GNOME Bugzilla – Bug 439070
Changing letters on selection
Last modified: 2011-02-15 17:28:23 UTC
Please describe the problem: When text is selected - it is changed visually. The ligature fl (fl U+FB02 LATIN SMALL LIGATURE FL) is changed to ß (U+00DF LATIN SMALL LETTER SHARP S). Steps to reproduce: 1. Download the following document: http://java.sun.com/docs/books/jvms/second_edition/ClassFileFormat-final-draft.pdf 2. Open page 54 which is numbered 146. 3. Select the word "reflective" on the second row from the top. Actual results: The letter changes Expected results: Just inverse the text Does this happen every time? Yes Other information: Evince 0.8.0 poppler 0.5.4 (cairo). This is what comes with Fedora Core Development (which will have to become soon Fedora Core 7) I am not sure that this is evince's bug. It might be poppler's. If necessary - I will refile it in fd.o. I am attaching a screen cast in gif format with the behavior.
Created attachment 88324 [details] The screencast as an animated gif
Confirm.
A similar problem was reported by a Debian user recently, it's reproducable with Evince 2.22.2 and Poppler 0.8.2: "Sample 1: This PDF: http://www.tim-taylor.com/papers/thesis/thesis.pdf [5 MB] renders fine with Evince. However, when I select text with the mouse, I see gibberish text in the highlighted region, in a different font (some kind of sans serif). The gibberish seems to be the original text where each character's encoding value is 2 less than it should be, e.g. "Copy" shows up as "Anmw", with each character positioned where the original was, so that a lot of characters overlap each other since the font and characters are different. This occurs when selecting text of any font in this document. With most PDFs I get the expected behavior of the same text being rendered, in the same font. Except for ff/fi ligatures, pdftotext converts the text portions of the PDF fine, and in Evince despite the gibberish appearing on screen, the selected text copies to clipboard fine (both PRIMARY upon select and CLIPBOARD upon Ctrl+C). Acrobat Reader 8 works fine when selecting text in this PDF (also except ligatures). Pdfinfo output: Title: main.dvi Creator: dvipsk 5.58f Copyright 1986, 1994 Radical Eye Software Producer: Acrobat Distiller Daemon 3.01 for Solaris 2.3 and later (SPARC) CreationDate: Wed Aug 25 10:36:44 1999 Tagged: no Pages: 317 Encrypted: no Page size: 595 x 842 pts (A4) File size: 5344024 bytes Optimized: no PDF version: 1.2 Sample 2: I also discovered in /usr/share/X11/doc/fonts/fonts.pdf (found in Debian package xorg-docs), that selecting works as expected, except when highlighting Courier-font XML text. The XML tag name and first word after tags (i.e. matching perl regexp /(?<=[<>])\w+?(?=[<>]|$)/) disappears (as if highlighting spaces) but the rest highlights fine; everything copies to clipboard fine. This is a "modern" PDF-1.4 output by pdflatex. This might be unrelated to the issue above; could file a separate bugreport if that's better." The original bug report: http://bugs.debian.org/481887
I'm experiencing similar problem with Evince 2.22.2 (poppler 0.8.7) on Debian system, ru_RU.UTF-8 locale. When selecting text in a PDF file, the selected text gets distorted. When copying the text to other applications (Emacs, GEdit) via X clipboard, the latin characters are copied nicely, but the cyrillics are replaced with umlauts. Problematic file example: http://zelmanov.ptep-online.com/ctan/lshort_russian.pdf I'm ready to provide any additional info. Andrey
*** Bug 465030 has been marked as a duplicate of this bug. ***
The problem is still present in evince 2.26.1, and this is not with some PDFs that I'm experiencing problems but with almost all (in particular with all LaTeX generated PDFs). I can give many others file causing problems. It would be nice if this problem were resolved, this is very annoying. Guillaume
Still present in 2.28.1 on Ubuntu.
Freshly reported, in regards to debugging the new UbuntuSans beta fonts; in the following PDF file (style guidelines), when text containing an 'fi' ligature is selected, the glyph is displayed as a square box instead: http://ubuntuone.com/p/7vH (direct link to a PDF file...)
Invisible text when going 1 line up and down during text selection ... http://tavvva.net/files/Evince-text-selection-bug.mp4 Since the original PDF document is a CV, I'll send it to the patch developer via E-mail only (if needed).
I forgot to mention the viewer version .... it's 2.28.1
Isn't it this bug in poppler: https://bugs.freedesktop.org/show_bug.cgi?id=6923 ?
(In reply to comment #11) > Isn't it this bug in poppler: > https://bugs.freedesktop.org/show_bug.cgi?id=6923 ? Yes, indeed. Thanks for pointing this out, since I was not aware of the minimal pdf Document showing the bug. I've attached a patch in the above bug that hopefully fixes the problem.
Update, in the comments there are several, IMO unrelated bugs, and these are bugs in poppler. The relevants bugs are: Problems in selections due to not correctly changing the font: https://bugs.freedesktop.org/show_bug.cgi?id=6923 Problems in selections and in text for some russian encoded pdfs: https://bugs.freedesktop.org/show_bug.cgi?id=34300 Problems in selections where ligatures are present: https://bugs.freedesktop.org/show_bug.cgi?id=34301 Some glyphs dissapear/reappear when moving the mouse while selecting text: https://bugzilla.gnome.org/show_bug.cgi?id=640466 Please, CC yourself to the corresponding bug you are interested in.