public static void displayImageInfo(ImageInfo info) {
ImageSize size = info.getSize();
Dimension2D dPt = size.getDimensionPt();
Dimension dPx = size.getDimensionPx();
log.debug(
info.getOriginalURI()
+ " "
+ info.getMimeType()
+ " "
+ Math.round(dPx.getWidth())
+ "x"
+ Math.round(dPx.getHeight()));
log.debug(
"Resolution:"
+ Math.round(size.getDpiHorizontal())
+ "x"
+ Math.round(size.getDpiVertical()));
log.debug(
"Print size: "
+ Math.round(dPt.getWidth() / 72)
+ "\" x"
+ Math.round(dPt.getHeight() / 72)
+ "\"");
}
public static CxCy scale(ImageInfo imageInfo, double xEmu, double yEmu) {
ImageSize size = imageInfo.getSize();
double iwEmu = toEmu(size.getWidthPx(), size.getDpiHorizontal());
double ihEmu = toEmu(size.getHeightPx(), size.getDpiVertical());
return scaleToFit(iwEmu, ihEmu, xEmu, yEmu);
}
/**
* Create a linked image part, and attach it as a rel of the specified source part (eg a header
* part).
*
* <p>The current behaviour is that the part is added to the package, but since the target mode of
* the rel is external, the part is redundant.
*
* @param wordMLPackage
* @param sourcePart
* @param url
* @return
* @throws Exception
*/
public static BinaryPartAbstractImage createLinkedImagePart(
OpcPackage opcPackage, Part sourcePart, URL url) throws Exception {
log.debug("Incoming url for linked image: " + url.toString());
ImageInfo info =
ensureFormatIsSupported(url, null, null, false); // final param doesn't matter in this case
ContentTypeManager ctm = opcPackage.getContentTypeManager();
String proposedRelId = sourcePart.getRelationshipsPart().getNextId();
// In order to ensure unique part name,
// idea is to use the relId, which ought to be unique
String ext = info.getMimeType().substring(info.getMimeType().indexOf("/") + 1);
BinaryPartAbstractImage imagePart =
(BinaryPartAbstractImage)
ctm.newPartForContentType(
info.getMimeType(),
createImageName(opcPackage, sourcePart, proposedRelId, ext),
null);
// NB: contents never populated
log.debug(
"created part "
+ imagePart.getClass().getName()
+ " with name "
+ imagePart.getPartName().toString());
imagePart.rels.add(
sourcePart.addTargetPart(
imagePart)); // want to create rel with suitable name; side effect is to add part
imagePart.getRelLast().setTargetMode("External");
opcPackage.getExternalResources().put(imagePart.getExternalTarget(), imagePart);
// if (!url.getProtocol().equals("file") && new File(url.toString() ).isFile()) {
// imagePart.rel.setTarget("file:///" + url);
// } else {
// imagePart.rel.setTarget(url.toString());
// }
imagePart.getRelLast().setTarget(url.toString());
imagePart.setImageInfo(info);
return imagePart;
}
/**
* Create an image part from the provided filePath image, attach it to the source part (eg the
* main document part, a header part etc), and return it.
*
* <p>Works for both docx and pptx.
*
* @param opcPackage
* @param sourcePart
* @param filePath
* @return
* @throws Exception
*/
public static BinaryPartAbstractImage createImagePart(
OpcPackage opcPackage, Part sourcePart, File imageFile) throws Exception {
final byte[] locByte = new byte[1];
// We are in the case that image is not load (no byte Array) so isLoad is false
ImageInfo info = ensureFormatIsSupported(imageFile, locByte, false);
ContentTypeManager ctm = opcPackage.getContentTypeManager();
// Ensure the relationships part exists
if (sourcePart.getRelationshipsPart() == null) {
RelationshipsPart.createRelationshipsPartForPart(sourcePart);
}
String proposedRelId = sourcePart.getRelationshipsPart().getNextId();
String ext = info.getMimeType().substring(info.getMimeType().indexOf("/") + 1);
BinaryPartAbstractImage imagePart =
(BinaryPartAbstractImage)
ctm.newPartForContentType(
info.getMimeType(),
createImageName(opcPackage, sourcePart, proposedRelId, ext),
null);
log.debug(
"created part "
+ imagePart.getClass().getName()
+ " with name "
+ imagePart.getPartName().toString());
FileInputStream fis = new FileInputStream(imageFile);
imagePart.setBinaryData(fis);
imagePart.rels.add(sourcePart.addTargetPart(imagePart, proposedRelId));
