/**
* @param pkg
* @throws IOException
* @throws SAXException
* @throws ParserConfigurationException
*/
public ReadonlySharedStringsTable(OPCPackage pkg)
throws IOException, SAXException, ParserConfigurationException {
ArrayList<PackagePart> parts =
pkg.getPartsByContentType(XSSFRelation.SHARED_STRINGS.getContentType());
PackagePart sstPart = parts.get(0);
readFrom(sstPart.getInputStream());
}
@Override
protected void commit() throws IOException {
PackagePart part = getPackagePart();
OutputStream out = part.getOutputStream();
writeTo(out);
out.close();
}
/**
* Called to update the embedded Excel workbook. As the format and structire of the workbook are
* known in advance, all this code attempts to do is write a new value into the first cell on the
* first row of the first worksheet. Prior to executing this method, that cell will contain the
* value 1.
*
* @throws org.apache.poi.openxml4j.exceptions.OpenXML4JException Rather than use the specific
* classes (HSSF/XSSF) to handle the embedded workbook this method uses those defeined in the
* SS stream. As a result, it might be the case that a SpreadsheetML file is opened for
* processing, throwing this exception if that file is invalid.
* @throws java.io.IOException Thrown if a problem occurs in the underlying file system.
*/
public void updateEmbeddedDoc() throws OpenXML4JException, IOException {
Workbook workbook = null;
Sheet sheet = null;
Row row = null;
Cell cell = null;
PackagePart pPart = null;
Iterator<PackagePart> pIter = null;
List<PackagePart> embeddedDocs = this.doc.getAllEmbedds();
if (embeddedDocs != null && !embeddedDocs.isEmpty()) {
pIter = embeddedDocs.iterator();
while (pIter.hasNext()) {
pPart = pIter.next();
if (pPart.getPartName().getExtension().equals(BINARY_EXTENSION)
|| pPart.getPartName().getExtension().equals(OPENXML_EXTENSION)) {
// Get an InputStream from the pacage part and pass that
// to the create method of the WorkbookFactory class. Update
// the resulting Workbook and then stream that out again
// using an OutputStream obtained from the same PackagePart.
workbook = WorkbookFactory.create(pPart.getInputStream());
sheet = workbook.getSheetAt(SHEET_NUM);
row = sheet.getRow(ROW_NUM);
cell = row.getCell(CELL_NUM);
cell.setCellValue(NEW_VALUE);
workbook.write(pPart.getOutputStream());
}
}
// Finally, write the newly modified Word document out to file.
this.doc.write(new FileOutputStream(this.docFile));
}
}
@Override
protected void commit() throws IOException {
XmlOptions xmlOptions = new XmlOptions(DEFAULT_XML_OPTIONS);
Map<String, String> map = new HashMap<String, String>();
map.put(STRelationshipId.type.getName().getNamespaceURI(), "r");
xmlOptions.setSaveSuggestedPrefixes(map);
PackagePart part = getPackagePart();
OutputStream out = part.getOutputStream();
_presentation.save(out, xmlOptions);
out.close();
}
public MediaType detect(TikaInputStream input, Metadata metadata) throws IOException {
ZipFile zip = new ZipFile(input.getFile());
for (ZipEntry entry : Collections.list(zip.entries())) {
// Is it an Open Document file?
