Jeremias Maerki [Mon, 3 Aug 2009 06:43:45 +0000 (06:43 +0000)]
Reverted rev 798521.
Reason: Loss of accuracy. PDF is operating at points rather than millipoints. Especially when it comes to prepress features people will want exact sizes in the target PDFs.
Jeremias Maerki [Sun, 2 Aug 2009 19:41:37 +0000 (19:41 +0000)]
Bugzilla #47311:
Added an initial set of extensions for prepress support (fox:bleed, fox:crop-offset, fox:crop-box and fox:scale). This is currently supported only by PDF and Java2D renderers.
Submitted by: Peter Coppens <pc.subscriptions.at.gmail.com>
Changes to patch by jeremias:
- crop box expands from trim box rather than the bleed box.
- fixed positioning in AWT preview
Additional changes included:
- disabled the border around every page in Java2DRenderer.
- if background painting is enabled for the Java2DRenderer, it uses the bleed box rather than the trim box.
AFPEventProducer was moved months ago but this file not adjusted. That cause an IllegalStateException: Event model doesn't contain the definition for org.apache.fop.afp.AFPEventProducer
Added support for AFP font embedding. Note: this changes the default behaviour. Like with PDF and PS, all fonts are embedded by default unless matched in the "referenced-fonts" section in the configuration.
Added support for embedding external AFP form maps (form defs) using the afp:include-form-map extension.
Fixed a small problem with AFP-related events.
DefaultFOPResourceAccessor got a fallback to the user agent's base URI if no category base URI is specified.
Adrian Cumiskey [Sun, 26 Jul 2009 10:13:37 +0000 (10:13 +0000)]
Fixes GUMP build problems, the automated e-mail was starting to annoy so I scratched an itch :).
* Updated xmlgraphics-commons to trunk version.
* Fixed ImageRendered constructor call in PDFGraphics2D and AFPGraphics2D to be compatible with the new parameter that has been added in xmlgraphics-commons.
Merged clean-up-only changes made in the ChangingIPDHack branch. Those correspond to the following revision numbers:
785291
785632
787733
787769
788371
793645
Some cleanups and attempts at improving code-readability and -extensibility:
* added inner holder class to BreakingAlgorithm to contain everything related to the fitness classes. Improves readability of both the code and the trace messages.
* extracted blocks of code in BreakingAlgorithm.findBreakingPoints() into protected methods. Following the already existing handleBox(), added handleGlueAt() and handlePenaltyAt() to provide extension points to subclasses. Extraction of the code-blocks related to the node-recovery mechanism.
* extracted blocks of code in BreakingAlgorithm.considerLegalBreak() into protected methods: deactivateNode(), activateNode() and forceNode()
* factored the functionality to obtain the index of the first box into KnuthSequence... just seemed to make sense; may be useful elsewhere.
* some (but not all :() javadoc fixups
* minor changes to LineLayoutManager and PageBreakingAlgorithm to take into account above changes
Adrian Cumiskey [Fri, 19 Jun 2009 11:04:13 +0000 (11:04 +0000)]
No code change here, just fixing the indentation of the if (useInclude) block that somehow got messed up by a previous commit/merge (http://svn.apache.org/viewvc/xmlgraphics/fop/trunk/src/java/org/apache/fop/afp/AFPResourceManager.java?revision=746664&view=markup).
AFP Output: Add support for fractional font-sizes, and make sure that, for non-registered sizes, the substitution will always map to the smaller size if there is also a larger one available
ApacheCon Europe 2009 is over, commented out link.
Increased max heap size allocated to Cocoon, otherwise an OutOfMemoryError will occur when generating the compliance PDF (I finally found out why I was having this error and how to fix it).
Changed meaning of '-v' option to 'verbose', which will print FOP's version and proceed. Added a '-version' option to simply print the version then exit, following Java practices.
Change made as per user demand, see thread here:
http://markmail.org/thread/e7juy7wvj7ujkltx
Bugzilla #47031:
PDFGraphics2D.writeClip doesn't generate a clip command anymore when the clip path is empty.
