CentraDoc 6.5 - Summary of Changes since 6.3.3

Table Of Contents

Introduction
Enhancements and Upgrades
Repairs and Enhancements by Ticket #
Some Unticketed Issues
Some Known Issues
References

Introduction

CentraDoc 6.5 is the first system-wide release since 6.3.3. Several versions of 6.4 were delivered to different clients with different development schedules and needs, so many of the features here are already in use and may be familiar.

This document is intentionally brief and not comprehensive. Other useful documents are the pvsdkini.txt present in the src/pvsdk directory, and the riptest.txt in the src/ut directory.

Most of the issues are listed here by ticket number. If you have any questions about a specific ticket or issue, please contact Liberty Technology Systems. Liberty can provide any additional information or assistance you may need.

Enhancements and Upgrades

The 6.5 RIP is built on a variable channel color model, so it will support RGB, CMYK and DeviceN color models without rebuilding.
The RIP now has very good support for the PDF transparency model. Many problem areas were fixed, and many advanced use cases are now correct.
The RIP now has cdocrip.dll and riptestd.exe projects as an alternative to the Preview SDK.
The Color Manager has extensive enhancements, including:
- Separate profile settings for color modes (Fill, Stroke, Image, Text).
- Separate blocking of PDF ICC input by color mode (Fill, Stroke, Image, Text).
- Color Ink manager for named DeviceN color output.
- Color Separation database for mapping named input colors to output colorants.
Preview SDK and the CDocRIP API have functions to open arbitrary FFs objects
OS/X XCode projects are now provided - just like the dsp and mak files, these are automatically built from the core .bld files.
CDoc 6.5 uses Little cms 1.19 plus a number of security & safety patches.
CDoc 6.5.2 has been updated to use FreeType 2.4.2, plus a few minor patches
CDoc 6.5.2 has been updated to use ZLib 1.2.5

