Graphics2D Bridge for Apache PDFBox
Using this library you can use any Graphics2D API based SVG / graph / chart library to embed those graphics as vector drawing in a PDF. In combination with PDFBox PDFRenderer/PageDrawer you can also "rerender" PDF pages and change certain aspects. E.g. change the color mapping and perform an overfill. Now also it's possible to setup masking of all draw()/fill() calls. Think of this like an additional clipping, just that this supports bitmap images and complete complex drawings (XForm's) as alpha masks. See here how that works.
The following features are supported:
- Drawing any shape using
draw...()
andfill...()
methods from Graphics2D. - Drawing images. The default is to always lossless compress them. You could plugin your
own
Image
->PDImageXObject
conversion if you want to encode the images as jpeg. - All
BasicStroke
attributes. Paint
:Color
. You can specify your own color mapping implementation to special map the (RGB) colors toPDColor
. Beside using CMYK colors you can also use spot colors.GradientPaint
,LinearGradientPaint
andRadialGradientPaint
. There are some restrictions:GradientPaint
always generates acyclic gradients.
TexturePaint
.
- Drawing text. By default, all text is drawn as vector shapes, so no fonts are embedded. RTL languages are supported. It's possible to use fonts, but this loses some features (especially RTL support) and you must provide the TTF files of the fonts if the default PDF fonts are not enough.
The following features are not supported (yet):
(Alpha-)Composite
with a rule different thenAlphaComposite.SRC_OVER
.copyArea()
. This is not possible to implement.hit()
. Why would you want to use that?setXORMode()
. Their is no blend mode in PDF which would allow to emulate this, so this is not possible to be implemeted.
This library is available through Maven:
<dependency>
<groupId>de.rototor.pdfbox</groupId>
<artifactId>graphics2d</artifactId>
<version>0.41</version>
</dependency>
This library targets Java 1.6 and should work with Java 1.6. But at the moment it is only tested with Java 8, Java 11 and Java 17.
public class PDFGraphics2DSample
{
public static main(String[] argv)
{
PDDocument document = new PDDocument();
PDPage page = new PDPage(PDRectangle.A4);
document.addPage(page);
/*
* Creates the Graphics and sets a size in pixel. This size is used for the BBox of the XForm.
* So everything drawn outside (0x0)-(width,height) will be clipped.
*/
PdfBoxGraphics2D pdfBoxGraphics2D = new PdfBoxGraphics2D(document, 400, 400);
/*
* Now do your drawing. By default all texts are rendered as vector shapes
*/
/* ... */
/*
* Dispose when finished
*/
pdfBoxGraphics2D.dispose();
/*
* After dispose() of the graphics object we can get the XForm.
*/
PDFormXObject xform = pdfBoxGraphics2D.getXFormObject();
/*
* Build a matrix to place the form
*/
Matrix matrix = new Matrix();
/*
* Note: As PDF coordinates start at the bottom left corner, we move up from there.
*/
matrix.translate(0, 20);
PDPageContentStream contentStream = new PDPageContentStream(document, page);
contentStream.transform(matrix);
/*
* Now finally draw the form. As we not do any scaling, the form drawn has a size of 5,5 x 5,5 inches,
* because PDF uses 72 DPI for its lengths by default. If you want to scale, skew or rotate the form you can
* of course do this. And you can also draw the form more then once. Think of the XForm as a stamper.
*/
contentStream.drawForm(xform);
contentStream.close();
document.save(new File("mysample.pdf"));
document.close();
}
}
See also manual drawing and drawing SVGs. The testdrivers are only smoke tests, i.e. they don't explicit test the result, they just run and test if the their are crashes. You have to manually compare the PDF result of the testdriver with the also generated PNG compare image.
When rendering a text in a PDF file you can choose two methods:
- Render the text using a font as text.
- Render the text using TextLayout as vector graphics.
Rendering a text using a font is the normal and preferred way to display a text:
- The text can be copied and is searchable.
- Usually it takes less space then when using vector shapes.
- When printing in PrePress (Digital / Offset Print) the RIP usually handles text special to ensure the best possible reading experience. E.g. RGB Black is usually mapped to a black with some cyan. This gives a " deeper" black, especially if you have a large black area. But if you use a RGB black to render text it is usually mapped to pure black to avoid any printing registration mismatches, which would be very bad for reading the text.
- Note: When rendering a text using a font you should always embed the needed subset of the font into the PDF. Otherwise not every (=most) PDF viewers will be able to display the text correctly, if they don't have the font or have a different version of the font, which can happen across different OS and OS versions.
- Note: Not all PDF viewer can handle all fonts correctly. E.g. PDFBox 1.8 was not able to handle fonts right. But nowadays all PDF viewers should be able to handle fonts fine.
