Enhancements for text-selection

[Benjamin Eltzner]

This is rather a feature request: The mode of marking text by dragging a frame is in my view not ideal and it would be preferable if the text were marked following the line format, as is standard in office software and other viewers like evince.

[Adam Reichold (adamreichold)]

Updated the bug description to describe the current state of the program.

[Andreas Buchinger (andreas-buchinger)]

Additional to this I think it would be perfect if we have both options - block selection and real text selection honoring line breaks as usual with editors.

Additionally I do not find the current implemented way very intuitive. All pdf viewers I'm aware of use left mouse button to make a selection and highlight the selected text portion. But copying to clipboard is made afterwards with <CTRL> <C>.
This is very convenient as you see the selected text portions before copying. Currently you select a rectangle and hope the portion you like is within there and you have to check the clipboard afterwards if the clipboard is really filled with the pieces you wanted to have. At least selection should be highlighted.

Moreover pdf viewer usually have a button to start 'text selection mode'. With qpdfview I do not find such button even in 'View tool bar' setting. Although I'm perfectly happy with a keyboard accelerators instead of a button I think most people are used to such 'start text selection' button. Probably worth to add.

And while I'm there - think it would be better if all tool bars/buttons are on by default as most people will not find out how to enable f.i. 'fit to page' button. And it's easier to switch off buttons then adding ones.

Regards, Andi

[Adam Reichold (adamreichold)]

Hello Andreas,

Thanks for your suggestions. Maybe this is a good place to note that we decided to wait with this until the recently proposed support for tagged PDF is merged in Poppler so that the text-selection process is actually backed by structural information and not done using heuristics. This is also the reason why the selected text isn't highlighted before copying, we don't "know" the exact relation between the selected rectangle and the text that is extracted since the heuristics are a one-way street. (There is also is support for extracting all text elements on a page and process them according to one's interface needs, but as I said we'll like to wait until this is actually backed by structural data.)

Concerning the way to activate this (assuming a current program version): Ctrl+C activates text-selection mode which ends with copying the text that is suspected to be displayed in the selected rectangle into the clipboard. So we do have a (text-)selection mode, even though it does work a little bit differently than in other viewers and is also used to select images. But this is something very different from being unintuitive. There also is a menu entry (Edit / Copy to clipboard) and hence one can add a tool bar button in the usual way. This is all described in the help, including the keyword for adding a button to the tool bar.

We also prefer a simple and clean interface as a default. We try to give users a lot of configurability, but we also expect them to take a few moments to adjust the program to their personal preferences.

Best regards, Adam.

[Andreas Buchinger (andreas-buchinger)]

Hi Adam,
thank you very much for your response and for your work on qpdfview. I like qpdfview very much and eventually it will be my preferred pdf reader. Probably it will make it to the default pdf reader for future releases of our platform (eComStation - OS/2) too. There are only a few glitches I found so far with the lastest release 0.4.3 which I tested. One of them is the missing 'text selection mode' button. Reading your comment it seems this is only missing in our (eCS) version as I have 'copyToClipboardMode' in my 'Ansicht-Werkzeugleiste' (toolbar settings?) setting. 'copyToClipboardMode' and 'addAnnotationMode' do not have any effect with our version and I thought this is general problem. But have to talk to Silvan who made our port if there are some missing resources in our version which lead to not showing up these buttons.

Regarding 'simple and clean' and 'take moments to adjust' the program to personal preferences - I for myself do not have a problem with this and I have adjusted it to my needs. Valid arguments. But if you are seeking for 'world domination' :-) :-) :-) maybe you will change your mind on that.

Btw. I can change the size of the settings dialog but after closing and reopening it it starts again with the default size (to small for my liking). Another thing I stumbled over is the 'Vorgaben' (Default ?) button in the settings dialog - this resets a lot of settings (all?) not only that one of the current pane (notebook page) which I had expected. My 2 cents.

Greetings, Andi

[Adam Reichold (adamreichold)]

Hello Andreas,

Since this is beginning to go off the topic, I would invite you to discuss general development topics on the mailing list at [1].

But concerning your questions: Did you try to add the 'copyToClipboardMode' action to the edit or view tool bar? The reason I ask is, that it is only available in the edit tool bar. Silvan appears to be quite thorough, so I would be surprised if he had missed this.

Concerning world domination: I don't think it's on the agenda. Thanks to the hard work of the people maintaining libraries like Poppler, IMHO Free Software does not really need to centralize so much.

The settings dialog not remembering its size is probably best described as a missing feature. If you could open a separate wishlist bug, I'll implement it as soon as I find the time.

The observed behavior of the "Defaults" button is as you describe and for the moment it is intended that way. Again I'd say a separate wishlist bug to add a "Defaults on tab" button to the settings dialog would be a way forward.

Best regards, Adam.

[1] https://launchpad.net/~qpdfview

[Adam Reichold (adamreichold)]

Just to tidy thing up: I added the implementation and the wishlist bugs for both the settings dialog remembering its size and the restricted defaults button. Feel to comment on the implementation in those bug reports.