Repairs and Enhancements by Ticket #

#987 gxc->image_explicit4stencil bomb in cleaning up color spaces.
#987 PDF output library, allow for very large type 1 glyph definitions.
#987 GhostScript distiller pathological pattern trap extended to handle stencil (colorless) patterns.
#986 Handle nested soft masks in PDF transparency.
#985 Sometimes PDF transparency requires CMYK rendering (as does overprint). Added -cmyk2rgb option to riptest: render as CMYK but write RGB. Only works for JPEG & TIFF. CDocRIP.c has the implementation details.
#984 DvCharEm Type 3 was blocking colored characters. Pass glyph name down to driver for data collection. Clean up d0 / d1 issue in PS output driver.
#981 Handle Kyocera PRESCRIBE directives in PostScript files. Detect and ignore ridiculous BoundingBox.
#979 Allow PDF images to have invalid mask entries, without error.
#973 Enhancements to the page count & statistics driver module. Added to shipping kit.
#972 Workaround for PostScript invalidrestore caused by findcolorrendering failure.
#971 Image emulation made assumptions about color space depth causing coloring issues in the RIP.
#970 Allow for Resource dictionary XRef Range error accepted by Acrobat.
#967 DvCharEm ligatures converted to character sequences are converted back to the Unicode ligature character.
#966 Several PreviewSDK enhancements * LS_OpenFFs API call * LS_OpenPSBuffer better support for PDF * Support for GCO class inheritance across DLL boundaries (IMPORT_PARENT) * Additional symbols exported from the PvSDK dll * pvsdk64.dll and pvsdk64.lib * pvsdkl.h dynamic link stubs automatically generated from pvsdkl.d
#965 Automatically turn on FreeTypeHinting for a set of "Tricky" Chinese TrueType fonts that will not render correctly with hinting off.
#963 A minor syntax change in the RIP for gcc 3.2 C++ compatiblity.
#962 A -text file option has been added to the riptest program, as an example of unicode text extraction via the CDocRIP API.
#961 PostScript driver will attempt to use meaningful glyph names when capturing fonts. This enables character data recovery later.
#959, #937 Ignore wide (multitarget) ToUnicode mappings for fonts with narrow input encoding. This is a Snow Leopard bug; both ToUnicode mappings are valid, but using both is not. (Other bad ToUnicode mappings have been found with dozens of target characters for one input character.) Also added default external font mappings for standard fonts on Mac OS/X.
#958 Fixed PostScript output driver polarity issue with grabbed CCITT image where BlackIs1 is False.
#957, #955 Ignore unknown layer names in optional content stream directives. Also, form XObjects may now reference resources from the page (old style).
#956, #906 DvImgGrab received wrong size information if automatic image resize triggered. Added ini switches for controlling image resizer at runtime.
#954 Fixed issues with capturing PDF Type 3 fonts with DvCharEm mechanism.
#950 The rgn clipper would choke on a single point path.
#949 riptest required a CMYK ICC profile if using -cmyk color model; it rendered images incorrectly if no CMYK profile was found; the color manager will now use the default RGB profile with internal conversion to CMYK if a CMYK ICC profile is not available.
#948, #931, #870, #789, #741 Improve handling of PDF files with trailing garbage and other broken PDF recovery strategy improvements.
#945 Added Preview SDK APIs for image extraction.
#943 Many updates to the symbol-utf16 mapping Unicode to narrow.
#941 More PDF and PostScript embedded ICC profiles
- Handled additional case of DefaultRGB and DefaultCMYK style PDF colorspaces
- Implemented undocumented PostScript DSC wrapper for ICC profile
#940 Added PDF soft mask caching for some performance improvements of files that reuse the same softmasks repeatedly. This is configurable; see pvsdkini.txt PdfSMaskCacheMaxK and PdfSMaskCacheMaxN REMOVED r9865: false positive cache hits in some cases.
#939 riptest now supports -monotext and -polytext in addition to -aliased flag, to control how text is rendered. Default is antialiased (8 bit) glyphs; -monotext uses single bit glyphs, -polytext uses paths.
#936, #892 Attempts to reduce GC memory leakage for special case PostScript. (The GC build is not the normal build, this is for extreme cases).
#935 #911 and #856 Draw the ".notdef" character (usually a box) instead of ignoring it.
#933 More color related enhancements
- riptest -scan mode reports colorants in use
- SmarterTransform=Y now preserves CMYK 1/1/1/1
- Added ini settings to block color manager PdfBlockICC[MODEL][MODE]=Y where
  - MODEL is one of Gray, RGB, CMYK, and All
  - MODE is one of Fill, Stroke, Image, Text, and All
- Added ini option AggNativeGradients=Y or N to enable/disable RIP axial & gouraud shaders
- Added uni option AggGradientDiffusion=Y or N to help smooth out native RIP axial
- Note that DeviceN may not mix Real separations with colorant DB separations. If any colors in a DeviceN colors are not in the colorant db, then the ones that are will be ignored. This is a function of how DeviceN alternate colors can be specified in input - there may not be a clear relationship between inputs and outputs. The Altona PDF test file has relevant test cases.
#932 Reduced resolution of PDF output coordinates to save space.
#927 The magic Windows EMF DPI resolution for text etc is 96. This is now the default DPI for PreviewSDK, and the EMF header rewrite.
#926 Several PostScript Idiom and Shader issues:
- Convert FreeHand EPS PostScript linear shading idioms to "real" shaders.
- PostScript idioms were broken in #445, repaired.
- Fixed PostScript shfill clipping glitch.
- Shading RGB doesn't produce the same results as Shading CMYK. If the end points are converted, colors in between may not interpolate the same. If source color model is different from the device color model, the fix creates more in between points to get better interpolation.
#924, #920 Various security checks:
- PDF reader checking for ftell errors.
- More careful string and allocation math, checking for overflows and sign errors.
- Added additional checks for run away recursion.
#922 Some color management fixes:
- Skip Color Manager transforms between ICC profiles with the same description.
- SmarterTransform was not plugged in correctly for some common cases.
- Fix SmarterTransform preserve K mode.
#921 Heuristic checks for Snow Leopard PDF bogus ToUnicode font bug.
#918 PostScript DSC BoundingBox parsing error if missing space after ':'.
#917 Trapping additional weird unicode values in override-utf16.
#915 Fixed some problems with PDF Type 3 fonts with Unicode encodings.
#912 Added gxc->pdf_substitute_all to disable encoding checks - forces all embedded fonts to be offered to the driver.
#910 Work around for PDF lying about Symbol font using ANSI encoding.
#909 Adding additional smarts to the Axial rendering function processing to reduce the number of Axial slices. Includes some related changes to PDF output of triangle shaders.
#907 Fixed some incorrect pattern checks from the file fuzzing fixes.
#903 Some riptest color manager setup fixes.
#902, #879 PostScript currentcmykcolor must return the last setcmykcolor, not anything calculated from the color manager, otherwise some PostScript files get confused and disable CMYK processing.
902 Several fixes to RIP PDF soft mask implementation.
#899 Fixed invalid matrix checks in PDF pattern drawing.
#898, #888 Handle PDF Axial shading with matrix offsets (matrix [4] & [5]).
#896 Removed the code from #597 that "squares" images that are "nearly square".
#887 More confusion over "Symbol" font (PSDVFONT) and symbolic font. PSDVFONT symbol == 2 now means "ZapfDingbats" encoding.
#885 PDF excessive pattern trap didn't clean up correctly, causing incorrect output.
#884, #882 Font mapper repairs, fixes for autostyle code.
#881 Added some clipping smarts to Radial shader for pathologically excessive files.
#878 and #855 Handle Undocumented (wrong) PDF ToUnicode syntax.
#875 Fixed some subtle bugs in the 64 bit build of the PostScript engine.
#872, #860 Handle PDF /Rotate 360 and other odd values.
#868 Fixed problem with JasPer (PDF JPX Image Decoder) caused by temporary file name buffer being far too small.
#861 Trap single bit chroma key image and convert to simple stencil mask image.
#859 Various RIP & gradient fixes:
- Flush alpha before beginning any native shader.
- Pick up overprint settings for RIP native axial shader.
- Repair bomb with CMYK colorspace & RIP native axial shader.
- Added new log utilities, useful for tracking ini and color manager settings.
#857 Fix nocurrentpoint error in rgn clipping code for example PDF file.
#853 Handle PostScript TBCP file with extra leading whitespace.
#852 Fixes for Color Mode issues with Separation colors.
#851 Allow PDF ToUnicode to be Identity-H instead of stream.
#850 Fix PostScript issues with proc style files.
#849 Implement Color Separation DB - see riptest.txt for more info.
#847 Fix CCITT decoder bad cleanup case.
#845 DvImgGrab for JPX to sidestep JasPer for client specific handler.

