Thank you, this is really disturbing :-(
I'll investigate that.
Tilman
Am 07.09.2018 um 15:39 schrieb Daniel Wildschut:
Hello, we use PDFBox to fill in PDF Forms and stumbled on a potential
bug while sanitizing the input.
We call PDFont.encode to check beforehand if a given character can be
inserted using the given font.
However we noticed that the results of the method call can change
depending on what other strings have been checked before.
Apparently PDType1Font stores previous results in a codeToBytesMap,
which then causes the unexpected behavior.
I'd say that the key used in "codeToBytesMap.put(code, bytes);" is
wrong; you probably want to use the method parameter "unicode" instead.
I tested 2.0.11, the current 2.0.x branch and the 3.0.x branch and was
able to reproduce the problem with all of them.
Code to reproduce:
public class PDFBoxEncodeTest
{
public static void main( final String[] args )
{
final PDType1Font font = PDType1Font.HELVETICA_BOLD;
tryEncode(font, "\u0080");
tryEncode(font, "€");
tryEncode(font, "\u0080");
}
private static void tryEncode(final PDFont font, final String str) {
try {
font.encode(str);
System.out.println("Character " + str.codePointAt(0) + "
can be encoded in Font " + font);
} catch (final IOException | IllegalArgumentException e) {
System.out.println("Character " + str.codePointAt(0) + "
cannot be encoded in Font " + font + ": " + e.getMessage());
}
}
}
Expected output:
Character 128 cannot be encoded in Font PDType1Font Helvetica-Bold:
U+0080 ('.notdef') is not available in this font Helvetica-Bold
encoding: WinAnsiEncoding
Character 8364 can be encoded in Font PDType1Font Helvetica-Bold
Character 128 cannot be encoded in Font PDType1Font Helvetica-Bold:
U+0080 ('.notdef') is not available in this font Helvetica-Bold
encoding: WinAnsiEncoding
Actual output:
Character 128 cannot be encoded in Font PDType1Font Helvetica-Bold:
U+0080 ('.notdef') is not available in this font Helvetica-Bold
encoding: WinAnsiEncoding
Character 8364 can be encoded in Font PDType1Font Helvetica-Bold
Character 128 can be encoded in Font PDType1Font Helvetica-Bold
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscr...@pdfbox.apache.org
For additional commands, e-mail: users-h...@pdfbox.apache.org