imagePart.setImageInfo(info);
return imagePart;
}
private ICC_Profile buildICCProfile(
final ImageInfo info, final ColorSpace colorSpace, final ByteArrayOutputStream iccStream)
throws IOException {
if (iccStream != null && iccStream.size() > 0) {
if (log.isDebugEnabled()) {
log.debug("Effective ICC profile size: " + iccStream.size());
}
final int alignment = 4;
final int padding = (alignment - iccStream.size() % alignment) % alignment;
if (padding != 0) {
try {
iccStream.write(new byte[padding]);
} catch (final IOException ioe) {
throw new IOException("Error while aligning ICC stream: " + ioe.getMessage());
}
}
ICC_Profile iccProfile = null;
try {
iccProfile = ColorProfileUtil.getICC_Profile(iccStream.toByteArray());
if (log.isDebugEnabled()) {
log.debug("JPEG has an ICC profile: " + iccProfile.toString());
}
} catch (final IllegalArgumentException iae) {
log.warn(
"An ICC profile is present in the JPEG file but it is invalid ("
+ iae.getMessage()
+ "). The color profile will be ignored. ("
+ info.getOriginalURI()
+ ")");
return null;
}
if (iccProfile.getNumComponents() != colorSpace.getNumComponents()) {
log.warn(
"The number of components of the ICC profile ("
+ iccProfile.getNumComponents()
+ ") doesn't match the image ("
+ colorSpace.getNumComponents()
+ "). Ignoring the ICC color profile.");
return null;
} else {
return iccProfile;
}
} else {
return null; // no ICC profile available
}
}
/**
* Create a <wp:inline> element suitable for this image, which can be _embedded_ in
* w:p/w:r/w:drawing.
*
* @param filenameHint Any text, for example the original filename
* @param altText Like HTML's alt text
* @param id1 An id unique in the document
* @param id2 Another id unique in the document
* @param cx Image width in twip
* @param link true if this is to be linked not embedded None of these things seem to be exposed
* in Word 2007's user interface, but Word won't open the document if any of the attributes
* these go in (except @ desc) aren't present!
* @throws Exception
*/
public Inline createImageInline(
String filenameHint, String altText, int id1, int id2, long cx, boolean link)
throws Exception {
ImageSize size = imageInfo.getSize();
Dimension2D dPt = size.getDimensionPt();
double imageWidthTwips = dPt.getWidth() * 20;
log.debug("imageWidthTwips: " + imageWidthTwips);
long cy;
log.debug("Scaling image height to retain aspect ratio");
cy = UnitsOfMeasurement.twipToEMU(dPt.getHeight() * 20 * cx / imageWidthTwips);
// Now convert cx to EMU
cx = UnitsOfMeasurement.twipToEMU(cx);
log.debug("cx=" + cx + "; cy=" + cy);
return createImageInline(filenameHint, altText, id1, id2, cx, cy, link);
}
public static CxCy scale(ImageInfo imageInfo, PageDimensions page) {
double writableWidthTwips = page.getWritableWidthTwips();
log.debug("writableWidthTwips: " + writableWidthTwips);
ImageSize size = imageInfo.getSize();
Dimension2D dPt = size.getDimensionPt();
double imageWidthTwips = dPt.getWidth() * 20;
log.debug("imageWidthTwips: " + imageWidthTwips);
long cx;
long cy;
boolean scaled = false;
if (imageWidthTwips > writableWidthTwips) {
log.debug("Scaling image to fit page width");
scaled = true;
cx = UnitsOfMeasurement.twipToEMU(writableWidthTwips);
cy =
UnitsOfMeasurement.twipToEMU(
dPt.getHeight() * 20 * writableWidthTwips / imageWidthTwips);
} else {
log.debug("Scaling image - not necessary");
cx = UnitsOfMeasurement.twipToEMU(imageWidthTwips);
cy = UnitsOfMeasurement.twipToEMU(dPt.getHeight() * 20);
}
log.debug("cx=" + cx + "; cy=" + cy);
return new CxCy(cx, cy, scaled);
}
/** {@inheritDoc} */
public void activateLayout() {
initialize();
FOUserAgent userAgent = pageSeq.getUserAgent();
ImageManager imageManager = userAgent.getFactory().getImageManager();
String uri = getExternalDocument().getSrc();
Integer firstPageIndex = ImageUtil.getPageIndexFromURI(uri);
boolean hasPageIndex = (firstPageIndex != null);
try {
ImageInfo info = imageManager.getImageInfo(uri, userAgent.getImageSessionContext());
Object moreImages = info.getCustomObjects().get(ImageInfo.HAS_MORE_IMAGES);