- Note:
TextAttribute.LIGATURES
is currently not supported. - Note:
TextAttribute.BACKGROUND
is currently not supported. - Note: There is no Bidi support at the moment. See the problems PDFBox has with rendering RTL languages at the moment.
On the other site rendering a text using vector shapes has the following properties:
- The text is always displayed the same. They will be no differences between the PDF viewers.
- The text is not searchable and can not be copied.
- Note: Vector shapes take more space than a embedded font.
- Note: You may want to manually alter the color mapping to e.g. ensure a black text is printed using pure CMYK black. If you do not plan to print the PDF in offset or digital print you can ignore that. This will make no difference for your normal desktop printer.
- Note: When using Apache Batik to draw SVGs the text will always be drawn as vector shape. Batik always converts texts to vector shapes first and then applies the transforms on it (if there are any). So PdfBoxGraphics2D never even gets a chance to draw the text using a PDF font. In theory, this could be solved by installing an appropriate text painter on the Batik bridge context. But no one has created such a text painter yet.
If you want to get a 1:1 mapping of your Graphics2D
drawing in the PDF you should use the
vector mode. If you want
to have the text searchable and only use LTR languanges (i.e. latin-based)
you may try the text mode. For this mode to work you need the font files (.ttf / .ttc) of the fonts
you want to use and
must register it with this library. Using the normal Java font API it is not possible to access the
underlying font
file. So a manual mapping of Font to PDFont is needed.
The font mapping is done using the PdfBoxGraphics2DFontTextDrawer
class. There you register
the fonts you have. By
default the mapping tries to only use fonts when all features used by the drawn text are supported.
If your text uses a
features which is not supported (e.g. RTL text) then it falls back to using vectorized text.
If you always want to force the use of fonts you can use the
class PdfBoxGraphics2DFontTextForcedDrawer
. But this
is unsafe and not recommend, because if some text can not be rendered using the given fonts it will
not be drawn at all
(e.g. if a font misses a needed glyph).
If you want to use the default PDF fonts as much as possible to have no embedded fonts you can use
the class
PdfBoxGraphics2DFontTextDrawerDefaultFonts
. This class will always use a default PDF font, but
you can also
register additional fonts.
public class PDFGraphics2DSample
{
public static main(String[] argv)
{
/*
* Document creation and init as in the example above
*/
// ...
/*
* Register your fonts
*/
PdfBoxGraphics2DFontTextDrawer fontTextDrawer = new PdfBoxGraphics2DFontTextDrawer();
try
{
/*
* Register the font using a file
*/
fontTextDrawer.registerFont(new File("..path..to../DejaVuSerifCondensed.ttf"));
/*
* Or register the font using a stream
*/
fontTextDrawer.registerFont(
PDFGraphics2DSample.class.getResourceAsStream("DejaVuSerifCondensed.ttf"));
/*
* You already have a PDFont in the document? Then make it known to the library.
*/
fontTextDrawer.registerFont("My Custom Font", pdMyCustomFont);
/*
* Create the graphics
*/
PdfBoxGraphics2D pdfBoxGraphics2D = new PdfBoxGraphics2D(document, 400, 400);
/*
* Set the fontTextDrawer on the Graphics2D. Note:
* You can and should reuse the PdfBoxGraphics2DFontTextDrawer
* within the same PDDocument if you use multiple PdfBoxGraphics2D.
*/
pdfBoxGraphics2D.setFontTextDrawer(fontTextDrawer);
/* Do you're drawing */
/*
* Dispose when finished
*/
pdfBoxGraphics2D.dispose();
/*
* Use the result as above
*/
// ...
}
finally
{
/*
* If you register a font using a stream then a tempfile
* will be created in the background.
* Close the PdfBoxGraphics2DFontTextDrawer to free any
* tempfiles created for the fonts.
*/
fontTextDrawer.close();
}
}
}
You can also complete customize the font mapping if you derive
from PdfBoxGraphics2DFontTextDrawer
:
class MyPdfBoxGraphics2DFontTextDrawer extends PdfBoxGraphics2DFontTextDrawer
{
@Override
protected PDFont mapFont(Font font, IFontTextDrawerEnv env)
throws IOException, FontFormatException
{
// Using the font, especially the font.getFontName() or font.getFamily() to determine which
// font to use... return null if the font can not be mapped. You can also call registerFont() here.
// Default lookup in the registered fonts
return super.mapFont(font, env);
}
}
This allows you to load the fonts on demand.
By default the content stream data is compressed using the zlib default level 6. If you want to get the maximum compression out of PDFBox you should set a system property before generating your PDF:
System.setProperty(Filter.SYSPROP_DEFLATELEVEL,"9");
If you want to create complex PDF reports with text and graphs mixed it is recommend to not use PDFBox and this library directly, as both are very low level. Instead you should use OpenHtmlToPdf. OpenHtmlToPdf allows you to build your reports using HTML ( which you can generate with any template engine you like, e.g. Apache FreeMarker) and place custom graphs (which are draw using Graphics2D using this library) with <object> HTML tags.
