Examples with TemporaryResources org.apache.tika.io.TemporaryResources used on opensource projects

Example 26 with TemporaryResources

use of org.apache.tika.io.TemporaryResources in project tika by apache.

the class GDALParser method parse.

@Override
public void parse(InputStream stream, ContentHandler handler, Metadata metadata, ParseContext context) throws IOException, SAXException, TikaException {
    if (!ExternalParser.check("gdalinfo")) {
        return;
    }
    // first set up and run GDAL
    // process the command
    TemporaryResources tmp = new TemporaryResources();
    TikaInputStream tis = TikaInputStream.get(stream, tmp);
    String runCommand = processCommand(tis);
    String output = execCommand(new String[] { runCommand });
    // now extract the actual metadata params
    // from the GDAL output in the content stream
    // to do this, we need to literally process the output
    // from the invoked command b/c we can't read metadata and
    // output text from the handler in ExternalParser
    // at the same time, so for now, we can't use the
    // ExternalParser to do this and I've had to bring some of
    // that functionality directly into this class
    // TODO: investigate a way to do both using ExternalParser
    extractMetFromOutput(output, metadata);
    applyPatternsToOutput(output, metadata, getPatterns());
    // make the content handler and provide output there
    // now that we have metadata
    processOutput(handler, metadata, output);
}

Example 27 with TemporaryResources

use of org.apache.tika.io.TemporaryResources in project tika by apache.

the class TiffParser method parse.

public void parse(InputStream stream, ContentHandler handler, Metadata metadata, ParseContext context) throws IOException, SAXException, TikaException {
    TemporaryResources tmp = new TemporaryResources();
    try {
        TikaInputStream tis = TikaInputStream.get(stream, tmp);
        new ImageMetadataExtractor(metadata).parseTiff(tis.getFile());
        new JempboxExtractor(metadata).parse(tis);
    } finally {
        tmp.dispose();
    }
    XHTMLContentHandler xhtml = new XHTMLContentHandler(handler, metadata);
    xhtml.startDocument();
    xhtml.endDocument();
}

Example 28 with TemporaryResources

use of org.apache.tika.io.TemporaryResources in project tika by apache.

the class OutlookPSTParser method parseMailAttachments.

private void parseMailAttachments(XHTMLContentHandler xhtml, PSTMessage email, EmbeddedDocumentExtractor embeddedExtractor) throws TikaException {
    int numberOfAttachments = email.getNumberOfAttachments();
    for (int i = 0; i < numberOfAttachments; i++) {
        File tempFile = null;
        try {
            PSTAttachment attach = email.getAttachment(i);
            // Get the filename; both long and short filenames can be used for attachments
            String filename = attach.getLongFilename();
            if (filename.isEmpty()) {
                filename = attach.getFilename();
            }
            xhtml.element("p", filename);
            Metadata attachMeta = new Metadata();
            attachMeta.set(Metadata.RESOURCE_NAME_KEY, filename);
            attachMeta.set(Metadata.EMBEDDED_RELATIONSHIP_ID, filename);
            AttributesImpl attributes = new AttributesImpl();
            attributes.addAttribute("", "class", "class", "CDATA", "embedded");
            attributes.addAttribute("", "id", "id", "CDATA", filename);
            xhtml.startElement("div", attributes);
            if (embeddedExtractor.shouldParseEmbedded(attachMeta)) {
                TemporaryResources tmp = new TemporaryResources();
                try {
                    TikaInputStream tis = TikaInputStream.get(attach.getFileInputStream(), tmp);
                    embeddedExtractor.parseEmbedded(tis, xhtml, attachMeta, true);
                } finally {
                    tmp.dispose();
                }
            }
            xhtml.endElement("div");
        } catch (Exception e) {
            throw new TikaException("Unable to unpack document stream", e);
        } finally {
            if (tempFile != null)
                tempFile.delete();
        }
    }
}

Example 29 with TemporaryResources

use of org.apache.tika.io.TemporaryResources in project tika by apache.

the class ISArchiveParser method parse.

@Override
public void parse(InputStream stream, ContentHandler handler, Metadata metadata, ParseContext context) throws IOException, SAXException, TikaException {
    TemporaryResources tmp = TikaInputStream.isTikaInputStream(stream) ? null : new TemporaryResources();
    TikaInputStream tis = TikaInputStream.get(stream, tmp);
    try {
        if (this.location == null) {
            this.location = tis.getFile().getParent() + File.separator;
        }
        this.studyFileName = tis.getFile().getName();
        File locationFile = new File(location);
        String[] investigationList = locationFile.list(new FilenameFilter() {

            @Override
            public boolean accept(File dir, String name) {
                return name.matches("i_.+\\.txt");
            }
        });
        XHTMLContentHandler xhtml = new XHTMLContentHandler(handler, metadata);
        xhtml.startDocument();
        parseInvestigation(investigationList, xhtml, metadata, context);
        parseStudy(stream, xhtml, metadata, context);
        parseAssay(xhtml, metadata, context);
        xhtml.endDocument();
    } finally {
        if (tmp != null) {
            tmp.dispose();
        }
    }
}

Example 30 with TemporaryResources

use of org.apache.tika.io.TemporaryResources in project tika by apache.

the class JournalParser method parse.

public void parse(InputStream stream, ContentHandler handler, Metadata metadata, ParseContext context) throws IOException, SAXException, TikaException {
    TikaInputStream tis = TikaInputStream.get(stream, new TemporaryResources());
    File tmpFile = tis.getFile();
    GrobidRESTParser grobidParser = new GrobidRESTParser();
    grobidParser.parse(tmpFile.getAbsolutePath(), handler, metadata, context);
    PDFParser parser = new PDFParser();
    parser.parse(new FileInputStream(tmpFile), handler, metadata, context);
}