pdf-tools icon indicating copy to clipboard operation
pdf-tools copied to clipboard
verified publisher icon vedang

two-column pdf: cannot select a single column

Open rdiaz02 opened this issue 2 years ago • 6 comments

Describe the bug In some two-column pdfs, selection extends over the two columns. In other words, we cannot select a single column.

Steps to Reproduce the behaviour Steps to reproduce the behaviour:

  1. Open the provided file (providing two pages, as an example; the full reference is from here: https://onlinelibrary.wiley.com/doi/full/10.1111/jeb.14202 "Environmental variance in male mating success modulates the positive versus negative impacts of sexual selection on genetic load" by Tschol, Reid, Bocedi, 2023, J. Evol. Biol)
  2. In the first page provided, try to select text to highlight
  3. Both columns are selected.

What is the expected behaviour? That we can select text in either one of the columns, not extending over both columns. (For example, with Okular I can properly select just the text of the column I want).

Desktop Please complete the following information:

  • OS: Debian GNU Linux
  • Emacs Version: GNU Emacs 29.0.92 (build 1, x86_64-pc-linux-gnu, GTK+ Version 3.24.37, cairo version 1.16.0) of 2023-07-06
  • Poppler Version: libpoppler-dev: 22.12.0-2+b1

Your pdf-tools install Please complete the following information:

  • pdf-tools Version: Status: Installed in ‘pdf-tools-20230611.239/’ (unsigned). Delete Version: 20230611.239 Commit: c69e7656a4678fe25afbd29f3503dd19ee7f9896
  • pdf-tools customization / configuration that you use
(use-package pdf-tools
  :ensure t
  :commands (pdf-sync-forward-search
	     try-pdf-outline-on-left
	     pdf-view-current-page
	     pdf-outline-on-left
	     )
  :init
  (pdf-loader-install)
  :mode (("\\.pdf\\'" . pdf-view-mode))
  :hook (TeX-after-compilation-finished . TeX-revert-document-buffer)
  :config
  (progn 

    (setq-default pdf-view-display-size 'fit-page) ; fit page by default

    (setq pdf-annot-default-text-annotation-properties
	  (quote
	   ((icon . "Note")
	    (color . "#ff0000")
	    (label . "X") ;; for reviews
	    (popup-is-open))))
    (add-hook 'pdf-annot-list-mode-hook
	      (lambda ()
		(pdf-annot-list-follow-minor-mode)) ;; C-c C-f for enabling/disabling
	      )
    (bind-key "M-w" 'pdf-view-kill-ring-save pdf-view-mode-map)
    (define-key pdf-view-mode-map (kbd "h")
		'pdf-annot-add-highlight-markup-annotation)
    (define-key pdf-view-mode-map (kbd "t")
		'pdf-annot-add-text-annotation)
    (define-key pdf-view-mode-map (kbd "D")
		'pdf-annot-delete)
    (add-hook 'pdf-view-mode-hook
              (lambda()
		(setq-local split-width-threshold  1)
		)
	      )
    (defun pdf-outline-on-left ()
      "Split pdf into two windows, display outline on left, 1/3 of original size"
      (interactive)
      (let ((hlineso (window-total-width)))
	(split-window-horizontally (round (* 0.33 hlineso)))
	(display-buffer-same-window (pdf-outline-noselect) nil)
	)
      )
    ;; do not fail
    (defun try-pdf-outline-on-left ()
      (interactive)
      (condition-case nil
	  (pdf-outline-on-left)
	(error nil)
	)
      )
    (define-key pdf-view-mode-map (kbd "O")
		'pdf-outline-on-left)
    )
  )

Additional context

  • If you are reporting a bug, please try and attach an example PDF file where I can reproduce the bug. Attached. ex1.pdf

  • If you can attach screenshots or recordings, that is a great help: attached b1

  • Please try reproducing the bug yourself on Vanilla Emacs before reporting the problem. Done with emacs -Q and then

(require 'package)
(package-initialize)
(use-package pdf-tools)
(pdf-loader-install)

The screenshot is precisely with this minimal configuration.

rdiaz02 avatar Sep 16 '23 17:09 rdiaz02

Just adding this comment, to bump this issue: I continue experiencing these problems, in PDFs that are annotated just fine with Okular.