Version 0.42 (not yet released):
- Upgraded PDFBox to 2.0.27
Version 0.41:
- #45 Copy & paste error in drawImage() call forwarding. sy1 should be passed for sy1, not sy2... Thanks @fransbouwmans for pointing this out. This affected one specifc drawImage() overload.
Version 0.40:
- Messed up the access permissions for
PdfBoxGraphics2DPaintApplier.PaintApplierState::setupLuminosityMasking
. They are now accessible.
Version 0.39:
- Extended the
PdfBoxGraphics2DPaintApplier
to allow overridingPdfBoxGraphics2DPaintApplier::applyPaint(Paint, PdfBoxGraphics2DPaintApplier.PaintApplierState)
. - Made a
PdfBoxGraphics2DPaintApplier.PaintApplierState::setupLuminosityMasking
, which allows to setup a masking for the next fill / draw operations. You can mask all fill / draw operations using a bitmap or XForm based mask. - Initial basic support for opacity in gradients. At the moment this does not correctly work for every case. More work is needed here. Thanks @iocoker for bringing this up. (#37)
Version 0.38:
- Fix for line width of stroks in the case there is a rotation transform on the Graphics2D. Thanks @jonmccracken-wf for reporting and fixing this bug (#36)
- Upgrade to PDFBox 2.0.26
Version 0.37:
- Make PdfBoxGraphics2DFontTextDrawer.getFontMetrics::stringWidth() behave like the JDK. Thanks @Lorgod (#35)
Version 0.36:
- API breakage: Changed the
IPdfBoxGraphics2DColorMapper::mapColor()
signatur to get a IColorMapperEnv instead of a PDPageContentStream. The env provides access to thePDPageContentStream
and thePDResources
- It is now possible to get the associated PDResources by using
PdfBoxGraphics2D::getResources()
.
Version 0.35:
- Temporary workaround for PDFBOX-5361. PDFBOX currently sets the OverprintMode (/OPM) in the Extended Graphics State as float - which is a spec violation.
- Fixed some dependency bugs in the extended tests, thanks @pgrt (#33, #34).
Version 0.34:
- Per default specify the rendering
hint
RenderingHints.KEY_FRACTIONALMETRICS=RenderingHints.VALUE_FRACTIONALMETRICS_ON
. We always render into a vector space, so we don't have to round to ints while drawing text. Thanks to @paatero for bringing this up (#32). - Upgrade to PDFBox 2.0.25
- It is not allowed to make state changes after defining a path and before clipping / drawing, at least acording to the PDF specification PDFBOX-5322. Fixed the ordering here.
Version 0.33:
- Don't crash when drawString() is called with an empty string #31. Thanks @Kischloren for the report.
Version 0.32:
- It is now possible to draw within a marked content sequence. PdfBoxGraphics2D got a new
drawInMarkedContentSequence() method for this. This is usefull if you want to mark some parts of a drawing so that you can later do some special processing on it. Or if you simply want to provide
accesibility information for the content you draw. See also e.g. https://www.w3.org/TR/WCAG20-TECHS/PDF21.html or the section 14.8.4.3.3 in the PDF specification. - Upgrade to PDFBox 2.0.24
Version 0.31:
- Support for colors with overprint.
- New PdfBoxGraphics2DColor class to allow using any kind of color. E.g. PDSeperation based colors.
Version 0.30:
- Clip invalid miter limit values #29. Thanks to @kiwiwings for reporting this.
- Added a new module for extended-tests. This module will contain tests with 3rdparty library which by themself depend on pdfbox-graphics2d. It also now contains a new class DebugCodeGeneratingGraphics2d (by @kiwiwings) which helps to create isolated testcases.
- Upgrade to PDFBox 2.0.22
Version 0.29:
- Fix a bug where the AlphaComposite alpha value would be mixed with a color alpha value when drawing images. When setting a transparent color this had resulted in a invisible image. Thanks to @kiwiwings for reporting this.
- Initial support for
TextAttribute.UNDERLINE
andTextattribute.STRIKETHROUGH
when using a font to render a text.
Version 0.28:
- Fix handling of AttributedString (off-by-one error). Thanks for @kiwiwings for pointing out the error and providing a fix #27.
- Upgrade to PDFBox 2.0.21
- Respect that default fonts may not allow to be embedded. (PDFBox 2.0.21 now respects the flags within a TTF font, so we also must do this)
- When painting an image with an AlphaComposite the alpha is now respected correctly.
Version 0.27:
- Internal API breakage to implement getFontMetrics().stringWidth() correctly in the case a PDFont is used to draw the text #16. Thanks to @megri for reporting this problem.