Call FFsJpx_SetGrabbing(TRUE); or DvImgGrab for JPX won't work.
#840 Another fix for LS_TextExtract() (from #757).
#839 Consolidate PostScript special file handling for utilities like psbbox().
#837 Ignore PDF syntax error for names that are too long, truncate and continue.
#835 Implemented RIP native Axial (linear) and Gouraud (triangle) gradients.
#831 Use gxc->emulate_winding_threshold to disable fill rule repair for large polygons.
#830 Added option for riptest -cfg filename for specific ini settings file.
#828 Fixed font mapping issues related to Wide character capture in psdvps2.
#824 Handle EPS header in psisetbuffersource().
#823 More fixes for fuzzed PDF files.
#822 Implement Color Manager object modes (Fill, Stroke, Image, Text).
#819 Preview SDK LS_TextExtract() wasn't handling multiple pages correctly.
#816 Attempt to continue drawing PDF page after syntax error in PDF input.
#814 PDF Output incremental Type 1 fonts.
#812 More fixes for fuzzed PDF files.
#809 More PDF annotation highlight appearance and placement fixes.
#808 More fixes for vertical alignment & placement of CJK-V text.
#807 Fix for PostScript chunk size oddities in embedded Type42 font fragments.
#803 Various fixes for fuzzed PDF files.
#802 Added missing DvFontEm signal in Type 3 font change.
#797 Fix for PDF IccBased Lab profile.
#796 Various font & encoding issues:
- Improved support for PDF Unicode Wingdings font.
- Use AlphaBlend instead of PATMASK for DvGlyph operations to Windows EMF file.
- Added etc/encoding/zapfdingbats-utf16 for conversion from Unicode ZapfDingbats to narrow (when using Monotype Sorts or similar font).
- ... and UnicodeDingbats=Y enhancement to avoid converting ZapfDingbats to narrow (when NOT using Monotype Sorts or similar font, and using comprehensive Unicode font instead).
- Support for synthetic Bold font display on Windows.
- Use FreeType to synthesize Bold and Outline font glyphs (Improves stroked & filled character rendering).
- Smarter Unicode language character set detection.
#793 FreeType patch for font with bad upm setting causing tiny font output.
#792 Fix for incorrect color mode state processing.
#790 Fixed another PDF annotation placement issue.
#788 PSDVFONT symbol field not initialized (and is not generally useful).
#787 Preview SDK pvsdk.dll wasn't calling cleanup function, causing leaks.
#783 Handle unusual font names when generating Type 1 fonts for PDF output.
#781 Fix for Windows image with alpha.
#780 Some fixes for Windows font name processing.
#778 Implement profiler module on Mac OS X.