boolean hasMoreImages = moreImages != null && !Boolean.FALSE.equals(moreImages);
Dimension intrinsicSize = info.getSize().getDimensionMpt();
ImageLayout layout = new ImageLayout(getExternalDocument(), this, intrinsicSize);
PageSequence pageSequence = new PageSequence(null);
transferExtensions(pageSequence);
areaTreeHandler.getAreaTreeModel().startPageSequence(pageSequence);
if (log.isDebugEnabled()) {
log.debug("Starting layout");
}
makePageForImage(info, layout);
if (!hasPageIndex && hasMoreImages) {
if (log.isTraceEnabled()) {
log.trace("Starting multi-page processing...");
}
URI originalURI;
try {
originalURI = new URI(URISpecification.escapeURI(uri));
int pageIndex = 1;
while (hasMoreImages) {
URI tempURI =
new URI(
originalURI.getScheme(),
originalURI.getSchemeSpecificPart(),
"page=" + Integer.toString(pageIndex + 1));
if (log.isTraceEnabled()) {
log.trace("Subimage: " + tempURI.toASCIIString());
}
ImageInfo subinfo =
imageManager.getImageInfo(
tempURI.toASCIIString(), userAgent.getImageSessionContext());
moreImages = subinfo.getCustomObjects().get(ImageInfo.HAS_MORE_IMAGES);
hasMoreImages = moreImages != null && !Boolean.FALSE.equals(moreImages);
intrinsicSize = subinfo.getSize().getDimensionMpt();
layout = new ImageLayout(getExternalDocument(), this, intrinsicSize);
makePageForImage(subinfo, layout);
pageIndex++;
}
} catch (URISyntaxException e) {
getResourceEventProducer().uriError(this, uri, e, getExternalDocument().getLocator());
return;
}
}
} catch (FileNotFoundException fnfe) {
getResourceEventProducer().imageNotFound(this, uri, fnfe, getExternalDocument().getLocator());
} catch (IOException ioe) {
getResourceEventProducer().imageIOError(this, uri, ioe, getExternalDocument().getLocator());
} catch (ImageException ie) {
getResourceEventProducer().imageError(this, uri, ie, getExternalDocument().getLocator());
}
}
private void makePageForImage(ImageInfo info, ImageLayout layout) {
this.imageLayout = layout;
curPage = makeNewPage(false, false);
fillPage(info.getOriginalURI());
finishPage();
}
/** {@inheritDoc} */
@Override
public Image loadImage(final ImageInfo info, final Map hints, final ImageSessionContext session)
throws ImageException, IOException {
if (!MimeConstants.MIME_JPEG.equals(info.getMimeType())) {
throw new IllegalArgumentException(
"ImageInfo must be from a image with MIME type: " + MimeConstants.MIME_JPEG);
}
ColorSpace colorSpace = null;
boolean appeFound = false;
int sofType = 0;
ByteArrayOutputStream iccStream = null;
final Source src = session.needSource(info.getOriginalURI());
final ImageInputStream in = ImageUtil.needImageInputStream(src);
final JPEGFile jpeg = new JPEGFile(in);
in.mark();
try {
outer:
while (true) {
int reclen;
final int segID = jpeg.readMarkerSegment();
if (log.isTraceEnabled()) {
log.trace("Seg Marker: " + Integer.toHexString(segID));
}
switch (segID) {
case EOI:
log.trace("EOI found. Stopping.");
break outer;
case SOS:
log.trace("SOS found. Stopping early."); // TODO Not sure if
// this is safe
break outer;
case SOI:
case NULL:
break;
case SOF0: // baseline
case SOF1: // extended sequential DCT
case SOF2: // progressive (since PDF 1.3)
case SOFA: // progressive (since PDF 1.3)
sofType = segID;
if (log.isTraceEnabled()) {
log.trace("SOF: " + Integer.toHexString(sofType));
}
in.mark();
try {
reclen = jpeg.readSegmentLength();
in.skipBytes(1); // data precision
in.skipBytes(2); // height
in.skipBytes(2); // width
final int numComponents = in.readUnsignedByte();
if (numComponents == 1) {
colorSpace = ColorSpace.getInstance(ColorSpace.CS_GRAY);
} else if (numComponents == 3) {
colorSpace = ColorSpace.getInstance(ColorSpace.CS_LINEAR_RGB);
} else if (numComponents == 4) {
colorSpace = ColorSpaces.getDeviceCMYKColorSpace();
} else {
throw new ImageException(
"Unsupported ColorSpace for image "
+ info
+ ". The number of components supported are 1, 3 and 4.");
}
} finally {
in.reset();
}
in.skipBytes(reclen);
break;
case APP2: // ICC (see ICC1V42.pdf)
in.mark();
try {
reclen = jpeg.readSegmentLength();
