Examples with PDStream org.apache.pdfbox.pdmodel.common.PDStream used on opensource projects

Example 1 with PDStream

use of org.apache.pdfbox.pdmodel.common.PDStream in project lucene-solr by apache.

the class ReducePDFSize method main.

public static void main(String[] args) throws IOException {
    if (2 != args.length) {
        throw new RuntimeException("arg0 must be input file, org1 must be output file");
    }
    String in = args[0];
    String out = args[1];
    PDDocument doc = null;
    try {
        doc = PDDocument.load(new File(in));
        doc.setAllSecurityToBeRemoved(true);
        for (COSObject cosObject : doc.getDocument().getObjects()) {
            COSBase base = cosObject.getObject();
            // if it's a stream: decode it, then re-write it using FLATE_DECODE
            if (base instanceof COSStream) {
                COSStream stream = (COSStream) base;
                byte[] bytes;
                try {
                    bytes = new PDStream(stream).toByteArray();
                } catch (IOException ex) {
                    // IIUC then that will leave the original (non-decoded / non-flated) stream in place?
                    throw new RuntimeException("can't serialize byte[] from: " + cosObject.getObjectNumber() + " " + cosObject.getGenerationNumber() + " obj: " + ex.getMessage(), ex);
                }
                stream.removeItem(COSName.FILTER);
                OutputStream streamOut = stream.createOutputStream(COSName.FLATE_DECODE);
                streamOut.write(bytes);
                streamOut.close();
            }
        }
        doc.getDocumentCatalog();
        doc.save(out);
    } finally {
        if (doc != null) {
            doc.close();
        }
    }
}