Examples with RecordFormatException org.apache.poi.util.RecordFormatException used on opensource projects

Example 6 with RecordFormatException

use of org.apache.poi.util.RecordFormatException in project poi by apache.

the class EscherTextboxRecord method serialize.

@Override
public int serialize(int offset, byte[] data, EscherSerializationListener listener) {
    listener.beforeRecordSerialize(offset, getRecordId(), this);
    LittleEndian.putShort(data, offset, getOptions());
    LittleEndian.putShort(data, offset + 2, getRecordId());
    int remainingBytes = thedata.length;
    LittleEndian.putInt(data, offset + 4, remainingBytes);
    System.arraycopy(thedata, 0, data, offset + 8, thedata.length);
    int pos = offset + 8 + thedata.length;
    listener.afterRecordSerialize(pos, getRecordId(), pos - offset, this);
    int size = pos - offset;
    if (size != getRecordSize())
        throw new RecordFormatException(size + " bytes written but getRecordSize() reports " + getRecordSize());
    return size;
}

Example 7 with RecordFormatException

use of org.apache.poi.util.RecordFormatException in project poi by apache.

the class HSSFShapeFactory method createShapeTree.

/**
     * build shape tree from escher container
     * @param container root escher container from which escher records must be taken
     * @param agg - EscherAggregate
     * @param out - shape container to which shapes must be added
     * @param root - node to create HSSFObjectData shapes
     */
public static void createShapeTree(EscherContainerRecord container, EscherAggregate agg, HSSFShapeContainer out, DirectoryNode root) {
    if (container.getRecordId() == EscherContainerRecord.SPGR_CONTAINER) {
        ObjRecord obj = null;
        EscherClientDataRecord clientData = ((EscherContainerRecord) container.getChild(0)).getChildById(EscherClientDataRecord.RECORD_ID);
        if (null != clientData) {
            obj = (ObjRecord) agg.getShapeToObjMapping().get(clientData);
        }
        HSSFShapeGroup group = new HSSFShapeGroup(container, obj);
        List<EscherContainerRecord> children = container.getChildContainers();
        // skip the first child record, it is group descriptor
        for (int i = 0; i < children.size(); i++) {
            EscherContainerRecord spContainer = children.get(i);
            if (i != 0) {
                createShapeTree(spContainer, agg, group, root);
            }
        }
        out.addShape(group);
    } else if (container.getRecordId() == EscherContainerRecord.SP_CONTAINER) {
        Map<EscherRecord, Record> shapeToObj = agg.getShapeToObjMapping();
        ObjRecord objRecord = null;
        TextObjectRecord txtRecord = null;
        for (EscherRecord record : container) {
            switch(record.getRecordId()) {
                case EscherClientDataRecord.RECORD_ID:
                    objRecord = (ObjRecord) shapeToObj.get(record);
                    break;
                case EscherTextboxRecord.RECORD_ID:
                    txtRecord = (TextObjectRecord) shapeToObj.get(record);
                    break;
                default:
                    break;
            }
        }
        if (objRecord == null) {
            throw new RecordFormatException("EscherClientDataRecord can't be found.");
        }
        if (isEmbeddedObject(objRecord)) {
            HSSFObjectData objectData = new HSSFObjectData(container, objRecord, root);
            out.addShape(objectData);
            return;
        }
        CommonObjectDataSubRecord cmo = (CommonObjectDataSubRecord) objRecord.getSubRecords().get(0);
        final HSSFShape shape;
        switch(cmo.getObjectType()) {
            case CommonObjectDataSubRecord.OBJECT_TYPE_PICTURE:
                shape = new HSSFPicture(container, objRecord);
                break;
            case CommonObjectDataSubRecord.OBJECT_TYPE_RECTANGLE:
                shape = new HSSFSimpleShape(container, objRecord, txtRecord);
                break;
            case CommonObjectDataSubRecord.OBJECT_TYPE_LINE:
                shape = new HSSFSimpleShape(container, objRecord);
                break;
            case CommonObjectDataSubRecord.OBJECT_TYPE_COMBO_BOX:
                shape = new HSSFCombobox(container, objRecord);
                break;
            case CommonObjectDataSubRecord.OBJECT_TYPE_MICROSOFT_OFFICE_DRAWING:
                EscherOptRecord optRecord = container.getChildById(EscherOptRecord.RECORD_ID);
                if (optRecord == null) {
                    shape = new HSSFSimpleShape(container, objRecord, txtRecord);
                } else {
                    EscherProperty property = optRecord.lookup(EscherProperties.GEOMETRY__VERTICES);
                    if (null != property) {
                        shape = new HSSFPolygon(container, objRecord, txtRecord);
                    } else {
                        shape = new HSSFSimpleShape(container, objRecord, txtRecord);
                    }
                }
                break;
            case CommonObjectDataSubRecord.OBJECT_TYPE_TEXT:
                shape = new HSSFTextbox(container, objRecord, txtRecord);
                break;
            case CommonObjectDataSubRecord.OBJECT_TYPE_COMMENT:
                shape = new HSSFComment(container, objRecord, txtRecord, agg.getNoteRecordByObj(objRecord));
                break;
            default:
                shape = new HSSFSimpleShape(container, objRecord, txtRecord);
        }
        out.addShape(shape);
    }
}

Example 8 with RecordFormatException

use of org.apache.poi.util.RecordFormatException in project tika by apache.

the class EMFParser method parse.