if (entry.getName().equals("mimetype")) {
InputStream stream = zip.getInputStream(entry);
try {
return fromString(IOUtils.toString(stream, "UTF-8"));
} finally {
stream.close();
}
} else if (entry.getName().equals("_rels/.rels")
|| entry.getName().equals("[Content_Types].xml")) {
// Office Open XML File
// As POI to open and investigate it for us
try {
OPCPackage pkg = OPCPackage.open(input.getFile().toString());
input.setOpenContainer(pkg);
PackageRelationshipCollection core =
pkg.getRelationshipsByType(ExtractorFactory.CORE_DOCUMENT_REL);
if (core.size() != 1) {
throw new IOException(
"Invalid OOXML Package received - expected 1 core document, found " + core.size());
}
// 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('.'));
return fromString(docType);
} catch (InvalidFormatException e) {
throw new IOException("Office Open XML File detected, but corrupted - " + e.getMessage());
}
} else if (entry.getName().equals("buildVersionHistory.plist")) {
// This is an iWork document
// Reset and ask
zip.close();
zip = new ZipFile(input.getFile());
return IWorkPackageParser.identifyType(zip);
} else if (entry.getName().equals("META-INF/")) {
// Java Jar
return MediaType.application("java-archive");
}
}
return MediaType.APPLICATION_ZIP;
}
/** save and commit numbering */
@Override
protected void commit() throws IOException {
XmlOptions xmlOptions = new XmlOptions(DEFAULT_XML_OPTIONS);
xmlOptions.setSaveSyntheticDocumentElement(
new QName(CTNumbering.type.getName().getNamespaceURI(), "numbering"));
Map map = new HashMap();
map.put("http://schemas.openxmlformats.org/markup-compatibility/2006", "ve");
map.put("urn:schemas-microsoft-com:office:office", "o");
map.put("http://schemas.openxmlformats.org/officeDocument/2006/relationships", "r");
map.put("http://schemas.openxmlformats.org/officeDocument/2006/math", "m");
map.put("urn:schemas-microsoft-com:vml", "v");
map.put("http://schemas.openxmlformats.org/drawingml/2006/wordprocessingDrawing", "wp");
map.put("urn:schemas-microsoft-com:office:word", "w10");
map.put("http://schemas.openxmlformats.org/wordprocessingml/2006/main", "w");
map.put("http://schemas.microsoft.com/office/word/2006/wordml", "wne");
xmlOptions.setSaveSuggestedPrefixes(map);
PackagePart part = getPackagePart();
OutputStream out = part.getOutputStream();
ctNumbering.save(out, xmlOptions);
out.close();
}
/**
* Called to test whether or not the embedded workbook was correctly updated. This method simply
* recovers the first cell from the first row of the first workbook and tests the value it
* contains.
*
* <p>Note that execution will not continue up to the assertion as the embedded workbook is now
* corrupted and causes an IllegalArgumentException with the following message
*
* <p><em>java.lang.IllegalArgumentException: Your InputStream was neither an OLE2 stream, nor an
* OOXML stream</em>
*
* <p>to be thrown when the WorkbookFactory.createWorkbook(InputStream) method is executed.
*
* @throws org.apache.poi.openxml4j.exceptions.OpenXML4JException Rather than use the specific
* classes (HSSF/XSSF) to handle the embedded workbook this method uses those defeined in the
* SS stream. As a result, it might be the case that a SpreadsheetML file is opened for
* processing, throwing this exception if that file is invalid.
* @throws java.io.IOException Thrown if a problem occurs in the underlying file system.
*/
public void checkUpdatedDoc() throws OpenXML4JException, IOException {
Workbook workbook = null;
Sheet sheet = null;
Row row = null;
Cell cell = null;
PackagePart pPart = null;
Iterator<PackagePart> pIter = null;
List<PackagePart> embeddedDocs = this.doc.getAllEmbedds();
if (embeddedDocs != null && !embeddedDocs.isEmpty()) {
pIter = embeddedDocs.iterator();
while (pIter.hasNext()) {
pPart = pIter.next();
if (pPart.getPartName().getExtension().equals(BINARY_EXTENSION)
|| pPart.getPartName().getExtension().equals(OPENXML_EXTENSION)) {
workbook = WorkbookFactory.create(pPart.getInputStream());
sheet = workbook.getSheetAt(SHEET_NUM);
row = sheet.getRow(ROW_NUM);
cell = row.getCell(CELL_NUM);
assertEquals(cell.getNumericCellValue(), NEW_VALUE, 0.0001);
}
}
}
}
public Table(PackagePart part, PackageRelationship rel) throws IOException {
super(part, rel);
readFrom(part.getInputStream());
}
/**
* Like POIXMLDocumentPart constructor
*
* @param part
* @param rel_ignored
* @throws IOException
*/
public ReadonlySharedStringsTable(PackagePart part, PackageRelationship rel_ignored)
throws IOException, SAXException, ParserConfigurationException {
readFrom(part.getInputStream());
}
