Examples with LimitedContentWriter org.olat.core.util.io.LimitedContentWriter used on opensource projects

Example 11 with LimitedContentWriter

use of org.olat.core.util.io.LimitedContentWriter in project openolat by klemens.

the class PowerPointDocument method readContent.

@Override
public FileContent readContent(VFSLeaf leaf) throws IOException, DocumentException {
    if (log.isDebug())
        log.debug("read PPT Content of leaf=" + leaf.getName());
    try (BufferedInputStream bis = new BufferedInputStream(leaf.getInputStream())) {
        LimitedContentWriter oStream = new LimitedContentWriter(100000, FileDocumentFactory.getMaxFileSize());
        extractText(bis, oStream);
        return new FileContent(oStream.toString());
    } catch (Exception e) {
        throw new DocumentException("Can not read PPT Content. File=" + leaf.getName(), e);
    }
}

Example 12 with LimitedContentWriter

use of org.olat.core.util.io.LimitedContentWriter in project openolat by klemens.

the class WordDocument method readContent.

@Override
protected FileContent readContent(VFSLeaf leaf) throws IOException, DocumentException {
    LimitedContentWriter sb = new LimitedContentWriter((int) leaf.getSize(), FileDocumentFactory.getMaxFileSize());
    try (InputStream bis = new BufferedInputStream(leaf.getInputStream())) {
        POIFSFileSystem filesystem = new POIFSFileSystem(bis);
        Iterator<?> entries = filesystem.getRoot().getEntries();
        while (entries.hasNext()) {
            Entry entry = (Entry) entries.next();
            String name = entry.getName();
            if (!(entry instanceof DocumentEntry)) {
            // Skip directory entries
            } else if ("WordDocument".equals(name)) {
                collectWordDocument(leaf, filesystem, sb);
            }
        }
        return new FileContent(sb.toString());
    } catch (Exception e) {
        log.warn("could not read in word document: " + leaf + " please check, that this is not an docx/rtf/html file!");
        throw new DocumentException(e.getMessage());
    }
}

Example 13 with LimitedContentWriter

use of org.olat.core.util.io.LimitedContentWriter in project openolat by klemens.

the class PdfBoxExtractor method extractTextFromPdf.

private FileContent extractTextFromPdf(VFSLeaf leaf) throws IOException, DocumentAccessException {
    if (log.isDebug())
        log.debug("readContent from pdf starts...");
    PDDocument document = null;
    BufferedInputStream bis = null;
    try {
        bis = new BufferedInputStream(leaf.getInputStream());
        document = PDDocument.load(bis);
        if (document.isEncrypted()) {
            try {
                document.decrypt("");
            } catch (Exception e) {
                log.warn("PDF is encrypted. Can not read content file=" + leaf.getName());
                LimitedContentWriter writer = new LimitedContentWriter(128, FileDocumentFactory.getMaxFileSize());
                writer.append(leaf.getName());
                writer.close();
                return new FileContent(leaf.getName(), writer.toString());
            }
        }
        String title = getTitle(document);
        if (log.isDebug())
            log.debug("readContent PDDocument loaded");
        PDFTextStripper stripper = new PDFTextStripper();
        LimitedContentWriter writer = new LimitedContentWriter(50000, FileDocumentFactory.getMaxFileSize());
        stripper.writeText(document, writer);
        writer.close();
        return new FileContent(title, writer.toString());
    } finally {
        if (document != null) {
            document.close();
        }
        if (bis != null) {
            bis.close();
        }
    }
}

Example 14 with LimitedContentWriter

use of org.olat.core.util.io.LimitedContentWriter in project OpenOLAT by OpenOLAT.

the class PowerPointOOXMLDocument method readContent.

@Override
public FileContent readContent(VFSLeaf leaf) throws IOException, DocumentException {
    File file = ((JavaIOItem) leaf).getBasefile();
    LimitedContentWriter writer = new LimitedContentWriter(100000, FileDocumentFactory.getMaxFileSize());
    try (ZipFile wordFile = new ZipFile(file)) {
        List<String> contents = new ArrayList<>();
        for (Enumeration<? extends ZipEntry> entriesEnumeration = wordFile.entries(); entriesEnumeration.hasMoreElements(); ) {
            ZipEntry entry = entriesEnumeration.nextElement();
            String name = entry.getName();
            if (name.startsWith(SLIDE) && name.endsWith(".xml")) {
                contents.add(name);
            }
        }
        if (contents.size() > 1) {
            Collections.sort(contents, new PowerPointDocumentComparator());
        }
        for (String content : contents) {
            if (writer.accept()) {
                ZipEntry entry = wordFile.getEntry(content);
                InputStream zip = wordFile.getInputStream(entry);
                OfficeDocumentHandler dh = new OfficeDocumentHandler(writer);
                parse(new ShieldInputStream(zip), dh);
                zip.close();
            }
        }
    } catch (DocumentException e) {
        throw e;
    } catch (Exception e) {
        throw new DocumentException(e.getMessage());
    }
    return new FileContent(writer.toString());
}

Example 15 with LimitedContentWriter

use of org.olat.core.util.io.LimitedContentWriter in project openolat by klemens.

the class ExcelOOXMLDocument method parseSheets.

private String parseSheets(Map<String, String> sharedStrings, VFSLeaf leaf) throws IOException, DocumentException {
    try (InputStream stream = leaf.getInputStream();
        ZipInputStream zip = new ZipInputStream(stream)) {
        ZipEntry entry = zip.getNextEntry();
        LimitedContentWriter writer = new LimitedContentWriter(100000, FileDocumentFactory.getMaxFileSize());
        while (entry != null) {
            if (writer.accept()) {
                String name = entry.getName();
                if (name.startsWith(SHEET) && name.endsWith(".xml")) {
                    OfficeDocumentHandler dh = new OfficeDocumentHandler(writer, sharedStrings);
                    parse(new ShieldInputStream(zip), dh);
                }
            }
            entry = zip.getNextEntry();
        }
        return writer.toString();
    } catch (DocumentException e) {
        throw e;
    } catch (Exception e) {
        throw new DocumentException(e.getMessage());
    }
}