@Override
public void parse(InputStream stream, ContentHandler handler, Metadata metadata, ParseContext context) throws IOException, SAXException, TikaException {
    EmbeddedDocumentExtractor embeddedDocumentExtractor = null;
    XHTMLContentHandler xhtml = new XHTMLContentHandler(handler, metadata);
    xhtml.startDocument();
    try {
        HemfExtractor ex = new HemfExtractor(stream);
        long lastY = -1;
        long lastX = -1;
        //derive this from the font or frame/bounds information
        long fudgeFactorX = 1000;
        StringBuilder buffer = new StringBuilder();
        for (HemfRecord record : ex) {
            if (record.getRecordType() == HemfRecordType.comment) {
                AbstractHemfComment comment = ((HemfCommentRecord) record).getComment();
                if (comment instanceof HemfCommentPublic.MultiFormats) {
                    if (embeddedDocumentExtractor == null) {
                        embeddedDocumentExtractor = EmbeddedDocumentUtil.getEmbeddedDocumentExtractor(context);
                    }
                    handleMultiFormats((HemfCommentPublic.MultiFormats) comment, xhtml, embeddedDocumentExtractor);
                } else if (comment instanceof HemfCommentPublic.WindowsMetafile) {
                    if (embeddedDocumentExtractor == null) {
                        embeddedDocumentExtractor = EmbeddedDocumentUtil.getEmbeddedDocumentExtractor(context);
                    }
                    handleWMF((HemfCommentPublic.WindowsMetafile) comment, xhtml, embeddedDocumentExtractor);
                }
            } else if (record.getRecordType().equals(HemfRecordType.exttextoutw)) {
                HemfText.ExtTextOutW extTextOutW = (HemfText.ExtTextOutW) record;
                if (lastY > -1 && lastY != extTextOutW.getY()) {
                    xhtml.startElement("p");
                    xhtml.characters(buffer.toString());
                    xhtml.endElement("p");
                    buffer.setLength(0);
                    lastX = -1;
                }
                if (lastX > -1 && extTextOutW.getX() - lastX > fudgeFactorX) {
                    buffer.append(" ");
                }
                String txt = extTextOutW.getText();
                buffer.append(txt);
                lastY = extTextOutW.getY();
                lastX = extTextOutW.getX();
            }
        }
        if (buffer.length() > 0) {
            xhtml.startElement("p");
            xhtml.characters(buffer.toString());
            xhtml.endElement("p");
        }
    } catch (RecordFormatException e) {
        //POI's hemfparser can throw these for "parse exceptions"
        throw new TikaException(e.getMessage(), e);
    } catch (RuntimeException e) {
        //convert Runtime to RecordFormatExceptions
        throw new TikaException(e.getMessage(), e);
    }
    xhtml.endDocument();
}

Example 9 with RecordFormatException

use of org.apache.poi.util.RecordFormatException in project tika by apache.

the class WMFParser method parse.

@Override
public void parse(InputStream stream, ContentHandler handler, Metadata metadata, ParseContext context) throws IOException, SAXException, TikaException {
    XHTMLContentHandler xhtml = new XHTMLContentHandler(handler, metadata);
    xhtml.startDocument();
    try {
        HwmfPicture picture = new HwmfPicture(stream);
        //to determine when to keep two text parts on the same line
        for (HwmfRecord record : picture.getRecords()) {
            Charset charset = LocaleUtil.CHARSET_1252;
            //This fix should be done within POI
            if (record.getRecordType().equals(HwmfRecordType.createFontIndirect)) {
                HwmfFont font = ((HwmfText.WmfCreateFontIndirect) record).getFont();
                charset = (font.getCharSet() == null || font.getCharSet().getCharset() == null) ? LocaleUtil.CHARSET_1252 : font.getCharSet().getCharset();
            }
            if (record.getRecordType().equals(HwmfRecordType.extTextOut)) {
                HwmfText.WmfExtTextOut textOut = (HwmfText.WmfExtTextOut) record;
                xhtml.startElement("p");
                xhtml.characters(textOut.getText(charset));
                xhtml.endElement("p");
            } else if (record.getRecordType().equals(HwmfRecordType.textOut)) {
                HwmfText.WmfTextOut textOut = (HwmfText.WmfTextOut) record;
                xhtml.startElement("p");
                xhtml.characters(textOut.getText(charset));
                xhtml.endElement("p");
            }
        }
    } catch (RecordFormatException e) {
        //POI's hwmfparser can throw these for "parse exceptions"
        throw new TikaException(e.getMessage(), e);
    } catch (RuntimeException e) {
        //convert Runtime to RecordFormatExceptions
        throw new TikaException(e.getMessage(), e);
    } catch (AssertionError e) {
        //POI's hwmfparser can throw these for parse exceptions
        throw new TikaException(e.getMessage(), e);
    }
    xhtml.endDocument();
}