[Andreas Buchinger (andreas-buchinger)]

Thank you Adam. You are correct. I added it to the wrong tool bar. Sorry I got confused by the English help together with my German dialogs. But now you know what stupid mistakes users are able to make ;-)

[Benjamin Eltzner (b-eltzner)]

I would like to add the Poppler meta-bug concerning tagged PDF support for convenience:

https://bugs.freedesktop.org/show_bug.cgi?id=64813

(There has been some development very recently, so I'm keeping my hopes up.)

[Martin Spacek (mspacek)]

I too would love to see this! The somewhat funny text selection is probably the biggest hangup a new user (like me) might have. Otherwise, I'm very impressed. qpdfview already become my evince replacement.

Doesn't seem like there's much holding up that poppler meta-bug...

[Yanpas (yanpaso)]

I was trying to understand how to highlight text for 10 minutes! There must be two instruments Selector (for high-lighting) and Hand (glove) for dragging. Very unhandy

[[e-mail] (az-k)]

This ticket seems to be open for years with little to no progress. Is there any possibility this feature is high on the priority list?

[Adam Reichold (adamreichold)]

Hello,

Am 16.10.2015 um 23:00 schrieb <email address hidden>:
> This ticket seems to be open for years with little to no progress. Is
> there any possibility this feature is high on the priority list?

The issue is certainly of a high priority but that does not change the
fact that it also is a complicated one. I also have decided - after
trying it out - *not* to implement text layout detection heuristics
within qpdfview. Therefore I will need to work upstream with Poppler
first to port the structured elements work to the Qt front-end.

Regards, Adam.

[MichaelB (mrbou)]

Hi Adam,

Would be nice to have this feature to select & Copy text in the pdf. We need all days to select and copy text from pdf file to another tool (like CMS, Text editor). But at this time we cannot use qpdfview for that, we have to use evince where we can select and copy text. Note: We use qpdfview in a company environment like main pdf reader software :)

Thanks for your work

[Antoni Marcinek (antoni-marcinek)]

Hi Adam,

I just moved from acroread to qpdfview after extensive search for a good PDF
reader for Linux. Most of popular choices (i.e. Okular, Evince) were not
satisfactory either due to dependencies and size (Okular) or to lack of
features (Evince - e.g. tabs, configurability, poor rendering). The qpdfview is
just awesome. Unfortunately it does have a problem which enforces me to keep
another viewer in parallel - the text selection. Do you know what is the state
of a development in poppler that keeps you from implementing the feature in
qpdfview?

Regards,
Antoni

[Oliver R. (oliverr)]

Hi Adam,

I just like to repeat Antoni's question.

Thanks & regards,
Oliver

[Adam Reichold (adamreichold)]

Hello Antoni, hello Oliver,

the last time I looked, the Structured Document API was only implemented for the glib frontend which is why an application using the Qt front-end is currently forced to implement heuristics which try to guess the text structure from its geometry, which is not an approach I find viable for a small scale project like qpdfview.

But then again, I have checked for some time due to being blocked on other things. As is usually the case with free software, patches are still welcome...

Best regards, Adam.

[Cedric Eveleigh (cedrico)]

I would really appreciate if this text selection issue got solved.

Thanks,
Cedric

[qhaz (qhaz)]

Just to echo the requests of so many others . . . I too wish the text selection issue could also be implemented.

Thank you to all who contribute to the ongoing development of this great application.

Qhaz

[Amit Keret (amitkeret)]

Hi Adam,

It's been almost two years since the last comment here... I tried looking online for current support for Structured Document API on QT but was unsuccessful.
Any chance this feature will implemented? I second the opinion that this is the only major feature missing from qpdfview

Thanks

[Adam Reichold (adamreichold)]

Hello Amit,

> I tried looking online for current support for Structured Document API on QT but was unsuccessful.

The code is available in Poppler's core but neither me nor anybody else has had any spare time to expose it in the Qt API.

> Any chance this feature will implemented? I second the opinion that this is the only major feature missing from qpdfview

I somewhat agree, but this does not create any free spare time on my end. I am sorry that people are left wanting for this, but we really need more code contributions to Poppler and/or qpdfview for this to happen.

Best regards,
Adam

[Lonnie Lee Best (launchpad-startport)]

Ok, I just discovered how to copy text:

Ctrl+c OR Menu > Edit > Copy to Clipboard puts qpdview into text selection mode.

Then, while dragging the rectangle around the text you want to copy, the document blinks constantly during that drag.

After you finish the click-drag, and release your finger from the mouse-button, a context menu immediately appears, which features the option to "Copy text" to you clipboard.

This isn't the end of the world. It does work. I agree with others that I do like how other PDF viewers often allow you to select text the way everyone expects: by just clicking the left-mouse-button and dragging over the lines you want to select, and then right-click what you've highlighted and click copy (or ctrl-c to copy to clipboard).

However, that feature must be really tricky to do, because a lot of other PDF viewers are not perfect in this regard. Maybe that's the reason why the qpdfview developers thought this method, offered in qpdfview, would be more reliable. I'm still trying to decide. There are probably good reasons why they decided to do it this way.

I will say this. Even if you we stick with this method of copying over the normal way you'd expect, I think this particular feature is worthy of being an icon on the tool bar, and an option in qpdfview's universal context-menu (the menu that pops up anywhere you right-click.

This is a frequent task for me in PDFs, and took me a little bit to figure out the qpdfview way.