// Check for ICC profile
final byte[] iccString = new byte[11];
in.readFully(iccString);
in.skipBytes(1); // string terminator (null byte)
if ("ICC_PROFILE".equals(new String(iccString, "US-ASCII"))) {
in.skipBytes(2); // chunk sequence number and total
// number of chunks
final int payloadSize = reclen - 2 - 12 - 2;
if (ignoreColorProfile(hints)) {
log.debug("Ignoring ICC profile data in JPEG");
in.skipBytes(payloadSize);
} else {
final byte[] buf = new byte[payloadSize];
in.readFully(buf);
if (iccStream == null) {
if (log.isDebugEnabled()) {
log.debug("JPEG has an ICC profile");
final DataInputStream din =
new DataInputStream(new ByteArrayInputStream(buf));
log.debug("Declared ICC profile size: " + din.readInt());
}
// ICC profiles can be split into several
// chunks
// so collect in a byte array output stream
iccStream = new ByteArrayOutputStream();
}
iccStream.write(buf);
}
}
} finally {
in.reset();
}
in.skipBytes(reclen);
break;
case APPE: // Adobe-specific (see 5116.DCT_Filter.pdf)
in.mark();
try {
reclen = jpeg.readSegmentLength();
// Check for Adobe header
final byte[] adobeHeader = new byte[5];
in.readFully(adobeHeader);
if ("Adobe".equals(new String(adobeHeader, "US-ASCII"))) {
// The reason for reading the APPE marker is that
// Adobe Photoshop
// generates CMYK JPEGs with inverted values. The
// correct thing
// to do would be to interpret the values in the
// marker, but for now
// only assume that if APPE marker is present and
// colorspace is CMYK,
// the image is inverted.
appeFound = true;
}
} finally {
in.reset();
}
in.skipBytes(reclen);
break;
default:
jpeg.skipCurrentMarkerSegment();
}
}
} finally {
in.reset();
}
final ICC_Profile iccProfile = buildICCProfile(info, colorSpace, iccStream);
if (iccProfile == null && colorSpace == null) {
throw new ImageException("ColorSpace could not be identified for JPEG image " + info);
}
boolean invertImage = false;
if (appeFound && colorSpace.getType() == ColorSpace.TYPE_CMYK) {
if (log.isDebugEnabled()) {
log.debug(
"JPEG has an Adobe APPE marker. Note: CMYK Image will be inverted. ("
+ info.getOriginalURI()
+ ")");
}
invertImage = true;
}
final ImageRawJPEG rawImage =
new ImageRawJPEG(
info, ImageUtil.needInputStream(src), sofType, colorSpace, iccProfile, invertImage);
return rawImage;
}
/**
* @param bytes
* @param imageFile
* @return
* @throws Exception
* @throws FileNotFoundException
* @throws IOException
* @throws InterruptedException
*/
private static ImageInfo ensureFormatIsSupported(
URL url, File imageFile, byte[] bytes, boolean isLoad) throws Docx4JException {
FileOutputStream fos;
// ImageInfo can also tell us what sort of image it is
ImageInfo info = null;
boolean imagePreloaderFound = true;
try {
try {
info = getImageInfo(url);
// Debug ... note that these figures
// aren't necessarily accurate for EPS
displayImageInfo(info);
} catch (org.apache.xmlgraphics.image.loader.ImageException e) {
// Assume: The file format is not supported. No ImagePreloader found for /tmp/img55623.img
// There is no preloader for eg PDFs.
// (To use an image natively, we do need a preloader)
imagePreloaderFound = false;
log.warn(e.getMessage());
}
if (imagePreloaderFound
&& (info.getMimeType().equals(ContentTypes.IMAGE_TIFF)
|| info.getMimeType().equals(ContentTypes.IMAGE_EMF2) // ImageInfo
|| info.getMimeType().equals(ContentTypes.IMAGE_WMF)
|| info.getMimeType().equals(ContentTypes.IMAGE_PNG)
|| info.getMimeType().equals(ContentTypes.IMAGE_JPEG)
|| info.getMimeType().equals(ContentTypes.IMAGE_GIF)
// || info.getMimeType().equals(ContentTypes.IMAGE_EPS)
|| info.getMimeType().equals(ContentTypes.IMAGE_BMP))) {
// TODO: add other supported formats
// If its a format Word supports natively,
// do nothing here
log.debug(".. supported natively by Word");
} else if (imageFile != null && bytes != null) {
// otherwise (eg if its an EPS or PDF), try to convert it
// Although the Word UI suggests you can embed an EPS
// directly, Word actually converts it to an EMF;
// Word is unable to read a plain EPS image part.