Submitted by: Francois Fernandes <f.fernandes.at.web.de>
Bugzilla #47000:
Added a custom text painter for rendering SVG text using text operators when rendering to PostScript or EPS. Text is no longer painted as shapes, thus creating much smaller files.
Bugzilla 47027: Correction for interpretation of pixel-lengths, as suggested by Jeremias.
+ minor changes to the fotreetest-classes: distinguish between fatal errors due to malformed testcases, and genuine assertion failures.
Some cleanup:
* extraction of the footnote-related loops in getNextKnuthElements() into private methods
to facilitate step-over when debugging (no additional breakpoints needed to skip those loops)
* merging of the private doPhase3() methods, so theoretically, it becomes possible to do
column-balancing, even with a last page condition, and to eliminate a significant amount
of duplicate lines of code
* other minor changes, like renaming stray Hungarians, renaming some variables to make the
code slightly more self-explanatory, removal of unused import, and whatnot...
Jeremias Maerki [Fri, 27 Mar 2009 14:19:00 +0000 (14:19 +0000)]
Improved multi-threading testbed a bit:
- Prints out a report with some statistics at the end.
- Worker threads moved into their own thread group for easier profiling.
Adrian Cumiskey [Wed, 25 Mar 2009 10:32:23 +0000 (10:32 +0000)]
Added some accessor methods to some public member variables. I have left the public member variables exposed for now as changing them to private/protected may possibly break something external to FOP.
Simplified the setting of filters in the AbstractPDFStream hierarchy. Instead of unnecessarily re-defining the setupFilterList method, sub-classes just define a getDefaultFilterName method
Set svn:eol-style property on page-position-last_break-before_bugzilla46489.xml test case. Simplified it by removing superfluous content and references to fonts that are not available by default.
Fixed small indentation issue in ElementListUtils.
Jeremias Maerki [Mon, 23 Mar 2009 07:54:43 +0000 (07:54 +0000)]
Deprecated font base url methods on user agent level in favor of the methods on the FontManager accessible via FopFactory as proposed in:
http://mail-archives.apache.org/mod_mbox/xmlgraphics-fop-dev/200903.mbox/%3C20090310161419.8FA8.60BA733C@jeremias-maerki.ch%3E
Add startsWithForcedBreak() utility method to ElementListUtils.
Avoid generating "in-between" penalties if the content list that will be appended already starts with a forced break.
Adrian Cumiskey [Fri, 20 Mar 2009 15:18:54 +0000 (15:18 +0000)]
A little font configuration cleaning.
* Extracted business logic from PrintRendererConfigurator and created a new FontDetector and FontAdder in org.apache.fop.fonts package.
* Created a new FontInfoConfigurator that factors out a lot of the static stuff in PrintRendererConfigurator.
* PDFDocumentGraphics2DConfigurator now uses a new FontInfoConfigurator() instead of static PrintRendererConfigurator.buildFontListFromConfiguration() call.
* static method updateReferencedFonts() in PrintRendererConfigurator moved to an instance method in FontManager.
Bugzilla 46828: Activation of the possibility to use CachedRenderPagesModel to conserve memory in case of large documents with a lot of cross-references. (area tree will be serialized to disk to avoid keeping it entirely in memory)
Thanks to Dario Laeria for submitting and testing.
Jeremias Maerki [Tue, 10 Mar 2009 16:16:31 +0000 (16:16 +0000)]
Restored ability to specify any URI base URI (URL), not just file URLs. For file URLs and file paths there's still a check whether the directory exists.
Enabled FontBaseBadTestCase.
Jeremias Maerki [Mon, 9 Mar 2009 08:12:45 +0000 (08:12 +0000)]
Restored compatibility with Java VMs that don't support nio Charsets for codepages such as Cp500 (EBCDIC). The simplifications for the fallback case may look wrong but the code basically represents the state before revision 746664 which introduced this problem.