Trouble with loading text from PDFs- "ti" "tt" "ft"

[Faolind]

I am having an issue when loading a PDF document. The document is read properly when loaded in Adobe Reader, but when processed intro Affinity the following Text errors occur-

"ti" becomes "I " (space intentional) or "Ł " or "È"
"ti" (bold intentional) becomes "Ĉ"
"tt" becomes "N"
"ft" becomes "J " (space intentional)

I may have missed some examples. Whats also interesting is that this error is not consistent, some words are corrupted, others are not, and there doesn't seem to be a pattern as to which are corrupted or not. The error becomes permanent if saved as a pdf from affinity (that is, Adobe reads the text error and not the intended text after Affinity creates a new PDF save file), however, simply closing the document without saving and opening it in Reader shows no error.

This issue has appeared on versions of Affinity both 1 year old and current. I have searched and not found anyone else discussing this issue.

I am using Win10 64bit, and when loading into affinity I select "Load All Pages" Estimate/Estimate, and  "Group lines of text into text frame."
Affinity claims all fonts used by the document are available, in the options dialog box.

I am really not sure what to do, but as you can imagine, with over 100 pages of text, the rate of typos is far too much to simply address each time I have to reload a PDF file into Affinity.

Edit- I am using only Affinity Publisher, and also I have tried checking "Favor Editable Text over Fidelity" (It didn't help).

Edited November 2, 2021 by Faolind
Edit for additional useful information

[Faolind]

Tiny bump because this post was invisible a long while while it got approved.

[Joachim_L]

Welcome to the Affinity forums @Faolind!

Are the fonts used in the PDF installed on your system? Could you upload a sample page of that PDF for others to check? An alternative would be trying to save to text from a capable PDF reader.

[Faolind]

9 hours ago, Joachim_L said:

Welcome to the Affinity forums @Faolind!

Are the fonts used in the PDF installed on your system? Could you upload a sample page of that PDF for others to check? An alternative would be trying to save to text from a capable PDF reader.

All fonts are installed, I've just checked. One of the fonts in question is Calibri, another is Papyrus. As calibri is ubiquitous, I didn't think to check.

The document itself is business-sensitive. However, let me see if I can find an unobtrusive page...

Attached- "Page 7" is the altered page, exported by Affinity.
                  "Page 7 (Original)" is the original text, exported via Reader as Print To PDF (I don't have access at current to another export method outside of Affinity, so while this erases the original encoding, it is the best I have for comparison). You may note a couple small errors similar to the Affinity export, this is because this has happened multiple times, and the first time we simply tried to correct it manually, clearly we missed some on that first pass.
 

Page 7.pdf Page 7 (original).pdf

[Alfred]

14 hours ago, Faolind said:

I have searched and not found anyone else discussing this issue.

I’m surprised you haven’t found any other discussions of this or similar issues. The character sequences “ti”, “tt” and “ft” are commonly replaced by ligatures, but if those ligatures are used in the PDF file and your installed copy of the font in question doesn’t have the expected glyphs at the specified codepoints you’ll get the effect that you describe.

15 hours ago, Faolind said:

Edit- I am using only Affinity Publisher, and also I have tried checking "Favor Editable Text over Fidelity" (It didn't help).

That only affects layout (e.g. keeping chunks of text together for easier editing) so the issue that you’ve described is unrelated.

[Faolind]

21 minutes ago, Alfred said:

I’m surprised you haven’t found any other discussions of this or similar issues. The character sequences “ti”, “tt” and “ft” are commonly replaced by ligatures, but if those ligatures are used in the PDF file and your installed copy of the font in question doesn’t have the expected glyphs at the specified codepoints you’ll get the effect that you describe.

Is there a way to fix this? Calibri shouldn't have this sort of problem, right? Also is there a way to verify this is the issue, by checking my font files, for example?

[Alfred]

14 minutes ago, Faolind said:

Is there a way to fix this? Calibri shouldn't have this sort of problem, right? Also is there a way to verify this is the issue, by checking my font files, for example?

Unless you have access to a font editor, I suspect that the simplest way to check would be to create some sample text and export it to PDF twice: once with the font subsetting option switched on and once with it switched off. Then open the resultant PDF files and look for differences between them.

[kenmcd]

4 hours ago, Faolind said:

Is there a way to fix this? Calibri shouldn't have this sort of problem, right? Also is there a way to verify this is the issue, by checking my font files, for example?

Nothing to do with the fonts.

Your "original" PDF above has no text in it.
It is a bunch of oddly spaced drawings.
Try to "find" any word; you will get nothing in the search results.
Try to highlight text to copy it - nothing.

I have no idea how Affinity is getting any text out of that.

Also look at the spacing in the first paragraph.
It is not left-justified.
It is not fully-justified.
It has a bunch of weird spaces between the words.
So Affinity is guessing how to put it back together into actual text.

The issue appears to be with the original source.

 

[Faolind]

26 minutes ago, LibreTraining said:

Your "original" PDF above has no text in it.
It is a bunch of oddly spaced drawings.

