Creating bookmarks From Page Text doesn't work properly in v10

evdbogaard · Post by **evdbogaard** » Wed Jul 19, 2023 2:25 pm

In scanned and ocr-ed documents I always create bookmarks to chapter and paragraph headings using 'Create Bookmarks From Page Text', but in version 10 this doesn't work properly any more.

The first time it finds text with the selected properties (usually a bold or italic font of a certain point size) it creates a bookmark that points to the correct place in the document. But the next time it finds a paragraph heading (or sometimes the third or fourth time) it creates a bookmark that point to somewhere higher than it should (a couple of lines above the paragraph heading it should point to). The text of the bookmark however is correct so I guess it does find the correct place, at least to copy the bookmark text from. But for some reason the created bookmark does not point to the correct place on the page.

In version 9 I never had issues with this, so this must be something that has changed in version 10.
Any ideas?

My current version is 10.0.1 build 371.

Regards,
Ed

Post by **Tracker Supp-Stefan** » Wed Jul 19, 2023 2:52 pm

Hello Ed,

Can we please have samples of a document before and after you add the bookmarks to it?
Do you know the exact build of V9 that you were using before where this worked correctly?

Could it be that the OCR text layer is not slightly larger/smaller - and you just need to increase the font size intervals for the detection a bit?

Kind regards,
Stefan

evdbogaard · Post by **evdbogaard** » Thu Jul 20, 2023 2:48 pm

Hi Stefan,

I have prepared a freshly scanned document, one version ocr-ed by PDFX v10 and one ocr-ed by ABBYY Hot Folder and then created te bookmarks from text. I also made a document comparing these by side to show where to the bookmarks point. This last one is a little over 16 MB.

I would like to send you the files so you can check them out, bur prefer not to do so on a public website. Can you send me an emailadres where I can send them to?

Regards,
Ed

Post by **TrackerSupp-Daniel** » Thu Jul 20, 2023 5:20 pm

Hello, evdbogaard

You can send them to us via support@pdf-xchange.com. Please also include a link to this topic in that email, so we know what they are in relation to.

Kind regards,

evdbogaard · Post by **evdbogaard** » Mon Sep 11, 2023 6:56 am

I installed the just released build 10.1.0.380 and the problem still exists.
My last used build of V9 that did not have this issue was build 9.5.368.0.

Hopefully there will be a solution soon, otherwise iI have to revert back to 368 (I use the create bookmarks from text a lot).

Ed

Post by **Paul - Tracker Supp** » Mon Sep 11, 2023 6:31 pm

Hi, evdbogaard

I am also seeing some strange results with a document I tested here. I have asked the devs to take a look and will post here what we find.

Thanks for bringing this to our attention.

Kind regards,
Paul - Tracker Supp

Post by **Paul - Tracker Supp** » Mon Sep 11, 2023 6:44 pm

Hi, Paul - Tracker Supp

It is confirmed that you found a bug. We will squash it. There is a ticket for this now, and while for internal use only, should you refer to RT#6596: Creating bookmarks From Page Text doesn't work properly in v10 here we can get you an update.

I hope that helps. I expect this to be fixed in the next release.

Kind regards,
Paul - Tracker Supp

PHK · Post by **PHK** » Tue Sep 12, 2023 12:41 pm

evdbogaard wrote: ↑Mon Sep 11, 2023 6:56 am I installed the just released build 10.1.0.380 and the problem still exists.
My last used build of V9 that did not have this issue was build 9.5.368.0.

Hopefully there will be a solution soon, otherwise iI have to revert back to 368 (I use the create bookmarks from text a lot).

Ed

Good catch!

Post by **TrackerSupp-Daniel** » Tue Sep 12, 2023 6:42 pm

igorlima · Post by **igorlima** » Fri Oct 06, 2023 11:52 am

HI all. Just want to add in that I hope too this gets fixed in the next build.