Current settings:

  • pdf-tools: pdf-tools-20240429.407
  • Emacs 30.0.93 (with both gtk and lucid: GNU Emacs 30.0.93 (build 1, x86_64-pc-linux-gnu, GTK+ Version 3.24.43, cairo version 1.18.2) of 2024-12-23 and GNU Emacs 30.0.93 (build 1, x86_64-pc-linux-gnu, X toolkit, cairo version 1.18.2, Xaw scroll bars) of 2024-12-26)
  • OS: Debian GNU Linux
  • poppler: lipoppler-dev: libpoppler-dev:amd64 24.08.0-2

rdiaz02 avatar Jan 01 '25 12:01 rdiaz02

Just adding this comment, to bump this issue: I continue experiencing these problems, in PDFs that are annotated just fine with Okular.

Current settings:

* pdf-tools: `pdf-tools-20240429.407`

* Emacs 30.0.93 (with both gtk and lucid: `GNU Emacs 30.0.93 (build 1, x86_64-pc-linux-gnu, GTK+ Version 3.24.43, cairo version 1.18.2) of 2024-12-23`  and `GNU Emacs 30.0.93 (build 1, x86_64-pc-linux-gnu, X toolkit, cairo version 1.18.2, Xaw scroll bars) of 2024-12-26`)

* OS: Debian GNU Linux

* poppler: lipoppler-dev: `libpoppler-dev:amd64   24.08.0-2 `

Can you try evince? It uses the same poppler backend as us and if it suffers from the same problem it is probably a poppler-glib bug.

aikrahguzar avatar Jan 01 '25 13:01 aikrahguzar

Yes, both evince and atril (a fork of evince) present this same problem.

For the sake of checking what Okular (that does not show the problem) and evince are using here, this is the output of pmap; so the culprit would be libpoppler-glib ?

pmap $(pidof okular) | grep poppler
   844K r---- libpoppler.so.140.0.0
  1744K r-x-- libpoppler.so.140.0.0
   768K r---- libpoppler.so.140.0.0
   360K r---- libpoppler.so.140.0.0
    24K rw--- libpoppler.so.140.0.0
   188K r---- libpoppler-qt5.so.1.36.0
   272K r-x-- libpoppler-qt5.so.1.36.0
    80K r---- libpoppler-qt5.so.1.36.0
    20K r---- libpoppler-qt5.so.1.36.0
     4K rw--- libpoppler-qt5.so.1.36.0
   112K r---- okularGenerator_poppler.so
   136K r-x-- okularGenerator_poppler.so
    36K r---- okularGenerator_poppler.so
    16K r---- okularGenerator_poppler.so
     4K rw--- okularGenerator_poppler.so

pmap $(pidof evince) | grep poppler
    844K r---- libpoppler.so.140.0.0
   1744K r-x-- libpoppler.so.140.0.0
    768K r---- libpoppler.so.140.0.0
    360K r---- libpoppler.so.140.0.0
     24K rw--- libpoppler.so.140.0.0
    124K r---- libpoppler-glib.so.8.26.0
    224K r-x-- libpoppler-glib.so.8.26.0
    104K r---- libpoppler-glib.so.8.26.0
     24K r---- libpoppler-glib.so.8.26.0
      4K rw--- libpoppler-glib.so.8.26.0

rdiaz02 avatar Jan 01 '25 21:01 rdiaz02

Searching for reports of this problem, I've found https://bugs.launchpad.net/ubuntu/+source/poppler/+bug/594205 (also https://gitlab.freedesktop.org/poppler/poppler/-/issues/409) . In the first link it seems to be marked as solved as the problem would be due to misordered columns. Yet, the file "Example.pdf", mentioned in that bug report, is problematic for Evince but not for Okular. (I haven't been able to locate the two other files mentioned in this thread: global.pdf and bookreview.pdf).

And in https://bugs.launchpad.net/ubuntu/+source/poppler/+bug/33288 file HE-News-Winter-2009.pdf is again problematic for Evince but not Okular.

My understanding, then, is that this is a known issue. So maybe we can close this?

rdiaz02 avatar Jan 01 '25 21:01 rdiaz02

Yes, both evince and atril (a fork of evince) present this same problem.

For the sake of checking what Okular (that does not show the problem) and evince are using here, this is the output of pmap; so the culprit would be libpoppler-glib ?

Yes, as far as I understand, poppler has a bunch of higher-level libraries that integrate with different toolkits. okular uses the qt frontend which is better maintained at the moment than the glib one which pdf-tools and evince use.

Closing the bug is up to you but leaving it open might be better for other people who might bump into this issue.

aikrahguzar avatar Jan 02 '25 05:01 aikrahguzar

Understood, thank you. I'll leave it open then, but of course feel free to close if deemed a better option.

rdiaz02 avatar Jan 02 '25 14:01 rdiaz02