- Reverted back to PDFBox 2.0.19 because of rendering issues PDFBOX-4886.
Version 0.26:
- Added a CMYK color mapper , which converts the paint colors to CMYK using an ICC Profile. Thanks to @larrylynn-wf for providing this feature #22.
- Upgrade to PDFBox 2.0.20
- Initial support for Apache PDFBox TilingPaint. Thanks to @p1xel. Currently this is not clean and also not correct in many cases.
Version 0.25:
- Upgrade to PDFBox 2.0.17
- Correctly handle GradientPaint fractions.
- Correctly handle SVG LinearGradientPaint's in ObjectBoundingBox mode #19. Thanks to @larrylynn-wf for the report and the idea how to fix it.
- Internal API breakage to support non quadratic SVG gradients correctly #19.
Version 0.24:
- Upgrade to PDFBox 2.0.16
Version 0.23:
- Correctly handle even odd winding rules when clipping and filling shapes.
Version 0.22:
- Upgrade the PDFBox version to 2.0.15
Version 0.21:
- Provide the current XORMode color in the IPaintEnv. And document that XORMode is not working as it's not possible to emulate. Thanks @gredler for pointing this out #14. But you can do whatever you want with that information in your IPdfBoxGraphics2DPaintAplier subclass.
- Upgrade the PDFBox version to 2.0.14
- Handle PDFBox ShadingPaint's.
Version 0.20:
- Handle null transforms in drawImage() correctly. I.e. dont throw a NullPointerException, just ignore the not existing transform.
- Cache the different environments for the mapper/drawer/applier. This is a minor memory saving.
Version 0.19:
- You can now influence the shape fill/draw operations by setting a custom IPdfBoxGraphics2DDrawControl. This allows to do different things like e.g. draw an overfill for shapes (i.e. make shapes have a additional border). This can be useful if you need to preprocess a PDF for pre-press.
Version 0.18:
- setPaint(null) will cause the following fillXXX() and drawXXX() operations to be ignored. This allows in combination with PDFRenderer/PageDrawer to extract parts of a PDF page. E.g. you can draw only certain seperation colors into the resulting PDF if you filter the paints in PageDrawer.getPaint() and extract a seperation color from a PDF in that way.
- New class PdfBoxGraphics2DCMYKColor() which derives from java.awt.Color to be able to specify a CMYK color when painting.
- The default PdfBoxGraphics2DColorMapper now also supports mapping of "legacy" old iText 2 CMYKColor's.
Version 0.17:
- Upgrade the PDFBox version to 2.0.12
Versoin 0.16:
- Added new method
disposeDanglingChildGraphics()
to cleanup all dangling child graphics. This allows to use this graphics adapter with old legacy code which does not correctly calldispose()
on the graphics it used.
Version 0.15:
- Upgrade the PDFBox version to 2.0.11
Version 0.14:
- Don't write invalid path commands into the stream, as this will break rendering in Acrobat Reader. Thanks @FabioVassallo #12
Version 0.13:
- Ugraded the PDFBox version to 2.0.9
Version 0.12:
- Don't share resources between XForm's, as Acrobat Reader does not like that.
Version 0.11:
- Support Batik SVG PatternPaint. Thanks @vipcxj for pointing this out and providing a testfile .
- Compress embedded image ICC Profile Data
Version 0.10:
- Don't export the same extended graphics state over and over again. Same for shadings. #8
Version 0.9:
- Compress the content stream generated for the XForm.
- When drawing the same image multiple times, it is only encoded once now.
Version 0.8:
- Implemented
PdfBoxGraphics2DFontTextDrawerDefaultFonts
to allow preferring default PDF fonts over vectorized text #5.
Version 0.7:
- Bugfixes on the font based text support. Now also gradients can be used to paint text.
Version 0.6:
- Implemented basic support for using fonts to render texts.
Version 0.5:
- Fixed
getClip()
andclip(Shape)
handling. Both did not correctly handle transforms. This bug was exposed by Batik 1.9 and found by @ketanmpandya. Thanks @ketanmpandya #2, OpenHtmlToPdf #99
Version 0.4:
- Initial support for basic
AlphaComposite
. Thanks @FabioVassallo #1 - When drawing a shape with a zero or negative size don't use
PDShadings
, as they won't work. Thanks @FabioVassallo #1
Version 0.3:
- Fix for a NPE when calling
setClip()
with null. - Upgrade to PDFBox 2.0.5, replacing the usage of
appendRawCommands()
withsetMiterLimit()
.
Version 0.2:
- The paint applier (Mapping of
java.awt.Paint
to PDF) can be customized, so you can map special paints if needed. - Support for
TexturePaint
Licenced using the Apache Licence 2.0.