I'm having this problem with digital documents too (not only digitalized)
I loved that function and how it always pointed exactly where the bookmarks is (the first line of the page);

Edit: In Version: 10.1.1, build 381, Portable (Sep 19 2023; 14:46:00) the bug still persists

Edit 2: But in version Version: 9.5, build 368.0, Portable (Apr 6 2023; 09:59:30) this bug doesnt exist indeed
Edit 3: I meant "the first line of the page view" not necessarily of the page per se)

evdbogaard · Post by **evdbogaard** » Fri Oct 06, 2023 12:44 pm

My experience too: not yet solved in build 381.

Post by **Tracker Supp-Stefan** » Fri Oct 06, 2023 3:03 pm

Hello evdbogaard, and all,

Thanks for adding your posts!
I can see that the ticket has been opened, and a Work Item assigned to a developer.
So while I still do not have an exact date - we would be looking at getting that resolved as soon as we can!

Kind regards,
Stefan

evdbogaard · Post by **evdbogaard** » Mon Oct 30, 2023 8:20 am

Not yet solved in build 382.

Mon Oct 30, 2023 11:03 am

Hello evdbogaard,

You are correct - this ticket has not been marked as resolved yet.

Kind regards,
Stefan

evdbogaard · Post by **evdbogaard** » Fri Feb 09, 2024 1:58 pm

At first sight this seemed to be solved in build 384 and 385.

However in build 385 in a lot of cases all the words of the headers to be bookmarked are placed in separate bookmarks, like this example:

Example of incorrect bookmarking.png

For inspection by the dev. team this is the first page of the original pdf :

Pdf of Example of incorrect bookmarking.pdf

If build 384 did it any better, I don't know because 385 came just days after 384 and I updated immediately.

Hopefully this info helps to locate and solve the (remaining) issue.

Kind regards,
Ed

Post by **TrackerSupp-Daniel** » Fri Feb 09, 2024 5:43 pm

Hello, evdbogaard

If you could please provide the Boorkmarks from page text settings that you used, that would be quite helpful yes. Without them, it is hard to discern where the problem is.

Kind regards,

evdbogaard · Post by **evdbogaard** » Thu Feb 15, 2024 3:38 pm

Hi Dan,

I have been playing with the bookmarks again and created 4 variants of the presets from scratch.

The issue with the document in question is that it had 2 hierarchies of fonts to base the bookmarks on. The look alike, but are different in font sizes. So I created a preset containing both hierarchies and that gave me some unexpected results.

Then I made separate presets for each hierarchy: Preset_1 containing the font hierarchy used on pages 1-7 en Preset_2 for the hierarchy used on page 8-27.
Preset_1 gave the correct result.
Preset_2 gave the correct result from page 8-27, but it turned out that page 1-7 contained one of the fonts from preset_2 too. Those bookmarks I can correct manually. The basis bookmarking is correct.

Then I tried to combine both hierarchies in two versions in one single preset. It is the combination that gives incorrect results.

I send the bookmarked pdf's and the used presets (AGBPresets_name.presets) to 'Support@tracker-software.com' for further inspection.

Regards,
Ed

Post by **TrackerSupp-Daniel** » Thu Feb 15, 2024 4:42 pm

Hello, evdbogaard

I notice in these presets that you have the same font selected, and the only different appears to be the size, in the combined presetns, you have two separate hierarchies, one which focuses on fonts around 14pt (and 10pt for the children), and another which focuses around 12pt (with 9pt for the children).

image.png

I expect this is why you are seeing the incorrect split, because of the hierarchy you have defined, all "main titles" which are ~14pt, would be "Below" all main titles around 12pt. Likewise, the children below them would suffer from the same ordering inconsistency because you have two separate "trees" defined by this search.
To rectify this, instead of making a separate tree for these specific cases, increase the font size "tolerance" range for your bookmarks. This will ensure that they are all defined under one tree.

image(1).png

I will sent an email reply with an example for you.

Kind regards,