Mmmh, I am trying to find a smaller file as mine is a 16Mb-weight.

waiting for a sample document (< than 10Mb),
you may find here a java thread dump about this issue.

Thread 25274: (state = IN_JAVA)
 - org.icepdf.core.util.content.b.a(byte[], int, int) @bci=2105 (Compiled frame; information may be imprecise)
 - org.icepdf.core.util.content.c.a(byte[], int, int) @bci=3 (Compiled frame)
 - org.icepdf.core.util.content.a.c() @bci=89 (Compiled frame)
 - org.icepdf.core.util.content.a.a() @bci=82 (Compiled frame)
 - org.icepdf.core.util.content.a.a() @bci=45 (Compiled frame)
 - org.icepdf.core.util.content.a.a() @bci=102 (Compiled frame)
 - org.icepdf.core.util.content.NContentParser.parseTextBlocks(byte[][]) @bci=189 (Compiled frame)
 - org.icepdf.core.pobjects.Page.getText() @bci=147, line=1449 (Interpreted frame)
 - org.icepdf.core.pobjects.Document.getPageText(int) @bci=27, line=1120 (Interpreted frame)

FYI, it works using Document#getPageViewText method

Thanks Felix, we'll take a look at it. The code path is slightly different for the two methods. We'll get a fix for this as soon as we get a sample. You can also email product.support@icesoft.com if the file is too large to attach to the case.

Hi Patrick, I droped a test file here, https://dl.dropboxusercontent.com/u/2853503/123456.pdf.

Problem happens on page n°79 (or 78 0-based indexed).
I just tested with your applet and did a search on the whole document
it is still hanging on page n°79 (running for more than 15 min).

Thanks,
Nicolas

The text extraction parser is a cut down version of the full page content parser with the intent on being faster because things like data and clipping parsing are ignored. In this particular file there is an imbedded image on the 79th page that contains a character stream that puts the parser into an infinite loop.

I've touched up the text extraction code to properly parse the inline images to avoid choking on " character in this case or any other operator that might be in the image stream. Inline images are very small and aren't use much anymore so I'm not to worried about any performance lose.

Checked in fix on the trunk and 5.0.1 branch.

Cool you have corrected it !
May you tell me when the 5.0.4 version will be release ? (previously announced on november 1st).