Also some performance improvments for embedded CMYK -> RGB profile FastCMYK=N
#777 Implement PDF annotation driver signaling via

gxc_>want_DvAnnot, DvBeginAnnot, DvEndAnnot
#776 Handle PostScript Type 2 function with single item instead of array.
#773 Fix for function logic (affecting shader results).
#771 Added Windows EMF header rewrite for predictable DPI settings.
#770 Fixed soft mask save/restore/reset issues.
#769 RGB overprint emulation blend modes: OverprintDarken=Y uses Darken, N uses Multiply; Multiply can be a problem if drawn areas overlap and are redrawn.
#768 Add riptest -page all option.
#767 Install dummy procs for Xerox style PostScript issues.
#766 Improvements for Preview SDK PostScript performance.
#763 Allow EPS file boundaries to be short without comment or error.
#760 Allow PDF image with too many decode elements without comment or error.
#759 Fix Preview SDK LS_PostScript() encoding issues.
#758 Fixes for PDF ToUnicode oddities with Unicode Surrogates.
#757 Fix Preview SDK LS_TextFile() implementation.
#754 Fixes for PDF annotation placement.
#752 Support PostScript DSC %AI5_ArtSize via

int psaiartsize(char * fname, float oWH[2]);
#751 Fixes for vertical alignment & placement of CJK-V text.
#748 Improve detection and handling of disk full condition.
#746 Handle odd case of PDF Encrypt settings (EncryptMetaData false & R < 3)
#745 Handle external TrueType font with missing glyph names.
#739 Fixes for PDF clipping optimization logic.
#736 PDF API enhancement Pdf_GetDcFormat (for checking AI compatibility)
#732 Fixed many issues with the text driver.

Some Unticketed Issues

Added alternative GCO class import mechanism for dlls, using #define IMPORT_ClassName instead of existing declspec mechanism. The riptestd project and prebind4dll.c / prebind2dll.c modules use this.
Added etc/encoding/symbol-utf16 for converting Unicode Symbol encoding to narrow for Windows Symbol font which doesn't support Unicode.
Added etc/encoding/override-utf16 for catching common problematic Unicode characters.

Some Known Issues

Overprint is no longer the default in the RIP. Correct overprinting is device specific - RGB can emulate somewhat, CMYK can overprint correctly if there are no separation colors, otherwise DeviceN printing is required. Refer to the riptest options for more information.
The RIP does not currently support 16 bit pixels components.
CDoc does not support Color Rendering Intents embedded in PDF file graphic states.
ZLib 1.2.5 is available but has not been integrated.
FreeType 2.4.0 is available but has not been integrated.