.addContent(single);
        analyseSheet(sheet, single, sn, efa);
    }
}

---