I'm afraid your conclusion is wrong. As I stated, that is a snippet via "Print to PDF" which "prints" the displayed PDF page as a series of images. as I stated-
 

5 hours ago, Faolind said:

this erases the original encoding

but I don't have another way to snip out a non-sensitive page from the rest of the sensitive material.

So, While I appreciate your help, your conclusion is misinformed.

[kenmcd]

1 hour ago, Faolind said:

but I don't have another way to snip out a non-sensitive page from the rest of the sensitive material.

Can't you just Export-to-PDF that one page from the other application?
Even printing to a PDF printer should preserve the text and embed the fonts properly.
So, you are right, I do not understand what you are doing, or why.

Calibri does have standard ligatures for ti, tt, and ft.
Your PDF writing application may or may not be encoding those in a way that can be imported.
The only way to tell is by looking at the actual original PDF text to see the codes behind it.
Then we can tell if it is a problem with APub or the writing app.

Note: one work-around is to disable the standard ligatures in the source app.
That way individual characters (not ligatures) get embedded in the PDF.
Then when re-opened in the target app those standard ligatures will automatically re-appear in the text.

[Old Bruce]

On 11/1/2021 at 7:15 PM, Faolind said:

... Whats also interesting is that this error is not consistent, some words are corrupted, others are not, and there doesn't seem to be a pattern as to which are corrupted or not. ...

I am going to guess that somehow the corrupted text (versus the not corrupted text) is not applying the ligatures in the original application. The original application is the one which is generating the PDF that you are opening with Publisher. 

I have to ask why you are using a PDF instead of the actual text and images.

[Alfred]

17 hours ago, LibreTraining said:

Even printing to a PDF printer should preserve the text and embed the fonts properly.

Some (many?) PDF printer drivers ‘print’ a series of images, as the OP has stated. That’s why I suggested testing the font(s) by exporting to PDF from an Affinity app.

[kenmcd]

1 hour ago, Alfred said:

Some (many?) PDF printer drivers ‘print’ a series of images, as the OP has stated.

Which ones only print images?
I have a bunch of PDF printers installed and I am not aware of any which only print images.

[Alfred]

7 minutes ago, LibreTraining said:

Which ones only print images?

I’ve encountered several over the years but I’m afraid I can’t name any off the top of my head.

[kenmcd]

1 minute ago, Alfred said:

I’ve encountered several over the years but I’m afraid I can’t name any off the top of my head.

Yeah, I tend to focus on which ones give me the most control over font embedding (i.e to force embed everything, etc.) and do not pay much attention to the others.
So I was just curious. Some of the ones I have probably work that way. :-)

 

[Herschel]

I have had this problem too.It only occurs with Afinity products. All of the "Real Adobe" sofware reads PDF files correctly.

[Michael_7]

I, too, have this problem on Mac, when exporting PDF from Pages using Calabri font.

After export, when I copy and paste from the export PDF, words like "software" become "So4ware".  Thus "ft", becomes "4".  Weird and awful experience.

[MikeTO]

44 minutes ago, Michael_7 said:

I, too, have this problem on Mac, when exporting PDF from Pages using Calabri font.

After export, when I copy and paste from the export PDF, words like "software" become "So4ware".  Thus "ft", becomes "4".  Weird and awful experience.

Hi Michael and welcome to the forums. This does seem to be a bug in Publisher.

You can avoid it by editing the document in Pages, selecting all of its text, and choosing Format > Font > Ligature > Use None. Also when opening a PDF in Publisher, ensure you select Favour editable text over fidelity.

Here's a test file for Serif in case this isn't a known bug. Calibri must be variable since it's not available in Publisher so the font is being substituted with Helvetica. It doesn't seem to matter what it's substituted with, but the font substitution combined with the ligature doesn't work. Changing the font in Pages to something else that works in Publisher resolves the problem, as does turning off ligatures in Pages.

test4.pagestest4.pdf

[Michael_7]

Quote

You can avoid it by editing the document in Pages, selecting all of its text, and choosing Format > Font > Ligature > Use None. 

Wow, this seems to work in Pages.app!  I'm just learning about Font Ligatures for the first time.

Thank you Mike, thank you!

[MikeTO]

Here is another report of this ligature conversion bug, this time from a Windows user.

 

[KinoYarov]

Today, I encountered a recurring issue with the Calibri font in my PDFs. Initially created in Adobe InDesign, the PDFs were later edited using Affinity Publisher and exported again. While the PDFs appeared fine upon opening, transferring their text into programs like Word resulted in certain characters displaying as symbols.

This proved to be a significant setback for me, especially since one of these PDFs was my CV! Considering the possibility of text recognition software processing my CVs incorrectly, I'm quite displeased with the situation.

As a solution, I've opted to switch the font in all my PDFs from Calibri to another font to prevent further complications.

[walt.farrell]

15 minutes ago, KinoYarov said:

As a solution, I've opted to switch the font in all my PDFs from Calibri to another font to prevent further complications

A simpler approach, that you might try: Ensure that in your Export settings you turn off the Subset Fonts option when you create your PDF file.