[
https://issues.apache.org/jira/browse/PDFBOX-5849?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17864695#comment-17864695
]
Tilman Hausherr commented on PDFBOX-5849:
-----------------------------------------
It's the Devangari "locl" feature :-(
!screenshot-2.png!
However you can switch it off by calling {{boldT.setEnableGsub(false);}} in the
snapshot (not in 3.0.2).
> ArrayIndexOutOfBoundsException in Apache PDFBox 3 in connection with Font
> -------------------------------------------------------------------------
>
> Key: PDFBOX-5849
> URL: https://issues.apache.org/jira/browse/PDFBOX-5849
> Project: PDFBox
> Issue Type: Bug
> Components: FontBox, Rendering
> Affects Versions: 3.0.2 PDFBox
> Reporter: Leonard Wicke
> Assignee: Tilman Hausherr
> Priority: Major
> Labels: fontbox, pdfbox
> Attachments: screenshot-1.png, screenshot-2.png
>
>
> *Affected Versions*
> PDFBox 2.0.30 is not affected - so its likely that also no other version of
> major-release 2 is affected.
> PDFBox 3.0.2 is affected.
> It appears to us that this is a bug that is new with major-release 3.
> *Description*
> We are using Apache PDFBox 3.0.2 in our software and have the following issue.
> We want to write a String using the font FreeSansBold.
> The font is loaded via PDType0Font#load from a TTF-file.
> If we load the font with embedSubset=true than the following exception occurs:
> {code:java}
> java.lang.ArrayIndexOutOfBoundsException: Index 2941 out of bounds for length
> 2912
> at
> org.apache.fontbox.ttf.TTFSubsetter.addCompoundReferences(TTFSubsetter.java:500)
> at org.apache.fontbox.ttf.TTFSubsetter.getGIDMap(TTFSubsetter.java:147)
> at
> org.apache.pdfbox.pdmodel.font.TrueTypeEmbedder.subset(TrueTypeEmbedder.java:336)
> at org.apache.pdfbox.pdmodel.font.PDType0Font.subset(PDType0Font.java:304)
> at
> org.apache.pdfbox.pdmodel.PDDocument.subsetDesignatedFonts(PDDocument.java:1046)
> at org.apache.pdfbox.pdmodel.PDDocument.save(PDDocument.java:1034)
> at org.apache.pdfbox.pdmodel.PDDocument.save(PDDocument.java:988)
> at org.apache.pdfbox.pdmodel.PDDocument.save(PDDocument.java:944){code}
> The reason is the question-mark-character "?". The character "!" also causes
> an exception.
> Letters like a-zA-Z dont.
> This character is first correctly identified as Glyph-ID 34 but then in
> PDAbstractContentStream#encodeForGsub converted to 2914 by
> GsubWorkerForDevanagari.
> This glyph does not exist for this font and causes the exception later in the
> code when saving the document when subsetting the fonts.
> The exception does not occur when writing the text in the PDPageContentStream.
> If we load the font with embetSubst=false then no exception occurs but the
> character is not visible/skipped in the pdf.
> I have only found old and fixed issues with ArrayIndexOutOfBoundsExceptions
> (https://issues.apache.org/jira/browse/PDFBOX-4946).
> *Code that causes the exception*
> With Apache PDFBox 3 a new functionality during showTextInternal in
> PDPageContentStream was added = encodeForGsub
> This causes the glyphs of the character to be modified - to a glyph that does
> not exist.
> *Code to redproduce*
> you need the font FreeSansBold or another font that causes this problem
> {code:java}
> PDDocument document = new PDDocument();
> File boldF = new File("src/test/resources/fonts", "FreeSansBold.ttf");
> TrueTypeFont boldT = new TTFParser().parse(new
> RandomAccessReadBufferedFile(boldF));
> PDFont bold = PDType0Font.load(document, boldT, true);
> PDPage page = new PDPage(PDRectangle.A4);
> PDPageContentStream contentStream = new PDPageContentStream(document, page,
> PDPageContentStream.AppendMode.APPEND, true, true);
> contentStream.setFont(bold, 11);
> contentStream.beginText();
> contentStream.newLineAtOffset(50, 50);
> contentStream.showText("?");
> contentStream.endText();
> contentStream.close();
> document.addPage(page);
> document.save(new File("Test.pdf"));{code}
> *Questions*
> Is this a bug in Apache PDFBox / FontBox ?
> It appears that the code in question is only executed for Fonts of classÂ
> PDType0Font - is it possible to load a font using another class to avoid this
> bug ?
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org
For additional commands, e-mail: dev-h...@pdfbox.apache.org