// (TODO: detect failure)
log.debug(".. attempting to convert to PNG");
// If image haven't been load (using function createImagePartFromFilePath), we load it
if (isLoad == false) {
// So first, we create tmpFile
File tmpImageFile = File.createTempFile("img", ".img");
fos = new FileOutputStream(tmpImageFile);
// Now we get the inputStream, which is represented by imageFile in this case
FileInputStream bais = new FileInputStream(imageFile);
// We convert
convertToPNG(bais, fos, density);
// We don't forget to change locFile because the new image file is the converted image
// file!!
imageFile = tmpImageFile;
} // Else image has been load in an array (using function cretaImagePart)
else {
ByteArrayInputStream bais = new ByteArrayInputStream(bytes);
fos = new FileOutputStream(imageFile);
convertToPNG(bais, fos, density);
}
fos.close();
fos = null;
// We need to refresh image info
getImageManager().getCache().clearCache();
info = getImageInfo(new URL("file://" + imageFile.getAbsolutePath()));
// Debug ...
displayImageInfo(info);
} else {
throw new Docx4JException("Unsupported linked image type.");
}
} catch (Exception e) {
throw new Docx4JException("Error checking image format", e);
}
return info;
}
/**
* Create an image part from the provided byte array, attach it to the source part (eg the main
* document part, a header part etc), and return it.
*
* <p>Works for both docx and pptx.
*
* <p>Note: this method creates a temp file (and attempts to delete it). That's because it uses
* org.apache.xmlgraphics
*
* @param opcPackage
* @param sourcePart
* @param bytes
* @return
* @throws Exception
*/
public static BinaryPartAbstractImage createImagePart(
OpcPackage opcPackage, Part sourcePart, byte[] bytes) throws Exception {
// Whatever image type this is, we're going to need
// to know its dimensions.
// For that we use ImageInfo, which can only
// load an image from a URI.
// So first, write the bytes to a temp file
File tmpImageFile = File.createTempFile("img", ".img");
FileOutputStream fos = new FileOutputStream(tmpImageFile);
fos.write(bytes);
fos.close();
log.debug("created tmp file: " + tmpImageFile.getAbsolutePath());
ImageInfo info = ensureFormatIsSupported(tmpImageFile, bytes, true);
// In the absence of an exception, tmpImageFile now contains an image
// Word will accept
ContentTypeManager ctm = opcPackage.getContentTypeManager();
// Ensure the relationships part exists
if (sourcePart.getRelationshipsPart() == null) {
RelationshipsPart.createRelationshipsPartForPart(sourcePart);
}
String proposedRelId = sourcePart.getRelationshipsPart().getNextId();
String ext = info.getMimeType().substring(info.getMimeType().indexOf("/") + 1);
// System.out.println(ext);
BinaryPartAbstractImage imagePart =
(BinaryPartAbstractImage)
ctm.newPartForContentType(
info.getMimeType(),
createImageName(opcPackage, sourcePart, proposedRelId, ext),
null);
log.debug(
"created part "
+ imagePart.getClass().getName()
+ " with name "
+ imagePart.getPartName().toString());
FileInputStream fis = new FileInputStream(tmpImageFile);
imagePart.setBinaryData(fis);
imagePart.rels.add(sourcePart.addTargetPart(imagePart, proposedRelId));
imagePart.setImageInfo(info);
// Delete the tmp file
// As per http://stackoverflow.com/questions/991489/i-cant-delete-a-file-in-java
// the following 3 lines are necessary, at least on Win 7 x64
// Also reported on Win XP, but in my testing, the files were deleting OK anyway.
fos = null;
fis = null;
if (Docx4jProperties.getProperty(
"docx4j.openpackaging.parts.WordprocessingML.BinaryPartAbstractImage.TempFiles.ForceGC",
true)) {
System.gc();
}
if (tmpImageFile.delete()) {
log.debug(".. deleted " + tmpImageFile.getAbsolutePath());
} else {
log.warn("Couldn't delete tmp file " + tmpImageFile.getAbsolutePath());
tmpImageFile.deleteOnExit();
// If that doesn't work, see "Clean Up Your Mess: Managing Temp Files in Java Apps"
// at devx.com
}
return imagePart;
}